Expanded Bed Adsorption - Thermo

Surface energetics of adsorbent-biomass interactions during expanded bed chromatography. Implications for process performance by Rami Reddy Vennapusa

A thesis submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Biochemical Engineering Approved, Thesis committee Prof. Dr. Marcelo Fernández-Lahore Prof. Dr. Jürgen Fritz Prof. Dr. Briger Anspach

Date of Defense: September 3, 2008

School of Engineering and Science

ORIGINALITY STATEMENT ‘I hereby declare that this submission is my own work and to the best of knowledge it contains no materials previously published or written by another researcher, or substantial proportions of material which have been accepted for the award of any other degree or diploma at Jacobs University or any other educational institutions, except where due acknowledgement is made in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from my thesis supervisor in the project’s design and conception or in style, presentation and linguistic expression is acknowledged.’

Signature ……………………………. Date ……………………..

II

Abstract Common limitations encountered during the direct recovery of bioproducts from an unclarified feedstock are related to the presence of biomass in such processing systems. Biomass related effects can be described as biomass-to-support interaction and cell-to-cell aggregation. In the current thesis work biomass related effects were studied in an important integrated primary unit operation mode viz Expanded bed adsorption (EBA), which was proved to suffer from the detrimental effects by the presence of biomass. Current work involves the investigation and understanding of the biomass interaction and aggregation onto various EBA process surfaces at local or molecular level. In doing so Streamline materialsTM of various chemistries were taken as process surface and intact yeast cell, yeast homogenates, and disrupted bacterial paste were employed as model colloids to understand their deposition and subsequent aggregation. Deposition and aggregation was studied with surface energetics according to XDLVO theory. These predictions based on the application of XDLVO theory were confirmed by independent experimental methods, like biomass deposition experiments and laser diffraction spectroscopy. Biomass components and beaded adsorbents were characterized by contact angle determinations with three diagnostic liquids and zeta potential measurements. Subsequently, free energy of interaction vs. distance profiles between interacting surfaces was calculated in aqueous media provided by its operating mobile phase. The effect of various chromatographic conditions based on the mode of operation was explored in relation to yeast interaction and aggregation. Calculations indicated that the interaction and aggregation is mainly due to the existence of a reversible secondary energy minimum. The extent and depth of pocket varied based on the operating process conditions for different interacting pairs. Understanding biomass-related effects will overcome or at least mitigate the process limitations. Exploring the effect of various types of additives for their ability to inhibit

III

either biomass deposition, cell aggregation, or a combination of both effects, a non ionic polymer PVP 360 was found to alleviate biomass deposition on weak anion exchangers. The predictions made by the XDLVO theory were well correlated with the physicochemical parameter α, in relation to ion exchangers where only interaction is happening. On the other hand a discrete modifications of XDLVO energies was observed with the lump parameter α for hydrophobic and pseudo affinity process surfaces where interaction and aggregation is taking place. Establishing a correlation defined a safe operational windows for EBA process when U ≤ |50| kT and α ≤ 0.15. Fundamental knowledge which could predict feedstock behaviour during primary unit operations of downstream processing would alleviate the current bottleneck during processing of bioproducts.

IV

Acknowledgements I would like to thank to all those who gave me the possibility and helping me to complete this exciting milestone. First, I owe my deepest thanks to my thesis supervisor Prof. Dr. Marcelo Fernandez Lahore offering me the opportunity to carry out my PhD work under his guidance. I also would like to express my sincere appreciation for his inspiring discussions, valuable suggestions, and constant encouragement throughout this work. I greatly appreciate all the learning’s I got from him related to professional and personal life, which are going to have great impact on my future career. I would like express my earnest gratitude for his personal kindness and care starting from the day one during my stay in Germany. There are many more to mention thank you very much for every thing. I would like thank my thesis jury members, Prof.Dr. Jürgen Fritz, Prof.Dr. Briger Anspach for accepting to be in my PhD defense committee and agreed to review my PhD thesis i gratefully acknowledge for their interest. Many thanks to Prof.Dr. Mathias Winterhalter and Prof.Dr. Ryan Richards for showing their interest to be as a part of my PhD proposal committee. My sincere thanks to Dean Prof.Dr. Kramer, former Dean Prof.Dr.Haerendel for giving me the possibility and supporting my application here at Jacobs University to continue my doctoral studies. I am indebted for the financially support from the Jacobs foundation. I would like to greatly appreciate the timely suggestions and help from the Prof.Dr. H.C. van der Mei, Dr. Carl Bolster, Prof.Dr. Gang Chen, Prof.Dr. Canan Tari and Prof.Dr. Mariano Grasselli from the other Universities. All suggestions added a great value to carry out the work in appropriate way. I would like thank all my current and former group members Aasim, Ayse, Benjamin, Doreen, Leo, Noor, Rustem, Marcelo, Sabinae, Sissi for their great cooperation. A special thanks to Leo, Marcelo, Marcel, Sissi for the great time when on outing for a social gathering. Thanks to all the Indian friends at JUB.

V

Not only the academic support important to conclude a thesis but also the emotional support especially when you are so far away from home. I would like thank Marcelo and his family members for being to my host family and supporting emotionally during my stay away from my home. My most profound thanks, my most heartfelt appreciation; my deepest gratitude goes to my family without whom none of this could have been accomplished. To my mum and dad, thanks for your unwavering confidence in me, for your love and sacrifice and for the moral energy. Thanks you so much for all the prayers and taking interest in my progress. Actually I have no words to thanks my dad and mom for their enumerous moral support. Thanks to my brothers Vasu and Kesav and their families for their love and encouragement. Deepest appreciation to my brother Kesav a doctor by profession, who took care of my sanity during course of my PhD. I also would like thank all my family members my grandfather, to my memory of grandmother and my siblings for constant caring and great moral support. Friends are every thing once you cross the sea; I have many friends back from India and here to thank who have directly and indirectly helped during the work. Thank you everyone. Finally the most important one Lord God almighty. With out his blessings nothing would have been possible today. I am infinitely grateful to my God for being my courage and refuge. Since there are no words to thank God for taking care of me all though my current life as heartfelt appreciation the current work is dedicated in his name.

VI

Dedication

To Lord Sri Venkateswara Swamy TTD, Tirumala India.

VII

Table of contents ABSTRACT ………………………………………………………………….

III

ACKNOWLEDGEMENTS …………………………………………………

V

1 GENERAL INTRODUCTION ……………………………………………

11

1.1 Introduction to Biotechnology …………………………………………..

11

1.2 Downstream Processing …………………………………………………

12

1.2.1 Process integration ………………………………………………...

13

1.2.2 Expanded bed adsorption …………………………………………

14

1.2.3 Operating principle ……………………………………………….

15

1.3 Problem statement and Research objective …………………………….

17

1.4 Goal of the work ………………………………………………………….

21

1.5 References ………………………………………………………………..

23

2.0 RESULTS / ORGANIZATION OF THESIS …………………………..

27

2.1 Assessing adsorbent-biomass interactions during expanded bed adsorption Onto ion exchangers utilizing surface energetics ………………………

28

2.1.1 Abstract …………………………………………………………….

28

2.1.2 Introduction ………………………………………………………..

29

2.1.3 Theory ………………………………………………………………

31

2.1.4 Materials and Methods ……………………………………………

35

2.1.5 Results and discussions ……………………………………………

38

2.1.6 Conclusion ………………………………………………………….

57

2.1.7 Acknowledgements ………………………………………………...

58

2.1.8 Nomenclature ………………………………………………………

59

2.19 References ……………………………………………………………

61

VIII

2.2 Colloid deposition experiments as a diagnostic tool for biomass attachment onto bioproduct adsorbent surfaces …………………………………………..

65

2.2.1 Abstract ………………………………………………………………

65

2.2.2 Introduction ………………………………………………………….

66

2.2.3 Materials and Methods ……………………………………………..

69

2.2.4 Results and Discussions ………………………………………….....

72

2.2.5 Conclusion …………………………………………………………...

84

2.2.6 Acknowledgements ……………………………………………….....

85

2.2.7 Nomenclature ………………………………………………………..

85

2.2.8 References ……………………………………………………………

87

2.3 Surface energetics to assess biomass attachment onto hydrophobic interaction adsorbents in expanded beds ……………………………………………… 90 2.3.1 Abstract ……………………………………………………………….

90

2.3.2 Introduction ……………………………………………………..........

91

2.3.3 Materials and Methods ………………………………………………

93

2.3.4 Results and Discussions ……………………………………………… 97 2.3.5 Conclusions …………………………………………………………… 112 2.3.6 Acknowledgements …………………………………………………… 123 2.3.7 Nomenclature …………………………………………………………. 124 2.3.8 References …………………………………………………………….. 125 2.4 Surface energetics to assess biomass attachment onto immobilized metal affinity adsorbents in expanded beds ……………………………………………… 128 2.4.1 Abstract …………………………………………………………….....

128

2.4.2 Introduction …………………………………………………………..

129

2.4.3 Materials and Methods ………………………………………………

131

2.4.4 Results and Discussions ……………………………………………...

135

2.4.5 Conclusions …………………………………………………………...

151

2.4.6 Acknowledgements …………………………………………………… 151 2.4.7 Nomenclature ………………………………………………………… 152 2.4.8 References …………………………………………………………….. 153 IX

2.5 Surface energetics to assess biomass deposition onto fluidized chromatographic supports …………………………………………………………………......

156

2.5.1 Abstract ……………………………………………………………….

156

2.5.2 Introduction …………………………………………………………..

157

2.5.3 Materials and Methods ………………………………………………

159

2.5.4 Results and Discussions ……………………………………………… 163 2.5.5 Conclusions …………………………………………………………… 174 2.5.6 Acknowledgements …………………………………………………… 175 2.5.7 Nomenclature …………………………………………………………. 175 2.5.8 References …………………………………………………………….. 177 2.6 The effect of chemical additives on biomass deposition onto beaded chromatographic supports ………………………………………………...

179

2.6.1 Abstract ……………………………………………………………….

179

2.6.2 Introduction …………………………………………………………..

180

2.6.3 Materials and Methods ………………………………………………. 183 2.6.4 Results and Discussions ………………………………………………. 187 2.5.5 Conclusions …………………………………………………………… 202 2.6.6 Acknowledgements …………………………………………………… 202 2.6.7 References ……………………………………………………………

203

3.0 GENERAL CONCLUSIONS AND REMARKS ………………………….. 207 4.0 Appendix …………………………………………………………………….. 214

X

Introduction

1 General Introduction 1.1 Introduction to Biotechnology Biotechnology is known to exist as such since the late 17th century. This “traditional” biotechnology was mostly concerned with processing of food e.g. wine, beer, cheese and other diary products. In the late 19th century a new wave in the biotechnology industry started when complex organic molecules like antibiotics and enzymes were produced for the first time by biosynthesis (Enfors and Häggström 2005). With the knowledge gained on microbial physiology, biochemistry and genetics it was possible to think about genetic manipulations of the cells during the 70’s and the so called “genetic engineering” was born. The advances in genetic engineering led to a new era in biotechnology with products like insulin, erythropoietin, and interferon. These biopharmaceutical products have a high market value. In the current century, the biopharmaceutical industry in one the fastest growing sectors in the global economy (Pavlou and Reichert 2004). Proteins constitute an important class of biopharmaceutical products, but also have food and biotechnology applications (Headon and Walsh 1994). Advances in the recombinant DNA and cell culture technology have permitted the large scale production of virtually any protein by fermentation routes at increased titers (Walsh 2006), thereby shifting the bottleneck in biopharmaceutical process development to the purification of such bioproducts (Smith 2005; Thiel 2004). Also microbial bioprocess can be divided in main two parts: a) the fermentation step, as the (bio) synthesizing step, and b) downstream processing, for the primary recovery and purification of the desired product (Ref. Figure 2). Since considerable efforts were made on the genetic manipulation of cells and on the improvement of fermentation strategies, a considerable increase in production level was already accomplished. However, optimized downstream processes have to be designed for the subsequent recovery of these products so as to match “upstream” performance.

11

Introduction 1.2 Downstream processing The most cost intensive component of biotech-processing is DOWNSTREAM PROCESSING (DSP), which accounts for ≥ 50-85% of the total processing cost. Increasing competition in biotech markets, and the development of new (niche) markets, which are possible due to the utilization of recombinant DNA technology and modern cell culture techniques in the industry have triggered the development of novel and efficient (bio) separation technologies (Gupta and Mattiasson 1994). Biotechnological process fluids are generally of complex nature and contain solid (biological) particles of various sizes, as well as solutes of various molecular masses and chemistries (Anspach et al. 1999; Thömmes 1997). The required purity of the products in the biotechnological industry ranges from partially purified concentrates e.g. food enzymes to highly purified preparations e.g. purity demand ≥ 99.99% in the case of therapeutic proteins used for intravenous dosage. A direct consequence of the latter is that purification processes comprise a relatively large number of unit operations whose complexity depends on the final product purity required (Wheelright 1991). It is common place to observe downstream processes having a total number of processing steps between seven and fourteen (Bonnerjea et al. 1986; Fish and Lilly 1984; Wheelright 1991). Each additional step or unit operation will affect the overall process economy by increasing operational cost and process time. Additional steps will also produce a certain degree of product loss and thus, the overall yield after a certain processing “train” will substantially decrease. For example, assuming single step yields in the range 70-95%, ≈ 60% of the product will be lost after six processing steps (Maitra and Verma 2003) (Ref. Figure 1). Therefore, the process economics, yield and time are interrelated and an optimum balance between them has to be found in order to design a successful downstream process.

12

Introduction

100

100 95

80

Individual step 85.5 72.6 61.7

60

46.3 41.6

Yield (%) 40 20

G F

SL S DI CL SR AR UP IF /C O NC CH RO M #1 CH RO M #2

Pr od uc t

0

Global process

Processing steps Figure 1: Series of steps and their yields in Downstream Processing. 1.2.1 Process Integration Integration is the creation of link between previously separate unit operations or combining individual steps in to one unit operation, by which product losses and process economics can be minimized. Process integration has been actively researched in the field of biochemical engineering over the last decade and these efforts continue today. The reason is that “integration” could be one of the keys for the rational, cost-effective and productive design of (bio) separation processes. From the preceding paragraphs it is understood that increasing the processing steps would lead to a suboptimal process. Draeger and Chase in the year 1994 (Chase 1994) presented a novel integrated concept based on fluidized adsorbent beads for the direct sequestration of bioproducts. Expanded bed adsorption (EBA) was introduced as an advantageous unit operation; details are given in following sections (McCormick 1993).

13

Introduction 1.2.2 Expanded bed adsorption (EBA) Expanded bed adsorption is an integrative primary recovery technology which allows the direct capture of targeted species from an unclarified biological feedstock. The classical route to downstream processing is illustrated in Figure 2. Within the previous scheme, three independent unit operations i.e. solid-liquid separation, volume (buffer) reduction, and (partial) fractionation can be ideally replaced with a single EBA “capture” step. Therefore, EBA allowed to reduce process cost, time and product losses in many processing examples (Anspach et al. 1999; Walter and Feuser 2002). For adequate performance, EBA relies on the formation of a stable -and perfectly classified- fluidized bed or “expanded” bed. This has to occur even in the presence of a turbid feedstock. Fermentation broth

Intracellular product

Secreted product

Cell disruption

Solid liquid separation

Volume reduction

Primary recovery (Isolates, concentrates and stabilizes)

EBA

Partial purification/Fractionation Intermediate purification

Polishing

Removes bulk impurities and purifies active compounds Final high-level active pure bioproducts

Figure 2: Downstream processing steps for a biotech prouct. 14

Introduction 1.2.3 Operating principle Standard chromatographic columns (“packed beds”) are characterized by adsorbent beads which are physically confined within the bed. On the other hand, EBA systems allow the introduction of a crude feedstock e.g. containing biological particulates, without the danger of clogging. This is due to the fact that the particulate matter and cell or cell debris can flow within the inter-particular space created upon the solid-liquid fluidization. EBA columns are fed from the bottom while a movable adapter is held away from the adsorbent bead population, thus letting the same to “expand”. As buffer is pumped from below, these beads become not only fluidized but also classified according to their size and density. This fluidization and classification occurs when their sedimentation velocity equals to the upward liquid velocity (Figure 3b). To accentuate this effect commercial adsorbent beads have been modified to include inert quartz or metal alloy cores, and beads have a defined size/density range (StreamlineTM GE Healthcare, Uppsala, Sweden; FastLineTM Upfront, Copenhagen, Denmark). Adsorbent beads used in EBA have a size range within 50 to 400 mm. When stable fluidization / classification occurs (“expansion”) the local mobility of the matrix particles is reduced (Figure 4). Therefore, EBA systems mimic packed bed chromatography in the sense of creating a number of equilibrium stages (“plates”) alongside the column length (Hubbuch et al. 2005). EBA mode of operation is shown in Figure 3.

15

Introduction

a) Sedimented bed

b) Equilibration Classified Fluidization

c) Sample Application

d) Elution

+ Bio-product

cell

Figure 3: The unclarified feedstock when applied to the EBA column. The particulates and the cell debris are supposed to move freely around the adsorbent beads and eventually leave through the top of the column. The compound of interest interacts with the beads via specific ligands and becomes adsorbed. Afterwards, the matrix is allowed to settle and the plunger is moved down flow. Elution can be performed either in the packed bed mode or alternatively in the expanded bed mode at decreased superficial velocity (Lihme et al. 1999).

Particle size gradient

Particle density gradient

+

+

Plug flow

Figure 4: The phenomenon of proper fluidization and classification during the expanded bed adsorption.

16

Introduction 1.3 Problem statement and Research objective The majority of unit operations applied in the downstream processing have a long history in the field of chemical and process engineering. All the platform technologies established are used in the above mentioned area to the best, however in the bioprocess engineering scenario these established technologies were not able to be utilized fully or may not be mimicked completely due to the complexity of biological feedstock. Cells types commonly utilized in biotechnology as hosts for the production of recombinant proteins include bacteria, yeast, filamentous fungi, plant cells, insect cells, and mammalian cells. The different adsorbent types used in EBA for product sequestration include ion exchange materials (cation- or anion- exchangers), hydrophobic interaction beads, and IMAC supports, among others like immuno- and pseudo- affinity. When a biological feedstock is loaded into the EBA system with the aim of selectively capturing the bioproduct of interest, it is usually observed that bed hydrodynamics and stability is compromised due to biomass attachment to the chromatographic beads (Anspach et al. 1999; Hubbuch et al. 2005). Biomass attachment may derive from cell-to-support interactions and eventually from cell-to-cell aggregation (Fernandez-Lahore et al. 2000) (Figure 5) . The influence of various biomass-types like Saccharomycess cervisiae, yeast cell homogenate, E.coli homogenate, and mammalian cells onto several commercial adsorbents (Streamline DEAE, SP, Phenyl, Chelating and Base matrix) has been studied (Fernandez-Lahore et al. 2000; Feuser et al. 1999; Poulin et al. 2008; Smith et al. 2002). But most of these studies were restricted to anion exchangers. Moreover, no effort has been made until now to elucidate the physicochemical mechanisms promoting biomass deposition under process conditions. In 1994, Erickson et al. (Erickson JC et al. 1994) described the deposition of CHO cells from a cell culture onto fluidized controlled pore glass beads, which were coated with Protein A in order to allow for the recovery of monoclonal antibodies. The extent of cell deposition is known to differ with the adsorbent type, cell type, solution chemistry, and operational conditions (Fernandez-Lahore et al. 2000). Many authors have emphasized the fouling effects of various biomasses types, especially on anion-exchangers. This is due to the strong electrostatic attraction which develops between the cells (usually negative) and the adsorbent beads (positively charged) (Feuser et

17

Introduction al. 1999; Lin et al. 2001). Biomass deposition on other adsorbent beads like cationexchangers, hydrophobic interaction materials, and Chelating beads was overlooked until now. All these materials have the potential to interact with biomass and hence the danger of impaired EBA performance can not be ruled out. For example, a significant fouling on a cation-exchanger by hybridoma cells in culture was reported during the

recovery of

murine IgG1 (Ameskamp et al. 1999). Some authors also showed the fouling of hydrophobic (Smith et al. 2002) and pseudo-affinity chromatography beads (Poulin et al. 2008) under real process conditions. It is clearly demonstrated that biomass deposition will result in deteriorated EBA process performance (Fernandez-Lahore et al. 2000). The adverse effects of the biomass interactions was experimentally determined and described as follows (Feuser et al. 1999; GEHealthCare 2001-04; GEHealthCare 2002-11): 1) The bed stability may be reduced due to the formation of channels and stagnant zones during sample application, and beyond (Figure 5). 2) The sorption performance of a bed where poor hydrodynamics is confirmed will deteriorate. 3) During product elution in the packed bed mode, resin particles to which biomass has adsorbed may show hindered sedimentation, poor packing quality, and distorted elution peaks i.e. diluted product. 4) The product solution can be contaminated with biomass, which is co-eluted under desorption conditions. This leads to degradation of product quality. 5) The life expectancy of the chromatographic beads may be reduced due to irreversible fouling effects and harsh regeneration conditions. 6) Biomass interaction would result in increased buffer consumption in order to remove and wash away sticky biological particles.

18

Introduction

Bio-product

Interaction Bead cells

Aggregation Bead

Figure 5: Interacting expanded system causing impaired hydrodynamics and decreasing sorption performance.

Local level (distance 1.5 Å) Bio-product

Bead cells

Figure 6: Illustration signifying biomass interaction to adsorbent (at a local level).

The biomass deposition phenomena is hampering the industrial utilization of EBA since its introduction in 1994 (Curbelo et al. 2003). Some advancement was made in the 90’s to alleviate such limitation with partial success. Several methods were developed to analyze the extent of biomass–adsorbent interactions. The methods include finite bath adsorption, pulse response and residence distribution analysis (Hubbuch et al. 2005). All these techniques can only provide an overall indication of the state of fouling. Few recent studies attempted to understand this phenomenon more in detail (Lin et al. 2006). The aforementioned diagnostic methods address the degree of interaction of biomass to a

19

Introduction limited range of material types, particularly anion-exchangers. For example, zeta potential was introduced as a significant parameter for process design, due to the obvious ionic interaction prevailing in ion-exchange systems. However, this approach cannot explain the interaction and aggregation of biomass onto hydrophobic and pseudo-affinity beads as electrostatic interactions play a minor role in such cases. This is due to processing conditions under which high-conductivity buffers are employed (Gallardo-Moreno et al. 2002; Klotz et al. 1985). Exploring further in this direction Peter Brixius during his doctoral work in Jülich and Novo Nordisk A/S (Brixius 2003) addressed the existence of some other forces like Van der Waals and hydrophobic forces involved in the adhesion of biomass apart from the electrostatic forces. However, Brixius’ work mainly dealt with chargemediated attraction forces onto anion exchangers. Some insight on physicochemical parameters affecting the adhesion of biomass on ion exchanger adsorbents was provided (Vergnault et al. 2004). A few authors also tried to understand fouling on chromatographic beads utilizing confocal laser microscopy (Siu et al. 2006). Also manufacturers have tried to alleviate biomass interaction by introducing novel type of equipment for EBA by designing novel bead structures (Viloria-Cols et al. 2004). Among the various methods tried to overcome the interaction of biomass to process surfaces thermal pretreatment of biomass before loading on the column was reported in literature (Ng et al. 2007). Despite all these efforts, a comprehensive picture of the interfacial forces acting between cells and beads was unavailable until now. Particularly, previous work has focused on cellto-bead interaction but the role of cell-to-cell aggregation was neglected since this phenomenon can not be captured by existing methods, like the biomass-impulse test earlier developed by Feuser (Feuser et al. 1999). Today, we have realized the importance of aggregation under certain processing conditions (Fernandez-Lahore et al. 2000). Fouling is a common phenomena in the integrated process where there direct contact between crude feedstock and reactive solids e.g., membrane operations, magnetic separations, direct capture techniques (Bierau et al. 2001; Theodossiou et al. 2001; Ventura et al. 2008)

20

Introduction 1.4 Goal of the work Complementing all the above research findings by different authors on the biomass adhesion, current work further progressed with the objective to have more quantitative fundamental understanding at local level between biomass and adsorbent bead, which are commonly utilized in EBA technology. It was targeted to determine the basic underlying phenomenon of the interfacial forces (Lifshitz-van der Waals, hydrophobic attractive or hydrophilic repulsive and electrostatic) at micrometer scale between a biological particle and process surface or between two biological particles (Figure 6). Understandings the phenomena at molecular level will allow developing an improved process performance of EBA making the process more robust and less complex in the process scenario with all its added advantages. Additionally the fundamental understanding could help to propose a universal tool for process/material design when direct sequestration is in focus. For having this comprehensive picture, surface thermodynamics was utilized. XDLVO theory was used to determine the interactions and aggregation onto the process surface. XDLVO calculations were performed via experimental determination of contact angles and zeta potentials values for the interacting surfaces or particles. Experimental XDLVO quantitative information was validated independently with the biomass deposition experiments (Tari et al. 2008) and laser diffraction experiments. Under the frame of current research work the following aspects were studied 1) Interaction of three different biomass types intact yeast, yeast homogenates and E.coli homogenates with the Streamline ion exchangers. Aggregation of only intact yeast was studied with this type. 2) Interaction and aggregation of Saccharomycess cervisiae with the Streamline hydrophobic and chelating (pseudo affinity) supports. 3) Influence of chemical additives on the interaction and aggregation of Saccharomycess cervisiae with different Streamline materials.

21

Introduction Surface energetics or physicochemical properties of the above-mentioned supports and biofoulants are studied in detail at various process conditions in order to have a clear picture in the problem-creating scenario while downstream processing. The biomass adhesion on the different substrata has been studied by many authors (Absolom et al. 1983; Bos et al. 1999) by applying classical DLVO (CDLVO) and extended DLVO (XDLVO) theory, which was proven to be more advantageous and can be applied to biological particles (Bos et al. 1999). The detailed theoretical part of XDLVO is described in chapter I within this thesis.

22

References 1.5 References Absolom DR, Lamberti FV, Policova Z, Zingg W, Van Oss CJ, Neumann AW. 1983. Surface thermodynamics of bacterial adhesion. Appl Environ Microbiol 46(1):90-7. Ameskamp N, Priesner C, Lehmann J, Lütkemeyer D. 1999. Pilot scale recovery of monoclonal antibodies by expanded bed ion exchange adsorption. Bioseparation 8(1):169-188. Anspach FB, Curbelo D, Hartmann R, Garke G, Deckwer WD. 1999. Expanded-bed chromatography in primary protein purification. J Chromatogr A 865(1-2):129-144. Bierau H, Hinton RJ, Lyddiatt A. 2001. Direct process integration of cell disruption and fluidised bed adsorption in the recovery of labile microbial enzymes. Bioseparation 10(1-3):73-85. Bonnerjea J, Oh S, Hoare M, Dunnill P. 1986. Protein Purification: The Right Step at the Right Time. Nat Biotech 4(11):954-958. Bos R, Van der Mei HC, Busscher HJ. 1999. Physico-chemistry of initial microbial adhesive interactions--its mechanisms and methods for study. FEMS Microbiol Rev 23(2):179-230. Brixius PJ. 2003. On the influence of feedstock properties and composition on process development of expanded bed adsorption. Dusseldorf, Germany: Heinrich Heine University. Chase HA. 1994. Purification of proteins by adsorption chromatography in expanded beds. Trends Biotechnol 12(8):296-303. Curbelo DR, Garke G, Guilarte RC, Anspach FB, Deckwer WD. 2003. Cost Comparison of Protein Capture from Cultivation Broths by Expanded and Packed Bed Adsorption. Eng Life Sci 3(10):406-415. Enfors S, Häggström L. 2005. Bioprocess Technology - Fundamentals and Applications A textbook for introduction of the theory and practice of biotechnical processes. 1-350 p. Erickson JC, Finch JD, Greene DC. 1994. Direct capture of recombinant proteins from animal cell culture media using a fluidized bed adsorber. In: Griffiths B, Spier RE, BertholdW, editors. Animal cell technology:Products for today, prospects for tomorrow. Oxford: Butterworth-Heinemann:557-560.

23

References Fernandez-Lahore HM, Geilenkirchen S, Boldt K, Nagel A, Kula MR, Thommes J. 2000. The influence of cell adsorbent interactions on protein adsorption in expanded beds. J Chromatogr A 873(2):195-208. Feuser J, Walter J, Kula MR, Thommes J. 1999. Cell/adsorbent interactions in expanded bed adsorption of proteins. Bioseparation 8(1-5):99-109. Fish NM, Lilly MD. 1984. The Interactions Between Fermentation and Protein Recovery. Nat Biotech 2(7):623-627. Gallardo-Moreno AM, Gonzalez-Martin ML, Perez-Giraldo C, Garduno E, Bruque JM, Gomez-Garcia AC. 2002. Thermodynamic Analysis of Growth Temperature Dependence in the Adhesion of Candida parapsilosis to Polystyrene. Appl Environ Microbiol 68(5):2610-2613. GEHealthCare. 2001-04. Cost comparison: expanded bed adsorption (EBA) vs conventional recovery in the industrial scale processing of proteins. Application note STREAMLINE expanded bed adsorption. p 1150-21 AA. GEHealthCare. 2002-11. A comparison of STREAMLINE expanded bed adsorption with the combined techniques of filtration and conventional fixed bed chromatography for the capture of an Fc-fusion protein from CHO cell culture. Application note STREAMLINE expanded bed adsorption. p 1144-87 AB. Gupta MN, Mattiasson B. 1994. Novel technologies in downstream processing. Chem Ind 17:673-675. Headon DR, Walsh G. 1994. The industrial production of enzymes. Biotechnol Adv 12(4):635-646. Hubbuch J, Thommes J, Kula MR. 2005. Biochemical engineering aspects of expanded bed adsorption. Adv Biochem Eng Biotechnol 92:101-23. Klotz SA, Drutz DJ, Zajic JE. 1985. Factors Governing Adherence of Candida Species to Plastic Surfaces. Infect Immun 50(1):97-191. Lihme A, Zafirakos E, Hansen M, Olander M. 1999. Simplified and more robust EBA processes by elution in expanded bed mode. Bioseparation 8(1):93-97. Lin

DQ,

Fernandez-Lahore

HM,

Kula

MR,

Thommes

J.

2001.

Minimising

biomass/adsorbent interactions in expanded bed adsorption processes: a methodological design approach. Bioseparation 10(1-3):7-19.

24

References Lin DQ, Zhong LN, Yao SJ. 2006. Zeta potential as a diagnostic tool to evaluate the biomass electrostatic adhesion during ion-exchange expanded bed application. Biotechnol Bioeng 95(1):185-91. Maitra SS, Verma AK. 2003. End of Small Volume High Value Myth in Biotechnology, Process Design for a Mega-plant Producing gamma Interferon for Mega Profit. IE (I) Journal CH 84. McCormick DK. 1993. Expanded Bed Adsorption. Nat Biotech 11(9):1059-1059. Ng MYT, Tan WS, Abdullah N, Ling TC, Tey BT. 2007. Direct purification of recombinant hepatitis B core antigen from two different pre-conditioned unclarified Escherichia coli feedstocks via expanded bed adsorption chromatography. J Chromatogr A 1172(1):47-56. Pavlou AK, Reichert JM. 2004. Recombinant protein therapeutics-success rates, market trends and values to 2010. Nat Biotech 22(12):1513-1519. Poulin F, Jacquemart R, DeCrescenzo G, Jolicoeur M, Legros R. 2008. A Study of the Interaction of HEK-293 Cells with Streamline Chelating Adsorbent in Expanded Bed Operation. Biotechnol Prog 24(1):279-282. Siu SC, Boushaba R, Topoyassakul V, Graham A, Choudhury S, Moss G, TitchenerHooker NJ. 2006. Visualising fouling of a chromatographic matrix using confocal scanning laser microscopy. Biotechnol Bioeng 95(4):714-723. Smith C. 2005. Striving for purity: advances in protein purification. Nat Meth 2(1):71-77. Smith MP, Bulmer MA, Hjorth R, Titchener-Hooker NJ. 2002. Hydrophobic interaction ligand selection and scale-up of an expanded bed separation of an intracellular enzyme from Saccharomyces cerevisiae. J Chromatogr A 968(1-2):121-128. Tari C, Vennapusa RR, Cabrera RB, Fernandez-Lahore M. 2008. Colloid deposition experiments as a diagnostic tool for biomass attachment onto bioproduct adsorbent surfaces. J Chem Technol Biotechnol 83:183-191. Theodossiou I, Sondergaard M, Thomas OR. 2001. Design of expanded bed supports for the recovery of plasmid DNA by anion exchange adsorption. Bioseparation 10(13):31-44. Thiel KA. 2004. Biomanufacturing, from bust to boom...to bubble? Nat Biotech 22(11):1365-1372. Thömmes J. 1997. Fluidized bed adsorption as a primary recovery step in protein purification. Adv Biochem Eng/Biotechnol 58:185-230.

25

References Ventura AM, Fernandez Lahore HM, Smolko EE, Grasselli M. 2008. High-speed protein purification by adsorptive cation-exchange hollow-fiber cartridges. J Membr Sci 321(2):350-355. Vergnault H, Mercier-Bonin M, Willemot RM. 2004. Physicochemical parameters involved in the interaction of Saccharomyces cerevisiae cells with ion-exchange adsorbents in expanded bed chromatography. Biotechnol Prog 20(5):1534-42. Viloria-Cols ME, Hatti-Kaul R, Mattiasson B. 2004. Agarose-coated anion exchanger prevents cell-adsorbent interactions. J Chromatogr A 1043(2):195-200. Walsh G. 2006. Biopharmaceutical benchmarks 2006. Nat Biotech 24(7):769-776. Walter J, Feuser J. Novel approach and technology in expanded bed adsorption techniques for primary recovery of proteins at large technical scale; 2002; Florida. Downstream EBA ’02, Sweden, Amersham Biosciences. p 37-39. Wheelright SM. 1991. Protein purification: Design and scale up of downstream processing. New York: Oxford University Press.

26

Results

2.0 Organization of dissertation The dissertation is organized in the form of the manuscripts originated during the course of my PhD work Assessing adsorbent-biomass interactions during expanded bed adsorption onto ion exchangers utilizing surface energetics R.R. Vennapusa, S.M. Hunegnaw, R.B. Cabrera, M. Fernandez-Lahore, published in Journal of Chromatography A. 2008, 1181, (1-2), 9-20. Colloid deposition experiments as a diagnostic tool for biomass attachment onto bioproduct adsorbent surfaces C.Tari†, R.R. Vennapusa†, R.B. Cabrera, M. Fernandez-Lahore, published in Journal of Chemical Technology & Biotechnology, 2008, 83, 183-191. Surface energetics to assess biomass attachment onto hydrophobic interaction adsorbents in expanded beds R.R. Vennapusa, C. Tari, R. B. Cabrera, M. Fernandez-Lahore. Biochemical Engineering Journal (accepted). Surface energetics to assess biomass attachment onto immobilized metal affinity adsorbents in expanded beds R.R. Vennapusa, M. Aasim, R.B. Cabrera, M. Fernandez-Lahore. Biotechnology and Bioprocess Engineering (Submitted). Surface energetics to assess microbial adhesion onto fluidized chromatography adsorbents R.R. Vennapusa, S. Binner, R.B. Cabrera, M. Fernandez-Lahore. Engineering in Life Sciences (accepted) The effect of chemical additives on biomass deposition onto beaded chromatographic supports R.R. Vennapusa, M. Fernandez-Lahore. Journal of Biotechnology (Submitted). †: Equal authorship

27

Results

2.1 Assessing adsorbent-biomass interactions during expanded bed adsorption onto ion-exchangers utilizing surface energetics Rami Reddy Vennapusa, Sara M. Hunegnaw, Rosa B. Cabrera, and Marcelo FernándezLahore Downstream Processing Laboratory, Jacobs University, Campus Ring 1, D-28759, Bremen, Germany. 2.1.1 Abstract Biomass adhesion onto an adsorbent matrix or “interaction” as well as biological particle co-adhesion or “aggregation” can severely affect the overall performance of many directcontact methods for downstream processing of bioproducts. Studies to quantitatively describe this biomass-adsorbent interaction were developed utilizing surface energetics. An indirect thermodynamic approach via contact angle and zeta potential measurements was utilized. Intact yeast cells, yeast homogenates, and disrupted bacterial paste were employed as model system. Various surfaces that are relevant to biochemical and environmental applications were characterized. The extended Derjaguin, Landau, Verwey, Overbeek (XDLVO) theory was found to appropriately predict biomass adhesion behaviour. It was observed that cell attachment onto anion exchange supports is promoted by strong and close interaction within a secondary energy minimum followed by moderate multilayer cell aggregation. On the other hand, cell interaction with cation exchange materials can take place within a reversible secondary energy minimum and at longer separation distance. The influence particle charge and size, as well as the influence of the nature of the material under study were summarized in the form of energy vs. distance profiles. These investigations lead to many process-related conclusions: a) Process buffer conductivity windows can be recommended for anion-exchange chromatography (AEX) vs. cationexchange chromatography (CEX) systems, b) Increased hydrodynamic shear is required to prevent biomass attachment onto AEX as compared to CEX, and c) Aggregation phenomena is a function of contact time and biomass concentration. Understanding biomass-adsorbent interaction at the particle (local) level is opening the pave for optimized operation of Expanded Bed Adsorption methods at the process (macro) scale. A universal methodological approach is presented to guide both process and material design.

28

Results 2.1.2 Introduction The current key component of biotech manufacturing is product downstream processing. Recovery and purification processes comprise a relatively large number of unit operations, which complexity depends on the final product purity required and which typically account for ≥ 50-85% of the total bioprocessing cost. The required purity of the products in the life science industry ranges from partially purified biocatalysts to highly purified therapeutic agents. In any case, bioprocess fluids are generally of complex nature and contain suspended solids like biological particles of various sizes, as well as solutes of various molecular masses and chemical structures. Moreover, the need for additional unit operations during downstreaming will cause a degree of product loss and will substantially decrease overall process yield. For example, assuming single step yields in the range 7095%, ~ 60% of the product will be lost after six processing steps. Expanded Bed Adsorption (EBA) has been proposed as an “integrative” downstream processing technology allowing the direct capture of targeted species from an unclarified feedstock e.g. a cell containing fermentation broth. This unit operation has the potential to combine solids removal, product concentration, and partial purification in a single processing step. The application of EBA implies, however, that intact cell particles or cell debris present in the feedstock will interact –in a minor or larger extent- with fluidized adsorbent beads. It is already known that interaction between biomass and the adsorbent phase may lead to the development of poor system hydrodynamics and therefore, impaired sorption performance under real process conditions (Anspach et al. 1999; Hubbuch et al. 2005). Detrimental processing conditions can also be expected in any other downstream operation where direct contacting between a crude feedstock and a reactive solid phase is supposed to occur (Bierau et al. 2001; Theodossiou et al. 2001). Moreover, biomass interaction would result in increased buffer consumption in order to remove and wash away sticky biological particles (GEHealthCare 2001; Northelfer and Walter 2002). These phenomena i.e. decreased sorption performance and buffer consumption is detrimental to cost-efficient processing utilizing direct sequestration unit operations. Earlier studies on biomass-adsorbent interactions were restricted to simple diagnostic tests to determine the extent of cell –or cell debris- attachment to the desired chromatographic supports (Feuser et al. 1999). The development of residence time distribution methods as

29

Results applied to turbid feedstock, and their subsequent application to evaluate system hydrodynamics under real process conditions, has established a clear picture of the deleterious potential of biomass-adsorbent interactions (Fernandez-Lahore et al. 1999). Further studies pointed out that interactions between (positively charged) anion exchangers and (negatively charged) biological particles resulted the most problematic system to deal with (Fernandez-Lahore et al. 2000; Lin et al. 2001). Due to the obvious electrostatic nature of such interaction, a single property of these interacting bodies i.e. the zeta potential has been recently proposed for a better understanding and prediction of biomass-adsorbent interactions (Lin et al. 2003; Lin et al. 2006). Other investigations on microbial adhesion to solid surfaces have lead to similar conclusions in the sense that electrostatic interaction between microbial cells and process surfaces is an important factor affecting such phenomena (Mills et al. 1994; Vergnault et al. 2007). These conclusions, however, were based on proving biomass adhesion on a single material type in solutions of different ionic strength. Furthermore, these studies were restricted to ion-exchangers, to yeast cells having a certain degree of hydrophobic character, and to an experimental evaluation based on the microbial-adhesion-to-solvents test. On the other hand, some studies have found a better correlation between surface energy, calculated by the three liquid contact angle method, and microbial adhesion on different solid supports at constant solution chemistry (Li and Logan 2004). Taken into consideration the complexity of interfacial phenomena at the (sub) micrometer scale, a more comprehensive approach would consider interaction forces other than those purely electrostatic in nature and would employ principles of colloid chemistry to explain biomass-adsorbent attachment at the local (particle) level (Van Oss 1994). It is known that biological particles like microbial cells can be considered “soft” colloidal particles and thus their adhesion to substrata should be studied as a physicochemical phenomenon. It is evident that, besides hydrodynamic effects, biomass adhesion to process supports has the potential to be strongly influenced by long-range (electrodynamic Lifshitz – van der Waals, electrostatic) and short-range (acid-base) interfacial interactions. Within the classical DLVO (Derjaguin, Landau, Verwey, Overbeek) theory, Lifshitz-Van der Waals (LW) and electrostatic interactions (EL) are considered while in the extended approach (XDLVO) the so called acid-base (AB) component is also accounted for. Application of these principles to process science would lead to the development of appropriate tools for better bioprocess

30

Results design and prediction and would guide the development of improved materials for downstream processing. The later is especially true when direct sequestration methods are in focus. Accurate understanding and prediction of interfacial forces during biomass adhesion onto process supports require the utilisation of quantitative models which, in turn, require experimental measurements to be performed. EL interactions arise from the existence of overlapping double layers of counter-ions near charged surfaces in aqueous media and are accessible by determination of the zeta potential. On the other hand, LW interactions are caused by the specific alignment and coupling of molecular dipoles. Additionally, the extended approach has been adopted to explain cell-surface interactions in the presence of other forces like hydrophobic (Van Oss 1995), hydration (Strevett and Chen 2003), and electrostatic (Camesano and Logan 2000). LW and AB forces are experimentally accessible via contact angle measurement with three diagnostic liquids. The aim of this paper was to contribute to a more in-depth understanding of biomassadsorbent adhesion and to propose a universal tool for process / material design. In doing so, the physicochemical properties of biomass-derived material, taken as colloidal particles, vs. the physicochemical properties of the adsorbent beads, taken as a process surface, were determined indirectly via contact angle and zeta potential measurements. Subsequently, total interfacial interaction energy values were calculated as a function of surface distance in aqueous media e.g. process buffer. Calculated interaction energy values were correlated to process performance. 2.1.3 Theory 2.1.3.1 Total interaction energy The total interaction energy between a colloidal particle and a solid surface can be expressed in terms of the classical DLVO theory as: DLVO LW EL U mwc = U mwc + U mwc

(1)

where UDLVO is the total interaction energy in aqueous media, ULW is the LW interaction term, and UEL is the EL interaction term. The subscript m is utilised for the chromatographic matrix (adsorbent bead), w refers to the watery environment, and c to the 31

Results colloidal (cell) particle. This classical DLVO approach can be extended to include a third short-range (≤ 5 nm) Lewis AB term so as to include “hydrophobic attractive” and “hydrophilic repulsive” forces into account (Van Oss 1994). XDLVO LW EL AB U mwc = U mwc + U mwc + U mwc

(2)

where UDLVO is the total interaction energy and UAB is the AB interaction term. 2.1.3.2 Lifshitz-van der Waals acid-base approach Surface energy parameters (tensions) can be calculated from contact angle measurements on the colloidal particles and adsorbent surface utilising the LW-AB approach. These parameters (or components) can be determined by performing contact angle measurements utilising three probe liquids (i.e. two high-energy polar liquids and one high-energy nonpolar liquid) with known surface tension parameters and employing the extended Young’s equation:

(1 + cos θ )γ lTOT

(

= 2 γ sLW γ lLW + γ s+ γ l− + γ s− γ l+

)

(3)

where θ is the contact angle, γLW is the LW surface tension parameter, γ+ is the electronacceptor parameter, and γ- is the electron-donor parameter. The subscript s and l is utilised for solid and liquid respectively. The polar AB component is given by:

γ AB = 2 γ + γ −

(4)

and the total surface tension of a pure substance can be represented by the sum of the polar AB and the non-polar LW surface tension parameters. The later represents a single electrodynamic property of a certain material.

2.1.3.3 Free energy of interaction The mentioned surface energy parameters can be employed to evaluate the free energy of interaction between two defined surfaces (ΔGLW - ΔGAB) e.g. the cell particles and the adsorbent bead (interaction) or between two cells (aggregation). ΔG represents here the interaction energy per unit area between two (assumed) infinite planar surfaces bearing the properties of the adsorbent bead and the cell or two cells, respectively. Moreover, contacts

32

Results between any of these two surfaces are evaluated at the so-called minimum cut-off distance (h0) i.e. the distance between the outer electron shells of adjoining non-covalently interacting molecules. The value for h0 is commonly assumed to be 0.158 nm. However, the mentioned LW and AB interaction energy components follow a unique decay profile with surface separation distance. Dimensions of the preceding equations are Joule. Nevertheless, for interaction energies the kT scale is preferred since 1 kT represents the Brownian motion energy of a microbial particle. The ULW energy-distance profile can be expressed according to the existing geometric constraints in order to obtain the actual interaction energy as: LW (h ) = − U mwc

LW (h ) = U cwc

⎛ h Rc A ⎡ Rc + ln⎜⎜ ⎢ + 6 ⎣ h h + 2 Rc ⎝ h + 2 Rc

− ARc Rm 6h (Rc + Rm )

⎞⎤ ⎟⎟⎥ ⎠⎦

(Sphere-Plate)

(5a)

(Sphere-Sphere)

(5b)

where Rc and Rm are the radius of the interacting bodies i.e. ~ 5 μm for yeast and ∼ 200 μm for the adsorbent bead. A is the Hamaker constant that can be obtained from ΔGLW, as calculated from contact angle measurements, according to

A = −12π h02 ΔG LW

(

LW ΔGmwc = 2 γ wLW − γ mLW

(6a)

)( γ

LW c

− γ wLW

)

(6b)

The UAB energy-distance profile can be expressed according to the existing geometric constraints in order to obtain the actual interaction energy as: ⎡h − h⎤ AB U mwc (h) = 2π Rc λ ΔG AB exp ⎢ 0 ⎥ ⎣ λ ⎦

(Plate-Sphere)

(7a)

⎡h − h⎤ AB U cwc ( h ) = π Rc λ ΔG AB exp ⎢ 0 ⎣ λ ⎥⎦

(Sphere-Sphere)

(7b)

where λ is a characteristic decay length for AB interactions in water (λ ~ 0.6 nm) and where:

33

Results

AB ΔGmwc = 2 γ w+

(γ

− m

)

+ γ c− − γ w− + 2 γ w−

(γ

+ m

) (

+ γ c+ − γ w+ − 2 γ m+ γ c− + γ m− γ c+

) (8)

In order to account for UEL energy-distance profile the following expression can be employed assuming either plate-sphere or sphere-sphere geometry, respectively: ⎡ 2ζ ζ ⎤ 1 + exp(− κ h ) EL ( h ) = π ε 0ε r Rc (ζ m2 + ζ c2 ) ⎢ 2 m c 2 ln U mwc + ln{1 − exp(− 2κ h )}⎥ (9a) ⎣ ζ m + ζ c 1 − exp(− κ h ) ⎦ EL (h ) = U cwc

⎤ π ε 0ε r Rc Rm (ζ m2 + ζ c2 ) ⎡ 2ζ mζ c 1 + exp(− κ h ) ln + ln{1 − exp(− 2κ h )}⎥ (9b) ⎢ 2 2 ( Rc + Rm ) ⎦ ⎣ ζ m + ζ c 1 − exp(− κ h )

where ε0εr is the dielectric permittivity of the suspending fluid, ζm is the zeta potential of the adsorbent bead, and ζc is the zeta potential of the cell particle. Zeta potential values are measured by electrophoretic mobility experiments. κ is the inverse Debye screening length and can be calculated on the basis of the relationship below:

κ=

e 2 ∑ ni z i2

(10)

ε r ε 0 kT

where e is the electron charge, ni is the number concentration of ion i in solution, zi is the valence of ion i, k is the Boltzman constant, and T is the absolute temperature.

34

Results 2.1.4 Materials and Methods 2.1.4.1 Materials Chromatographic matrices and columns were purchased from GE Healthcare (Munich, Germany). Solvents utilised for contact angle measurements (1-bromonaphtalene and formamide) were obtained from Fluka (Buchs, Switzerland) with 99% and 99.5% purity, respectively. Water was Milli-Q quality. All other chemicals were of analytical grade. The goniometric system (OCA 20) was obtained from DataPhysics Instruments GmbH (Filderstadt, Germany). Zeta potential was measured with a Zetasizer Nano ZS from Malvern Instruments (Worcestershire, United Kingdom). 2.1.4.2 Biomass Yeast cells (Saccharomyces cerevisiae) were cultivated, harvested at late exponential phase, and washed three times with dilute buffer solutions (Ganeva et al. 2004). Fresh E. coli DH5α biomass was produced according to standard methods (Sambrook and Russell 2006). Cell disruption was performed by bead milling, as previously described (FernandezLahore et al. 1999). 2.1.4.3 Surface preparation for contact angle measurements Preparation of the intact yeast cells for contact angle measurements was performed essentially according to Henriques et al. (Henriques et al. 2002). Fresh (washed) cells were suspended to 10% (w/v) in 50 mM citrate phosphate buffer (pH 3, 5, 7 or 9). Suspended cells were further allowed to equilibrate in the respective buffer for 30 minutes and the suspension was poured onto an agar plate containing 10% glycerol and 2% agar-agar. The plate was allowed to dry for 24-36 hours at room temperature on a properly levelled surface and free from dust. Agarose-based adsorbent beads harbouring various ligand chemistries were thoroughly equilibrated in 50 mM sodium acetate (pH 4) or 20 mM phosphate buffer (pH 7). Once equilibrated, matrix beads were frozen in liquid nitrogen and crushed mechanically. Crushing efficiency was assessed by microscopic examination and particle size determination. Crushed matrix was made 40% (w/v) in buffer and allowed to remain in

35

Results contact with the liquid phase for additional 30 minutes with gentle mixing. Fine bead fragments were poured onto a glycerol-containing agar plate and allowed to dry, as described before. Immobilized biomass or adsorbent fragments (< 10 μm diameter) on squares pieces of the agar-supported surface were utilised for measuring the contact angles. 2.1.4.4 Contact angle measurements For the contact angle estimation, the sessile drop technique was utilized (Sharma and Rao 2002). Data acquisition and analysis was performed utilising SCA20 commercial software (DataPhysics Instruments GmbH, Filderstadt, Germany). Measurements were performed at room temperature, using three different diagnostic liquids: water, formamide and 1bromonaphtalene. Assays were performed in triplicate and at least 20 contact angles per samples were measured. Contact angles were measured for biomass samples as a function of pH in the range from 3.0 to 7.0. Measurements for adsorbent materials were performed at pH 4.0 and pH 7.0 in diluted buffer solutions. The solution chemistry employed reflected common process conditions. 2.1.4.5 Zeta potential determinations Particle zeta potential was determined for the cell particles and for the chromatographic supports under study. Biomass-derived particles were suspended to 1% (w/v) in 20 mM phosphate or citrate-phosphate buffers. Fragmented Sepharose beads were utilized instead of Streamline beads due to their lower density and to avoid sedimentation during measurements. Particles were contacted with buffer until equilibrium was reached and further diluted to appropriate particle count (~200) before measuring the zeta potential. Zeta potentials were calculated from the electrophoretic mobility data as per the Smoluchowski’s equation (Ottewill and Shaw 1972). All the measurements were done in triplicate.

36

Results 2.1.4.6 Biomass-pulse experiments Experiments were performed on an ÄKTA Explorer system (GE-Healthcare, Munich, Germany) utilising a modified XK-10 chromatographic column filled with the sample adsorbent (2.0 ml) and irrigated from the bottom with the mobile phase. The solid phase was fluidised at a fluid velocity of 7.5-8.3⋅10-4 m⋅s-1 in order to promote the formation of a stable expanded bed. A biomass pulse (~ 2 ml of 0.03% w/v biomass suspension) were loaded into the system through a three way injection port. Cell concentration in the pulse before and after passage through the expanded bed was detected on-line by measuring the optical density at 600 nm. Results were expressed as Cell Transmition Index (CTI) (Feuser et al. 1999). 2.1.4.7 Partition experiments Solid-liquid partitioning experiments were performed with adsorbent beads and biomass in glass flasks (4 cm height, 1.5 cm diameter), which were closed with plastic caps. Chromatographic beads (0.5 ml) were contacted with a cell suspension (2.0 ml of 0.03% w/v) under gentle orbital stirring. Samples were taken after 15 min and 3 h to evaluate the fast and slow phases of cell deposition (Fernandez-Lahore et al. 2000). The optical density of the samples was evaluated by absorbance at 600 nm. The fraction of non-bound cells or biomass particles to each material type was defined as the Cell Partition Index (CPI).

37

Results 2.1.5 Results and Discussions 2.1.5.1 Contact angle measurements The diagnostic liquids water, formamide, and 1-bromonaphtalene were employed to measure contact angles on lawns of hydrated biomass or crushed agarose beads utilising the sessile drop technique. The surface free energy components γLW and γAB, as well as, the electron-donating and electron-accepting parameters for these liquids can be found in the literature (Bos et al. 1999). Biomass and adsorbent fragments were equilibrated in 20 mM phosphate buffer pH 7, which provides a chemical environment similar to that found under ion-exchange sorption processing in industrial practice. In this work, biomass or crushed adsorbent lawns were prepared on agar layers (Henriques et al. 2002). This method permits contact angle measurements under the assumption that only bound water is present on the sample surface and proved to be suitable to handle a variety of materials by forming an even and homogeneous surface. Table 1: Contact angle measurements for agarose-based beaded supports. Determinations were performed on lawns of crushed agrose-based adsorbents in in 20mM Phosphate buffer, pH 7.

Support type

Contact angle (θ) Water

Formamide

1-Bromonaphtalene

Sepharose 4B

9.5 ± 2

10 ± 1

44 ± 1

Q Sepharose XL

12 ± 1

14 ± 2

52 ± 1

9.6 ± 3

13 ± 2

41 ± 1

6.7 ± 3

13 ± 1

39 ± 1

DEAE Sepharose SP Sepharose

Table 1 shows the contact angle values for the anion-exchanger DEAE-Sepharose, the cation exchanger SP-Sepharose, and the agarose base material 4B-Sepharose. An additional composite ion-exchanger, Q-Sepharose-XL was also included. Sepharose materials were utilised for obtaining small particles suitable for contact angle measurements since Streamline materials have a difficult-to-brake quartz core but similar chemical structure. In the later case, particle diameter was lower than 10μm to assure no interference with

38

Results measurement of contact angles. Contact angle values with polar liquids were similar for all the chromatographic supports under consideration. Low values were observed for water (712°) and formamide (10-14°), which reflects the general very hydrophilic nature of the beads under study. However, 1-bromonaphtalene contact angles were able to discriminate between Agarose- based beads (Sepharose series) and the Agarose-Dextran composite (XL material). The contact angle value with the apolar solvent for the later was 20% higher, which might indicate an increased hydrophilic character for XL-Sepharose in comparison with standard materials. This is in agreement with the known higher hydrophilic character of Dextran T-70 in comparison with polymeric Agarose as judged by the free energy of interaction of these molecules in water: ΔGsws was reported as -9.2 mJ·m-2 for agarose and +17.6 mJ·m-2 for Dextran (Van Oss 1994). Moreover, comparison between 1bromonaphtalene contact angle values for functionalised vs. non-functionalised Sepharose materials showed decreased values for the first group i.e. 44° (base material) vs. 39°- 41° (SP and DEAE materials, respectively). This might reflect an increased hydrophobic character of the functionalised adsorbent due to the influence of ligand immobilisation chemistry. Contact angles were determined with the various adsorbents and all the three liquids at pH 4 but no major changes were observed (data not shown). Table 2: Contact angle measurements for biological materials. Determinations were performed in 20mM Phosphate buffer at pH 7. Biomass type

Contact angle (θ) Formamide

Water

Intact yeast cells

1-Bromonaphtalene

Yeast homogenate

15 ± 2 18 ± 1

14 ± 1 22 ± 2

54 ± 1 53 ± 1

Bacterial homogenate

28 ± 4

30 ± 2

54 ± 3

Table 2 shows the contact angle values obtained for biomass types which are relevant to process situations: intact yeast cells, disrupted yeast cells, and disrupted bacterial cells. Saccharomyces cerevisiae and Escherichia coli were employed as model biomass systems. Cell disruption was accomplished by bead milling which generated yeast cell fragments with a size ~ 2-3 μm and bacterial fragments with a size ~ 1 μm. As opposed to the observed trend when analysing the adsorbent materials, contact angle values for the apolar 39

Results liquid were almost the same i.e. 54° with 1-bromonaphtalene. This apolar liquid can probe the ability of biomass surfaces to exert LW forces, as expressed by the Hamaker constant. Although it is not the case for the experimental work performed here, previous investigations among large collections of different microorganisms may impede the generalisation of Hamaker constant values for biomass adhesion analysis (Sharma and Rao 2002). When considering the biomass type, contact angle values with water varied from 15° (intact yeast cells) to 18° (yeast debris) and 28° (bacterial homogenate). Similarly, formamide contact angles varied between 14° (yeast cells) to 22° (yeast debris) and 30° (bacterial debris). This indicates that polar liquids can be employed as main discriminators between biomass types. For the biomass types studied in this work contact angle values with both polar liquids were: Intact yeast cells < Yeast homogenate << Bacterial homogenate This sequence can be interpreted in terms of decreased hydrophilic/hydrophobic ratio. The discrimination capability found with polar liquids is in agreement with previous reports by others (Van der Mei et al. 1998) and further supports the need for experimental evaluation of contact angles with the biomass material present in the actual industrial feedstock. This is due to the fact that no general assumption can be made regarding the hydrophilic or hydrophobic character of a certain species/strains. Contact angles were also performed with the various biomass types and the three diagnostic liquids at pH 4 but no major changes were observed. Some changes were noticed, however, at pH 9 (data not shown). 2.1.5.2 Surface free energy calculations Contact angle data has allowed the calculation of the LW and AB surface free energy components by application of the modified Young-Dupré equation (Table 3 and 4). Neglecting at this point the influence of electrostatic interactions and structural characteristics of the cell wall, the nature of the biomass surface was directly quantitated according to (Van Oss and Good 1988; Volpe and Siboni 1997):

γ sv = γ svLW + γ svAB

40

Results In which the AB component equals

γ svAB = 2 γ sv− γ sv+ The electron-donating ( γ sv− ) and the electron-aceptor ( γ sv+ ) free energy components give an indication on the biomass or material ability to exert acid-base interactions on an scale taking water as an arbitrary reference. Since 1-bromonaphtalene is apolar ( γ lvAB = 0), this liquid can be utilised to calculate the LW component of the biomass/material:

γ

LW sv

⎧⎪ γ (cos θ + 1) ⎫⎪ = ⎨ lv ⎬ 2 ⎪⎩ ⎪⎭

2

On the other hand, since water and formamide are polar, these liquids can be employed in combination to calculate the electron-donating and electron-accepting parameters of the sample surface from:

γ lv (cos θ + 1) − 2 γ svLW γ lvLW = 2 γ sv− γ lv+ + 2 γ sv+ γ lv− According to van Oss (Van Oss 1997) the hydrophilic/ hydrophobic character of a certain material can be defined in terms of the variation of the free energy of interaction between two moieties of that material immersed in water. This is given after:

(

TOT ΔG sws = −2 γ sLW − γ wLW

) − 4( γ 2

+ s

γ s− + γ w+ γ w− − γ s+ γ w− − γ s− γ w+

)

Calculation of ΔGsws yielded positive values for all the adsorbents i.e. > +25 mJ·m-2, which demonstrates their hydrophilic nature. However, further inspection of Table 3 showed that a lower γLW value was obtained for the XL material and therefore the more polar character of this composite in comparison with the Agarose beads can be confirmed. Moreover, an increased value for the electron-acceptor parameter characterises the composite adsorbent. On the basis of the contact angle values obtained with polar liquids and the calculation of the corresponding surface free energy parameters, the tested adsorbents can be considered to have a polar character according to the following series: Q-XL > Beaded agarose > DEAE = SP

41

Results Calculation of ΔGsws also yielded positive values for all the biomass types under consideration (Table 4). Concerning microorganisms it is generally accepted that ΔG > 0 characterises hydrophilic cell surfaces and therefore, the tendency of these biological particles to aggregate in aqueous environments is very limited. This is particularly true for cells suspended in dilute buffer solutions, which are common during adsorption onto ionexchangers. Table 3: Surface energy parameters for agarose and beaded chromatographic supports calculated from contact angle measurements at pH 7. Surface energy parameters [mJ m-2]

Support type

Agarosea Dextran T-70a Sepharose 4B Q Sepharose XL DEAE Sepharose SP Sepharose

γLW

γ+

γ-

γAB

γTOT

ΔGsws

40.9 41.8 32.8 28.9 34.1 35.0

0.1 1.0 2.9 3.9 2.3 2.0

23.8 47.2 53.6 53.2 54.5 55.7

2.9 13.7 24.9 28.8 22.3 21.1

43.9 55.5 57.7 57.8 56.7 56.4

-9.2 +17.6 +28.1 +26.6 +30.7 +31.9

(a) Taken from van Oss (Van Oss 1994). Table 4: Surface energy parameters for several common biomass types at pH 7 as calculated from contact angle measurements. Surface energy parameters (mJ m-2)

Biomass type γLW

γ+

γ-

γAB

γTOT

ΔGsws

Intact yeast cells

27.9

4.4

51.5

30.1

58.3

+24.3

Yeast homogenate

28.4

3.3

53.2

26.4

55.2

+28.1

Bacterial homogenate

27.9

2.7

49.2

23.1

51.3

+26.0

42

Results

2.1.5.3 Interfacial free energy of adhesion: interaction and aggregation Biomass adhesion to process supports can be considered as a complex phenomenon including at least two distinct phases: a) The interaction phase characterised by rapid kinetics, where a biological particle approaches the adsorbent bead, and b) The aggregation phase characterised by cell to cell clumping. The later phase shows lower kinetics and it is affected by the process contact time and the biomass concentration in the feedstock (Fernandez-Lahore et al. 1999). Table 5 depicts the values for interfacial free energy of interaction between several ionexchangers and model biomass particles, at closest distance of approximation. ΔGLW values for agarose-based chromatographic supports were very similar, irrespective of the ligand chemistry. This may indicate the strong influence of the base material e.g. cross-linked agarose on the calculated LW energy component. As expected LW energy values were negative, indicating an attractive interaction. The composite material Q-XL showed a 30% decreased ΔGLW value (-0.9 mJ·m-2), which again indicates the different structural nature of the later. On the basis of experimental ΔGLW values it was possible to calculate an average Hamaker constant for agarose-based materials equal to 4⋅10-21 J or 0.34 kT. This is in agreement with commonly assumed values for microbial systems (~ 0.49 kT). Variations in ΔGLW values within a range ± 50% were observed when comparing beaded agarose supports with other biomass-interacting materials, like polystyrene, ceramic, or glass. Repulsive forces were found to play a role during biomass interaction phenomena. This forces, based on electron donor / electron acceptor or Lewis acid-base, can be seen as responsible for abnormalities found in the DLVO theoretical interpretation of interfacial interactions in aqueous media. Table 5 shows an average value for the studied chromatographic supports in the range +26 to +30 mJ·m-2. These values for the AB component are 20 times higher than those found for the attractive LW component. AB forces are known to surpass other DLVO forces by as much as two decimal orders of magnitude and therefore are extremely important in understanding biomass-support interactions. The decay with distance of the AB interaction energy is assumed to describe the distance dependence of the boundary layer ordering. Moreover, ΔGAB values were shown to change when other systems were examined. For example, the E. coli / PES system

43

Results showed a low value (+2.5 mJ·m-2) and the yeast / Q-Hyper Z or the mammalian cell / glass showed moderate values (~ +20 mJ·m-2). From these data it becomes clear that acid-base forces exerted in dilute buffer solutions have the potential to strongly influence biomass interactions during normal processing conditions. Moreover, ΔGAB forces are the dominant component of the calculated total interfacial free energy of interaction (ΔGTOT as per Table 5) in several process systems of biochemical and environmental importance. Table 5: Interfacial free energy of interaction between biomass and process materials. Calculations were performed assuming interactions under process buffer conditions at pH 7. Biomass type

Support

ΔGLW

ΔG (mJ·m-2) ΔGAB

ΔGTOT

Intact yeast cells

Agarose beads XL-Q DEAE SP

-1.3 -0.9 -1.4 -1.5

+27.6 +26.3 +28.7 +29.7

+26.3 +25.5 +27.4 +28.0

Yeast homog.

Agarose beads XL-Q DEAE SP

-1.4 -0.9 -1.5 -1.6

+29.7 +28.3 +30.9 +31.9

+28.3 +27.3 +29.3 +30.2

Bacterial homog.

Agarose beads

-1.3

+28.6

+27.3

XL-Q DEAE SP

-0.9 -1.4 -1.5

+27.3 +29.7 +30.7

+26.5 +28.3 +29.0

PES

-2.0

+2.5

+0.5

S. cerveviseae

Q-Hyper Z

-0.7

+18.4

+17.7

Mammalian c cells

Glass

-2.5

+20.9

+17.3

a

E. coli

b

a

b

c

Taken from (Gallardo-Moreno et al. 2002) , taken from (Vergnault et al. 2004), taken from (Li and Logan 2004; Van Oss 1994).

44

Results Cell to cell aggregation was also characterised in terms of free surface energy components (Table 6). Following the general trends already described in the preceding paragraphs, ΔGLW values showed attraction between particles (~ -1.6 mJ·m-2) while repulsion was dominating as judged by ΔGAB values in the range +25 - +29 mJ·m-2. Therefore, in the absence of other attractive forces like charge-mediated effects, these biological particles would have a tendency to coexist as discrete entities suspended in aqueous media. Table 6: Interfacial free energy of aggregation between biomass particles. Calculations were performed assuming interaction under process buffer conditions at pH 7. Biomass type ΔGLW

ΔG [mJm-2] ΔGAB

ΔGTOT

Intact yeast cells

-1.5

25.1

23.6

Yeast homog.

-1.7

29.0

27.3

Bacterial homog.

-1.5

26.7

25.2

2.1.5.4 Electrostatic double layer forces The zeta potential is an important parameter, which is commonly employed to characterise biomass-adsorbent electrostatic interactions (Lin et al. 2003). This parameter has also been extensively applied to understand many colloidal systems, the surface charge properties of materials, and the microbial adhesion in porous media. The zeta potential, as measured by (micro) electrophoresis, gives information on the surface charge of a certain particle-type as a function of solution chemistry i.e. ionic strength, electrolyte composition, and pH. Table 7 summarises zeta potential values for the chromatographic supports under investigation. In diluted aqueous buffers (≤9 mS/cm, pH 7) agarose-based materials showed a slightly negative charge (~ -2 mV). This is in agreement with the value reported for soluble polysaccharides like Dextran (-0.05 mV) (Van Oss 1994). Beads harbouring positively charged ligands, under similar conditions, showed zeta potential values ~ +12 mV (DEAE) and ~ +17 mV (Q-XL). On the contrary, the cation exchanger SP showed strongly negative zeta potential values (∼ -30 mV). Zeta potential measurements approached minimal values close to zero, when the ionic strength of the medium was

45

Results increased. These values keep constant by decreasing the pH down to 4 in diluted buffer solution. Table 7: Zeta potential values for beaded adsorbents. Zeta potential (mV) pH (-) Cond. (mS/cm)

4

7

≤9

~15

≤9

~15

-3.2a +27a / +18c +15a / +24c / +13d -14a / -29c

nd nd

-2.0a + 15a / +14c +8.7a / +21c / +11d -24a / -30c

nd +8d

Support type

Sepharose 4B Q Sepharose XLb DEAE Sepharosee SP Sepharosee

nd nd

+6d nd

Zeta potentials values for silica particles (Si-m) were reported as -36.5 mV by Li and Logan (Li and Logan 2004). nd: not determined a Own determinations b Experiments were performed in sodium acetate buffer at pH (1mS/cm) and sodium/potassium phosphate buffer at pH 7 (4 mS/cm). c Published values after Lin et al.(Lin et al. 2006)as performed in 50 mM sodium phosphate buffer pH 7.2 or values after Lin et al (Lin et al. 2003) in 10 mM KNO3 pH 7.2. d Zeta potential measurements according to Lin et al (Lin et al. 2006)in 50 mM phosphate buffer and sodium chloride as added salt. e Experiments were performed in 20 mM sodium/potassium phosphate buffer at pH 4 and pH 7. Cell or cell debris particles are also known to bear surface charge. Table 8 depicts zeta potential values for several model biomass types like intact yeast cells, yeast homogenate particles, and bacterial debris. It can be observed that intact yeast cells have negative zeta potential values raging from ∼ -30 mV to ∼ -10 mV in salt solution from 1.0 to 100 mM, respectively, at neutral pH. At lower pH (∼ 4) zeta potential values ranged from ∼ -18 mV to

∼ -2 mV. Biological particles originated by cell disruption have had a tendency to be

less negative (yeast debris ∼-12 mV) or more negative (bacterial debris ∼ -30 mV) than intact yeast and bacterial cells, respectively. This fact reflects both the influence of

46

Results feedstock treatment and the biological nature of the feedstock (Brixius 2003; Lin et al. 2007). Values reported here for zeta potential of biomass and process materials were obtained by a meta-analysis of the current literature and confirmed by own measurements. Special emphasis was placed on conditions that are significant in industrial practice, like feedstock loading (buffer conductivity ≤ 2-9 mS·cm-1) and product elution (buffer conductivity ≥ 15 mS·cm-1). Extremes of pH (4 and 7) were considered to evaluate to potential effect of pH change on electrostatic interactions. Table 8: Zeta potential values for biological particles. Zeta potential (mV) pH (-) Cond. (mS/cm)

4

7

~2-4

~15

~2-4

-7a / -7b / -9c

-2c

Yeast homogenate

-4a / -3d

nd

-15a / -9b / 21c -12a / -14d

E. coli homogenate+

-10a / -5d

nd

-30a / -35d

~15

Biomass type

Intact yeast cells

-16c / -7d -5d -22d

nd: Not determined. (+) Published values for intact E.coli cells are -17 at pH 4 and -27 at pH 7 in 50 mM phosphate buffer (Lin et al. 2006). Mammalian cells were reported to have zeta potential values ~ -25 mV (Van Oss 1994). a Own measurements in 20mM sodium/potassium phosphate buffer. b Data from Lin et al.(Lin et al. 2006) c According to Kang et al. (Kang and Choi 2005) d Taken from Lin et al.(Lin et al. 2007) 2.1.5.5 Adhesion and interaction phenomena: free energy vs. distance profiles Integration of the various existing interfacial forces between an adsorbent bead and a biological particle (interaction) or between two biological particles (aggregation) can be performed by calculating Energy (U) vs. distance (H) profiles. Figure 1 depicts such interfacial energy curve for the adhesion of an intact yeast cell onto a DEAE Streamline™ bead. From this figure, it can be realised how different interfacial forces can contribute to

47

Results cell-bead interaction. While LW and EL forces are attractive in this case, AB forces are repulsive. The total energy profile obtained after application of the DLVO theory would predict an infinite (primary) energy pocket where a cell would be irreversibly trapped in close contact with the adsorbent bead. However, previous experimental findings have shown that intact cell yeasts can be detached from adsorbent beads by applying an increased shear stress to the system. Calculations based on the XDLVO theory better explain this phenomenon by showing a secondary energy minimum having a finite magnitude.

Figure 1: Interfacial free energy components as a function of distance for a DEAE functionalized adsorbent particle and an intact yeast cell in aqueous media: (—) LW: (—) EL: (—) AB. Total interaction energy profiles are shown according to the DLVO theory (—) and the (—) XDLVO theory.

In order to better elucidate the appropriateness of the XDLVO theory to predict microbial adhesion within the frame of biochemical engineering systems, calculation were run with several cell / support pairs. Figure 2 depicts the total energy vs. distance profiles for selected agarose and non-agarose based materials and several biomass types. Adsorbent beads suitable for expanded bed operation showed, in agreement with previous reports,

48

Results strong interaction with intact cells. It is now clear that this interaction occurs at a distance of 4-5 nm and within (secondary) energy wells between -200 kT for DEAE (taken as a reference) and -400 kT for Q-Hyper-Z, a ceramic composite material. These findings can explain why the utilisation of dynamic flow distribution and the introduction of denser particles can alleviate biomass adhesion to fluidised adsorbent: the increased shear / hydrodynamic stress provide enough energy to detach cell particles from the (finite) secondary minimum. Other system behaved differently. The PES / bacteria pair, which is know as a strong interacting system (Absolom et al. 1983), showed an infinite primary minimum. This is due to the hydrophobic nature of the solid substrate and the low contribution of repulsive AB forces. On the other hand, the mammalian / glass pair showed a moderate secondary energy well (-60 kT) occurring at 15-20 nm distance. This is also in agreement with hydrodynamic limitations found during the purification of monoclonal antibodies onto porous glass cation exchangers in the fluidised mode (Thommes et al. 1995).

Figure 2: Total interaction free energy profiles for several process systems according to the extended approach: (—) DEAE/yeast cells, (—) PES/bacterial cells, (—) Q ceramic/yeast cells, and (—) glass particles/mammalian cells.

49

Results Interaction energy as a function of biomass type in the feedstock can be observed in Figure 3. Calculation performed for intact yeast cells, yeast homogenate, and bacterial homogenate suggest that a much strong interaction would occur between the intact cell and the DEAE Streamline adsorbent bead than with any of the two homogenates. This is in fully agreement with experimental evidence reported earlier by Fernandez Lahore et al (Fernandez-Lahore et al. 2000; Fernandez-Lahore et al. 1999). Additionally, from this figure the effect of particle size on the overall energy vs. distance profile can be understood. Lin et al. (Lin et al. 2003) have realised the importance of biological particle size, besides the obvious electrostatic effects between two opposite charged spheres, during biomass interactions in EBA. Both factors, in addition to the contribution of LW, AB, and BR forces are nicely summarised in a single U vs. H curve.

Figure 3: Total energy vs. distance profiles for an anion-exchange (DEAE) chromatographic support and various types of biomass-derived particles. (—) intact yeast cells; (—) yeast homogenate; (—) bacterial homogenate.

50

Results Process operational parameters and process materials are of prime importance to biomass adhesion effects. Figure 4 depict the energy vs. distance profiles for agarose based supports in contact with intact yeast cells. Interaction of yeast cells with these adsorbent beads, as judged by the depth of the energy pocket, can be ordered as follows: Q XL(-370 kT) > DEAE (-200 kT) >> BASE (-12 kT) ≈ SP (-10 kT) The depth of the energy pocket correlates with the proximity of such interaction between cells and adsorbent beads, as follows: Q XL (4 nm) > DEAE (5 nm) > BASE (10 nm) > SP (20 nm ) The strongest and closest interactions predicted by the XDLVO theory in the mentioned scale, therefore, fully correlates with previous work on biomass interaction and hydrodynamics in expanded beds. In the same line of thought (Fernandez-Lahore et al. 2001), it has been reported that ionic strength operation windows could help in alleviating hydrodynamic and sorption performance constraint during EBA operation with anionexchangers. Figure 5 shows energy vs. distance calculations for various buffer conditions i.e. low vs. high pH and low vs. high conductivity within the range expected to occur during real operations. Moreover, with ion-exchange operations it was found that an increased conductivity reduced the depth of the energy pocket from -200 kT to – 40 kT. Charge-masking effects and double layer compression mainly dominate this effect. The influence of the pH, within the range 4 to 7, was only marginal for yeast / anion-exchanger interacting pair. Calculations performed with experimental data gathered from CEX materials and intact yeast cells revealed an opposite behaviour. The later showed the development of a secondary energy well at low to very-low salt concentration in the running buffer (data not shown). This situation could lead to unexpected biomass interactions with materials known as “non-interacting” (SP) and with mobile phase compositions that are not suspected to promote impaired hydrodynamic conditions.

51

Results

Figure 4: The influence of the functional ligand on the total interfacial energy when an intact yeast cell and an agarose-based bead are the interacting bodies. (—) DEAE, (—) QXL,(—) SP, (—) BASE.

Figure 5: Energy vs. distance curves showing the influence of the buffer pH and conductivity on the interaction between an anion-exchange bead and an intact yeast cell.(—) low conductivity pH 7, (—) high conductivity pH 7, (—) low conductivity pH 4,(•••) high conductivity pH 4. Low conductivity: 2 ms cm-1, High conductivity: 14 mS cm-1.

52

Results 2.1.5.6 Aggregation phenomena: the cell to cell interface As mentioned before, biomass adhesion kinetics recognises two main phases as revealed by partitioning experiments (Fernandez-Lahore et al. 2000). The second, slow, biomass concentration dependent phase was linked to cell-to-cell aggregation (homo-coagulation). Therefore, it is also important to study the interfacial free energy between two cell particles to have a clear picture of biomass adhesion onto surfaces during bioprocessing. Mathematical expressions derived for sphere-sphere contact were utilised. Figure 6 depicts cell aggregation at pH 4 and 7. The figure also presents calculated profiles for low and high salt concentrations at these two pH values. It can be observed that at low salt concentrations -and with little influence of the pH value- the secondary energy pocket is almost inexistent. At high salt concentration, however, the depth of the energy trap increases moderately (∼510 kT) due to the compression of the double layer at increased ion concentration in the solution. Consequently, for particles repelling each other by charge-mediated effect the probability of aggregation is higher than in diluted buffer. This situation is analogous to the interaction between a (negatively charged) cation-exchanger bead and a (negatively charged) cell particle, as mentioned before. Interaction effects of this kind, however, are expected to have more impact on the retention of intact cells than on the adhesion of cell debris since the size of the particles involved is higher in the first case. Moreover, aggregation might be worsened by the presence of bivalent ions since they were reported to depress the monopolar electron-donor parameter of the surface tension. This would result in depressing their mutual repulsion (Van Oss et al. 1987). According to these studies, the calculated Hamaker constant was 1.4.10-21 for biomass aggregation.

53

Results

Figure 6: Interfacial energy between two yeast cells as function of solution chemistry. (—) 2 mS cm-1 pH 7, (•••) 14 mS cm-1 pH 7, (—) 2 mS cm-1 pH 4, (—) 14 mS cm-1 pH 4.

2.1.5.7 The energy pocket as predictor of process performance A correlation between the magnitude and sign of the secondary energy pocket and standard indexes employed to evaluate biomass adhesion onto fluidised adsorbent was established. These indexes, as obtained by cell pulse experiments or partition experiments are known to correlate with the quality of bed fluidization and sorption performance in EBA. This correlation is depicted in Figure 7. Three main groups of data points can be observed: a) A first group (U values > -20 kT) showed almost complete cell transmition (CTI ≥ 90%). Process conditions represented by these points are not expected to create hydrodynamic disturbances and thus, maximised sorption performance will be reached. This group is represented by the cation exchangers in dilute buffer solutions, and the anion exchangers at moderate-high salt containing buffers. The agarose-base material is also included here. b) A second group (-20 kT < U < -200 kT) showed a linear correlation with the cell transmition index, from 40% to 90%. Process conditions represented by this group require 54

Results process optimisation for appropriate sorption performance. Interventions via operational window optimisation or solution chemistry design are mandatory. In the worst cases, material engineering could restore process performance. This group is mainly composed by several anion exchangers operating in diluted buffer at pH 4 – 7 and by the cation exchangers in moderate-high salt containing buffers. This group also contain adsorbent materials other than ion-exchangers (data not shown). c) A third group was obtained at strongly negative energy pocket (≤ -300 kT) where cell transmission was extremely reduced e.g. < 30%. The later situation is most commonly associated with a complete bed collapse upon feedstock application. This group is represented by strongly adhesive systems, especially when AEX brush-type composite materials are employed in diluted buffer solutions or when DEAE supports are subjected to long contact times. The effect of hydrodynamic forces on the detachment of biological particles on chromatographic supports and other process materials can be explained on the basis of the finite value obtained for the calculated minimum of energy. A finite (secondary) minimum is present at 4-5 nm distance between the interacting bodies, even in those cases where strong biomass attachment is known to occur and to cause impaired sorption performance. Under such circumstances, detachment of adhering cells will be promoted upon applying enough energy to overcome the energy pocket. Since interaction energies are proportional to the particle radius, the effect of the mentioned energy secondary minimum is expected to be more significant for larger particles. This is in agreement with the strong biomass adhesion observed for intact yeast cells (4 μm diameter) onto fluidised beads as compared to cell debris (< 1μm diameter). The surface energetics approach presented in this work can be useful in guiding process developments. Calculations can be performed easily utilising a personal computer and commercial software. This assists in finding conditions for reduced interaction and aggregation with a minimum of experimental effort. Moreover, our approach can help in the development of novel (less interacting) materials for direct capture applications. In a recent publication, Kang and Choi (Kang and Choi 2005) have demonstrated the effect of surface modification as a controlling factor in microbial adhesion. These authors, in

55

Results agreement with this work, have also explained the interaction between microbial cells and solid substrates on the basis of the XDLVO theory. The surface energetic approach has a universal nature as predictor of process performance since it is not restricted to the type of material, the nature / size of the biomass particles, and the environmental conditions prevailing within the running phase. As such, it is not restricted to process situations that are dominated by coulomb-type interactions. For example, Gallardo-Moreno etal. (Gallardo-Moreno et al. 2002) have found a good correlation between thermodynamic prediction and adhesion behaviour of Candida parapsilosis to polystyrene. It is worth to mention at this point that interactions other than the ones described here may influence interaction between biological and/or polymeric particles. For example, steric interaction may arise between a polymer-coated surface, which is the case for some microbial and adsorbent surfaces. A crude feedstock may also contain variable amounts of bridging cations or macromolecular polymers (Dainiak et al. 2002; Mattiasson et al. 1996).

56

Results

Figure 7: Correlation graph between the depth of the secondary energy pocket and the cell or partition transmission index (CTI). (•) DEAE, (▲) Q- XL, (▼) Q- Hyper D, (■) SP, (×) Base.

2.1.6 Conclusions A universal approach was developed to understand biomass adhesion to process supports, with special reference to downstream processing systems where the direct sequestration of targeted species is intended from the crude feedstock. Besides the influence of coulombtype interactions, this approach takes into account other forces so as to present a more comprehensive visualisation of the underlying thermodynamic phenomena of adhesion. This is conveniently performed by utilisation of energy-distance profiles. In this way, the distance and strength of interaction can be explored for support-biomass interaction, as well as, for cell-cell aggregation. The LW interaction, which is predominantly attractive in microbial systems, was not influenced by the ionic strength but both the range and magnitude of the EL interactions decrease with increasing ionic strength due to shielding of surface charges. AB interactions were found to be a function of the nature of the process solid phase onto which cell

57

Results adhesion took place. As a consequence, depending on the liquid phase ionic strength and the nature of the process material, a (finite) secondary minimum may exist allowing the reversible capture of biological particles. At the distance at which the energy pocket occurs, a correlation exists between the depth of such energy minima and the degree of biomass entrapment as described by the “Cell transmition index” and the “Cell partition index”. This correlation is valid for both cation and anion exchangers under a variety of common operational conditions, which are relevant to industrial practice. According to the XDLVO methodological approach interactions are predicted to be alleviated by working within operational windows at moderate conductivity values for AEX i.e. when employing diluted buffer for sample application. On the contrary, high conductivity values might hamper CEX operations i.e. under elution conditions were high salt concentration are commonly utilised. The evaluation of the complete range of interfacial forces, as proposed here, represents a first step to global modelling. This would further establish a link between shear / hydrodynamic effects and cell adhesion onto process surfaces. Calculations required are simple to produce and are based on two experimental measurements that are contact angle measurements and zeta potential determinations. This approach is useful for process design where reduced optimisation time would be required. But particularly the method provides an excellent tool for novel material design. This in not only restricted to the development of improved expanded bed adsorbents. Reactive solid phases utilised in other direct-capture unit operations like finite bath systems, separations based on magnetic particles, macroporous systems, and big-beads packed beds can be tailored with assistance of the surface energetics approach. 2.1.7 Acknowledgements This work was partially funded by the BID 1201/OC AR 649 PICT 08352 and the start-up grant from Jacobs University [IUB] (2130-90050). The authors would like thank Dr. H. C. van der Mei for valuable discussions.

58

Results

2.1.8 Nomenclature A

Hamker constant [J] / [kT]

CPI

Cell partition index [-]

CTI

Cell transmission index [-]

BR

Born repulsion

DEAE

Diethylaminoethyl-

EBA

Expanded Bed Adsorption

∆G

Total interfacial free energy [mJ⋅m-2]

ho

Distance at a closet approach [m]

H

Distance [m]

IC

Intact cells

IEX

Ion exchange Chromatography

PES

Plastic

R

Radius [m]

Si-m

Silica

SP

Sulphopropyl-

T

Temperature [K]

U

Total interaction energy [kT]

Greek letters γ

Surface tension [mJ⋅m-2]

γ+

Electron-acceptor component of surface tension (Lewis acid) [mJ⋅m-2]

γ-

Electron-donor component of surface tension (Lewis base) [mJ⋅m-2]

λ

Characteristic decay length [m]

ε0

Permittivity of vacuum [J⋅m-1⋅V-2]

59

Results εr

Relative permittivity or dielectric constant for water [-]

κ

Inverse of Debye length [m-1]

k

Boltzmann constant [J⋅K-1]

ζ

Zeta potential [V]

Superscripts AB

Acid-Base

DLVO

Derjaguin, Landau, Verwey and Overbeek Theory

EL

Electrostatic

LW

Lifshitz-Van der Waals

TOT

Total

XDLVO

Extended DLVO Theory

Subscripts c

Cell particle

m

Chromatographic matrix

l

Liquid

s

Solid

v

Vapour

w

Aqueous buffer

60

References 2.1.9 References Absolom DR, Lamberti FV, Policova Z, Zingg W, Van Oss CJ, Neumann AW. 1983. Surface thermodynamics of bacterial adhesion. Appl Environ Microbiol 46(1):90-7. Anspach FB, Curbelo D, Hartmann R, Garke G, Deckwer WD. 1999. Expanded-bed chromatography in primary protein purification. J Chromatogr A 865(1-2):129-44. Bierau H, Hinton RJ, Lyddiatt A. 2001. Direct process integration of cell disruption and fluidised bed adsorption in the recovery of labile microbial enzymes. Bioseparation 10(1-3):73-85. Bos R, Van der Mei HC, Busscher HJ. 1999. Physico-chemistry of initial microbial adhesive interactions--its mechanisms and methods for study. FEMS Microbiol Rev 23(2):179-230. Brixius PJ. 2003. On the influence of feedstock properties and composition on process development of expanded bed adsorption. PhD thesis. Camesano TA, Logan BE. 2000. Probing electrostatic interactions using atomic force microscopy. Environ Sci Technol 34(16):3354-3362. Dainiak MB, Galaev IY, Mattiasson B. 2002. Polyelectrolyte-coated ion exchangers for cell-resistant expanded bed adsorption. Biotechnol Prog 18(4):815-20. Fernandez-Lahore HM, Geilenkirchen S, Boldt K, Nagel A, Kula MR, Thommes J. 2000. The influence of cell adsorbent interactions on protein adsorption in expanded beds. J Chromatogr A 873(2):195-208. Fernandez-Lahore HM, Kleef R, Kula M, Thommes J. 1999. The influence of complex biological feedstock on the fluidization and bed stability in expanded bed adsorption. Biotechnol Bioeng 64(4):484-96. Fernandez-Lahore HM, Lin DQ, Hubbuch JJ, Kula MR, Thommes J. 2001. The Use of IonSelective Electrodes for Evaluating Residence Time Distributions in Expanded Bed Adsorption Systems. Biotechnol. Prog. 17(6):1128-1136. Feuser J, Walter J, Kula MR, Thommes J. 1999. Cell/adsorbent interactions in expanded bed adsorption of proteins. Bioseparation 8(1-5):99-109. Gallardo-Moreno AM, Gonzalez-Martin ML, Perez-Giraldo C, Garduno E, Bruque JM, Gomez-Garcia AC. 2002. Thermodynamic analysis of growth temperature dependence in the adhesion of Candida parapsilosis to polystyrene. Appl Environ Microbiol 68(5):2610-3.

61

References Ganeva V, Galutzov B, Teissie J. 2004. Flow process for electroextraction of intracellular enzymes from the fission yeast, Schizosaccharomyces pombe. Biotechnol Lett 26(11):933-7. GEHealthCare. 2001. Cost comparison: expanded bed adsorption (EBA) vs conventional recovery in the industrial scale processing of proteins. Application note STREAMLINE expanded bed adsorption 18-1150-21 AA:1. Henriques M, Gasparetto K, Azeredo J, Oliveira R. 2002. Experimental methodology to quantify Candida albicans cell surface hydrophobicity. Biotechnol Lett 24:1111– 1115. Hubbuch J, Thommes J, Kula MR. 2005. Biochemical engineering aspects of expanded bed adsorption. Adv Biochem Eng Biotechnol 92:101-23. Kang S, Choi H. 2005. Effect of surface hydrophobicity on the adhesion of S. cerevisiae onto modified surfaces by poly(styrene-ran-sulfonic acid) random copolymers. Colloids Surf B Biointerfaces 10(46(2)):70-7. Li B, Logan BE. 2004. Bacterial adhesion to glass and metal-oxide surfaces. Colloids Surf B Biointerfaces 36(2):81-90. Lin DQ, Brixius PJ, Hubbuch JJ, Thommes J, Kula MR. 2003. Biomass/adsorbent electrostatic interactions in expanded bed adsorption: a zeta potential study. Biotechnol Bioeng 83(2):149-57. Lin DQ, Dong JN, Yao SJ. 2007. Target Control of Cell Disruption To Minimize the Biomass Electrostatic Adhesion during Anion-Exchange Expanded Bed Adsorption. Biotechnol Prog 23(1):162-7. Lin

DQ,

Fernandez-Lahore

HM,

Kula

MR,

Thommes

J.

2001.

Minimising

biomass/adsorbent interactions in expanded bed adsorption processes: a methodological design approach. Bioseparation 10(1-3):7-19. Lin DQ, Zhong LN, Yao SJ. 2006. Zeta potential as a diagnostic tool to evaluate the biomass electrostatic adhesion during ion-exchange expanded bed application. Biotechnol Bioeng 95(1):185-91. Mattiasson B, Galaev I, Garg N. 1996. Polymer-shielded dye-affinity chromatography. J Mol Recognit 9(5-6):509-14. Mills AL, Herman JS, Hornberger GM, Dejesus TH. 1994. Effect of Solution Ionic Strength and Iron Coatings on Mineral Grains on the Sorption of Bacterial Cells to Quartz Sand. Appl Environ Microbiol 60(9):3300-3306.

62

References Northelfer F, Walter JK. 2002. A comparsion of STREAMLINE expanded bed adsorption with the combined techniques of filtration and conventional fixed bed chromatography for the capture of an Fc-fusion protein from CHO cell culture. Application note streamline expanded bed adsorption 18(1144-87 AB):1. Ottewill RH, Shaw JN. 1972. Electrophoretic studies on polystyrene lattices. J Electroanal Chem 37:133-142. Sambrook J, Russell DW. 2006. The condensed protocols from Molecular cloning: a laboratory manual. Cold Spring Harbor, N.Y: Cold Spring Harbor Laboratory Press. v, 800 p. p. Sharma PK, Rao KH. 2002. Analysis of different approaches for evaluation of surface energy of microbial cells by contact angle goniometry. Adv Colloid Interface Sci 98(3):341-463. Strevett KA, Chen G. 2003. Microbial surface thermodynamics and applications. Res Microbiol 154(5):329-35. Theodossiou I, Sondergaard M, Thomas ORT. 2001. Design of expanded bed supports for the recovery of plasmid DNA by anion exchange adsorption. Bioseparation 10(13):31-44. Thommes J, Weiher M, Karau A, Kula M-R. 1995. Hydrodynamics and Performance in Fluidized Bed Adsorption. Biotechnol Bioeng 48(4):367-374. Van der Mei HC, Bos R, Busscher HJ. 1998. A reference guide to microbial cell surface hydrophobicity based on contact angles. Colloids Surf B Biointerfaces 11(4):213221. Van Oss CJ. 1994. Interfacial forces in aqueous media. New York: M. Dekker. viii,440p. p. Van Oss CJ. 1995. Hydrophobicity of biosurfaces - origin, quantitative determination and interaction energies. Colloids Surf B Biointerfaces 5:91-110. Van Oss CJ. 1997. Hydrophobicity and hydrophilicity of biosurfactants. Curr Opin Colloid Interface Sci 2:503-512. Van Oss CJ, Chaudhury MK, Good RJ. 1987. Monopolar surfaces. Adv Colloid Interface Sci 28(1):35-64. Van Oss CJ, Good RJ. 1988. Orientation of the water molecules of hydration of human serum albumin. J Protein Chem 7(2):179-83.

63

References Vergnault H, Mercier-Bonin M, Willemot RM. 2004. Physicochemical parameters involved in the interaction of Saccharomyces cerevisiae cells with ion-exchange adsorbents in expanded bed chromatography. Biotechnol Prog 20(5):1534-42. Vergnault H, Willemot R-M, Mercier-Bonin M. 2007. Non-electrostatic interactions between cultured Saccharomyces cerevisiae yeast cells and adsorbent beads in expanded bed adsorption: Influence of cell wall properties. Process Biochem 42(2):244-251. Volpe C, Siboni S. 1997. Some Reflections on Acid-Base Solid Surface Free Energy Theories. J Colloid Interface Sci. 195(1):121-36.

64

Results

2.2 Colloid deposition experiments as a diagnostic tool for biomass attachment onto bioproduct adsorbent surfaces 1

2

2

Canan Tari †, Rami Reddy Vennapusa †, Rosa B. Cabrera , and Marcelo Fernández2

Lahore . 1

Department of Food Engineering, Izmir Institute of Technology, Urla, Izmir 35430, Turkey. 2 Downstream Processing Laboratory, Jacobs University, Campus Ring 1, D-28759, Bremen, Germany. †These authors contributed equally to the work. 2.2.1 Abstract BACKGROUND: Detrimental processing conditions can be expected in any downstream operation where direct contacting between a crude feedstock and a reactive solid phase is supposed to occur. In this paper we have investigated the factors influencing intact yeast cells deposition onto anion- and cation- exchangers currently utilised for expanded bed adsorption of biotechnological products. The aim of this study was two-fold: a) To confirm previous findings relating biomass deposition with surface energetics according to the XDLVO theory; and b) To provide a simple experimental tool to evaluate biomass deposition onto process surfaces. RESULTS: Biomass deposition experiments were performed on automated workstation utilizing a packed-bed format. Two commercial ion-exchangers intended for the direct capture of bioproducts in the presence of suspended biological particles were employed. Intact yeast cells in the late exponential phase of growth were selected as model biocolloids. Cell deposition was systematically evaluated as a function of fluid phase conductivity and quantitatively expressed as a biomass deposition parameter (α). CONCLUSION: α ≤ 0.15 was established as criteria to reflect negligible biomass adhesion to the process support(s). Biomass deposition experiments further confirmed predictions made on the basis of free interfacial energy calculations as per the extended DLVO approach.

65

Results 2.2.2 Introduction Detrimental processing conditions can be expected in any downstream operation where direct contacting between a crude feedstock and a reactive solid phase is supposed to occur. This type of unit operations has the potential to combine solids removal, product concentration, and partial purification in a single processing step. However, it is already known that suspended biological particles will interact with adsorbent materials. In the particular case of Expanded Bed Adsorption (EBA) interaction phenomena may lead to the development of poor system hydrodynamics and therefore, impaired sorption performance under real process conditions (Hubbuch et al. 2005). Biomass deposition would also result in increased buffer consumption (Northelfer and Walter 2002). The principles of colloid chemistry can be applied to explain biomass-adsorbent attachment at the local (particle) level (Van Oss 1994). Biomass adhesion to process supports has the potential to be strongly influenced by long-range electrodynamic Lifshitz – van der Waals (LW) and electrostatic (EL) and short-range acid-base (AB) interfacial interactions. EL interactions arise from the existence of overlapping double layers of counter-ions near charged surfaces in aqueous media and are accessible by determination of the zeta potential. LW and AB forces are experimentally accessible via contact angle measurement with three diagnostic liquids. Earlier studies on biomass-adsorbent interactions pointed out that interactions between (positively charged) anion exchangers and (negatively charged) biological particles resulted the most problematic system to deal with. Due to the obvious electrostatic nature of such interaction, a single property of these interacting bodies i.e. the zeta potential has been recently proposed for a better understanding and prediction of biomass-adsorbent interactions (Lin et al. 2003; Lin et al. 2006). It is now understood that Coulomb-type interaction are predominant when the basic nature of the process material and the characteristics of the microbial species / strains is kept similar. Moreover, charge effects are only predominant in deposition systems where strongly charged materials are under consideration. Therefore, a single measure like the particle zeta-potential cannot be considered a universal approach to process / material design. Some studies have found a better correlation between surface energy, calculated by the three liquid contact angle method, and microbial adhesion on different solid supports at constant solution chemistry (Li and Logan 2004). 66

Results The mechanistic understanding of the transport and deposition of microbial cell onto natural and process surfaces has significant interest in various environmental and bioprocess situations. A better description of the factors controlling the transport of biological particles is important for the appropriate design of direct contact downstream operations, as well as, for the development of novel adsorbent materials. Traditionally, microbial deposition has been studied employing packed-beds. A population of biological particles is introduced into such systems and the suspended cell or cell-debris effluent is monitored as a function of process time. This type of experiments can provide useful and quantitative information when assessing factors like cell size and shape, microorganisms strain, growth phase, bead size, surface coatings, fluid velocity, and ionic strength on cell deposition onto process media (Tufenkji 2007). Mathematical models of microbial transport in porous media most commonly utilises the advection-dispersion equation as derived by mass balance principles (Brown and Jaffe 2001; Rijnaarts et al. 1996). A common approach to evaluate biomass deposition in laboratory packed-bed experiments employs the “clean-bed” filtration model or colloid filtration theory (CFT). This model is valid for steady-state systems which are initially free of biomass particles and where axial dispersion can be neglected (Pe ≥ 20) (Unice and Logan 2000). Within the CFT, mass transport phenomena are accounted by the “singlecollector contact efficiency” (η0) while the physicochemical phenomena related to biomass attachment are reflected by the “attachment efficiency parameter” (α) (Redman et al. 2004). At larger biomass loads, α values are controlled not only by cell-support interactions but also by the amount of previously attached biomass particles. This implies that attached biomass particles onto the process surface can effectively reduce deposition by a so called collector “blocking” effect (Rijnaarts et al. 1996). On the other hand, increased biomass attachment can result from cell-to-cell aggregation a phenomena known as system “ripening” (Nascimento et al. 2006). In this paper we have investigated the factors influencing intact yeast cells deposition onto anion- and cation- exchangers currently utilised for expanded bed adsorption of biotechnological products. These two systems represent examples of “interacting” vs. “noninteracting” situations, which are relevant in industrial practice. The aim of this study was two-fold: a) To confirm previous findings relating biomass deposition with surface

67

Results energetics according to the XDLVO theory, and b) To provide a simple experimental tool to evaluate biomass deposition onto process surfaces.

68

Results 2.2.3 Material and Methods 2.2.3.1 Materials Chromatographic matrices and columns were purchased from GE Healthcare (Munich, Germany). Water was Milli-Q quality. All other chemicals were of analytical grade. 2.2.3.2 Yeast cells Saccharomyces cerevisiae cells were cultivated, harvested at late exponential phase, and washed three times with 10 mM buffer solutions, as previously described (Ganeva et al. 2004). Cells were employed immediately after harvest and washing for deposition experiments. Intact yeast cell diameter was taken as 8 μm. A Hamaker constant value of 0.34 kT was utilized according to previous work based on contact angle determinations (Vennapusa et al. 2006). 2.2.3.3 Zeta-potential determination Zeta potential was measured with a Zetasizer Nano ZS from Malvern Instruments (Worcestershire, United Kingdom). Particles were contacted with 20 mM sodium phosphate buffer at pH 7.6 until equilibrium was reached and further diluted to appropriate particle count (~200 cells total count) before measuring the zeta potential. Zeta potentials were calculated from the electrophoretic mobility data as per the Smoluchowski’s equation (Ottewill and Shaw 1972). All the measurements were done in triplicate. 2.2.3.4 Biomass deposition experiments Biomass deposition experiments were performed in TricornTM glass chromatographic columns (5 mm internal diameter, 50 mm length) packed with Streamline™ adsorbents (GE Healthcare, Munich, Germany). These macroporous adsorbents are made of cross-linked agarose (6%) containing a crystalline quartz core. Average bead (collector) diameter was taken as 200 μm (Spherical particles with size distribution 100-300 μm). Matrix was uniformly packed as judged by residence time distribution studies, performed with 5% acetone as a tracer. Bed porosity was estimated as 0.4. Highly porous frits were utilised in order to allow for non-restricted passage of yeast cell through the system. Biological particle (soft colloid) deposition dynamics was studied by injecting a 4 ml biomass pulse (∼ 10 pore volumes). Cell concentration was adjusted to ≈ 6.4 X 10 7 cell· cm-3 by diluting the

69

Results cell suspension in order to reach an absorbance value at 600 nm (1 cm path length) ≈ 0.8AU. Cell number was determined employing a Coulter Counter (Multisizer™ 3, Beckman Coulter, CA, USA). Deposition experiments were run in an automated ÄKTA Explorer 100 system (GE Healthcare, Munich, Germany). A mobile phase composed of phosphate-based buffers (pH 7.6) at different conductivities (0.66, 2.0, 8.4, 14.0, and 38.6 mS/cm) was pumped at 76.4 cm·h-1. Particle breakthrough curves were obtained by monitoring the effluent suspension @ 600 nm. Regeneration of the material was performed by extensive treatment with 1 M sodium hydroxide followed by exhaustive rinsing with distilled water. 2.2.3.5 Parameter calculation Cell (colloid) deposition onto collector (adsorbent) beads can be described by model parameters. To quantitatively compare biomass breakthrough curves, the deposition rate coefficient (kd) was calculated for each packed-bed experimental run according to the following expression (Tufenkji et al. 2004):

kd = −

U ⎛C ln⎜ ε L ⎜⎝ C 0

⎞ ⎟⎟ ⎠

Equation 1

where U is superficial velocity, ε is the bed porosity, and L is the column length. The value for C/C0 corresponding to the initial “clean bed” condition i.e. C/C0 at 2 pore volumes was utilised for calculations. The deposition rate coefficient is directly related to the single-collector contact efficiency (η0) and the empirical attachment efficiency (α) according to the following expression: kd =

3 (1 − ε ) U α η0 2 dc ε

Equation 2

where dc is the diameter of a spherical collector, and η0 is the single-collector contact efficiency. η0 can be calculated from published correlations (Tufenkji and Elimelech 2004). The attachment efficiency (α) represents the fraction of collisions between biomass particles suspended in the fluid phase which results in attachment:

α =−

⎛C ⎞ dc 2 ln⎜⎜ ⎟⎟ 3 (1 − ε ) Lη 0 ⎝ C 0 ⎠

Equation 3

70

Results

Utilising experimental data from breakthrough of cells from packed beds the attachment efficiency parameter (α) can be calculated as α = kd / kd,fav (Redman et al. 2004).

71

Results 2.2.4 Results and Discussion 2.2.4.1 Physicochemical properties of the cell particles / beaded adsorbents The electrokinetic potential of the interacting particles e.g. the yeast cells vs. the beaded adsorbents were studied as a function of fluid phase conductivity. Figure 1 depicts zeta potential values for intact yeast cells, cation-exchange beads, and anion-exchange beads in phosphate buffers of varying conductivities. The zeta-potential has been reported as a main parameter affecting yeast cell deposition onto beaded adsorbents, particularly onto anionexchangers (Lin et al. 2006). Intact yeast cells, harvested at the late exponential phase of growth, showed zeta-potential values ≈ -25 mV at very low buffer concentration (0.7 mS.cm-1). At standard ion-exchange mobile phase composition e.g. ∼20 mM phosphate pH 7.6, zeta-potential values were -18 mV. Lower zeta-potential values were observed with increasing conductivity i.e. -6 mV at 34 mS⋅cm-1. A similar trend was observed when studying the effect of mobile phase conductivity on the electrokinetic behaviour of the cation-exchanger beads. SP Sepharose fragments were utilised for such studies in order to avoid errors derived from the settling of the intact adsorbent particles. Lower zeta-potential values obtained were -36 mV (0.7 mS⋅cm-1) while maximum values were -14 mV (34 mS⋅cm-1). A second factor recognized to influence biomass deposition onto process surfaces is cell or cell-debris size and shape (Hubbuch et al. 2006). In this study, both factors are kept constant since only intact yeast cells (8 μm diameter) of spherical shape were utilised as model biomass. Besides electrostatic forces (EL), electrodynamic Lifshitz – Van der Waals forces (LW) are known to mediate biomass interactions. The LW interaction, which is predominantly attractive in microbial systems, is not influenced by the ionic strength (Bos et al. 1999) but both the range and magnitude of the EL interactions decrease with increasing ionic strength due to shielding of surface charges. LW forces between intact yeast cells and agarose-based material can be described by a Hamaker constant (A). The value for A, in this particular system, was previously calculated as 0.34 kT from contact angle measurements; details will be published elsewhere. Obtained Hamaker constant value are in agreement with assumed values for various microbial systems (Bos et al. 1999).

72

Results

The so-called acid-base (AB) forces are also included in the XDLVO approach (Van Oss 1994). AB interactions were found to be a function of the nature of the process solid phase onto which cell adhesion took place. For agarose-based supports and yeast cells an average ΔGAB value was calculated as +30 mJ⋅m-2, indicating the repulsive nature of the AB component (Vennapusa et al. 2006). This value is valid at closest distance of approximation between the interacting bodies (1.57 Å).

Figure 1: Zeta potentials of intact yeast cells and adsorbent beads as a function of fluid phase conductivity at pH 7.6. ▲, intact yeast cells; ■, SP beads; ●, DEAE beads.

2.2.4.2 Biomass deposition experiments Deposition experiments were performed in an automated chromatographic system for increase throughput and convenience of use. Figure 2 depicts the schematic illustration of the chromatographic set up. In packed-bed systems, physical straining of bio-colloids is considered to be significant on the basis of geometrical consideration when dp/dc > 0.05 (Rijnaarts et al. 1996). In the system under study in this work dp/dc ≈ 0.04 and thus physical straining can be neglected. Although straining has been observed when dp/dc values were as low as 0.002 (Tufenkji et al. 2004) experiments performed with the cation-exchange material supported the previous assumption . No physical entrapment of the bio-colloids

73

Results was observed within the packed-bed system since almost quantitative recovery of cells was verified under (chemical) non-deposition conditions (Figure 3). Hydrodynamic forces were kept constant within the laminar regime (Re < 10) by maintaining a constant flow rate of 76.4 cm·h-1. Therefore, biomass adhesion was evaluated under carefully controlled experimental conditions. Figure 3 illustrates typical run cycles for both “interacting” and “non-interacting” deposition systems. Each cycle is composed of an equilibration phase (20 PV), a sample (suspended biomass) pulse (∼10 PV), a washing step with running buffer (15 PV), and (partial) regeneration with sodium hydroxide solution (20 PV). The breakthrough of biomass particles suspended in the effluent buffer can be observed in case b) which corresponds to the partial deposition of cells onto the packed chromatographic beads. On the contrary, a strong deposition of cell is characterised by the experimental profile observed in case a) where no suspended material is leaving the packed-bed system.

WASTE Detector COLUMN

PC

COLUMN-BYPASS COLLECTOR

SOLUTION

VALVE

PUMP INJECTION LOOP

Figure 2: Schematic illustration of chromatographic system set-up.

74

Results

(a)

(b)

Figure 3: Typical experimental data as obtained from packed-bed experiments utilizing chromatographic beads as colloid collectors. Intact yeast cells were employed as a model. (a) Favorable deposition onto DEAE functionalized beads, (b) Unfavorable deposition onto SP functionalized support. The arrows indicate (A) Cell pulse injection and (B) Column regeneration with 0.5 mol L-1 NaOH.

75

Results Employing the above described methodology, systematic studies were performed to evaluate yeast cell deposition onto anion- and cation-exchangers. Figure 4a shows the family of deposition curves obtained by variation of fluid phase conductivity when DEAEStreamline™ beads were utilised as collectors. These adsorbent beads are weak anion exchangers and thus they are positively charged. Results are presented as normalised concentration (C/C0) vs. pore volumes. In order to calculate C/C0 absorbance data was employed since cell concentration was linearly related to such measurement within the concentration range involved in this study. The length of the biomass pulse, equivalent to ∼ 10 PV, was sufficient to produce a semi-complete breakthrough of suspended biological particles i.e. the effluent cell concentration never reached C0. It can be observed that total cell deposition took place at very-low to low conductivity values e.g. almost no cells were detected when conductivity was ≤ 2.0 mS⋅cm-1. An increased conductivity of the mobile phase has allowed for a progressive increased in the number of cells leaving the system. These results can be explained considering a predominant role of EL forces in a system characterised by collectors and colloids harbouring opposite charges. Since other forces were kept constant, as well as the colloid size, the only mechanism expected to govern deposition is related to Coulomb-type effects. This is in agreement with previous studies focusing on zeta-potential as a diagnostic parameter for biomass / support interactions (Lin et al. 2006). Similar experiments were performed utilising SP- Streamline™ (negatively charged) beads. The biomass breakthrough curves are presented in Figure 4b. As it can be observed from this Figure, lower conductivity values in the fluid phase have resulted in negligible deposition of cells onto the cation-exchange collectors. However, increased conductivity (≥ 14 mS⋅cm-1) has promoted biomass deposition onto the cation-exchanger. This finding might have an impact on bioprocess design since this material has been considered as “noninteracting” with particulate feedstock (Feuser et al. 1999). However, deposition of intact yeast cells onto SP- Streamline™ beads can be inferred from XDLVO calculations as shown below. Practical consequences related to this behaviour during EBA capture of bioproducts could arise during product elution i.e. since high conductivity buffers are commonly employed, aggregative fluidization may develop resulting in a diluted product fraction.

76

Results

Figure 4: Normalized cell-effluent concentrations plotted against pore volumes pumped through the packed bed at different fluid phase conductivity values (pH 7.6). (a) DEAE yeast cells. (b) SP / yeast cells. × 0.66 mS cm-1; z 2.00 mScm-1; z 8.4 mScm-1 (5.5 mS cm-1 in fig. b); z 14.00 mS cm-1; z 36 mS cm-1.

77

Results 2.2.4.3 Parameter calculation Bio-colloid depositions experiments, as shown in Figure 4, depict cell breakthrough behaviour compatible with collector blocking or cell release from the packed-bed system. This is demonstrated by the fact that in most experiments the normalised cell concentration (C/C0) did not raise a steady-state value after the initial dispersive curve region. Therefore, the initial clean bed C/C0 for each experiment was taken for parameter calculations. This allowed the application of the colloid filtration theory which is valid under clean bed conditions (Redman et al. 2004). Table 1 present calculated values for both kd (s-1) and α (-) as a function of fluid phase conductivity. kd,fav data corresponds to the experimental run performed with the DEAEStreamline™ beads as collectors under lowest conductivity (0.6 mS⋅cm-1). Table 1: Calculated parameters from packed-bed experiments where chromatographic supports were employed as cell collectors. Calculation were performed according to (Redman et al. 2004).

DEAE Streamline TM - Intact yeast

Conductivity ( mS cm-1 ) 0.66

C/Co

kd

α

0.003

0.246

1.00

2.0 8.4 14.0 38.6

0.006 0.075 0.244 0.254

0.211 0.107 0.058 0.056

0.858 0.435 0.236 0.229

C/Co

kd

α

0.654 0.568 0.519 0.445 0.333

0.017 0.023 0.027 0.033 0.045

0.071 0.095 0.110 0.136 0.184

SP Streamline TM – Intact yeast

Conductivity ( mS cm-1 ) 0.66 2.0

5.5 14.0 38.6

78

Results Evidence is suggesting that simple models for calculating α based on deposition in the secondary energy minimum can result in accurate prediction of biomass attachment to porous media (Tufenkji 2007). For the DEAE-Streamline™ / yeast system, α values decreased from ∼ 1 at very low conductivity to 0.23 at ∼39 mS⋅cm-1. On the other hand, for the SP-Streamline™ / yeast system, α values increased from 0.07 (∼ 0.6 mS⋅cm-1) to 0.18 (∼ 39 mS⋅cm-1). As a reference α = 1, has the meaning of complete cell deposition onto the packed collectors. Figure 5 summarizes these results in a graphic form. Minimum α values were obtained for the anion-exchanger system, which are nearly equivalent to α values obtained for SP-Streamline™ beads at 39 mS⋅cm-1. These results indicate that biomass deposition experiments are an appropriate design tool to evaluate biomass deposition onto process surfaces. On the basis of the preceding experimental evidence, α is proposed as a diagnostic parameter that provides information on biomass attachment onto process surfaces. Applying the proposed methodology, changes in α can be effectively utilised to monitor biomass-support interactions even in such cases where such interaction was overlooked in the past (Feuser et al. 1999).

Figure 5: Changes in the attachment efficiency parameter (α) as a function of fluid phase conductivity. Deposition of intact yeast cell was studied for (z) DEAE and (z) SP chromatographic materials.

79

Results 2.2.4.4 Bio-colloid deposition in the secondary minimum Figure 6 depicts total interfacial energy profiles as a function of the distance between two interacting bodies (U vs. H) in aqueous media i.e. diluted buffer solutions. Calculations were performed considering the role of LW, AB, and EL forces, as previously reported (Bos et al. 1999) by utilising data from contact angle measurement and zeta potential determinations (Vennapusa et al. 2006). Therefore, the presented profiles are in accordance to the extended DLVO approach. Since the radius of the chromatographic beads is much higher than the radius of the intact yeast cell, total free energy was computed assuming a plane-to-sphere geometry. Figure 6a represents XDLVO energy profiles for the system where a DEAE-Streamline™ bead interacts with an intact yeast cell. It can be observed that a secondary energy minimum exists at a distance of ≈ 5nm, where deposition of the cell particle can occur. This is essentially a reversible interaction that can be overcome by sufficient energy input for example, in the form of shear or hydrodynamic stress. The magnitude (depth) of the energy pocket, however, increases upon modification (reduction) of the liquid phase conductivity. As a consequence, stronger deposition of cell particles is expected when working with diluted buffers than when working with buffers / salt solutions with higher conductivities. This situation is reflected by the biomass deposition experiments as presented in Figure 4a. Therefore, this kind of experiments can confirm the trends predicted by XDLVO calculations. Both U vs. H calculations and deposition experiments are in full agreement with the known biomass-interaction behaviour for DEAE-Streamline™ (Fernandez-Lahore et al. 2000; Fernandez-Lahore et al. 1999; Lin et al. 2001). Moreover, biomass deposition experiments can offer a simple way to access interfacial phenomena in aqueous media. These phenomena have relevance from the bioprocess point of view and have important consequences for appropriate process optimisation and material design.

80

Results

Figure 6: Calculated XDLVO total interaction energy as a function of separation distance. a) DEAE / yeast cells b) SP / yeast cells. A family of curves representing variations in the conductivity value of the fluid phase is shown (—) 2 mS cm-1; (—) 4 mS cm-1; (—) 9.55 mS cm-1; (—)15.1 mS cm-1; (—) 34 mS cm-1).

81

Results Figure 6b represents energy profiles for the system where a SP-Streamline™ bead interacts with an intact yeast cell. In this case, secondary energy pockets can also be observed. Minimum free energy, according to XDLVO calculations, occurs at distances between 30 nm and 7 nm. However, when compared to the minimum energy values observed for the anion-exchange / yeast system, the SP-Streamline™ material promotes the formation of much less deep pockets e.g. -5 to -30 kT (SP) as compared to -30kT to -1000 kT (DEAE). It can also be realised that an opposite interaction behaviour takes place for this system i.e. the depth of the energy pocket increases with the increase in fluid phase conductivity. This behaviour is again reflected by the biomass deposition experiments, confirming that these experiments are sensitive and able to reveal not previously recognised underlying phenomena. 2.2.4.5 Correlation with interaction energy profiles The correlation between the biomass deposition parameter (α) and the value of the free interfacial energy minimum (energy pocket) is shown in Figure 7. This picture summarises data gathered from ion-exchanger chromatographic matrices (DEAE-Streamline™ and SPStreamline™) interacting with intact yeast cells, under the range of fluid phase conductivity values reported in this work (0.66 – 38.6 mS⋅cm-1). In a previous study, (Vennapusa et al. 2006) we have demonstrated that energy pockets –as calculated by the XDLVO theoryshowing energy minima ≥ -20 kT are not deleterious during product capture for example, by expanded bed adsorption. From Figure 7, it can be proposed α ≤ 0.15 as a cut-off value for negligible biomass deposition. Therefore, as long as α is kept low enough efficient fluidisation and sorption performance can be anticipated.

82

Results

Figure 7: Correlation graph between the attachment efficiency parameter (α) and the total interaction energy according to the XDLVO approach.

Besides biomass deposition, as expressed by the deposition parameter (α), the cell breakthrough profiles observed in Figure 4 might indicated that a blocking phenomenon is also occurring in the systems under study. Blocking refers to the fact that cells which are already attached to the solid support can interfere with the attachment of further cells being contacted with the adsorbent beads. This phenomenon, usually accounted by the so called excluded area parameter (β) has important environmental implications (Bolster et al. 2001). During direct capture of bioproducts, due to the presence of much higher concentration of biomass (∼ 8 % wet weight) in contact with the beaded adsorbents, blocking can be supposed to play a less significant role. Therefore, modeling approaches have been kept simple enough so as to provide a single parameter (α), which reflects biomass attachment as an important event having practical consequences for the performance of a direct capturing unit operation. Biomass multilayer formation would also occur at long contact time and / or high biomass concentration, which is not the case for the biomass deposition experiment as described

83

Results here. This phenomenon could be more important in cases where cell-to-cell aggregation or “ripening” is favored, like during hydrophobic interaction chromatography. Understanding of cell-to-cell aggregation would require independent experimental methods to be evaluated (Ramachandran and Fogler 1998). 2.2.5 Conclusions Biomass deposition experiments were performed in an automated workstation utilizing a packed-bed format and intact yeast cells in the late exponential phase of growth as biomass model. Under carefully controlled experimental conditions, removal of biomass particles onto chromatographic collectors was assumed to be dependent on a) mass transfer i.e. the transfer of a biological particle from the bulk liquid phase to the adsorbent bead and b) the capture of a biological particle onto an adsorbent bead by interfacial forces. Straining was neglected since the size of the biomass-derived particles is much smaller than the process beads. Detachment was also supposed to be insignificant under laminar flow conditions. The description of “aggregation” (cell-to-cell), which implies multilayer adhesion, was omitted since this phenomenon takes place at high biomass concentrations and long contact times. Cell deposition was systematically evaluated as a function of fluid phase conductivity and quantitatively expressed as a biomass deposition parameter (α). Deposition onto commercial anion-exchanger beads was observed to increase with decreasing conductivity values in the mobile phase. The opposite behavior was observed when cation-exchange beads were utilized as collectors in the packed-bed system. In both cases, experimental deposition studies confirmed predictions based on the free energy of interaction according to the XDLVO theory. Coulomb-type interactions were dominating since EL forces are affected by the ionic strength of the aqueous media surrounding the interaction bodies. Other forces, which are relevant to the evaluation of biomass deposition, were kept constant. The evaluation of LW and AB forces is mandatory when comparing microbial strains and / or process materials apart from the model system employed in this work.

84

Results The biomass deposition parameter ranges from 0 to 1 and defines the probability that a bioparticle will adhere to a surface upon collision. α can be viewed as a single value when a fixed transport distance is assumed. A value for α ≤ 0.15 was established as criteria to reflect insignificant biomass adhesion to the process support(s). It should be pointed out that this α cut-off value is valid under the experimental conditions described in this work which are easily matched by utilization of commercial chromatographic systems and columns. The bio-colloid deposition experiment can be proposed as simple diagnostic tool for the evaluation of biomass interference during direct capture of bioproducts. Moreover, the biomass deposition method results in a novel approach for testing biomass / support compatibility, which is easy to implement in a standard chromatographic workstation. 2.2.6 Acknowledgements This work was partially funded by the BID 1201/OC AR 649 PICT 08352 and the start-up grant from Jacobs University [IUB / 2130-90050]. Dr. Canan Tari was supported by the International Centre for Transdisciplinary Science (ICTS) at Jacobs University. 2.2.7 Nomenclature A

Hamaker constant [J] / [kT]

AB

Acid Base

AU

Absorbance Units

CFT

Colloid Filtration Theory

C

Final concentration

Co

Initial concentration

dc

Diameter of collector [m]

DEAE

Diethylaminoethyl-

DLVO

Derjaguin, Landau, Verwey and Overbeek Theory

85

Results dp

Diameter of yeast cells [m]

EBA

Expanded bed adsorption

EL

Electrostatic interaction

H

Separation distance [m]

kd

Deposition rate coefficient [s-1]

kd fav

Deposition rate coefficient for favorable deposition [s-1]

L

Length of column [m]

PV

Pore volume [ml]

Pe

Peclet number [-]

Re

Reynolds number [-]

SP

Sulphopropyl-

U

Superficial fluid velocity [ms-1]

U

Total interaction energy [kT]

XDLVO

Extended DLVO

Greek letters

ε

Porosity [-]

α

Attachment efficiency parameter [-]

ηο

Single collector contact efficiency [-]

86

Results 2.2.8 Referances Bolster CH, Mills AL, Hornberger GM, Herman JS. 2001. Effect of surface coatings, grain size, and ionic strength on the maximum attainable coverage of bacteria on sand surfaces. J Contam Hydrol 50(3-4):287-305. Bos R, Van der Mei HC, Busscher HJ. 1999. Physico-chemistry of initial microbial adhesive interactions--its mechanisms and methods for study. FEMS Microbiol Rev 23(2):179-230. Brown DG, Jaffe PR. 2001. Effects of Nonionic Surfactants on Bacterial Transport through Porous Media. Environ Sci Technol 35(19):3877-3883. Fernandez-Lahore HM, Geilenkirchen S, Boldt K, Nagel A, Kula MR, Thommes J. 2000. The influence of cell adsorbent interactions on protein adsorption in expanded beds. J Chromatogr A 873(2):195-208. Fernandez-Lahore HM, Kleef R, Kula M, Thommes J. 1999. The influence of complex biological feedstock on the fluidization and bed stability in expanded bed adsorption. Biotechnol Bioeng 64(4):484-96. Feuser J, Walter J, Kula MR, Thommes J. 1999. Cell/adsorbent interactions in expanded bed adsorption of proteins. Bioseparation 8(1-5):99-109. Ganeva V, Galutzov B, Teissie J. 2004. Flow process for electroextraction of intracellular enzymes from the fission yeast, Schizosaccharomyces pombe. Biotechnol Lett 26(11):933-7. Hubbuch JJ, Brixius PJ, Lin DQ, Mollerup I, Kula MR. 2006. The influence of homogenisation conditions on biomass-adsorbent interactions during ion-exchange expanded bed adsorption. Biotechnol Bioeng 94(3):543-53. Hubbuch JJ, Thommes J, Kula MR. 2005. Biochemical engineering aspects of expanded bed adsorption. Adv Biochem Eng Biotechnol 92:101-23. Li B, Logan BE. 2004. Bacterial adhesion to glass and metal-oxide surfaces. Colloids Surf B Biointerfaces 36(2):81-90. Lin DQ, Brixius PJ, Hubbuch JJ, Thommes J, Kula MR. 2003. Biomass/adsorbent electrostatic interactions in expanded bed adsorption: a zeta potential study. Biotechnol Bioeng 83(2):149-57. Lin

DQ,

Fernandez-Lahore

HM,

Kula

MR,

Thommes

J.

2001.

Minimising

biomass/adsorbent interactions in expanded bed adsorption processes: a methodological design approach. Bioseparation 10(1-3):7-19.

87

Results Lin DQ, Zhong LN, Yao SJ. 2006. Zeta potential as a diagnostic tool to evaluate the biomass electrostatic adhesion during ion-exchange expanded bed application. Biotechnol Bioeng 95(1):185-91. Nascimento AG, Totola MR, Souza CS, Borges MT, Borges AC. 2006. Temporal and spatial dynamics of blocking and ripening effects on bacterial transport through a porous system: A possible explanation for CFT deviation. Colloids Surf B Biointerfaces 53(2):241-244. Northelfer F, Walter JK. 2002. A comparison of STREAMLINE expanded bed adsorption with the combined techniques of filtration and conventional fixed bed chromatography for the capture of an Fc-fusion protein from CHO cell culture. Application note STREAMLINE expanded bed adsorption 18 1144-87 AB. Ottewill RH, Shaw JN. 1972. Electrophoretic studies on polystyrene lattices. J Electroanal Chem 37:133-142. Ramachandran V, Fogler HS. 1998. Multilayer Deposition of Stable Colloidal Particles during Flow within Cylindrical Pores. Langmuir 14(16):4435-4444. Redman JA, Walker SL, Elimelech M. 2004. Bacterial Adhesion and Transport in Porous Media: Role of the Secondary Energy Minimum. Environ Sci Technol 38(6):17771785. Rijnaarts HHM, Norde W, Bouwer EJ, Lyklema J, Zehnder AJB. 1996. Bacterial Deposition in Porous Media Related to the Clean Bed Collision Efficiency and to Substratum Blocking by Attached Cells. Environ Sci Technol 30(10):2869-2876. Tufenkji N. 2007. Modeling microbial transport in porous media: Traditional approaches and recent developments. Adv Water Resour 30(6-7):1455-1469. Tufenkji N, Elimelech M. 2004. Correlation Equation for Predicting Single-Collector Efficiency in Physicochemical Filtration in Saturated Porous Media. Environ Sci Technol 38(2):529-536. Tufenkji N, Miller GF, Ryan JN, Harvey RW, Elimelech M. 2004. Transport of Cryptosporidium Oocysts in Porous Media: Role of Straining and Physicochemical Filtration. Environ Sci Technol 38(22):5932-5938. Unice KM, Logan BE. 2000. Insignificant role of hydrodynamic dispersion on bacterial transport. J Environ Engin 126(6):491-500. Van Oss CJ. 1994. Interfacial forces in aqueous media. New York: M. Dekker. viii,440 p.

88

Results Vennapusa RR, Cabrera R, Ganeva V, Fernandez-Lahore HM. 2006. Direct capture from electro-permeabilized yeast cells on expanded beds: a biomass-adsorbent interaction study via surface energetics. Book of abstracts, 6th European Symposium on Biochemical Engineering Science.

89

Results

2.3 Surface energetics to assess biomass attachment onto hydrophobic interaction adsorbents in expanded beds Rami Reddy Vennapusa1, Canan Tari2, Rosa Cabrera1, and Marcelo Fernandez-Lahore1* 1

Downstream Processing Laboratory, Jacobs University gGmbH, Campus Ring 1, D28759, Bremen, Germany. 2Department of Food Engineering, Izmir Institute of Technology, Urla, Izmir 35430, Turkey.

2.3.1 Abstract Cell-to-support interaction and cell-to-cell aggregation phenomena have been studied in a model system composed of intact yeast cells and Phenyl-Streamline adsorbents. Biomass components and beaded adsorbents were characterized by contact angle determinations with three diagnostic liquids and zeta potential measurements. Subsequently, free energy of interaction vs. distance profiles between interacting surfaces was calculated in the aqueous media provided by operating mobile phases. The effect of pH and ammonium sulphate concentration within the normal operating ranges was evaluated. Calculation indicated that moderate interaction between cell particles and adsorbent beads can develop in the presence of salt. Cell-to-cell aggregation was suspected to occur at high salt concentration and neutral pH. Predictions based on the application of the XDLVO approach were confirmed by independent experimental methods like biomass deposition experiments and laser diffraction spectroscopy. Understanding biomass attachment onto hydrophobic supports can help in alleviating process limitations normally encountered during expanded bed adsorption of bioproducts.

90

Results 2.3.2 Introduction Expanded Bed Adsorption (EBA) has been proposed as an integrative downstream processing technology allowing the direct capture of targeted species from an unclarified feedstock e.g. a cell containing fermentation broth. This unit operation has the potential to combine solids removal, product concentration, and partial purification in a single processing step. The application of EBA implies, however, that intact cell particles or cell debris present in the feedstock will interact –in a minor or larger extent- with fluidized adsorbent beads. It is already known that interaction between biomass and the adsorbent phase may lead to the development of poor system hydrodynamics and therefore, impaired sorption performance under real process conditions. Moreover, biomass interaction would result in increased buffer consumption in order to remove and wash away sticky biological particles. Biomass components can also mask binding sites thus reducing their availability to the targeted species. These phenomena i.e. decreased sorption performance and buffer consumption is detrimental to cost-efficient processing utilizing expanded bed adsorption (Fernandez-Lahore et al. 1999; Lin et al. 2001). Previous studies on biomass-adsorbent interactions were restricted to simple diagnostic tests to determine the extent of cell –or cell debris- attachment to the desired chromatographic supports (Feuser et al. 1999). More recently, a single property of the suspended biological particle i.e. the zeta potential has been proposed for a better understanding and prediction of biomass-adsorbent interactions during expanded bed adsorption. Since then a number of studies has been developed to illustrate the usefulness of this approach when adsorption is performed onto anion-exchangers (Lin et al. 2003; Lin et al. 2006). Such systems are obviously dominated by Coulomb-type interactions and therefore, non-electrostatic interactions are anticipated to play a minor role (Vergnault et al. 2007). Experimental evidence gathered by many authors has addressed the importance of nonelectrostatic forces for biomass adhesion to process surfaces in the broader context provided by a group of systems of technical and environmental relevance. For example, hydrophobic interaction as measured by partition tests has been proposed as a generalized assay to measure adhesion-potential of bacteria to low-energy surfaces (Stenstrom 1989). Complementarily, differences in the hydrophobic surface characteristics of bacterial strains 91

Results were revealed by hydrophobic interaction chromatography (Smyth et al. 1978). Recently, acid-base interactions have been employed to understand yeast deposition onto chemically modified substrates (Kang and Choi 2005). However, very little is known on biomass attachment onto chromatographic materials like hydrophobic interaction media (HIC) under real downstream process conditions. The mentioned chromatographic mode represents a widely utilized industrial operation (Mahn and Asenjo 2005), which is amenable for direct sequestration of bioproducts. Since sorption performance limitations were already observed due to biomass interference during HIC-based EBA, a better understanding and control of such phenomena is needed (Smith et al. 2002). A more comprehensive approach to understand biomass deposition onto chromatographic supports has been proposed by utilizing principles of colloid theory to explain biomassadsorbent attachment at the local (particle) level (Vennapusa et al. 2008). This approach is based on extended DLVO calculations performed via experimentally determination of contact angles and z-potential values for the interacting surfaces or particles. The comprehensive method takes into account several types of possible interaction forces i.e. Lifshitz-Van der Waals (LW) and acid-base (AB) and, therefore, it is not limited to those purely electrostatic in nature (EL). Biomass adhesion behavior onto chromatographic beads predicted on the basis of XDLVO calculations was validated by independent biomass deposition experiments (Tari et al. 2008). The aim of this paper was to contribute to a deeper understanding of biomass-adsorbent interactions to further open the pave for optimized EBA processing in industry. Studies targeted biomass adhesion to hydrophobic interaction materials which have not been extensively studied so far. The physicochemical properties of biomass-derived material, taken as colloidal particles, vs. the physicochemical properties of the adsorbent beads, taken as a process surface, were determined indirectly via contact angle and zeta potential measurements. Subsequently, total interfacial interaction energy values were calculated as a function of surface distance in aqueous media e.g. process buffer. Cell-to-support interactions and cell-to-cell aggregation phenomena were independently confirmed by colloid deposition experiments and laser diffraction spectroscopy, respectively.

92

Results 2.3.3 Materials and Methods 2.3.3.1 Materials Chromatographic matrices (Phenyl Sepharose FF, high substitution; Phenyl Streamline) and columns (Tricorn 5/50) were purchased from GE Health Care (Munich, Germany). αbromonaphtalene and formamide were obtained from Fluka (Buchs, Switzerland). Water was Milli-Q quality. All other chemicals were analytical grade. 2.3.3.2 Generation of biomass Yeast cells (Saccharomyces cerevisiae) wild strain was utilized. Five ml of 24 h culture were inoculated in 500 ml of 3.5 % (w/v) YES medium (yeast extract with supplements of yeast extract, 5 g l-1; glucose, 30 g l-1; 225 mg l-1 adenine, histidine, leucine, uracail and lysine hydrochloride) and grown at 30 oC. Cells are harvested at late exponential phase by centrifugation, and washed three times with 10mM phosphate buffer solutions, as previously described (Ganeva et al. 2004). Cells were employed immediately after preparation for further experimental measurements or routines. 2.3.3.3 Contact angle measurements Preparation of intact yeast cells for contact angle measurements was performed as described (Henriques et al. 2002). To evaluate the effect of pH, washed cells were suspended to 10% (w/v) in 20mM phosphate buffer, pH 7or 50mM sodium acetate buffer, pH 4 and to evaluate the effect of salt concentration, biomass was suspended in 20mM phosphate buffer (pH 7) and 50mM sodium acetate buffer (pH 4) containing added ammonium sulphate (0.2, 0.4, 0.8, 1.2, 1.6 and 2.0M). Cells were equilibrated in the appropriate buffer condition and the suspension subsequently poured onto agar plates containing 10% glycerol and 2% agar-agar. The plate was allowed to dry for 24-36 hours at room temperature on a properly leveled surface free from dust. Salt crystallization was avoided. Agar plates without cell spreads were utilized as control. Contact angles were measured as per the sessile drop method (Sharma and Rao 2002) utilizing a commercial goniometric system (OCA 20, Data Physics instruments GmbH, Filderstadt, Germany). The three diagnostic liquids α-bromonaphtalene, formamide, and

93

Results water were employed (Bos et al. 1999). All the measurements were performed in triplicate and at least 20 contact angles per samples were measured. Contact angle determination on buffer-equilibrated chromatographic beads was performed utilizing the same physicochemical conditions and experimental procedures described for cell particles. Previous to pouring onto the agar plates, matrix beads were frozen in liquid nitrogen and crushed mechanically. Crushing efficiency was assessed by microscopic examination and particle size determination so as to assure particle fragment diameters ≤ 10 μm. Phenyl Sepharose (high-sub) was utilized. Square pieces of the agar supported chromatographic bead fragments were utilized for measuring contact angles.

2.3.3.4 Zeta potential determination Zeta-potential measurements were performed with a ZetaSizer Nano-ZS (Malvern instruments, Worcestershire, United Kingdom), as previously described (Vennapusa et al. 2008). Zeta-potential values were gathered employing biomass pretreated as described before (under 2.3.3.3) and utilizing the same buffers utilized for contact angle determination. Zeta-potential values for crushed and equilibrated chromatographic beads were calculated from the electrophoretic mobility data according the Smoluchowski’s equation (Ottewill and Shaw 1972). Data was gathered under identical buffer compositions as shown for biomass related determinations. 2.3.3.5 Particle size determination and cell aggregation behavior Particle size determinations and cell aggregation studies were performed by laser diffraction employing a MasterSizer 2000, hydro 2000 G (Malvern instruments, Worcestershire, United Kingdom), according to manufacturer instructions. Cell aggregation was studied as a function of pH and ammonium sulphate concentration utilizing the buffers systems already described. For each condition, kinetic studies were performed within a time interval of 60 minutes (Voloshin et al. 2005). Measurements were performed utilizing cell suspensions having an optical density ≈ 0.1 for better reproducibility.

94

Results Visual inspection of aggregate formation was performed with a confocal laser scanning microscope, equipped with argon and helium/neon mixed gas laser with excitation wavelengths of 488 or 543 nm (LSM 510, Carl Zeiss, Oberkochen, Germany). Washed yeast cells in 20 mM phosphate buffer (pH 7) or buffered 1.6 M ammonium sulphate solution were mounted on glass slides and observed. Scans at a resolution of 1024 x 1024 pixels were taken in the line-averaging mode. Micrographs were stored in LSM format (Carl Zeiss LSM Image Browser). 2.3.3.6 Bio-colloid deposition experiments Biomass deposition experiments were performed automatically employing an ÄKTA Explorer 100 system (GE Health Care, Munich, Germany) as previously described (Tari et al. 2008). These experiments were run by introducing a population of yeast cells particles is introduced into a system composed of collector (adsorbent) beads; the suspended biomass effluent is monitored as a function of process time. This type of experiments can provide useful and quantitative information when assessing factors like cell size and shape, microorganisms strain, growth phase, bead size, surface coatings, fluid velocity, and ionic strength on cell deposition onto process media (Tari et al. 2008). A common approach to evaluate biomass deposition in laboratory packed-bed experiments employs the “clean-bed” filtration model (CBFM). In this case, mass transport phenomena are accounted by the “single-collector contact efficiency” (η0) while the physicochemical phenomena related to biomass attachment are reflected by the “attachment efficiency parameter” (α). Streamline Phenyl materials (high-sub) were packed in commercial chromatographic columns (5 mm internal diameter, 50mm length). The quality of the packing was evaluated by residence time distribution analysis employing 1% acetone as tracer (Bak and Thomas 2007). Biomass deposition studies were done by injecting a 4 ml biomass pulse (OD @ 600 nm ≈ 0.8 AU). Experiments were performed utilizing 20 mM phosphate buffer pH 7 or 50 mM acetate buffer pH 4. Buffers contained various amounts of ammonium sulphate as added salt (0.0, 0.4, 0.8, 1.2, 1.6, 2.0 M). The operational flow rate was 76.4 cm.h-1. Particle breakthrough curves were obtained by monitoring the effluent suspensions at 600 nm. On the basis of such data, the biomass deposition parameter (α) was calculated (Redman et al. 2004). Biomass deposition experiments were performed in triplicate and showed to be reproducible within ± 20%.

95

Results 2.3.3.7 Energy-distance profile calculations The total interaction energy between a colloidal particle and a solid surface can be expressed in terms of the extended DLVO theory as: XDLVO LW EL AB U mwc = U mwc + U mwc + U mwc

(1)

where UXDLVO is the total interaction energy in aqueous media, ULW is the LW interaction term, and UEL is the EL interaction term. The subscript m is utilized for the chromatographic matrix (adsorbent bead), w refers to the watery environment, and c to the colloidal (cell) particle. A third short-range (≤ 5 nm) Lewis AB term is included to account for “hydrophobic attractive” and “hydrophilic repulsive” interactions (Van Oss 2003).

Material surface energy parameters (tensions) can be calculated from contact angle measurements utilizing three diagnostic liquids, according to (Van Oss 1994). In turn, this data can be employed to evaluate the free energy of interaction between two defined surfaces (ΔGLW and ΔGAB). ΔG represents here the interaction energy per unit area between two (assumed) infinite planar surfaces bearing the properties of the adsorbent bead and the cell (interaction) or two cells (aggregation), respectively. Interaction between any of these two surfaces are evaluated at a closest distance of approximation (h0 ≈ 0.158 nm) (Bos et al. 1999). When integrated into mathematical expressions accounting the geometric constraints existing between two interacting bodies, ΔG values can be utilized to calculate the corresponding energy-distance profile (U vs. H). Details of this procedure were published (Bos et al. 1999; Vennapusa et al. 2008). ΔGLW are also related to the Hamaker constant, as follows: A = −12π h02 ΔG LW

(2)

UEL energy-distance profile can be calculated, assuming either plate-sphere or spheresphere geometry, upon experimental determination of particle zeta potential values. Zeta potential values are measured by electrophoretic mobility experiments (Vennapusa et al. 2008). Calculations were performed employing a commercial software package (GraphPad Prism, GraphPad Software Inc., San Diego, CA, USA).

96

Results 2.3.4 Results and discussions 2.3.4.1 Contact angle measurements and surface energy components The diagnostic liquids water, formamide, and α-bromonaphtalene were employed to measure contact angles onto homogeneous lawns of the materials under study i.e. intact yeast cells or crushed Phenyl-Sepharose beads. The sessile drop technique was employed. The utilization of the agar plate method assured that contact angle values were obtained for the mentioned materials in the hydrated state. Diagnostic liquids were chosen to have a higher surface tension than the sample materials so as to allow for stable drop formation and accurate contact angle determination. Both materials were carefully equilibrated with either 20 mM phosphate buffer (pH 7) or 50 mM acetate buffer (pH 4), which are buffers commonly encountered as mobile phases during hydrophobic interaction chromatography (HIC). Since conditions for binding proteins and macromolecules onto this particular chromatographic media are usually found at increased concentrations of ammonium sulphate i.e. within the range 0.2 - 2.0 M, this salt was included during sample preparation. Therefore, contact angles with three different liquids were performed as a function of pH and salt concentration so as to evaluate material(s) properties within the normal HIC operational range. Table 1 summarizes the contact angle values obtained after measurements performed onto homogeneous layers of intact yeast cells at pH 7 and pH 4. The agar plate technique utilized allowed the measurement of contact angles under the assumption that only bound water is present in the sample materials. Irrespective of pH (phosphate buffer pH 7 vs. acetate buffer pH 4) and salt concentration (the ammonium sulphate concentration increased from 0 to 2 M in the corresponding buffer solution), data gathered for contact angles measured with both water and formamide overall showed low and nearly constant values. Average values for water were ≈ 10 and for formamide ≈ 12. This indicates the very hydrophilic nature of the samples. On the contrary, contact angles values gathered with αbromonaphtalene decreased from ≈ 54 to ≈ 30 and from ≈ 46 to ≈ 30 at pH 7 and pH 4, respectively, upon addition of salt. A more progressive decrease in the contact angle values was observed -as a function of salt concentration- at pH 7 than at pH 4. In the later case, values for contact angles at varying salt concentrations tended to keep a constant level (≈ 30) a condition which differentiates from the contact angle measured in plain buffer

97

Results solution (≈ 46). This indicates that a non-polar liquid can be employed to discriminate between biomass types or conditions in relation to surface hydrophobic character (Butkus and Grasso 1998).

98

15.0 ± 1.0

12.4 ± 1.0

11.1 ± 0.6

9.0 ± 1.3

10.0 ± 0.5

9.3 ± 0.7

8.6 ±1.0

0.0

0.2

0.4

0.8

1.2

1.6

2.0

(◦)

(M)

pH 7

Water

(NH4)2SO4

8.3 ± 0.1

8.6 ± 0.6

10.9 ± 0.2

7.9 ± 1.0

11.5 ± 1.0

11.4 ± 1.0

12.0 ± 1.0

pH 4

99

9.5 ± 1.0

13.0 ± 2.6

12.0 ± 1.1

11.0 ± 1.7

14.0 ± 1.4

13.5 ± 1.0

14.0 ± 1.0

pH 7

(◦) pH 4

10.1 ± 0.2

10.2 ± 1.4

8.6 ± 0.9

9.6 ± 1.4

12.3 ± 2

13.4 ± 0.5

16.0 ± 1.0

Formamide

30.2 ± 1.0

31.0 ± 0.7

33.0 ± 2.4

33.4 ± 3.5

37.5 ± 2

44.0 ± 0.7

54.0 ± 1.0

pH 7

(◦)

29.8 ± 2.9

30.5 ± 2.9

30.5 ± 3.2

29.9 ± 3.2

30.3 ± 1.2

32.0 ± 1.0

46.0 ± 2.0

pH 4

α-Bromonaphtelene

Table 1: Contact angle values of intact yeast cells in 20 mM phosphate buffer pH 7, 50 mM acetate buffer pH 4 as a function of ammonium sulphate concentration.

Results

Results Table 2 shows contact angle values obtained by performing measurements onto layered fragments (< 10 µm) of the hydrophobic interaction media, Phenyl-Sepharose. This method was utilized since for soft gel particles other approaches e.g. the capillary raise method are difficult to implement. Moreover, measurements onto layered materials showed good reproducibility i.e within ±10% in triplicate measurements (Table 2). As described with biomass, a range of conditions was explored. At pH 7 contact angle values were ≈ 6-7 for water and ≈ 8-11 for formamide, irrespective of salt concentration. On the other hand, a step change in the contact angle with α-bromonaphtalene from ≈ 48 (no salt) to ≈ 30 (0.2 – 2.0 M ammonium sulphate) was noticed. At pH 4 recorded contact angle values were ≈ 7-8 with water and ≈ 9-10 with formamide but observed values with α-bromonaphtalene were progressively reduced from ≈ 36 (no salt) to ≈ 22 (2.0 M ammonium sulphate). As a whole, these results stressed the known hydrophilic nature of the chromatographic beads, which are composed by an agarose backbone. Contact angles values observed with the apolar liquid also indicate an increased hydrophobic character in the presence of ammonium sulphate.

100

6.0 ± 1.0

6.0 ± 1.0

6.0 ± 1.0

7.0 ± 1.0

7.0 ± 1.0

6.0 ± 1.0

7.0 ± 1.0

0.0

0.2

0.4

0.8

1.2

1.6

2.0

(◦)

(M)

pH 7

Water

(NH4)2SO4

8.0 ± 1.0

7.7 ± 0.5

7.0 ± 1.0

7.4 ± 0.5

8.0 ± 1.0

7.3 ± 0.5

7.0 ± 1.0

pH 4

101

11.0 ± 1.0

8.0 ± 1.0

10.0 ± 1.0

11.0 ± 1.0

8.0 ± 1.0

8.0 ± 1.0

10.0 ± 1.0

pH 7

(◦) pH 4

10.0 ± 1.0

9.0 ± 1.0

10.0 ± 1.0

9.0 ± 1.0

10.0 ± 1.0

10.0 ± 1.0

9.0 ± 1.0

Formamide

32.0 ± 3.5

30.3 ± 3.0

24.0 ± 2.5

30.7 ± 3.1

23.7 ± 2.5

28.0 ± 1.0

48.0 ± 4.8

pH 7

(◦)

23.6 ± 1.9

22.3 ±1.0

21.0 ± 1.0

23.0 ± 1.0

25.0 ± 2.2

28.5 ± 0.5

36.0 ± 3.5

pH 4

α-Bromonaphtelene

Table 2: Contact angle values for Phenyl-Sepharose particles in 20 mM phosphate buffer pH 7, 50 mM acetate buffer pH 4 as a function of salt concentration.

Results

Results Global analysis of contact angle data suggests a decrease in the contact angle values, as a function of ammonium sulphate concentration, measured with α-bromonaphtalene for cells and chromatographic beads. Contact angle values obtained for Phenyl-Sepharose with water and formamide were nearly constant irrespective of salt concentration. On the other hand, contact angles determined with the later two diagnostic liquids showed a trend to decrease when yeast cells were tested in the presence of salt Experimental contact angle determinations were utilized to calculate surface energy parameters for both biomass and chromatographic media according to the acid-base approach (Bos et al. 1999). Calculated parameters reflect the contribution of the various energy components i.e. Lifshitz-van-der-Waals and Acid-base (electron-acceptor, electrondonor) to the total surface energy of a defined material. Table 3 depicts the surface energy components (γ) calculated for layered intact yeast cells as a function of pH (7 and 4) and ammonium sulphate concentration (0 – 2.0 M). As a general trend it was observed that γLW increased (e.g. from 28 mJ⋅m-2 to 38 mJ⋅m-2 at pH 7 and from 32 mJ⋅m-2 to 39 mJ⋅m-2 at pH 4) while γAB decreased (e.g. from 30 mJ⋅m-2 to 18 mJ⋅m-2 at pH 7 and from 25 mJ⋅m-2 to 18 mJ⋅m-2 at pH 4) as salt concentration was increased. Table 4 shows surface energy components for crushed chromatographic media as a function of pH and salt concentration, as before. At pH 7, γLW increased from 31 mJ⋅m-2 (no salt) to 39 mJ⋅m-2 (0.4 - 2.0 M ammonium sulphate) while γAB decreased from 28 mJ⋅m-2 (no salt) to 17 mJ⋅m-2 (2.0 M ammonium sulphate). At pH 4 a similar trend was noticed: γLW increased from 36 mJ⋅m-2 (no salt) to 41 mJ⋅m-2 (1.2 - 2.0 M ammonium sulphate) while γAB decreased from 21 mJ⋅m2

(no salt) to 15 mJ⋅m-2 (2.0 M ammonium sulphate). As observed from Table 3 and 4, the

parameter ΔGiwi took always values +23-27 mJ⋅m-2 reflecting the hydrophilic nature of the yeast cells and the chromatographic beads. For comparison, the ΔGiwi of hydrophilic repulsion for Dextran T-150 is +41.2 mJ⋅m-2 (Van Oss 2003). Concerning the materials acid-base character, particularly noticeable was a decrease of the values of the electronacceptor parameter i.e. up to 60% when comparing γ- in the absence and presence of salt, respectively (Table 3 and Table 4). γ- values obtained via contact angle measurements more often pertain only to the global or averaged surface properties of the materials under study. Therefore, the agarose backbone onto which Phenyl ligands are immobilized is expected to have a major contribution to the overall material properties. On the other hand, differences in surface energy components might arise due to macromolecular changes within the cell

102

Results envelop which can occur as a function of pH and salt concentration. The observed AB repulsion in aqueous media often explains the formation of stable suspensions of biological particles or stable dispersions of proteins and polysaccharides (Wu et al. 1999).

103

28.0

33.0

35.6

37.4

37.9

38.0

38.5

0.2

0.4

0.8

1.2

1.6

2.0

38.6

38.6

38.6

38.6

38.6

37.8

31.7

pH 4

1.5

1.5

1.5

1.6

2.0

2.7

4.4

pH 7

1.5

1.5

1.5

1.5

1.4

1.5

2.9

pH 4

(mJ⋅m-2)

(mJ⋅m-2)

pH 7

γ+

γLW

0.0

(NH4)2SO4 (M)

54.3

54.8

54.8

54.7

54.0

53.2

51.5

pH 7

104

54.4

54.3

54.2

54.0

54.1

54.3

54.1

pH 4

(mJ⋅m-2)

γ-

18.0

18.0

18.2

18.6

20.7

24.1

30.1

pH 7

18.0

18.0

18.0

17.9

17.5

18.0

24.9

pH 4

(mJ⋅m-2)

γAB

56.6

56.0

56.0

56.0

56.5

56.9

58.3

pH 7

56.6

56.6

56.5

56.5

56.0

55.8

56.6

pH 4

(mJ⋅m-2)

γtot

+26.0

+27.1

+27.0

+27.2

+26.8

+26.0

+23.5

pH 7

+26.0

+26.0

+25.9

+25.7

+26.0

+26.7

+27.2

pH 4

(mJ⋅m-2)

ΔG iwi

Table 3: Surface energy parameters of intact yeast cells in 20 mM phosphate buffer pH 7, 50 mM acetate buffer pH 4 as a function of ammonium sulphate concentration.

Results

30.8

39.3

39.3

39.3

39.3

39.3

39.3

0.2

0.4

0.8

1.2

1.6

2.0

41.1

41.1

41.1

40.9

40.3

39.1

36.3

pH 4

1.3

1.4

1.3

1.3

1.4

1.4

3.5

pH 7

1.0

1.0

1.0

1.0

1.1

1.3

2.0

pH 4

(mJ⋅m-2)

(mJ⋅m-2)

pH 7

γ+

γLW

0

(NH4)2SO4 (M)

55.4

54.9

55.1

55.4

54.9

54.9

54.4

pH 7

105

55.0

54.8

54.6

54.9

54.8

55.0

54.4

pH 4

(mJ⋅m-2)

γ-

16.7

17.3

16.9

16.7

17.3

17.3

27.5

pH 7

14.8

15.0

15.3

15.3

16.0

17.1

20.8

pH 4

(mJ⋅m-2)

γAB

56.0

56.7

56.3

56.0

56.7

56.7

58.4

pH 7

56.0

56.2

56.4

56.2

56.3

56.3

57.1

pH 4

(mJ⋅m-2)

γtot

+27.3

+26.3

+26.8

+27.3

+26.3

+26.3

+26.5

pH 7

+26.2

+25.7

+25.4

+25.9

+26.0

+26.7

+26.5

pH 4

(mJ⋅m-2)

ΔG iwi

Table 4: Surface energy parameters of Phenyl-Sepharose particles in 20 mM phosphate buffer pH 7, 50 mM acetate buffer pH 4 as a function of ammonium sulphate concentration.

Results

Results 2.3.4.2 Cell-to-support interaction 2.3.4.2.1 Interfacial free energy interaction and energy-distance profiles Interaction between biomass particles and chromatographic beads can be understood by calculating interfacial free energy (U) vs. distance (H) profiles. These calculations are based on the experimental determination of contact angles with three diagnostic liquids and the additional information gathered from zeta potential determinations. Hydrophobic interaction chromatography is operated in a context characterized by an increased salt concentration (high ionic strength and conductivity) in the mobile phase, as well as, by uncharged beaded adsorbents. Therefore, it is expected that the information provided by contact angle determination will be more relevant to understand cell-to-support interactions than the information provided via z-potential determinations. Indeed, measurements of zeta potentials performed for Phenyl-Sepharose adsorbent particles under the experimental conditions reported in this work revealed very low values: -2.0 mV (base buffer) to -0.1 mV (high salt concentration). Zeta-potential values for yeast cells in diluted buffer solutions have been reported elsewhere but these values are expected to approach negligible values at high salt concentrations (Lin et al. 2006). This situation is radically different from the case of the ion-exchangers where, due to the low conductivity of the mobile phases and the charged nature of the adsorbents, z-potential has been established as a parameter describing biomass deposition onto process supports (Lin et al. 2006). Table 5 depicts the interfacial free energy of interaction between a biomass particles and a hydrophobic interaction bead, in aqueous media at pH 7 or pH 4, at closest distance of approximation (1.57 Å). ). The separation distance is determined by the balance between Born repulsion and van der Waals attractive forces. At pH 7 it can be observed that ΔGLW decreased from -1.1 mJ⋅m-2 (phosphate buffer) to -4.9 mJ⋅m-2 (salt containing buffer) indicating increasing LW attraction while ΔGAB increased from +27.2 (buffer) to +36.5 (salt) indicating enhanced repulsion by AB forces. At pH 4 a similar trend was noticed

106

-1.1

-3.4

-4.1

-4.6

-4.8

-4.8

-4.9

0.2

0.4

0.8

1.2

1.6

2.0

pH 7

0.0

(NH4)2SO4 (M)

ΔG LW (mJ⋅m-2)

-5.4

-5.4

-5.4

-5.3

-5.2

-4.7

-2.6

pH 4

107

+36.5

+36.3

+36.5

+36.5

+35.0

+33.4

+27.2

pH 7

ΔG AB (mJ⋅m-2)

+37.0

+36.7

+36.5

+36.5

+36.5

+36.1

+32.4

pH 4

+31.6

+31.5

+31.7

+31.9

+30.9

+30.0

+26.1

pH 7

ΔG Tot (mJ⋅m-2)

+31.6

+31.3

+31.0

+31.2

+31.3

+31.4

+29.8

pH 4

Table 5: Interfacial free energy of interaction between intact yeast cells and Phenyl-Sepharose in 20 mM phosphate buffer pH 7, 50 mM acetate buffer pH 4 as a function of ammonium sulphate concentration.

Results

Results The Hamaker constant (A) for the interaction pair Phenyl-Sepharose / yeast cells was calculated from ΔGLW according to Equation (2). When calculated for dilute buffer solution i.e. phosphate buffer pH 7 and acetate buffer pH 4, a value of 0.42 kT was obtained. The calculated value for A in buffers containing ammonium sulphate was 1.1 kT. Therefore, an influence of salt concentration but not of pH was observed on interaction Hamaker constant values; interaction refers to support-cell phenomena (Butkus and Grasso 1998). Utilizing the data provided before i.e. ΔGLW, ΔGAB, and zeta potential values, interaction energy (U) vs. distance (H) profiles were calculated according to the XDLVO approach. Figure 1(a/b) shows the calculated secondary energy pockets occurring at ≈ 5 nm upon interaction of a yeast cell and the adsorbent surface. Calculations assumed sphere-to-plate geometry. This is justified since the adsorbent particles are bigger than the yeast particles by the factor of ~40. The depth of such energy pockets shifted from low to moderate values ≈ -20-50 kT in dilute buffer solutions down to values ≈ -120 kT at high salt concentrations. A more gradual modification of the involved interaction energies took place at pH 7 than at pH 4. This is agreements with previous findings utilizing bacterial cells (Stenstrom 1989). Stronger interaction energies between cells and fluidized beads in the presence of ammonium sulphate might explain observed biomass interference during direct HIC / EBA capturing of bioproducts from a crude feedstock (Fernandez-Lahore et al. 2000). Application of the extended DLVO approach is justified since due to the very polar nature of the buffer solutions where cell-adsorbent interactions take place, these interactions are known to be strongly influenced by polar Lewis acid-base (AB) or electron-acceptor / electron-donor forces. Contributions by electric double layer (EL) forces and particularly contributions by apolar Lifshitz-van der Waals (LW) forces are also expected to occur. Important to the particular system considered here EL and AB forces decay exponentially with distance but as opposed to EL, the rate of decay of AB forces with distance is independent on low to moderate variations in the ionic strength. On the other hand, LW interactions decay gradually and proportional to the separation distance between two bodies. As observed from Table 5, LW interactions were promoted upon salt addition. On the other hand, the pronounced asymmetry of the polar properties of hydrophilic materials like agarose-based chromatographic supports or biological particles promotes a strong AB repulsion i.e. hydrophilic repulsion. Taken as a whole, calculations performed in relation to 108

Results interaction phenomena i.e. cell-to support interactions have shown hydrophilic AB repulsion, increased LW attraction, and marginal contribution of EL forces under standard operational conditions. The extended DLVO approach has served to explain the behavior of many other colloidal systems. Brandt and Childress have demonstrated that short-range interactions between synthetic membranes and bio-colloids can be better explained by taking into consideration the role of AB forces (Brant and Childress 2002). Van Oss and coworkers have studied the stability of a thixotropic suspension of 2 μm hectorite particles and concluded that Lewis acid-base interactions play a key role in the coagulation dynamics of such system (Grasso et al. 2002). 2.3.4.3 Biomass deposition experiments Biomass deposition experiments were performed to evaluate yeast cells attachment to hydrophobic interaction supports. This allowed an independent experimental verification of the predictions made on the basis of energy vs. distance calculations (Figure 1 a/b).

109

Results

Figure 1: Energy vs. distance profiles for interaction between intact yeast cells and hydrophobic interaction beads, at varying ammonium sulphate concentration. A) 20 mM phosphate buffer pH 7 b) 50 mM acetate buffer pH 4.

110

Results Figure 2 (a/b) depict the cell effluent profiles measured as a function of the chemical environment provided by the mobile phase. Ammonium sulphate concentration was systematically varied to observe its influence on cell attachment onto Phenyl-Streamline beads. Cell deposition was evaluated a pH 7 and 4. Biomass deposition experiments showed a profound effect of salt concentration on cell effluent profiles e.g. higher cell deposition with increased ammonium sulphate concentrations. From Figure 2 (a/b) it can also be noticed that and increased tendency exists for particles to be retained at pH 7 (a) that at pH 4 (b) when cell deposition was evaluated as a function of increasing ammonium sulphate concentration (0 – 2 M).

111

Results

Figure 2: Biomass deposition experiments as a function of salt concentration. a) Phosphate buffer pH 7 b) Acetate buffer pH 4.

112

Results This trend i.e. increased deposition with neutral pH and increased salt concentration is reflected by α, a lump parameter describing such phenomena (Table 6). For example utilizing either phosphate buffer pH 7 or acetate buffer pH 4, values for α were 0.065 and 0.031, respectively. When ammonium sulphate was included in the mobile phase at a concentration of 2.0 M, α values were 0.443 at pH 7 and 0.214 at pH 4. This “attachment efficiency” parameter depends on the experimental conditions set by the experimenter. In this case the method has been adapted to a chromatographic workstation that can operate in automatic mode. Therefore, the procedure can be implemented in any chromatographic laboratory and utilized to gather information without the need of more complicated experimental determinations like contact angle measurements or zeta potential estimations. Qualitative and quantitative evaluation of cell deposition experiments can reveal several underlying phenomena like cell-to-support attachment (interaction), prevention of cell depositions by already deposited biomass particles (blocking), and cell-to-cell ripening (aggregation). The biomass deposition experiment employs a bed of packed collectors which creates a more stable hydrodynamic situation in comparison with fluidized or expanded bed systems. Additionally, the biomass deposition experiment operates at a flow rate (~ 75 cm/h) that is lower than the flow rates expected during expanded bed operation (~ 300 cm/h). These experiments, however, were designed to confirm XDLVO calculations e.g. to obtain information related to cell deposition onto the solid surface. Biomass deposition experiments were run under optimized conditions with demonstrated sensibility to changes in XDLVO interactions. Studies performed as a function of superficial velocity were utilized to evaluate the interplay between cell-support attraction and cell detachment by hydrodynamic drag; no evidence of filtration effects was observed. This data will be published elsewhere .

113

0.677 0.561 0.493 0.234 0.129 0.071

0.4

0.8

1.2

1.6

2.0

pH 7

0.0

(NH4)2SO4 (M)

C/Co (-)

114

0.279

0.321

0.397

0.551

0.647

0.829

pH 4

0.443

0.343

0.243

0.118

0.097

0.065

pH 7

α (-)

0.214

0.190

0.155

0.100

0.073

0.031

pH 4

Table 6: Calculated lumped biomass-attachment parameter from biomass deposition experiments for Phenyl-StreamlineTM particles vs. intact yeast cells in 20 mM phosphate buffer pH 7, 50 mM acetate buffer pH 4 as a function of ammonium sulphate concentration.

Results

Results Figure 3 shows the correlation between the attachment efficiency parameter and the depth of the secondary free energy of interaction between a cell particle and a chromatographic bead. Points corresponding to hydrophobic interaction systems are presented within the frame of previous results gathered with ion-exchangers. It can be observed that conditions were no salt is present, and irrespective of pH and buffer chemical composition, are characterized by low deposition parameter values (≤ 0.15) which correlate with limited energy pockets (≤ |25-50| kT). However, by adding ammonium sulphate to the flowing phase an increase in α values was noticed. The magnitude of this increment depended on pH. For buffers at neutral pH the parameter α changed from ≈0.1 (0.4 M salt) to ≈0.45 (2.0 M salt). On the other hand, at pH 4 moderate changes in α were observed e.g. from ≈0.07 (0.4 M salt) to ≈0.21 (2.0 M salt). Therefore, cell deposition in the presence of ammonium sulphate generally resulted in α ≥ 0.15. The later criterion has been set as threshold for problem-free operation during direct capture of bioproducts from a crude feedstock (Tari et al. 2008). From a process performance point of view this could indicate hydrodynamic and sorption performance limitations from example, during expanded bed adsorption of bioproducts (Fernandez-Lahore et al. 2000). Sorption performance utilizing HIC / EBA systems has previously been reported (Smith et al. 2002). Until now, however, it has been difficult to correlate such behavior with simple cell transmission indexes (Feuser et al. 1999). Biomass-impulse experiments, however, have shown to correlate with ionexchanger sorption performance were electrostatic-driven cell-to-matrix interactions effects are predominant.

115

Results

Figure 3: Correlation between depth of free energy of interaction pocket and lumped attachment coefficient for several systems.

Analysis of the correlation between the depth of the interaction energy pockets and the attachment efficiency values for hydrophobic interaction materials in the presence of ammonium sulphate reveled differences with ion-exchange adsorbents. For HIC systems, a modification in α values correlated with discrete modifications in energy pocket values (Figure 3). Moreover, extreme values of both attachment efficiency and energy valleys were not observed. These results, as a whole, might indicate that total deposition of biomass particles is mediated not only by cell-to-matrix interaction but also by cell-to-cell aggregation phenomena (ripening). Deposition experiments also seem to indicate that ripening is occurring in a larger extent at pH 7 than at pH 4. Summarizing, for hydrophobic interaction systems modifications within a secondary interaction energy pocket occurred only from -70 kT to -120 kT but α values increased up to 0.45 when ammonium sulphate increased from 0 to 2 M (Figure 3).

116

Results Experiments performed to evaluate the influence of the age of the culture on cell attachment -as observed by biomass deposition experiments- showed increased α values when aged cells were employed. For example, in phosphate buffer pH 7 containing 1.0 M ammonium sulphate α increased from 0.20 to 0.36 when fresh cells were compared to an aged culture (data not shown). At pH 4 a similar trend was observed with α increasing for 0.14 to 0.26 when considering late exponential phase vs. one day aged culture. 2.3.4.4 Cell-to-cell aggregation Cell-to-cell aggregation might represent and important mechanism promoting overall cell attachment during biomass deposition experiments. Therefore, increased values for the lumped α parameter might indicate not only stronger cell-to-support interaction but enhanced cell-to-cell aggregation. Consequently, results from biomass deposition experiments will reveal conditions prevailing during real process performance where both interaction and aggregation phenomena can coexist. Contact angle and zeta potential determinations, as reported in this work and elsewhere (Lin et al. 2006) have been utilized to calculate energy vs. distance profiles between two intact yeast cells. Sphere-to-sphere geometry was assumed. These XDLVO calculations have indicated that: a) At closest distance of approximation ΔGLW took values between -1.5 mJ⋅m-2 (20 mM phosphate buffer pH 7) and -3.8 mJ⋅m-2 (50 mM acetate buffer pH 4) under the chemical environment provided by the buffering solutions employed. By adding increasing amounts of ammonium sulphate i.e. up to 2 M ΔGLW values decreased to -9.5 mJ⋅m-2, irrespective of system pH. Therefore, attraction between cell particles due to LW forces is similar at both pH values but increased with salt concentration (Table 7). Hamaker constant values were 0.6 kT (diluted buffer solution) and 2.0 kT (added salt ≥ 0.4 M) for yeast-to-yeast aggregation. b) Under similar conditions, ΔGAB showed more repulsion when calculating interfacial energy values at pH 4 (from +31.0 mJ⋅m-2 and up to +35.6 mJ⋅m-2 under buffer and added salt conditions, respectively) than when calculating interfacial energy values at pH 7 (from +25.0 mJ⋅m-2 and up to +36.0 mJ⋅m-2 under buffer and added salt

117

Results conditions, respectively). Therefore, the model biomass utilized in this work might have a tendency to be more stable e.g. less aggregation under acidic pH conditions due to enhanced repulsion by AB forces (Table 7). c) Coulomb-type interactions are repulsive in nature, but of marginal importance when salt concentration is higher than 0.1 M ammonium sulphate e.g. EL are irrelevant under normal processing conditions. d) Calculations performed to evaluate energy vs. distance profiles for interaction between two cells in aqueous media have shown secondary energy pockets taking values within the range -3 kT and -11 kT under diluted buffer conditions and ≈ -30 kT in the presence of 2.0 M ammonium sulphate (data not shown).

118

-1.5

-4.5

-6.7

-8.4

-8.8

-8.9

-9.5

0.0

0.2

0.4

0.8

1.2

1.6

2.0

-9.5

-9.5

-9.5

-9.5

-9.5

-8.8

-3.8

pH 4

119

+35.6

+36.0

+35.9

+35.5

+33.5

+30.5

+25.0

pH 7

(mJ⋅m-2)

(mJ⋅m-2)

(M)

pH 7

ΔG AB

ΔG LW

(NH4)2SO4

+35.6

+35.5

+35.4

+35.2

+35.6

+35.5

+31.0

pH 4

+26.0

+27.1

+27.0

+27.1

+26.8

+26.0

+23.5

pH 7

(mJ⋅m-2)

ΔG Tot

+26.0

+26.0

+25.9

+25.7

+26.0

+26.7

+27.2

pH 4

Table 7: Interfacial free energy of aggregation of intact yeast cells in 20 mM phosphate buffer pH 7, 50 mM acetate buffer pH 4 as a function of ammonium sulphate concentration.

Results

Results In order to elucidate cell aggregation behavior as a function of pH and salt concentration laser diffraction spectroscopic measurements were employed (Voloshin et al. 2005). The implementation of an independent method to specifically evaluate cell-to-cell aggregation can help in understanding (lumped) deposition coefficient values. For example, high α values in the absence of aggregation by light scattering can be attributed to strong cell-tosupport attachment. On the contrary, high α values and strong aggregation can indicate a combined effect during biomass deposition. Figure 4 depicts particle size for isolated yeast cells and formed aggregates, if any. Determinations were performed in 20 mM phosphate buffer pH 7 and in 50 mM acetate buffer pH 4, so as to reproduce the conditions found during biomass deposition experiments. Under these conditions, results indicated that cells were suspended without any association and existed as ≈ 8 μm particles (Figure 4a). This is in perfect agreement with the known size of intact yeast cells. Similar experiments performed in the presence of 1.6 M ammonium sulphate showed a faster cell-to-cell aggregation at pH 7 that at pH 4 at short contact times (10 min) (Figure 4b). Furthermore, longer contact times (45 min) promoted the formation of larger aggregates at pH 7 (≈ 400 μm) than at pH 4 (≈ 250 μm) (Figure 4c). Laser diffraction experiments performed in the presence of salt were also able to show the shrinkage of individual yeast cell to ≈ 5 μm (data not shown). Cell clumping in the presence of salt was confirmed by confocal microscopy (Figure 4d). Table 8 summarizes quantitative information obtained after laser diffraction spectroscopic evaluation of the samples. Results are expressed as percentiles. The d(0.1), d(0.5), and d(0.9) values shown in Table 8 are indicating that 10 %, 50% and 90% of the particles measured were less than or the equal to the size stated in each case. Sample replicates (n=5) have indicated that the shear exerted by the instrument during the measurement process was not promoting aggregate disruption (Table 8).

120

Results a)

b)

c)

d)

Figure 4: Laser diffraction experiments performed with intact yeast cells as a function of salt concentration, pH of the suspending buffer, and contact time. a) control yeast cells in plain buffer at pH 7 and 4; (b) Yeast cells after 10 minutes in contact with buffer containing 1.6 M (NH4)2SO4; (c) yeast cells after 45minutes in contact with buffer containing 1.6 M (NH4)2SO4; d) Visual aggregation of yeast cells suspended in 1.6 M of salt.

Table 8: Laser diffraction experimental data gathered for intact yeast cells as a function salt concentration, pH of the suspending buffer, and contact time. Time (min)

10/45

10

45

pH (*)

(NH4)2SO4 (M)

d (0.1) (µm)

d (0.5) (µm)

d (0.9) (µm)

7

-

4.3 ± 1

5.6 ± 1

7.6 ± 1

4

-

4.6 ± 1

6.2 ± 1

8.7 ± 0.5

7 4

1.6 1.6

4.7 ± 0.5 3.5 ± 0.5

160.5 ± 10 231.5 ± 20 5.3 ± 1 117.5 ± 5

7

1.6

5.2 ± 1

284.7 ± 15 409.0 ± 25

4

1.6

4.3 ± 0.5

(*) pH 7: 20 mM phosphate buffer; pH 4: 50 mM acetate buffer

121

18.4 ± 4

275.3 ± 15

Results 2.3.5 Conclusions A comprehensive approach to understand biomass deposition / adhesion onto process supports, with special emphasis on hydrophobic interaction surfaces have included interaction forces other than those purely electrostatic in nature and have utilized principles of colloid theory to explain biomass-adsorbent attachment at the local (particle) level. Within the classical DLVO theory approach, Lifshitz-Van der Waals (LW) and electrostatic interactions (EL) were considered. Other forces like acid-base (AB) interactions were included in the extended approach (XDLVO) so as to explain biomass interaction and aggregation phenomena. Interaction between biomass particles and chromatographic beads was understood by calculating interfacial free energy (U) vs. distance (H) profiles. These calculations were based on the experimental determination of contact angles with three diagnostic liquids and the additional information gathered from zeta potential determinations. Hydrophobic interaction chromatography is operated in a context characterized by an increased salt concentration (high ionic strength and conductivity) in the mobile phase, as well as, by uncharged beaded adsorbents. Therefore, it was expected that information provided by contact angle determination would be more relevant to understand cell-to-support interactions than the information provided via zeta potential determinations. Qualitative and quantitative evaluation of cell deposition experiments have revealed several underlying phenomena like cell-to-support sticking, prevention of cell depositions by already deposited biomass particles (blocking), and cell-to-cell aggregation (ripening). Analysis of the correlation between the depth of the interaction energy pockets and the deposition coefficient values for hydrophobic interaction materials in the presence of ammonium sulphate reveled differences with ion-exchange adsorbents. For HIC systems, modifications in α values were followed by discrete modifications in energy pocket depths. Moreover, extreme values of both deposition coefficients and energy valleys were not observed. These results, as a whole, might indicate that total deposition of biomass particles is mediated not only by cell-to-material interaction but mainly by cell-to-cell aggregation phenomena (ripening).

122

Results Cell-to-cell aggregation has represented and important mechanism promoting overall cell adhesion during biomass deposition experiments. These results would indicate that similar phenomena would impact on real process performance. Cell aggregation behavior, as a function of pH and salt concentration, was confirmed by laser diffraction spectroscopic measurements. Besides direct attachment of cells to the beaded support, cell aggregation has contributed to elevated α-parameter values, particularly at pH 7, during biomass deposition experiments. Summarising, it was demonstrated that both cell-to-adsorbent (interaction) and cell-to-cell (aggregation) phenomena are responsible to biomass deposition onto hydrophobic interaction chromatographic materials. Interaction and aggregation was inferred from XDLVO calculations on the basis of contact angle and zeta potential measurements. Moreover, experimental confirmation was obtained by independent methods like biomass deposition experiments and laser diffraction spectrometry. Further work is being performed in our laboratory in order to extent the observations reported in this paper to other adsorbent chemistries, biomass types of various characteristics, and broader operational windows. For example, cell debris shows stronger interactions with hydrophobic adsorbents than intact cells, because of the hydrophobic inner membrane. Additionally, the information provided by the XDLVO approach is being utilised to alleviate process limitations. 2.3.6 Acknowledgements CT was financially supported by TUBITAK, the Turkish Scientific and Technical Research Council, Ankara Turkey.

RRVP gratefully acknowledges a doctoral fellowship from

Jacobs University. The authors would like thank Professor Udo Fritsching and Ms. Lydia Achelis, Department of Process Technology, University of Bremen, for helpful assistance during laser diffraction measurements.

123

Results 2.3.7 Nomenclature AB

Acid-Base

DLVO

Classical DLVO theory (Derjaguin, Landau, Verwey and Overbeek)

HIC

Hydrophobic interaction chromatography

EBA

Expanded Bed Adsorption

EL

Electrostatic

LW

Lifshitz-Van der Waals

A

Hamaker constant [kT]

IC

Intact yeast cell particles

γ LW

Apolar or Lifshitz-van der Waals component of surface tension [mJ⋅m-2]

γ AB

Polar or acid–base component of surface tension [mJ⋅m-2]

γ

Electron-donor component of surface tension (Lewis base) [mJ⋅m-2]

−

γ+

Electron-acceptor component of surface tension (Lewis acid) [mJ⋅m-2]

ε

Dielectric constant of the medium [-]

R

Radius of the particle [m]

ζ

Zeta potential [mV]

κ

Inverse of Debye length [m]

H

Distance between surfaces, measured from outer edge [m]

XDLVO

Extended DLVO theory, according to Van Oss

ΔG

Interfacial free energy @ 1.57 Å approach [mJ⋅m-2]

U

Interfacial energy of interaction [kT]

k

Boltzmann constant [J⋅K-1]

T

Absolute temperature [K]

h0

Closest distance of approximation [1.57 Å]

α

Lumped biomass attachment coefficient [-]

124

References 2.3.7 References Bak H, Thomas ORT. 2007. Evaluation of commercial chromatographic adsorbents for the direct capture of polyclonal rabbit antibodies from clarified antiserum. J Chromatogr B 848(1):116-130. Bos R, Van der Mei HC, Busscher HJ. 1999. Physico-chemistry of initial microbial adhesive interactions--its mechanisms and methods for study. FEMS Microbiol Rev 23(2):179-230. Brant JA, Childress AE. 2002. Assessing short-range membrane-colloid interactions using surface energetics. J Membr Sci 203:257-273. Butkus MA, Grasso D. 1998. Impact of Aqueous Electrolytes on Interfacial Energy. J Colloid Interface Sci 200(1):172-181. Fernandez-Lahore HM, Geilenkirchen S, Boldt K, Nagel A, Kula MR, Thommes J. 2000. The influence of cell adsorbent interactions on protein adsorption in expanded beds. J Chromatogr A 873(2):195-208. Fernandez-Lahore HM, Kleef R, Kula M, Thommes J. 1999. The influence of complex biological feedstock on the fluidization and bed stability in expanded bed adsorption. Biotechnol Bioeng 64(4):484-96. Feuser J, Walter J, Kula MR, Thommes J. 1999. Cell/adsorbent interactions in expanded bed adsorption of proteins. Bioseparation 8(1-5):99-109. Ganeva V, Galutzov B, Teissie J. 2004. Flow process for electroextraction of intracellular enzymes from the fission yeast, Schizosaccharomyces pombe. Biotechnol Lett 26(11):933-7. Grasso D, Subramaniam K, Butkus M, Strevett K, Bergendahl J. 2002. A review of nonDLVO interactions in environmental colloidal systems. Rev Environ Sci Biotechnol 1(1):17-38. Henriques M, Gasparetto K, Azeredo J, Oliveira R. 2002. Experimental methodology to quantify Candida albicans cell surface Hydrophobicity. Biotechnol Lett 24:1111– 1115. Kang S, Choi H. 2005. Effect of surface hydrophobicity on the adhesion of S. cerevisiae onto modified surfaces by poly(styrene-ran-sulfonic acid) random copolymers. Colloids Surf B Biointerfaces 10(46 (2)):70-7.

125

References Lin DQ, Brixius PJ, Hubbuch JJ, Thommes J, Kula MR. 2003. Biomass/adsorbent electrostatic interactions in expanded bed adsorption: a zeta potential study. Biotechnol Bioeng 83(2):149-57. Lin

DQ,

Fernandez-Lahore

HM,

Kula

MR,

Thommes

J.

2001.

Minimising

biomass/adsorbent interactions in expanded bed adsorption processes: a methodological design approach. Bioseparation 10(1-3):7-19. Lin DQ, Zhong LN, Yao SJ. 2006. Zeta potential as a diagnostic tool to evaluate the biomass electrostatic adhesion during ion-exchange expanded bed application. Biotechnol Bioeng 95(1):185-91. Mahn A, Asenjo JA. 2005. Prediction of protein retention in hydrophobic interaction chromatography. Biotechnol Adv 23(5):359-368. Ottewill RH, Shaw JN. 1972. Electrophoretic studies on polystyrene lattices. J Electroanal Chem 37:133-142. Redman JA, Walker SL, Elimelech M. 2004. Bacterial Adhesion and Transport in Porous Media: Role of the Secondary Energy Minimum. Environ Sci Technol 38(6):17771785. Sharma PK, Rao KH. 2002. Analysis of different approaches for evaluation of surface energy of microbial cells by contact angle goniometry. Adv Colloid Interface Sci 98(3):341-463. Smith MP, Bulmer MA, Hjorth R, Titchener-Hooker NJ. 2002. Hydrophobic interaction ligand selection and scale-up of an expanded bed separation of an intracellular enzyme from Saccharomyces cerevisiae. J Chromatogr A 968(1-2):121-128. Smyth CJ, Jonsson P, Olsson E, Soderlind O, Rosengren J, Hjerten S, Wadstrom T. 1978. Differences in Hydrophobic Surface Characteristics of Porcine Enteropathogenic Escherichia-Coli with or without K88 Antigen as Revealed by Hydrophobic Interaction Chromatography. Infect Immun 22(2):462-472. Stenstrom TA. 1989. Bacterial hydrophobicity, an overall parameter for the measurement of adhesion potential to soil particles. Appl Environ Microbiol 55(1):142-147. Tari C, Vennapusa RR, Cabrera RB, Fernandez-Lahore M. 2008. Colloid deposition experiments as a diagnostic tool for biomass attachment onto bioproduct adsorbent surfaces. J Chem Technol Biotechnol 83:183-191. Van Oss CJ. 1994. Interfacial forces in aqueous media. New York: Marcel Dekker. viii,440p. p.

126

References Van Oss CJ. 2003. Long-range and short-range mechanisms of hydrophobic attraction and hydrophilic repulsion in specific and aspecific interactions. J Mol Recognit 16(4):177-190. Vennapusa RR, Hunegnaw SM, Cabrera RB, Fernandez-Lahore M. 2008. Assessing adsorbent-biomass interactions during expanded bed adsorption onto ion exchangers utilizing surface energetics. J Chromatogr A 1181(1-2):9-20. Vergnault H, Willemot RM, Mercier-Bonin M. 2007. Non-electrostatic interactions between cultured Saccharomyces cerevisiae yeast cells and adsorbent beads in expanded bed adsorption: Influence of cell wall properties. Process Biochem 42(2):244-251. Voloshin S, Shleeva M, Syroeshkin A, Kaprelyants A. 2005. The Role of Intercellular Contacts in the Initiation of Growth and in the Development of a Transiently Nonculturable State by Cultures of Rhodococcus rhodochrous Grown in Poor Media. Microbiology 74:420-427. Wu W, Giese RF, Van Oss CJ. 1999. Stability versus flocculation of particle suspensions in water-correlation with the extended DLVO approach for aqueous systems, compared with classical DLVO theory. Colloids Surf B Biointerfaces 14:47-55.

127

Results

2.4 Surface energetics to assess biomass attachment onto immobilized metal affinity adsorbents in expanded beds Rami Reddy Vennapusa, Muhammad Aasim, Rosa Cabrera, and Marcelo FernandezLahore* Downstream Processing Laboratory, School of Engineering and Science, Jacobs University Bremen gGmbH, Campus Ring 1, D-28759, Bremen, Germany.

2.4.1 Abstract Cell-to-support interaction and cell-to-cell aggregation phenomena have been studied in a model system composed of intact yeast cells and Chelating-StreamlineTM adsorbents. Biomass components and beaded adsorbents were mainly characterized by contact angle determinations

with

three

diagnostic

liquids.

Complementarily,

zeta

potential

measurements were performed. These experimental values were employed to calculate free energy of interaction vs. distance profiles in aqueous media. The effect of immobilized metal-ion type and buffer pH on the interaction energy was evaluated. Calculations indicated that moderate interaction between cell particles and adsorbent beads can develop due to the presence of Cu2+ ions onto the solid phase. The strength of interaction increased with buffer pH, within the range 6.0 to 8.3 e.g. secondary energy pockets increased from |15| to |60| kT. Cell-to-cell secondary energy minimum was ≥ |14| kT showing low-tomoderate tendency to aggregate, particularly at pH ≥ 8. Extended DLVO predictions were generally confirmed by biomass deposition experiments. However, an exception was found when working with immobilized Cu2+ at pH 8 since yeast cells were able to sequestrate such immobilized ions. Therefore, lower-than-expected values for the depositions coefficient (α) were observed. Understanding biomass attachment onto Chelating supports can help to better design and operate expanded bed adsorption of bioproducts.

128

Results 2.4.2 Introduction Expanded Bed Adsorption (EBA) has been proposed as an integrative downstream processing technology allowing the direct capture of targeted species from an unclarified feedstock. Therefore, combination of solids removal, product concentration, and partial purification in a single processing step became possible. During EBA operation intact cell particles or cell debris present in the feedstock can interact –depending on the characteristics of the bioprocess system- with fluidized / classified adsorbent beads. It is already known that interaction between biomass and the adsorbent phase may lead to the development of poor system hydrodynamics and therefore, impaired sorption performance under real process conditions. Moreover, biomass interaction would result in increased buffer consumption in order to remove and wash away sticky biological particles. Biomass components can also mask ligand sites thus reducing their availability to bind the targeted species. These phenomena i.e. decreased sorption performance and buffer consumption is detrimental to cost-efficient processing utilizing expanded bed adsorption (FernandezLahore et al. 2000). Immobilized metal affinity chromatography (IMAC) is based on the interaction of certain superficial amino-acid residues e.g. histidine, tryptophan, and cysteine with metallic ions fixed by chelation onto a solid support. This chromatographic mode is usually run employing mobile phases that contains buffer salts e.g. phosphates at pH ≈ 7.5 but in the presence of other salts e.g. sodium chloride in order to reduce ionic (non-specific) interactions. Therefore, the charge properties of the entities involved i.e. biological particles and chromatographic beads are anticipated to play a minor role concerning interaction or aggregation (Gallardo-Moreno et al. 2002; Klotz et al. 1985). Nevertheless, evidence of potential sorption performance limitations were already observed for IMAC-EBA systems (Poulin et al. 2008). Biomass effects during EBA with group-specific adsorbents are poorly understood. Expanded-bed adsorption techniques constitute a broad field of IMAC application. Downstream processing procedures from unclarified E. coli or yeast homogenates were developed for native, as well as, histidine-tagged proteins (Clemmitt and Chase 2000; Willoughby et al. 1999). Vaccine candidates for clinical studies i.e. a His6-tagged modified diphtheria toxin, expressed in E. coli, and a malaria-transmission-blocking vaccine,

129

Results secreted from S. cerevisiae were also processed employing this technique (Noronha et al. 1999). Generally, recoveries over 80% of the product were achieved in successful cases, but at least two major weak features must be further improved: low dynamic capacity and efficiency of Clean In Place (CIP) procedures for eliminating contaminants. The later are directly linked to biomass interference with the sorption process. The combination of IMAC and EBA techniques has potential to provide a unique approach to simplifying the whole downstream process, reduce the number of steps and start-up investment, and thus make the purification more economical. Experimental evidence has addressed the importance of non-electrostatic forces for biomass adhesion to process surfaces in the broader context provided by a group of systems of technical and environmental relevance. In this regard, a more comprehensive approach to understand biomass deposition onto chromatographic supports has been proposed by utilizing principles of colloid theory to explain biomass-adsorbent attachment at the local (particle) level (Vennapusa et al. 2008). This approach is based on extended DLVO calculations performed via experimentally determined contact angles and z-potentials for the interacting bodies. This comprehensive method takes into account several types of possible interaction forces. Lifshitz-Van der Waals (LW) and acid-base (AB) forces are considered and, therefore, the approach is not limited to those purely electrostatic in nature (EL). Moreover, biomass attachment behavior onto chromatographic beads predicted on the basis of XDLVO calculations was validated by independent biomass deposition experiments (Tari et al. 2008). The aim of this paper was to contribute to a deeper understanding of biomass-adsorbent interactions. This would further open the pave for optimized EBA processing in industry. In this work, studies have targeted biomass adhesion to Chelating materials that have not been extensively studied so far. The physicochemical properties of biomass-derived material (bio colloid particles) vs. the physicochemical properties of the adsorbent beads (the process surface) were determined indirectly via contact angle and zeta potential measurements. Subsequently, Gibbs free (interfacial) energy of interaction was calculated as a function of surface distance in aqueous media e.g. process buffer. Calculations were experimentally confirmed by independent biomass deposition experiments.

130

Results 2.4.3 Materials and Methods 2.4.3.1 Materials Chromatographic matrices and columns were purchased from GE Health Care (Munich, Germany). α-bromonaphtalene and formamide were obtained from Fluka (Buchs, Switzerland). Water was Milli-Q quality. All other chemicals were analytical grade. 2.4.3.2 Generation of biomass Yeast cells (Saccharomyces cerevisiae) were cultivated in shake-flasks, harvested at late exponential phase by centrifugation, and washed three times with 10mM phosphate buffer solutions, as previously described (Ganeva et al. 2004). Cells were employed immediately after preparation for further experimental procedures. 2.4.3.3 Contact angle measurements Preparation of intact yeast cells for contact angle measurements was performed as described (Henriques et al. 2002). Washed cells were suspended to 10% (w/v) in IMAC buffer [20mM phosphate buffer, pH 7.6 / 250 mM sodium chloride / 1 mM imidazol]. Cells were further equilibrated in the buffer solution and the suspension subsequently poured onto agar plates containing 10% glycerol and 2% agar-agar. The plate was allowed to dry for 24-36 hours at room temperature on a properly leveled surface free from dust. Salt crystallization was avoided. Agar plates without cell spreads were utilized as control. Contact angles were measured as per the sessile drop method utilizing a commercial goniometric system (OCA 20, Data Physics instruments GmbH, Filderstadt, Germany). The three diagnostic liquids α-bromonaphtalene, formamide, and water were employed (Bos et al. 1999). All the measurements were performed in triplicate and at least 20 contact angles per samples were measured. Contact angle determination on buffer-equilibrated chromatographic beads was performed utilizing the same physicochemical conditions and experimental procedures described for cell particles. Chelating Sepharose was utilized (GE Healthcare, Munich, Germany). Previous to pouring onto the agar plates, matrix beads were frozen in liquid nitrogen and crushed mechanically. Crushing efficiency was assessed by microscopic examination and particle size determination so as to assure particle fragment diameters ≤ 10 μm. Square 131

Results pieces of the agar-supported chromatographic bead fragments were utilized for measuring contact angles. 2.4.3.4 Zeta potential determination Zeta-potential measurements were performed with a ZetaSizer Nano-ZS (Malvern instruments, Worcestershire, United Kingdom), as previously described (Vennapusa et al. 2008). Zeta-potential values were gathered employing biomass pretreated under the conditions described before. Zeta-potential values for crushed and pre-equilibrated chromatographic beads were calculated from the electrophoretic mobility data according the Smoluchowski’s equation (Ottewill and Shaw 1972). Data was gathered under identical buffer compositions as shown for biomass related determinations. 2.4.3.5 Cell aggregation behavior Visual inspection of aggregate formation was performed with a confocal laser scanning microscope, equipped with argon and helium/neon mixed gas laser with excitation wavelengths of 488 or 543 nm (LSM 510, Carl Zeiss, Oberkochen, Germany). Washed yeast cells in control buffer [20 mM phosphate buffer, pH 7.6] or IMAC buffer [20mM phosphate buffer, pH 7.6 / 250 mM sodium chloride / 1 mM imidazol] were mounted on glass slides and visually inspected. Scans at a resolution of 1024 x 1024 pixels were taken in the line-averaging mode. Micrographs were stored in LSM format for further analysis (Carl Zeiss LSM Image Browser). 2.4.3.6 Bio-colloid deposition experiments Biomass deposition experiments were performed automatically employing an ÄKTA Explorer 100 system (GE Health Care, Munich, Germany) as previously described (Tari et al. 2008). Streamline Chelating was packed in commercial chromatographic columns (5 mm internal diameter, 50 mm length). The quality of the packing was evaluated by residence time distribution analysis employing 1% acetone as tracer (Bak and Thomas 2007). Chelating particles were loaded with metal ions utilizing standard procedures (Clemmitt and Chase 2000). Biomass deposition studies were done by injecting a 4 ml biomass pulse (OD ≈ 0.8 AU). Experiments were performed utilizing IMAC buffer at pH

132

Results 6.0, 7.6, and 8.3. The operational flow rate was 76.4 cm.h-1. Particle breakthrough curves were obtained by monitoring the effluent suspensions at 600 nm. On the basis of such data, the biomass deposition parameter (α) was calculated (Redman et al. 2004). 2.4.3.6 Energy-distance profile calculations The total interaction energy between a colloidal particle and a solid surface can be expressed in terms of the extended DLVO theory as: XDLVO LW EL AB U mwc = U mwc + U mwc + U mwc

(1)

where UXDLVO is the total interaction energy in aqueous media, ULW is the LW interaction term, and UEL is the EL interaction term. The subscript m is utilized for the chromatographic matrix (adsorbent bead), w refers to the watery environment, and c to the colloidal (cell) particle. A third short-range (≤ 5 nm) Lewis AB term is included to account for “hydrophobic attractive” and “hydrophilic repulsive” interactions (Van Oss 2003). Material surface energy parameters (tensions) can be calculated from contact angle measurements utilizing three diagnostic liquids (Van Oss 1994). In turn, this data can be employed to evaluate the free energy of interaction between two defined surfaces (ΔGLW and ΔGAB). ΔG represents here the interaction energy per unit area between two (assumed) infinite planar surfaces bearing the properties of the adsorbent bead and the cell (interaction) or two cells (aggregation), respectively. Interaction between any of these two surfaces are evaluated at a closest distance of approximation [h0 ≈ 0.158 nm] (Bos et al. 1999). When integrated into mathematical expressions accounting the geometric constraints existing between two interacting bodies, ΔG values can be utilized to calculate the corresponding energy-distance profile (U vs. H). Details of this procedure were published elsewhere (Bos et al. 1999; Vennapusa et al. 2008). ΔGLW is also related to the interaction Hamaker constant, as follows: A = −12π h02 ΔG LW

(2)

UEL energy-distance curves can be calculated, assuming either plate-sphere or spheresphere geometry, upon experimental determination of particle zeta potential values. Zeta potential values are measured by electrophoretic mobility experiments. Calculations were

133

Results performed employing a commercial software package (GraphPad Prism, GraphPad Software Inc., San Diego, CA, USA).

134

Results

2.4.4 Results and Discussions 2.4.4.1 Characterization of the Chelating chromatographic support The sessile drop technique was employed to measure contact angle values onto homogeneously deposited fragments crushed Chelating-Sepharose beads in the hydrated state. The diagnostic liquids water, formamide, and α-bromonaphtalene were utilized. Diagnostic liquids were chosen to have a higher surface tension than the sample materials so as to allow for stable drop formation and accurate contact angle determination. In addition, zeta potential measurements were performed. Zeta-potential measurements values were ≈ ≤ |10|mV, indicating that charge effect cannot be dominating for this system (data not shown). 2.4.4.2 The effect of immobilized-ion type on the stationary phase The chromatographic material was carefully equilibrated with 20 mM phosphate buffer (pH 7.6) containing sodium chloride 250 mM and 1 mM imidazol (IMAC buffer). This is a buffer with a typical composition to be utilized as mobile phase during immobilized metal affinity chromatography (IMAC). Conditions for binding proteins and macromolecules onto this particular chromatographic support are usually found within a pH range 7 to 8. The nature of the immobilized ion has also a profound influence on selectivity and binding. Therefore, contact angles with three different liquids were performed as a function of pH and metal-ion type e.g. Cu2+, Ni2+, Zn2+, and Co2+. Table 1 shows contact angle values obtained by performing measurements onto layered fragments (< 10 µm) of the Chelating-Sepharose adsorbent. The agar plate technique utilized allowed the measurement of contact angles under the assumption that only bound water is present in the sample materials. For unloaded (metal-ion free) material contact angle values were ≈ 7 for water, ≈ 10 for formamide, and ≈ 53 for α-bromonaphtalene at pH 7.6 in IMAC buffer. This is in agreement with values reported previously for agarosebased chromatographic beads (Vennapusa et al. 2008). These results stressed the known hydrophilic nature of this type of material. Upon loading the Chelating supports with metalions, distinct contact angle values were observed in some cases. By immobilizing Zn2+ and Cu2+, the contact angle value for α-bromonaphtalene increased to ≈ 57 or decreased to ≈ 45, respectively. Therefore, contact angle measurements were able to capture information related to the nature of the fixed ion. 135

Results

Table 1: Contact angles obtained on crushed Chelating Sepharose with and without immobilized metal ions. For IDA-Cu2+, determinations were performed a various pH values. Measurements were performed in 20 mM phosphate buffer, pH 7.6, containing sodium chloride.

Metal ion

pH

Contact angle (θ) [°] Water

Formamide

αBromonaphtalene

No metal

7.6

7.0 ± 0.6

9.8 ± 0.8

53.0 ± 1.0

Zn2+

7.6

8.0 ± 2.0

12.8 ± 2.0

57.0 ± 2.1

Ni2+

7.6

8.0 ± 1.7

11.4 ± 2.0

54.0 ± 3.4

2+

Co

7.6

10.0 ± 2.0

12.6 ± 3.1

54.0 ± 2.0

Cu2+

6.0

7.5 ± 0.5

9.5 ± 1.3

57.0 ± 5.5

Cu2+

7.6

10.2 ± 2.4

12.1 ± 0.9

45.0 ± 3.0

Cu2+

8.3

11.1 ± 0.4

11.0 ± 1.5

40.0 ± 2.2

Experimental contact angle determinations were utilized to calculate surface energy parameters for the chromatographic media according to the acid-base approach (Bos et al. 1999). Table 2 depicts the surface energy components (γ) calculated as a function immobilized metal-ion type. γLW values were ≈ 28 mJ⋅m-2 for the free Chelating matrix and the Ni2+ or Co2+ loaded beads. A decreased value for such parameter, however, was observed with Zn2+ (26.5 mJ⋅m-2). On the other hand, γLW values increased for Cu2+ (32.3 mJ⋅m-2). An interesting behavior was observed in relation with acid-base character of the metalimmobilized materials. Values taken by the electron-acceptor parameter (γ+) and the total acid-base parameter (γAB) allowed differentiating between beads harboring different metalions as follows : Zn2+ > Ni2+ ≈ [IDA] ≈ Co2+ >> Cu2+

136

Results γ+ took values between 3.0 and 5.0 mJ⋅m-2 while for most Sepharose adsorbent values ≈ 1-2 mJ⋅m-2 are typical (Vennapusa et al. 2008). Similar observations were made by (Bayramoglu et al. 2006) when comparing L-histidine affinity membranes with or without immobilised copper (II) ions. These variations in the acid-base character of the (metal ion) loaded chromatographic matrices are expected since IMAC adsorption is based on the interaction between an immobilised transition-metal ion (electron pair acceptors) and electron-donor groups e.g. on protein surfaces. For proteins, the apparent affinity for a metal chelate depends strongly on the metal ion involved in coordination. In the case of the iminodiacetic acid (IDA) chelator, the affinities of many retained proteins and their respective retention times are in the following order: Cu2+ > Ni2+ > Zn2+ ≈ Co2+ (GabercPorekar and Menart 2001). Table 2: Surface energy parameters calculated for Chelating Sepharose loaded with various metal ions, calculated according to the contact angle values reported in Table 1. Metal ion

Surface energy parameters [mJ·m-2]

pH γLW

γ+

γ-

γAB

γTOT

ΔGsws

No metal

7.6

28.5

4.4

53.8

30.7

59.2

+25.2

Zn2+

7.6

26.5

5.0

54.1

32.8

59.3

+25.0

Ni2+

7.6

28.0

4.5

53.9

31.0

59.0

+25.3

Co2+

7.6

28.0

4.4

53.5

30.7

58.7

+25.2

2+

Cu

6.0

26.5

5.3

53.3

33.5

59.9

+24.0

Cu2+

7.6

32.3

3.0

53.7

25.0

57.4

+26.3

Cu2+

8.3

34.6

2.4

53.3

22.5

57.1

+25.6

2.4.4.3 The effect of mobile-phase pH Adsorption of a protein to the IMAC support is performed at a pH at which imidazole nitrogen’s in histidyl residues are in the nonprotonated form, normally in neutral or slightly basic medium (Chaga 2001; Ueda et al. 2003).

137

Results Contact angle determinations were also performed as a function of pH, employing immobilized copper ions. Table 1 also summarizes the contact angle values obtained in IMAC buffer at pH 6.0, 7.6, and 8.3. Some tendencies were observed when pH raised: a) water contact angles increased from 7.5 to 11.0, and b) formamide contact angles increased from 9.5 to 11.1, but c) α-bromonaphtalene contact angles decreased from 57 to 40. Table 2 depicts the surface energy components (γ) calculated for the IMAC adsorbent at various pH values. As a general trend it was observed that γLW increased with pH (e.g. from 26.5 mJ⋅m-2 at pH 6.0 to 34.6 mJ⋅m-2 at pH 8.3) while γAB decreased (e.g. from 33.5 mJ⋅m-2 at pH 6.0 to 22.5 mJ⋅m-2 at pH 8.3). Clearly noticeable was the influence of pH on the γ+ since this parameter decreased from 5.3 mJ⋅m-2 at pH 6.0 to 2.4 mJ⋅m-2 at pH 8.3. As observed from Table 2, the parameter ΔGiwi took always values +24-26 mJ⋅m-2, irrespective of the type of transition metal ion immobilized or the buffer pH, reflecting the highly hydrophilic nature of the chromatographic beads. 2.4.4.4 Characterization of the yeast particles Contact angle determinations were performed on intact yeast cell lawns within the physicochemical environment provided by the IMAC buffer. Contact angle values were collected for three distinct pHs: 6.0, 7.6, and 8.3. Water and formamide contact angle values were observed to increase with pH e.g. from 7.6 to 12.2 and from 10.4 to 18, respectively. On the contrary α-bromonaphtalene contact angles dropped from 50 to 44. Overall, this indicates an increased hydrophobic character for the yeast particles at higher pH values [Refer to Table 3 (a)].

138

Results Table 3(a): Contact angles obtained on intact yeast cell lawns. Measurements were performed in 20 mM phosphate buffer, pH 6.0, 7.6, and 8.3. Buffers solution contained sodium chloride to a final concentration of 250 mM. pH

Contact angle (θ)

Water

Formamide

α-Bromonaphtalene

6.0

7.6 ± 0.9

10.4 ± 1.0

50 ± 1.3

7.6

11.7 ± 2.1

13.7 ± 2.2

49.3 ± 0.7

8.3

12.2 ± 1.6

18.0 ± 2.7

43.6 ± 1.9

Table 3(b) depicts the surface energy components (γ) calculated for yeast cells at various pH values. As a general trend it was observed that γLW increased with pH (e.g. from 30 mJ⋅m-2 at pH 6.0 to 33 mJ⋅m-2 at pH 8.3) while γAB decreased (e.g. from 29 mJ⋅m-2 at pH 6.0 to 22 mJ⋅m-2 at pH 8.3). Buffer pH has also had and influence on the γ+ which decreased from 3.8 mJ⋅m-2 (pH 6.0) to 2.3 mJ⋅m-2 (pH 8.3). These observations are is in general agreement with previous studies on the impact of aqueous electrolytes on total interaction energy parameters (Butkus and Grasso 1998; Van Oss 1994). Table 3(b): Surface energy parameters calculated for intact yeast cells, calculated according to the contact angle values reported in Table 3(a). Surface energy parameters [mJ m-2]

pH γLW

γ+

γ-

γAB

γTOT

ΔGsws

6.0

29.8

3.8

53.9

28.6

58.6

+25.9

7.6

30.3

3.5

53.4

27.3

57.6

+25.9

8.3

33.0

2.3

55.3

22.4

55.4

+29.2

139

Results Zeta potential measurements performed in IMAC buffer at various pHs were ≈ |8| mV (data not shown). Zeta potential values close to zero are expected due to the moderate-high concentrations of salt which are present in such buffers e.g. from 0.25 to 1.0 M in sodium chloride. 2.4.4.5 Biomass interaction phenomena Interaction between biomass particles and chromatographic beads can be understood by calculating interfacial free energy (U) vs. distance (H) profiles. These calculations are based on the experimental determination of contact angles with three diagnostic liquids and the additional information gathered from zeta potential determinations. Table 4 depicts the interfacial free energy of interaction between a biomass particles and a Chelating bead, in aqueous media at pH 7.6, at closest distance of approximation (1.57 Å). For the unloaded particle, ΔGLW and ΔGAB values were -1.1 mJ⋅m-2 and +28 mJ⋅m-2, respectively. Upon loading the matrix with different transition metal ions, variations in the preceding values were observed for Zn2+ (ΔGLW = -0.8 mJ⋅m-2 / ΔGAB = 27.4 mJ⋅m-2) and for Cu2+ (ΔGLW = -1.7 mJ⋅m-2 / ΔGAB = 29.6 mJ⋅m-2) but not for Ni2+ and Co2+. Interaction between Cu (II) loaded beads and yeast cells was further investigated as a function of buffer pH (Table 4). It was noticed that ΔGLW decreased from -0.76 mJ⋅m-2 at pH 6.0 to -2.6 mJ⋅m-2 at pH 8.3. AB values also followed a similar tendency e.g. ΔGAB increased from 26.6 mJ⋅m-2 at pH 6.0 to 32.6 mJ⋅m-2 at pH 8.3. This data is indicating that the pH values of the buffer where interaction is occurring are correlated with modification in interaction energies. The interaction Hamaker constant (A) for the pair Chelating-Sepharose / yeast cells was calculated from ΔGLW according to Equation (2). When calculated for IMAC buffer solution (pH 7.6) an average value of 0.35 kT was obtained. A was lower for Zn2+ but higher for Cu2+ i.e. 0.17 kT and 0.40 kT, respectively.

140

Results Table 4: The interfacial Gibbs energy of interaction between intact yeast cells and Chelating Sepharose, at closest distance of approximation. Interaction occurs in 20 mM phosphate buffer containing sodium chloride as added salt. Loaded metal ion

ΔG [mJ·m-2]

Buffer pH ΔGLW

ΔGAB

No metal

7.6

-1.1

27.9

Zn2+

7.6

-0.8

27.4

Ni2+

7.6

-1.0

27.8

Co2+

7.6

-1.0

27.7

2+

Cu

6.0

-0.76

26.6

Cu2+

7.6

-1.7

29.6

Cu2+

8.3

-2.6

32.6

Utilizing the data provided before i.e. ΔGLW, ΔGAB, and zeta potential values, interaction energy (U) vs. distance (H) profiles were calculated according to the XDLVO approach. Due to the relatively high conductivity of the mobile phase utilized in Chelating systems (≈ 30 mS/cm), zeta-potential values for both yeast particles and chromatographic beads were very low (< 8-10 mV). Figure 1(a) and Figure 1(b) shows the calculated secondary energy pockets occurring at ≈ 7 nm upon interaction of a yeast cell and the adsorbent surface. Calculations assumed sphereto-plate geometry. The depth of such energy pocket for the unloaded matrix showed a moderate value ≈ -20 kT at pH 7.6. Metal ion loaded systems showed a similar energy profile. However, the presence of immobilized Cu2+ resulted in an increased pocket depth ≈ -40 kT. In the later case, modification of pH has lead to a reduced (≈ -15 kT at pH 6.0) or increased (≈ -60 kT at pH 8.3) energy pocket. Therefore, more cell deposition would be expected when working with immobilized cupper ions at higher pH values. Data indicated that cell-to-support interactions can be strongly influenced by polar Lewis acid-base (AB) or electron-acceptor / electron-donor forces which are included in the XDLVO approach. This has served to explain the behavior of many other colloidal

141

Results systems. Brandt and Childress have demonstrated that short-range interactions between synthetic membranes and bio-colloids can be better explained by taking into consideration the role of AB forces (Brant and Childress 2002; Brant and Childress 2004). Van Oss and coworkers have studied the stability of a thixotropic suspension of 2 μm hectorite particles and concluded that Lewis acid-base interactions play a key role in the coagulation dynamics of such system (Grasso et al. 2002). Van Oss also has reviewed the importance of AB forces for the stability of many colloidal systems (Van Oss 1993; Van Oss 2003).

Figure1 (a): Free energy of interaction vs. distance profiles between intact yeast and Chelating Sepharose loaded with different metal-ion types. Calculations were performed assuming that interaction occurs in a buffer with a typical composition for immobilized metal affinity chromatography, at pH 7.6 / (—) No metal, (—) Zn2+, (---) Ni2+, (—) Co2+, (—) Cu2+

142

Results

Figure 1(b): Free energy of interaction vs. distance profiles between intact yeast and Chelating Sepharose loaded with Cu (II) ions. Calculations were performed assuming that interaction occurs in a buffer with a typical composition for immobilized metal affinity chromatography, at various pH values. (—) pH 6, (—) pH 7.6, (—) pH 8.3.

2.4.4.6 Biomass aggregation phenomena Contact angle and zeta potential determinations, as reported in this work have been utilized to calculate energy vs. distance profiles between two intact yeast cells. The data gathered on zeta potential showed that charge effects are of marginal importance for the Chelating system. Sphere-to-sphere geometry was assumed. Table V depicts the interfacial interaction free energy between two biological particles, in aqueous media as function of pH at the closest distance of approximation (1.57 Å). The tendency of ΔGLW and ΔGAB interaction energy of aggregation is similar to that of the interaction of yeast with IDA-Cu2+ as function of pH (Table IV). Upon modification of buffer pH from 6.0 to 8.3, the ΔGLW interaction energy decreased from -2.6 mJ⋅m-2 to -4.6 mJ⋅m-2 while ΔGAB changed from +28.4 mJ⋅m-2 to 33.8 mJ⋅m-2, respectively. Interaction free energies as function of distance were calculated according to the XDLVO model (Figure 2) . Calculations, in aqueous media, have indicated that secondary energy pockets can develop at H ≈ 5-7 nm. The depth

143

Results of such pockets indicated that attraction between cell particles is higher when the pH of the buffer increased i.e. U ≈ |14| at pH 8.3 and ≤ |8| at lower pH values. This trend was confirmed by microscopic observations. Figure 3 depicts freely suspended cells in 20 mM phosphate buffer in comparison with clumped cell in IMAC buffer. Cell aggregation was verified by independent laser diffraction experiments (Data not shown). Table 5: The interfacial Gibbs energy of aggregation between intact yeast cells, at closest distance of approximation. Interaction occurs in 20 mM phosphate buffer containing sodium chloride as added salt. Buffer pH

ΔGLW [mJ·m-2]

ΔGAB [mJ·m-2]

6.0

-2.6

28.4

7.6

-2.8

28.7

8.3

-4.6

33.8

144

Results

Figure 2: Free energy of interaction vs. distance profiles between intact yeast cells. Calculations were performed assuming that interaction occurs in a buffer with a typical composition for immobilized metal affinity chromatography, at various pH values. (—) pH 6, (—) pH 7.6, (—) pH 8.3.

a)

b)

Figure 3: Microscopic observation of yeast cell aggregation employing a confocal system: a) Cells in 20 mM phosphate buffer at pH 7.6, and b) Cells in sodium chloride containing buffer at the same pH.

145

Results 2.4.4.7 Biomass deposition experiments Figure 4(a) depicts the cell effluent profiles measured as a function of the chemical environment provided by the mobile phase. The transition metal ion immobilized on the IDA moiety was varied to observe the influence of ion type on cell attachment. Chelating Streamline beads were utilized. Biomass deposition experiments confirmed that an increased deposition of yeast cell occur when Cu2+ is the fixed metal ion, at pH 7.6.

Figure 4 (a): Cell effluent profiles obtained after biomass deposition experiments. Intact yeast cells were utilized as model bio-colloids. Runs were performed in phosphate-based buffer solutions, at pH 7.6. Chelating Sepharose was utilized as collector particles. Iminodiacetic beads were loaded with several metal ions: (z) No metal, (z) Zn2+, (z) Ni2+, (z) Co2+, (z) Cu+2.

This fact is reflected by the “attachment efficiency” parameter (α). This is a lumped number which depends on experimental conditions; the method has been adapted to a chromatographic workstation that can operate in automatic mode. α values for the unloaded material and the Zn2+, Ni2+, and Co2+ loaded support fall within the range 0.056-0.078. Upon Cu2+ immobilization, α increased to 0.172 i.e. more yeast particles were trapped within the collector bed (Table 4).

146

Results Table 6: Lumped attachment parameter (α) calculated from yeast deposition experiments utilizing metal-ion loaded Chelating Sepharose at various pH values. Runs were performed in 20 mM phosphate buffer, containing sodium chloride as added salt. Loaded metal-ion

Buffer pH

C/C0 [-]

α [-]

No metal

7.6

0.716

0.056

Zn+2

7.6

0.696

0.061

Ni+2

7.6

0.646

0.073

Co+2

7.6

0.628

0.078

Cu+2

6.0

0.300

0.202

Cu+2

7.6

0.358

0.172

Cu+2

8.3

0.413

0.148

Moreover, cell deposition was evaluated a pH 6.0, 7.6, and 8.3. This was performed employing chelated Cu2+ onto the chromatographic beads. As observed in Figure 4(b), cell breakthrough profiles indicated that yeast particles are less retained at pH 8.3 than at pH 7.6. This result apparently does not correlate with the predictions made by the application of the XDLVO theory. However, yeast cells can interact so strongly with Cu (II) ions that they could actually sequestrate these ions from the chromatographic matrix (Ting. and Sun 2000). Results from collector experiments, which correlate with actual bioprocess performance, indicated that Cu2+ systems could be prone to deleterious effect by biomass. This is reflected by α ≥ 0.15 (Tari et al. 2008). At pH 6.0 chemical groups present within the cell wall structure like carboxyl and phosphoryl groups could be responsible for cell interaction with the Cu2+ loaded beads.

147

Results

Figure 4(b): Cell effluent profiles obtained after biomass deposition experiments. Intact yeast cells were utilized as model bio-colloids. Runs were performed in phosphate-based buffer solutions, at several pH values. Chelating Sepharose, loaded with Cu (II) ions, was utilized as collector particles. (z) pH 6, (z) pH 7.6, (z) pH 8.3.

Cell-to-cell aggregation might represent and important mechanism promoting overall cell attachment during biomass deposition experiments. Therefore, increased values for the lumped α parameter might indicate not only stronger cell-to-support interaction but enhanced cell-to-cell aggregation. For the Chelating system, cell aggregation effects are predicted to be low-to-moderate according to XDLVO calculations. To evaluate whether cell aggregation or attrition–in the absence of cell-to-matrix interaction- is influencing cell breakthrough profiles, biomass deposition experiments were run with plain material as collectors but utilizing mobile phase compositions known to increase cell aggregation (Ljungh and Wadström 1982). Experimental runs showed that α values were fairly constant (α ≈ 0.05) irrespective of the presence of ammonium sulphate, a salt that induces cell-tocell aggregation (Figure 5).

148

Results

Figure 5: Cell effluent profiles obtained after biomass deposition experiments. Intact yeast cells were utilized as model bio-colloids. Runs were performed in phosphate-based buffer solutions, at pH 7.6. Unloaded Chelating Sepharose beads were utilized as collectors. (z) 20 mM phosphate buffer, (z) Buffer containing added ammonium sulphate (0.8 M), (z) Buffer containing added ammonium sulphate (1.6 M).

As a whole, biomass deposition experiments have indicated that Cu (II) would promote more cell deposition than other metal ions. Cell aggregation can occur in buffers containing moderate-to-high concentration of added salts. However, in the absence of cell-to-support interaction further deposition due to cell aggregation is not possible. Summarizing, IMAC systems where biomass deposition could play a role seem to be limited to a) immobilized Cu (II), b) at pH within the range 7.0 to 7.6, and c) with salt containing buffers. 2.4.4.8 Chelating systems in the context of EBA adsorbents Figure 6 shows the correlation between the attachment efficiency parameter (α) and the secondary-pocket-depth (free energy of interaction between a cell particle and a chromatographic bead). Points corresponding to ion-exchangers and hydrophobic interaction systems are presented as a reference (Tari et al. 2008). It can be observed that Chelating materials are generally characterized by low deposition parameter values (α ≤ 0.15) which correlate with limited energy pockets (≤ |20| kT). However, the effect of 149

Results immobilizing Cu2+ creates a situation characterized by increased deposition of yeast cells (α ≥ 0.15) and concomitant increased of the free energy of interaction (secondary) pocket (≈ |40| kT). From a process performance point of view this could indicate hydrodynamic and sorption performance limitations from example, during expanded bed adsorption of bioproducts. Sorption performance utilizing Chelating / EBA systems has been reported recently (Poulin et al. 2008) and characterized as moderate, in full agreement with this paper and previous work (Fernandez-Lahore et al. 2000).

Figure 6: Correlation analysis between the calculated (interaction) energy-pocket-depth vs. the experimental attachment coefficient (α). Data corresponding to Chelating systems in presented within the frame provided by other interaction pairs: (z) Cation exchange materials, (z) Anion exchange supports, Phenyl Sepharose beads (●) [pH 7] and (z) [pH 4], (z) Chelating Sepharose.

150

Results

2.4.5 Conclusions A

comprehensive

approach

to

understand

biomass

deposition

onto

Chelating

chromatogrpahic supports has utilized principles of colloid theory to explain biomassadsorbent attachment at the local (particle) level. Acid-base (AB) interactions were included in the extended approach (XDLVO) so as to explain biomass interaction and aggregation phenomena. Besides, Lifshitz-Van der Waals (LW) (EL) were considered. Electrostatic interactions played a minor role due to the high conductivity of the process buffer involved. Interaction between biomass particles and chromatographic beads was studied by calculating interfacial free energy (U) vs. distance (H) profiles. These calculations were based on the experimental determination of contact angles with three diagnostic liquids and the additional information gathered from zeta potential determinations. Qualitative and quantitative data from cell deposition experiments has revealed a predominant role of cell-to-support interaction. Cell-to-cell aggregation showed a less impact on total biomass deposition in Chelating systems. Analysis of the correlation between the depth of the interaction energy pockets and the deposition coefficient values for Chelating materials in the presence of sodium chloride at neutral pH reveled differences with ion-exchange and hydrophobic interaction adsorbents. The strength of biomass interaction was enhanced by having copper (II) ions immobilized onto the solid phase. Summarising, it was demonstrated that cell-to-adsorbent (interaction) and –to a lesser extent- cell-to-cell (aggregation) phenomena are responsible to biomass deposition onto Chelating chromatographic materials. Interaction and aggregation was inferred from XDLVO calculations on the basis of contact angle and zeta potential measurements. Moreover, experimental confirmation was obtained by independent methods like biomass deposition experiments and confocal microscopy.

2.4.6 Acknowledgements RRVP gratefully acknowledges a doctoral fellowship from Jacobs University. The authors would thank Dr. Carl Bolster for his valuable discussions during the work.

151

Results

2.4.7 Nomenclature AB

Acid-Base

DLVO

Classical DLVO theory (Derjaguin, Landau, Verwey and Overbeek)

CHE

Chelating [Iminodiacetic Group - IDA]

EBA

Expanded Bed Adsorption

EL

Electrostatic

LW

Lifshitz-Van der Waals

A

Hamaker constant [kT]

IC

Intact yeast cell particles

IMAC

Immobilized metal affinity chromatography

γ LW

Apolar or Lifshitz-Van der Waals component of surface tension [mJ⋅m-2]

γ AB

Polar or acid–base component of surface tension [mJ⋅m-2]

γ

Electron-donor component of surface tension (Lewis base) [mJ⋅m-2]

−

γ+

Electron-acceptor component of surface tension (Lewis acid) [mJ⋅m-2]

Ε

Dielectric constant of the medium [-]

R

Radius of the particle [m]

Ζ

Zeta potential [mV]

κ

Inverse of Debye length [m]

H

Distance between surfaces, measured from outer edge [m]

XDLVO

Extended DLVO theory, according to Van Oss

ΔG

Interfacial free energy @ 1.57 Å approach [mJ⋅m-2]

U

Interfacial energy of interaction [kT]

K

Boltzmann constant [J⋅K-1]

T

Absolute temperature [K]

h0

Closest distance of approximation [1.57 Å]

α

Lumped biomass deposition coefficient [-]

152

References

2.4.8 References Bak H, Thomas ORT. 2007. Evaluation of commercial chromatographic adsorbents for the direct capture of polyclonal rabbit antibodies from clarified antiserum. J Chromatogr B 848(1):116-130. Bayramoglu G, Celik G, Arica MY. 2006. Immunoglobulin G adsorption behavior of lhistidine ligand attached and Lewis metal ions chelated affinity membranes. Colloids Surf A Physicochem Eng Asp 287(1-3):75-85. Bos R, Van der Mei HC, Busscher HJ. 1999. Physico-chemistry of initial microbial adhesive interactions--its mechanisms and methods for study. FEMS Microbiol Rev 23(2):179-230. Brant JA, Childress AE. 2002. Assessing short-range membrane-colloid interactions using surface energetics. J Membr Sci 203:257-273. Brant JA, Childress AE. 2004. Colloidal adhesion to hydrophilic membrane surfaces. J Membr Sci 241(2):235-248. Butkus MA, Grasso D. 1998. Impact of Aqueous Electrolytes on Interfacial Energy. J Colloid Interface Sci 200(1):172-181. Chaga GS. 2001. Twenty-five years of immobilized metal ion affinity chromatography: past, present and future. J Biochem Biophys Methods 49(1-3):313-334. Clemmitt RH, Chase HA. 2000. Immobilised metal affinity chromatography of [beta]galactosidase from unclarified Escherichia coli homogenates using expanded bed adsorption. J Chromatogr A 874(1):27-43. Fernandez-Lahore HM, Geilenkirchen S, Boldt K, Nagel A, Kula MR, Thommes J. 2000. The influence of cell adsorbent interactions on protein adsorption in expanded beds. J Chromatogr A 873(2):195-208. Gaberc-Porekar V, Menart V. 2001. Perspectives of immobilized-metal affinity chromatography. J Biochem Biophys Methods 49(1-3):335-360. Gallardo-Moreno AM, Gonzalez-Martin ML, Perez-Giraldo C, Garduno E, Bruque JM, Gomez-Garcia AC. 2002. Thermodynamic Analysis of Growth Temperature Dependence in the Adhesion of Candida parapsilosis to Polystyrene. Appl Environ Microbiol 68(5):2610-2613. Ganeva V, Galutzov B, Teissie J. 2004. Flow process for electroextraction of intracellular enzymes from the fission yeast, Schizosaccharomyces pombe. Biotechnol Lett 26(11):933-7.

153

References Grasso D, Subramaniam K, Butkus M, Strevett K, Bergendahl J. 2002. A review of nonDLVO interactions in environmental colloidal systems. Rev Environ Sci Biotechnol 1(1):17-38. Henriques M, Gasparetto K, Azeredo J, Oliveira R. 2002. Experimental methodology to quantify Candida albicans cell surface Hydrophobicity. Biotechnol Lett 24:1111– 1115. Klotz SA, Drutz DJ, Zajic JE. 1985. Factors Governing Adherence of Candida Species to Plastic Surfaces. Infect Immun 50(1):97-191. Ljungh Å, Wadström T. 1982. Salt aggregation test for measuring cell surface hydrophobicity of urinaryEscherichia coli. Eur J Clin Microbiol 1(6):388-393. Noronha S, Kaufman J, Shiloach J. 1999. Use of Streamline chelating for capture and purification of poly-His-tagged recombinant proteins. Bioseparation 8(1):145-151. Ottewill RH, Shaw JN. 1972. Electrophoretic studies on polystyrene lattices. J Electroanal Chem 37:133-142. Poulin F, Jacquemart R, De Crescenzo G, Jolicoeur M, Legros R. 2008. A Study of the Interaction of HEK-293 Cells with Streamline Chelating Adsorbent in Expanded Bed Operation. Biotechnol Prog 24(1):279-282. Redman JA, Walker SL, Elimelech M. 2004. Bacterial Adhesion and Transport in Porous Media: Role of the Secondary Energy Minimum. Environ Sci Technol 38(6):17771785. Tari C, Vennapusa RR, Cabrera RB, Fernandez-Lahore M. 2008. Colloid deposition experiments as a diagnostic tool for biomass attachment onto bioproduct adsorbent surfaces. J Chem Technol Biotechnol 83:183-191. Ting. Y-P, Sun G. 2000. Use of polyvinyl alcohol as a cell immobilization matrix for copper biosorption by yeast cells. J Chem Technol Biotechnol 75(7):541-546. Ueda EKM, Gout PW, Morganti L. 2003. Current and prospective applications of metal ion-protein binding. J Chromatogr A 988(1):1-23. Van Oss CJ. 1993. Acid-base interfacial interactions in aqueous media. Colloids Surf A Physicochem Eng Asp 78:1-49. Van Oss CJ. 1994. Interfacial forces in aqueous media. New York: M. Dekker. viii,440p. p. Van Oss CJ. 2003. Long-range and short-range mechanisms of hydrophobic attraction and hydrophilic repulsion in specific and aspecific interactions. J Mol Recognit 16(4):177-190.

154

References Vennapusa RR, Hunegnaw SM, Cabrera RB, Fernandez-Lahore M. 2008. Assessing adsorbent-biomass interactions during expanded bed adsorption onto ion exchangers utilizing surface energetics. J Chromatogr A 1181(1-2):9-20. Willoughby NA, Kirschner T, Smith MP, Hjorth R, Titchener-Hooker NJ. 1999. Immobilised

metal

ion

affinity

chromatography

purification

of

alcohol

dehydrogenase from baker's yeast using an expanded bed adsorption system. J Chromatogr A 840(2):195-204.

155

Results

2.5 Surface energetics to assess microbial adhesion onto fluidized chromatography adsorbents Rami Reddy Vennapusa, Sabine Binner, Rosa Cabrera, and Marcelo Fernandez-Lahore* Downstream Processing Laboratory, Jacobs University Bremen gGmbH, Campus Ring 1, D-28759, Bremen, Germany.

2.5.1 Abstract Cell-to-support interaction and cell-to-cell aggregation phenomena have been studied in a model system composed of intact yeast cells and agarose-based chromatography adsorbent surfaces. Biomass components and beaded adsorbents were characterized by contact angle determinations with three diagnostic liquids and, complementarily, by zeta potential measurements. Such experimental characterization of the interacting surfaces has allowed the calculation of interfacial free energy of interaction in aqueous media vs. distance profiles. The extent of biomass adhesion was inferred from calculations performed assuming standard chromatographic conditions, but different adsorption modes. Several stationary support / mobile phase systems were considered i.e. ion-exchange, hydrophobic interaction, and pseudo-affinity. Calculated interaction energy minima revealed marginal attraction between cells and cation-exchangers or agarose-matrix beads (U ≤ |10-20| kT) but strong attraction with anion-exchangers (U ≥ |200-1000| kT). Other systems including hydrophobic interaction and chelating beads showed intermediate energy minima values (U

≈ |40-100| kT) for interaction with biological particles. However, calculations also showed that working conditions in the presence of salt can promote cell aggregation besides cell-tosupport interaction. Predictions based on the application of the XDLVO approach were confirmed by independent experimental methods like biomass deposition experiments and laser diffraction spectroscopy. Understanding biomass attachment onto chromatographic supports can help in alleviating process limitations normally encountered during direct (primary) sequestration of bioproducts.

156

Results

2.5.2 Introduction Expanded Bed Adsorption (EBA) has been proposed as an integrative downstream processing technology allowing the direct capture of targeted species from an unclarified feedstock e.g. a cell containing fermentation broth. This unit operation has the potential to combine solids removal, product concentration, and partial purification in a single processing step. The application of EBA implies, however, that intact cell particles or cell debris present in the feedstock will interact –in a minor or larger extent- with fluidized adsorbent beads. It is already known that interaction between biomass and the adsorbent phase may lead to the development of poor system hydrodynamics and therefore, impaired sorption performance under real process conditions. Moreover, biomass interaction would result in increased buffer consumption in order to remove and wash away sticky biological particles. Biomass components can also mask binding sites thus reducing their availability to the targeted species. These phenomena i.e. decreased sorption performance and buffer consumption is detrimental to cost-efficient processing utilizing expanded bed adsorption and other direct sequestration unit operations(Fernandez-Lahore et al. 1999). The deposition of microbial cells or biomass debris is related to the physico-chemical characteristics of the cell-surface components. These surfaces are in most cases of anionic nature due to the existence of negatively charged chemical groups like phosphate, carboxylate, and sulphate moieties. However, the cell envelop can also exert hydrophobic interaction due to the presence of S-layer proteins, amphipathic polymers, and lipids. Therefore, microbial deposition onto (process) surfaces will be driven by the polymeric components of the rigid outer boundary and eventually, by the presence of cell-surface appendages (if present). On the other hand, biomass deposition will be governed by the nature (material structure) and functionality (ligand type) of the surface e.g. the structure of the chromatographic support. Previous studies on biomass-adsorbent interactions were restricted to simple diagnostic tests to determine the extent of cell –or cell debris- attachment to the desired chromatographic supports (Feuser et al. 1999). The measurement of the zeta potential has been proposed for a better understanding and prediction of biomass-adsorbent interactions during ion-exchange expanded bed adsorption (Lin et al. 2006). Such systems are obviously dominated by Coulomb-type interactions and therefore, non-electrostatic interactions are anticipated to play a minor role (Vergnault et al. 2007). Recent studies have highlighted 157

Results that biomass deposition onto other chromatographic materials like hydrophobic interaction media (HIC) or immobilized metal-ion supports can occur under real downstream process conditions (Poulin et al. 2008; Smith et al. 2002). The mentioned chromatographic modes are significant in laboratory and industrial practice. Therefore, a better understanding and control of such interaction phenomena is needed. For better understanding yeast was chosen a model biomass to evaluate adhesion on to various chromatographic adsorbent surfaces. This model organism was also utilized in number of previous studies and therefore, comparison can be better established. Additionally yeast cells are major expression host system in many biotechnology industries .This mentions that yeast cells could be good model for proof-of- concept for investigations. A more comprehensive approach to understand biomass deposition onto chromatographic supports has been proposed by utilizing principles of colloid theory to explain biomassadsorbent attachment at the local (particle) level (Vennapusa et al. 2008). This approach is based on extended DLVO calculations performed via experimentally determination of contact angles and z-potential values for the interacting surfaces or particles. The comprehensive method takes into account several types of possible interaction forces i.e. Lifshitz-Van der Waals (LW) and acid-base (AB) and, therefore, it is not limited to those purely electrostatic in nature (EL). Biomass adhesion behavior onto chromatographic beads predicted on the basis of XDLVO calculations was validated by independent biomass deposition experiments (Tari et al. 2008). The aim of this paper was to understand biomass deposition onto a variety of chromatographic systems. Cell-to-support interaction and cell-to-cell aggregation were evaluated via a surface energetics approach. Overall deposition phenomena were independently confirmed by colloid deposition experiments and laser diffraction spectroscopy, respectively.

158

Results

2.5.3 Materials and Methods 2.5.3.1 Materials Chromatographic matrices and columns were purchased from GE Health Care (Munich, Germany). α-bromonaphtalene and formamide were obtained from Fluka (Buchs, Switzerland). Water was ultra pure quality. All other chemicals were analytical grade. 2.5.3.2 Generation of biomass Yeast cells (Saccharomyces cerevisiae FY 86, wild type, haploid) were cultivated in shakeflasks, harvested at late exponential phase by centrifugation, and washed three times with 10 mM phosphate buffer solutions, as previously described (Ganeva et al. 2004). Cells were employed immediately after preparation. 2.5.3.3 Contact angle measurements Preparation of intact yeast cells for contact angle measurements was performed as described (Henriques et al. 2002). Washed cells were suspended to 10% (w/v) in 20mM phosphate buffer, pH 7 (ion-exchange buffer / Buffer A / σ = 4 mS·cm-1), Buffer A containing 1.2 M ammonium sulphate (hydrophobic interaction buffer / Buffer B / σ = 145 mS·cm-1), and 20 mM phosphate buffer adjusted to pH 7.6 with 250 mM sodium chloride (Chelating buffer / Buffer C / σ = 30 mS·cm-1). Cells were subsequently poured onto agar plates containing 10% glycerol and 2% agar-agar. The plate was allowed to dry for 24-36 hours at room temperature on a properly leveled surface free from dust. Salt crystallization was avoided. Agar plates without cell spreads were utilized as control. Contact angles were measured as per the sessile drop method (Sharma and Rao 2002) utilizing a commercial goniometric system (OCA 20, Data Physics instruments GmbH, Filderstadt, Germany). The three diagnostic liquids α-bromonaphtalene, formamide, and water were employed. All the measurements were performed in triplicate and at least 20 contact angles per samples were measured. Contact angle determination on buffer-equilibrated chromatographic beads was performed utilizing the same physicochemical conditions and experimental procedures described for cell particles. Previous to pouring onto the agar plates, matrix beads were frozen in liquid nitrogen and crushed mechanically. Crushing efficiency was assessed by microscopic 159

Results examination and particle size determination so as to assure particle fragment diameters ≤ 10

μm. Sepharose adsorbent were utilized. Square pieces of the agar supported chromatographic bead fragments were utilized for measuring contact angles. 2.5.3.4 Zeta potential determination Zeta-potential measurements were performed with a ZetaSizer Nano-ZS (Malvern instruments, Worcestershire, United Kingdom), as previously described (Vennapusa et al. 2008). Zeta-potential values were gathered employing biomass pretreated as described before (under 2.5.3.3) and utilizing the same buffers utilized for contact angle determination. Zeta-potential values for crushed and equilibrated chromatographic beads were calculated from the electrophoretic mobility data according the Smoluchowski’s equation(Ottewill and Shaw 1972). Data was gathered under identical buffer compositions as shown for biomass related determinations. 2.5.3.5 Particle size determination and cell aggregation behavior Particle size determinations and cell aggregation studies were performed by laser diffraction employing a MasterSizer 2000, hydro 2000 G (Malvern instruments, Worcestershire, United Kingdom), according to manufacturer instructions. Cell aggregation was studied as a function of buffer composition at 45 minutes contact time (Voloshin et al. 2005). Measurements were performed utilizing cell suspensions having an optical density ≈ 0.1 for better reproducibility. Visual inspection of aggregate formation was performed with a confocal laser scanning microscope, equipped with argon and helium/neon mixed gas laser with excitation wavelengths of 488 or 543 nm (LSM 510, Carl Zeiss, Oberkochen, Germany). 2.5.3.6 Biomass deposition experiments Biomass deposition experiments were performed automatically employing an ÄKTA Explorer 100 system (GE Health Care, Munich, Germany) as previously described (Tari et al. 2008). Streamline materials were packed in commercial chromatographic columns (5 mm internal diameter, 50mm length). The quality of the packing was evaluated by residence time distribution analysis employing 1% acetone as tracer. Biomass deposition

160

Results studies were done by injecting a 4 ml biomass pulse (OD ≈ 0.8 AU). Experiments were performed utilizing the above described buffer solutions. The operational flow rate was 76.4 cm·h-1. Particle breakthrough curves were obtained by monitoring the effluent suspensions at 600 nm. On the basis of such data, a lumped deposition parameter (α) was calculated according to Redman et al. (Redman et al. 2004). 2.5.3.7 Energy-distance profile calculations The total interaction energy between a colloidal particle and a solid surface can be expressed in terms of the extended DLVO theory as: XDLVO LW EL AB U mwc = U mwc + U mwc + U mwc

(1)

where UXDLVO is the total interaction energy in aqueous media, ULW is the LW interaction term, and UEL is the EL interaction term. The subscript m is utilized for the chromatographic matrix (adsorbent bead), w refers to the watery environment, and c to the colloidal (cell) particle. A third short-range (≤ 5 nm) Lewis AB term is included to account for “hydrophobic attractive” and “hydrophilic repulsive” interactions (Van Oss 2003). Material surface energy parameters (tensions) can be calculated from contact angle measurements utilizing three diagnostic liquids, according to Van Oss (Van Oss 1994). In turn, this data can be employed to evaluate the free energy of interaction between two defined surfaces (ΔGLW and ΔGAB). ΔG represents here the interaction energy per unit area between two (assumed) infinite planar surfaces bearing the properties of the adsorbent bead and the cell (interaction) or two cells (aggregation), respectively. Interaction between any of these two surfaces are evaluated at a closest distance of approximation (h0 ≈ 0.158 nm) (Bos et al. 1999). When integrated into mathematical expressions accounting the geometric constraints existing between two interacting bodies, ΔG values can be utilized to calculate the corresponding energy-distance profile (U vs. H). Details of this procedure were published (Bos et al. 1999; Vennapusa et al. 2008). ΔGmwc and ΔGcwc was calculated according to Vennapusa et al (Vennapusa et al. 2008). ΔGLW are also related to the

Hamaker constant, as follows: A = −12π h02 ΔG LW

(2)

161

Results UEL energy-distance profile can be calculated, assuming either plate-sphere or spheresphere geometry, upon experimental determination of particle zeta potential values. Zeta potential values are measured by electrophoretic mobility experiments (Vennapusa et al. 2008). Calculations were performed employing a commercial software package (GraphPad Prism, GraphPad Software Inc., San Diego, CA, USA).

162

Results

2.5.4 Results and discussions 2.5.4.1 Contact angle measurements and surface energy components The diagnostic liquids water, formamide, and α-bromonaphtalene were employed to measure contact angles onto homogeneous lawns of the materials under study i.e. intact yeast cells or crushed Sepharose beads. The sessile drop technique was employed. The utilization of the agar plate method assured that contact angle values were obtained for the mentioned materials in the hydrated state. Diagnostic liquids were chosen to have a higher surface tension than the sample materials so as to allow for stable drop formation and accurate contact angle determination. Materials were carefully equilibrated with buffers commonly utilized in practice. Contact angle determinations with three different liquids were performed so as to consider conditions prevailing in ion-exchange, hydrophobic interaction, or pseudo-affinity chromatography. Table 1 shows contact angle values obtained by performing measurements onto layered fragments (< 10 µm) of the various chromatographic supports, including Q-XL, DEAE, SP, Chelating-Cu2+, Phenyl, and agarose-base matrix beads. At neutral pH, adsorbent contact angle values were ≈ 7-10 for water and ≈ 10-13 for formamide, when considering the base-matrix, chelating matrix, the anion-exchanger DEAE-Sepharose, and the cation-exchanger SP-Sepharose. On the other hand, for the mentioned adsorbents the contact angle values with α-bromonaphtalene were ≈ 39-45. Overall, these values indicate the very hydrophilic nature of such materials. For the composite support XL-Q Sepharose an increase in the contact angle value for αbromonaphtalene was noticed, which might indicate an even increased hydrophilic character due to the presence of superficial Dextran chains. As expected due to the presence of hydrophobic ligands, the Phenyl-Sepharose material showed decreased contact angle values with α-bromonaphtalene. These data indicate that similarities and differences between different supports can be actually observed on the basis of contact angle determinations. Published work have shown that the addition of salt can actually influence the contact angle values -and correspondingly the surface free energy componentsobtained for some types of mineral particles (Karagüzel et al. 2005).

163

Results Table 1: Contact angle measurements for agarose-based beaded supports. Determinations were performed in buffers which composition represents standard operational conditions. Support type

Contact angle (θ) (Degrees) Water

Formamide

Q-XL

12.0 ± 1.3

14.0 ± 2.0

αBromonaphtalene 52.0 ± 1.1

DEAE

9.6 ± 3.1

13.0 ± 2.1

41.0 ± 1.9

6.7 ± 3.3

13.0 ± 1.9

39.0 ± 1.5

10.2 ± 2.4

12.1 ± 0.9

45.0 ± 3.0

Phenyl

7.0 ± 1.0

10.0 ± 1.7

24.0 ± 2.5

Agarose bead

9.5 ± 2.1

10.0 ± 1.7

44.0 ± 1.0

SP Chelating-Cu

2+

Table 2 summarizes the contact angle values obtained after measurements performed onto homogeneous layers of intact yeast cells at pH close to 7. The agar plate technique utilized allowed the measurement of contact angles under the assumption that only bound water is present in the sample materials. Contact angles observed for water (≈ 10-15) and formamide (≈12-14) were similar when three different buffers were utilized. Such buffers corresponded to typical compositions found in ion-exchange, hydrophobic interaction, or pseudo-affinity chromatography. On the contrary, contact angles measured with the apolar liquid (α-bromonaphtalene) were able to distinguish between the various conditions tested. Notably, α-bromonaphtalene contact angle values decreased with buffer conductivity (54 at 4 mS·cm-1, 49 at 30 mS·cm-1, and 33 at 145 mS·cm-1). Therefore, changes in cell surface as a function of the chemical environment provided by the mobile phases can be detected utilizing this methodology. Further, these results indicates that a non-polar liquid can be employed to discriminate between biomass types or conditions in relation to surface hydrophobic character (Butkus and Grasso 1998).

164

Results Table 2: Contact angle measurements for yeast cells. Determinations were performed under conditions normally encountered in ion-exchange, hydrophobic interaction, and chelating systems. Yeast cell suspension

Contact angle (θ) (Degrees) Water

Formamide

αBromonaphtalene

IEX-type buffer

15.0 ± 2.2

14.0 ± 1.0

54.0 ± 1.8

HIC-type buffer

10.0 ± 0.5

12.0 ± 1.1

33.0 ± 2.4

Chelating-type buffer

11.7 ± 2.4

13.7 ± 1.7

49.3 ± 1.0

Experimental contact angle determinations were utilized to calculate surface energy parameters for chromatographic media and biomass particles according to the acid-base approach (Bos et al. 1999). Table 3 depicts the surface energy components (γ) calculated for various chromatographic supports under typical buffer conditions ion-exchange buffer / 20mM phosphate buffer pH 7.0 (Buffer A ), Hydrophobic interaction buffer (Buffer A containing 1.2 M ammonium sulphate and Chelating buffer (20 mM phosphate buffer adjusted to pH 7.6 with 250 mM sodium chloride and 1mM imidazole. These supports can be distinguished on the basis of their characteristic γLW and γ+ (electron acceptor component of the AB surface tension) parameters. According to such parameters, the chromatographic supports can be ordered as follows: γLW → [Phenyl] > [Chelating] ≈ [Agarose-matrix] ≈ [DEAE ≈ SP] > [Q-XL] γ+ → [Q-XL] >> [Agarose-matrix] ≈ [Chelating] > [DEAE ≈ SP] > [Phenyl] As a whole these results indicate that each of the materials studied possess characteristic surface energetic properties that are experimentally accessible via contact angle measurements with three diagnostic liquids. For example, Phenyl supports presented high γLW values (∼ 39 mJ·m-2) but low γ+ values (1.3 mJ·m-2) in comparison with the agarosebackbone bead. On the contrary, the composite material Q-XL showed the inverse tendency i.e. high γ+ values (3.9 mJ·m-2) but low γLW values (∼ 29 mJ·m-2).

165

Results Table 3: Surface energy parameters for beaded chromatographic supports calculated from contact angle measurements under standard buffer compositions. Surface energy parameters [mJ·m-2]

Support type γLW

γ+

γ-

γAB

γTOT

ΔGsws

Q-XL

28.9

3.9

53.2

28.8

57.8

+26.6

DEAE

34.1

2.3

54.5

22.3

56.7

+30.7

SP

35.0

2.0

55.7

21.1

56.4

+31.9

Chelating- CU2+

32.3

3.0

53.7

25.0

57.4

+26.3

Phenyl

39.3

1.3

55.1

16.9

56.3

+26.8

Agarose bead

32.8

2.9

53.6

24.9

57.7

+28.1

Modification in the surface characteristics of intact yeast particles were also noticed upon observation of the values taken by the γLW and γ+ parameters (Table 4), according to the chemical environment provided by the proposed mobile phases: γLW → [HIC Buffer] >> [Chelating Buffer] > [IEX Buffer] γ+ → [IEX Buffer] > [Chelating Buffer] >> [HIC Buffer] These results are indicative of an increased hydrophobic character for cells in the presence of high concentrations of ammonium sulphate (HIC Buffer) and indicative of an increased Lewis-acid character for cells in dilute phosphate buffer (IEX Buffer). Summarizing, surface energy parameters are able to characterize the various chromatographic systems under consideration. As observed from Table 3 and 4, the parameter ΔGsws took always values +24-31 mJ·m-2 reflecting the hydrophilic nature of the yeast cells and the chromatographic beads. For comparison, the ΔGsws of hydrophilic repulsion for Dextran T-150 is +41.2 mJ·m-2 (Van Oss 2003).

166

Results Table 4: Surface energy parameters for yeast cells calculated from contact angle measurements, under conditions provided by typical chromatographic mobile phases. Yeast cell suspension

Surface energy parameters [mJ·m-2] γLW

γ+

γ-

γAB

γTOT

ΔGsws

IEX-type buffer

27.9

4.4

51.5

30.1

58.3

+24.3

HIC-type buffer

37.9

1.5

54.8

18.2

56.0

+27.0

30.3

3.5

53.4

27.3

57.6

+25.9

Chelating-type buffer

2.5.4.2 Interfacial free energy of interaction and energy-distance profiles Interaction between biomass particles and chromatographic beads can be understood by calculating interfacial free energy (U) vs. distance (H) profiles. These calculations are based on the experimental determination of contact angles with three diagnostic liquids and the additional information gathered from zeta potential determinations. Table 5 depicts the interfacial free energy of interaction between a biomass particle and chromatographic adsorbent surfaces in aqueous media at closest distance of approximation (1.57 Å). Furthermore, this table also gives information on zeta-potential values for the mentioned adsorbents. Hydrophobic interaction and immobilized-metal affinity chromatography are operated in a context characterized by an increased salt concentration (high ionic strength and conductivity) in the mobile phase, as well as, by uncharged beaded adsorbents. Therefore, it is expected that the information provided by contact angle determination will be more relevant to understand cell-to-support interactions than the information provided via zpotential determinations. This situation is radically different from the case of the ionexchangers where, due to the low conductivity of the mobile phases and the charged nature of the adsorbents, zeta-potential has been established as a parameter describing biomass deposition onto process supports (Lin et al. 2006).

167

Results The Hamaker constant (A) for the interaction chromatographic systems under study were calculated from ΔGLW according to Equation (2). A values, according to the adsorption mode, can be ordered as follows: HIC (1.1 kT) > IMAC (0.40 kT) > IEX (0.34 kT) Table 5: Interfacial free energy of interaction between intact yeast cells and agarose beaded supports, at closest distance of approximation. Calculations were performed assuming interaction under typical bioprocess conditions. Zeta-potential values for the chromatographic beads are provided. ΔG [mJ·m-2]

Support

Zeta potential [mV]

ΔGLW

ΔGAB

Q-XL

-0.9

+26.3

+20.0

DEAE

-1.4

+28.7

+15.0

-1.5

+29.7

-30.0

-1.7

+29.6

-8.0

Phenyl

-4.8

+36.5

-0.1

Agarose bead

-1.3

+27.6

-2.0

SP Chelating- CU

2+

Figure 1: Total free energy of interaction as function of distance between intact yeast cells and several chromatographic supports, under conditions provided by commonly utilizing mobile phases. (—) Q XL, (—) DEAE, (—) Phenyl, (—) IMAC-Cu+2, (—) Base, (—) SP.

168

Results Figure 1 depicts interaction energy (U) vs. distance (H) profiles calculated according to the XDLVO approach, utilizing the data provided before (Table 5). Calculated secondary energy minima occurring at ≈ 5-10 nm upon interaction of a yeast cell and the adsorbent surface were observed. Calculations assumed sphere-to-plate geometry. The depth of such energy minima shifted from low to moderate values ≈ -5-20 kT in dilute buffer solutions for the agarose-base material or the cation-exchanger down to intermediate values ≈ -40-120

kT at high salt concentrations for Chelating and Phenyl supports. Anion exchangers showed, in 20 mM phosphate buffer, energy pockets in the range -200-400 kT. These values increased to > |1000| kT in 10 mM phosphate buffer. The information provided by analyzing U vs. H profiles is in full agreement with previous experimental data concerning biomass interaction onto expanded bed adsorbents (Fernandez-Lahore et al. 1999). Application of the extended DLVO approach is justified since due to the very polar nature of the buffer solutions where cell-adsorbent interactions take place; these interactions are known to be strongly influenced by polar Lewis acid-base (AB) or electron-acceptor / electron-donor forces. Contributions by electric double layer (EL) forces and particularly contributions by apolar Lifshitz-van der Waals (LW) forces are also expected to occur. The extended DLVO approach has served to explain the behavior of many other colloidal systems. Brandt and Childress have demonstrated that short-range interactions between synthetic membranes and bio-colloids can be better explained by taking into consideration the role of AB forces(Brant and Childress 2002). Van Oss and coworkers have studied the stability of a thixotropic suspension of 2 μm hectorite particles and concluded that Lewis acid-base interactions play a key role in the coagulation dynamics of such system (Grasso et al. 2002). 2.5.4.3 Interfacial free energy of aggregation and energy-distance profiles Contact angle and zeta potential determinations, as reported in this work have been utilized to calculate energy vs. distance profiles between two intact yeast cells. Table 6 shows

ΔGLW, ΔGAB, and zeta-potential determinations for yeast cells in various buffers. Sphere-tosphere geometry was assumed. These XDLVO calculations can be observed in Figure 2 and they have indicated that cell-to-cell aggregation might have the chance to occur under the conditions provided by the hydrophobic interaction buffer e.g. at high conductivity values. In the later case, a secondary energy minima (∼ -30 kT) can be anticipated at a distance of 5 169

Results nm. Cell clumping in the presence of salt was confirmed by confocal microscopy and laser diffraction studies (Data not shown). On the other hand, moderate energy minima were observed for aggregation of cells suspended in chelating buffer (∼ -8 kT @ 5 nm). However, moderate cell clumping could exist in buffers containing 250 mM sodium chloride i.e. IMAC buffer. The secondary interfacial free energy well between to yeast cell was of a very limited depth in the case of dilute phosphate buffer (∼ -4 kT @ 5 nm). These low attractive forces would be easily disrupted by the hydrodynamic stress normally encountered in most processing schemes. Therefore, cell-to-cell aggregation would rarely take place when biological particles are suspended in dilute buffer solution due to higher propensity of the cell complex to be disrupted by shear drag. The mentioned aggregation behavior of yeast particles suspended in buffers of various chemical compositions is also reflected by the values of the corresponding Hamaker constants, as follows: HIC (2.0 kT) > IMAC (0.65 kT) > IEX (0.34 kT) and therefore, LW forces can be considered of importance regarding aggregation phenomena, particularly for cells suspended in high conductivity buffers. Table 6: Interfacial free energy of aggregation between yeast cells at closest distance of approximation. Calculations were performed assuming standard chromatographic conditions. Zeta potential values are provided. ΔG [mJ·m-2]

Yeast cells

Zeta potentials [mV]

ΔGLW

ΔGAB

IEX-type buffer

-1.5

+25.1

-18.0

HIC-type buffer

-8.8

+35.9

-0.1

-2.8

+28.7

-8.0

Chelating-type buffer

170

Results

Figure 2: Interfacial free energy as function of distance between two yeast cells under mobile phases having typical chemical compositions for ion-exchange, hydrophobic interaction, and immobilised-metal ion affinity chromatography. (—) IEX buffer, (—) Chelating-Cu2+, (—) HIC.

2.5.4.4 Biomass deposition experiments Biomass deposition experiments were performed to evaluate overall yeast cells deposition onto ion-exchange, hydrophobic interaction, and pseudo-affinity chromatographic supports. This allowed an independent experimental verification of the predictions made on the basis of energy vs. distance calculations. Interaction phenomena taken place in each of these chromatographic systems are verified utilizing mobile phase standard compositions i.e. IEX buffer, IMAC buffer and HIC buffer. Figure 3 depict the cell effluent profiles measured for the various chromatographic systems i.e. diverse combinations of solid and mobile phases. Biomass deposition experiments showed a characteristic cell effluent profile for each of the systems under study. Particle retention was extremely high for Q-XL and DEAE materials, moderate for Phenyl and Chelating supports, and very low for the cation-exchanger and the base matrix. Biomass

171

Results deposition behavior was reflected by an “attachment efficiency” number (α), a lump parameter describing such phenomena. Qualitative and quantitative evaluation of cell deposition experiments revealed several underlying phenomena like cell-to-support attachment (interaction), prevention of cell depositions by already deposited biomass particles (blocking), and cell-to-cell ripening (aggregation). Cell-to-cell aggregation might represent and important mechanism promoting overall cell attachment during biomass deposition experiments. Therefore, increased values for the lumped α parameter might indicate not only stronger cell-to-support interaction but enhanced cell-to-cell aggregation. Consequently, results from biomass deposition experiments will reveal conditions prevailing during real process performance where both interaction and aggregation phenomena can coexist.

Figure 3: Biomass deposition experiments with intact yeast cells onto several process surfaces. Mobile phases of standard chemical compositions were employed. (z) QXL, (z) DEAE, (z) Phenyl, (z) IMAC Cu+2, (z) Base, (z) SP.

172

Results Figure 4 shows the correlation between the attachment efficiency parameter and the depth of the secondary free energy of interaction between a cell particle and a chromatographic bead. Points corresponding to the various chromatographic systems can be observed. Three main groups can be clearly distinguished: a) Group I characterized by α ≤ 0.15 and U ≤ |10-20| kT. This group contains systems were both cell-to-support interaction and cell-to-cell aggregation is negligible and therefore, overall biomass deposition phenomena can be neglected [CEX]. Underlying interaction mechanisms within Group I are repulsive EL and AB forces with moderate attractive LW forces. Cell-to-cell forces are predominantly repulsive mainly due to AB and EL components. b) Group II characterized by 0.2 ≤ α ≤ 0.4 and U ≈ |50-100| kT. This group is composed by systems were mixed phenomena occur i.e. there is a degree of biomass deposition onto the solid phase but additional cell entrapment can exists due to aggregation [Phenyl and Chelating]. Underlying interaction mechanisms within Group II attractive LW forces, moderately repulsive AB forces, and a negligible EL component. Cell-to-cell forces are predominantly attractive mainly due to LW forces. c) Group III characterized by α > 0.90 and U > |200-1000| kT. This group represents systems with strong cell-to-support interaction, mainly mediated by electrostatic attraction (EL) between the interacting bodies. AB and LW forces play a minor role in this case.

For expanded bed adsorption (EBA) which is the more studied system concerning biomass deposition evidence exits in the open literature regarding problem-free operation was consistently reported for CEX (Group I). On the contrary, biomass interference with appropriate sorption performance was noticed for AEX (Group III). Information regarding Group II chromatographic systems is scarce. However, recent reports also indicate that such process combinations are suffering from biomass compatibility limitation (Poulin et al. 2008; Smith et al. 2002).

173

Results

Figure 4: Correlation between the depth of energy pocket and lump factor alpha for several process systems.( z) Cation exchangers, (z) Base matrix, (z) Chelating Cu2+, (z) Phenyl,

[(z) DEAE, (z) Q-XL in 10mM PO4 buffer].

2.5.5 Conclusions A comprehensive approach to understand biomass attachment onto chromatography adsorbent surfaces with special emphasis on commonly utilized chromatographic systems have included several interaction forces, according to the XDLVO approach. These calculations were based on the experimental determination of contact angles with three diagnostic liquids and the additional information gathered from zeta-potential determinations. Qualitative and quantitative evaluation of cell adhesion experiments have revealed several underlying phenomena like cell-to-support sticking, prevention of cell depositions by already deposited biomass particles (blocking), and cell-to-cell aggregation (ripening). A correlation between the depth of the interaction energy pockets and the deposition coefficient values was established and three distinct groups defined.

174

Results In summary, it has been shown that both cell-to-adsorbent (interaction) and cell-to-cell (aggregation) phenomena are involved in biomass deposition effects onto chromatographic materials, under conditions normally encountered in practice. Interaction and aggregation was inferred from XDLVO calculations on the basis of contact angle and zeta potential measurements. Moreover, experimental confirmation was obtained by independent methods like biomass deposition experiments and laser diffraction spectrometry.

2.5.6 Acknowledgements RRVP gratefully acknowledges a doctoral fellowship from Jacobs University.

2.5.7 Nomenclature AB

Acid-Base

DEAE

Diethylaminoethyl [-]

DLVO

Classical DLVO theory (Derjaguin, Landau, Verwey and Overbeek)

EBA

Expanded Bed Adsorption

EL

Electrostatic

H

Distance between surfaces, measured from outer edge [m]

HIC

Hydrophobic interaction chromatography

IC

Intact yeast cell particles

IEX

Ion-exchange chrom. (AEX, anion exchange; CEX cation exchange)

IDA-Cu2+

Iminodiacetic-immobilized cupper ions (IMAC or Chelating)

LW

Lifshitz-Van der Waals

R

Radius of the particle [m]

SP

Sulphopropyl [-]

T

Absolute temperature [K]

U

Interfacial energy of interaction [kT]

XDLVO

Extended DLVO theory, according to Van Oss

A

Hamaker constant [kT]

γ LW

Apolar or Lifshitz-van der Waals component of surface tension [mJ⋅m-2]

175

Results

γ AB

Polar or acid–base component of surface tension [mJ⋅m-2]

γ

Electron-donor component of surface tension (Lewis base) [mJ⋅m-2]

−

γ+

Electron-acceptor component of surface tension (Lewis acid) [mJ⋅m-2]

ζ

Zeta potential [mV]

h0

Closest distance of approximation [1.57 Å]

α

Lumped biomass attachment coefficient [-]

ΔG

Interfacial free energy @ 1.57 Å approach [mJ⋅m-2]

176

References

2.5.8 Referances Bos R, Van der Mei HC, Busscher HJ. 1999. Physico-chemistry of initial microbial adhesive interactions--its mechanisms and methods for study. FEMS Microbiol Rev 23(2):179-230. Brant JA, Childress AE. 2002. Assessing short-range membrane-colloid interactions using surface energetics. J Membr Sci 203:257-273. Butkus MA, Grasso D. 1998. Impact of Aqueous Electrolytes on Interfacial Energy. J Colloid Interface Sci 200(1):172-181. Fernandez-Lahore HM, Kleef R, Kula M, Thommes J. 1999. The influence of complex biological feedstock on the fluidization and bed stability in expanded bed adsorption. Biotechnol Bioeng 64(4):484-96. Feuser J, Walter J, Kula MR, Thommes J. 1999. Cell/adsorbent interactions in expanded bed adsorption of proteins. Bioseparation 8(1-5):99-109. Ganeva V, Galutzov B, Teissie J. 2004. Flow process for electroextraction of intracellular enzymes from the fission yeast, Schizosaccharomyces pombe. Biotechnol Lett 26(11):933-7. Grasso D, Subramaniam K, Butkus M, Strevett K, Bergendahl J. 2002. A review of nonDLVO interactions in environmental colloidal systems. Rev Environ Sci Biotechnol 1(1):17-38. Henriques M, Gasparetto K, Azeredo J, Oliveira R. 2002. Experimental methodology to quantify Candida albicans cell surface Hydrophobicity. Biotechnol Lett 24:1111– 1115. Karagüzel C, Can MF, Sönmez E, Celik MS. 2005. Effect of electrolyte on surface free energy components of feldspar minerals using thin-layer wicking method. J. Colloid Interface Sci. 285(1):192-200. Lin DQ, Zhong LN, Yao SJ. 2006. Zeta potential as a diagnostic tool to evaluate the biomass electrostatic adhesion during ion-exchange expanded bed application. Biotechnol Bioeng 95(1):185-91. Ottewill RH, Shaw JN. 1972. Electrophoretic studies on polystyrene lattices. J Electroanal Chem 37:133-142. Poulin F, Jacquemart R, DeCrescenzo G, Jolicoeur M, Legros R. 2008. A Study of the Interaction of HEK-293 Cells with Streamline Chelating Adsorbent in Expanded Bed Operation. Biotechnol Prog 24(1):279-282.

177

References Redman JA, Walker SL, Elimelech M. 2004. Bacterial Adhesion and Transport in Porous Media: Role of the Secondary Energy Minimum. Environ Sci Technol 38(6):17771785. Sharma PK, Rao KH. 2002. Analysis of different approaches for evaluation of surface energy of microbial cells by contact angle goniometry. Adv Colloid Interface Sci 98(3):341-463. Smith MP, Bulmer MA, Hjorth R, Titchener-Hooker NJ. 2002. Hydrophobic interaction ligand selection and scale-up of an expanded bed separation of an intracellular enzyme from Saccharomyces cerevisiae. J Chromatogr A 968(1-2):121-128. Tari C, Vennapusa RR, Cabrera RB, Fernandez-Lahore M. 2008. Colloid deposition experiments as a diagnostic tool for biomass attachment onto bioproduct adsorbent surfaces. J Chem Technol Biotechnol 83:183-191. Van Oss CJ. 1994. Interfacial forces in aqueous media. New York: M. Dekker. viii, 440p. p. Van Oss CJ. 2003. Long-range and short-range mechanisms of hydrophobic attraction and hydrophilic repulsion in specific and aspecific interactions. J Mol Recognit 16(4):177-190. Vennapusa RR, Hunegnaw SM, Cabrera RB, Fernandez-Lahore M. 2008. Assessing adsorbent-biomass interactions during expanded bed adsorption onto ion exchangers utilizing surface energetics. J Chromatogr A 1181(1-2):9-20. Vergnault H, Willemot R-M, Mercier-Bonin M. 2007. Non-electrostatic interactions between cultured Saccharomyces cerevisiae yeast cells and adsorbent beads in expanded bed adsorption: Influence of cell wall properties. Process Biochem 42(2):244-251. Voloshin S, Shleeva M, Syroeshkin A, Kaprelyants A. 2005. The Role of Intercellular Contacts in the Initiation of Growth and in the Development of a Transiently Nonculturable State by Cultures of Rhodococcus rhodochrous Grown in Poor Media. Microbiology 74:420-427.

178

Results

2.6 The effect of chemical additives on biomass deposition onto beaded chromatographic supports Rami Reddy Vennapusa and Marcelo Fernandez-Lahore* Downstream Processing Laboratory, School of Engineering and Science, Jacobs University Bremen gGmbH, Campus Ring 1, D-28759, Bremen, Germany.

2.6.1 Abstract Common limitations encountered during the direct recovery of bioproducts from an unclarified feedstock are related to the presence of biomass in such processing systems. Biomass-related effects can be described as biomass-to-support deposition and cell-to-cell aggregation. In this work, a number of chemical additives were screened for their ability to inhibit either biomass deposition, cell aggregation, or a combination of both effects. Several interacting pairs were screened. These were composed of i. a commercial chromatographic matrix harbouring a variety of ligand types and ii. intact yeast cells -as a model biomass type. Studies were performed on the basis of partitioning tests and colloid deposition experiments. Results indicated that the incorporation of the synthetic polymer PVP 360 into the mobile phase has alleviated biomass deposition onto weak-anion exchanger beads by a factor of ≈3. This behaviour correlated well with calculations performed according to the XDLVO approach: the secondary (interaction) free energy pockets decreased from -230 kT to -100 kT in the absence and in the presence of PVP 360, respectively. Experiments performed in parallel demonstrated that total binding capacity for the model protein (BSA) decreased minimally, from 33.6 to 32.4 mg/ml. Other combinations of additives and adsorbents were tested. However, no solution chemistry was able to inhibit biomass deposition onto strong (composite) ion exchangers. Moreover, yeast cells deposition was only marginally decreased when hydrophobic interaction and pseudo-affinity supports were explored. The utilization of non-toxic polymers could help to avoid detrimental biomass deposition during expanded bed adsorption of bioproducts and other direct contact sequestration methods.

179

Results

2.6.2 Introduction Current bottleneck in the downstream processing of biological products can be alleviated by application of direct sequestration methods. For example, the utilization of expanded bed adsorption (EBA) may play an important role during product primary capture. This process strategy permits simultaneous solids separation and product concentration and (partial) purification. Therefore, an integrative technology presents a clear benefit as it reduces the number of process steps and contributes considerably to cost reduction, by saving on process times and capital demands (Anspach et al. 1999). To deliver appropriate sorption performance expanded bed systems have to allow for the formation of a stable or perfectly classified fluidized bed, even in the presence of a turbid feedstock. However, this is often not the case. It was early reported (Fernandez-Lahore et al. 1999; Feuser et al. 1999) that interactions between the biomass components and the fluidized chromatographic adsorbents may disturb the otherwise stable expansion of the bed, by changing its hydrodynamic characteristics. Moreover, biomass deposition can reduce the life expectancy of the (costly) matrix due to adsorbent fouling and due to the harsh regeneration conditions subsequently required to release the bound cellular material (Dainiak et al. 2002; Feuser et al. 1999). Deteriorated process performance in expanded bed systems generates an increased processing time and capital investment (Curbelo et al. 2003). Therefore, the biofouling of chromatographic supports is a significant technical challenge which has to be better understood to overcome the many limitations that have been addressed in the last years.

The deposition of microbial cells or biomass debris is related to the physico-chemical characteristics of the cell-surface components. These surfaces are in most cases of anionic nature due to the existence of negatively charged chemical groups like phosphate, carboxylate, and sulphate moieties. However, the cell envelop can also exert hydrophobic interaction due to the presence of S-layer proteins, amphipathic polymers, and lipids. Therefore, microbial deposition onto (process) surfaces will be driven by the polymeric components of the rigid outer boundary and eventually, by the presence of cell-surface appendages (if present). On the other hand, biomass deposition will be governed by the nature (material structure) and functionality (ligand type) of the surface e.g. the structure of the chromatographic support.

180

Results A semi-quantitative analysis of biomass interactions with several biomass-adsorbent pairs was performed by Feuser et al (Feuser et al. 1999). Since then, several studies were devoted to determine the importance and extent of biomass effects on the sorption performance of fluidized beads (Fernandez-Lahore et al. 1999; Fernandez-Lahore et al. 2001). In order to overcome such limitations a methodological design approach has been proposed to determine appropriate operational windows so as to reduce biomass adsorbent interactions to a minimum (Lin et al. 2001). Our group has recently made an attempt to understand – at the local level - the deposition of biomass particles onto several process surfaces. This novel approach might offer a more universal approach and valuable information to guide process and material design (Vennapusa et al. 2008). The shielding of chromatographic support surfaces with polymeric layers proved to inhibit non-specific interactions and therefore to be a helpful strategy to optimize separation methods, like high performance chromatography or capillary electrophoresis (Desilets et al. 1991; Petro and Berek 1993; Santarelli et al. 1988; Schomburg 1991). A similar strategy was attempted during expanded bed adsorption by covering fluidised beads with polyelectrolyte or agarose to reduce biomass interference (Dainiak et al. 2002; Viloria-Cols et al. 2004). Other methods implemented to reduce non-specific interaction of biological particles included a thermal treatment of the crude feedstock before contacting with the solid phase (Ng et al. 2007). However, biomass deposition or cell aggregation is still observed in many adsorbent-biomass systems e.g. with anion-exchangers, hydrophobic interaction beads and (some) pseudo-affinity supports (Fernandez-Lahore et al. 2000; Poulin et al. 2008; Silvino Dos Santos et al. 2002; Smith et al. 2002). There is room enough for investigations concerning the potential effects of solution chemistry changes on biomass deposition in bioprocessing systems. The mechanistic understanding of transport and deposition of microbial cell onto process surfaces has significant interest in various bioprocess situations. Traditionally, microbial deposition has been studied employing packed-beds of collector particles. A population of biological particles is introduced into such systems and the suspended biomass effluent is monitored as a function of process time. This type of experiments can provide useful and quantitative information when assessing factors like cell size and shape, microorganisms strain, growth phase, bead size, surface coatings, fluid velocity, and ionic strength on cell deposition onto process media (Tari et al. 2008). A common approach to evaluate biomass 181

Results deposition in laboratory packed-bed experiments employs the “clean-bed” filtration model (CBFM). In this case, mass transport phenomena are accounted by the “single-collector contact efficiency” (η0) while the physicochemical phenomena related to biomass attachment are reflected by the “attachment efficiency parameter” (α). This work has gathered information on the effect of several chemical additives, which were incorporated into chromatographic mobile phases, on biomass deposition onto chromatographic adsorbents. Additives belong to the group of synthetic polymers, nonionic surfactants, neutral detergents, and salts. Yeast cells were utilized as model biomass particles. Various combinations of commercial adsorbents and additives were screened for cell deposition via partition and biomass deposition experiments. The extended DerjaguinLandau-Verwey-Overbeek (XDLVO) theory was employed to explain the observed cell deposition behaviour.

182

Results

2.6.3 Materials and Methods 2.6.3.1 Materials Chromatography adsorbents and columns were purchased from GE Healthcare, Munich, Germany. Solvents utilised for contact angle measurements: α-bromonaphtalene (99% purity) and formamide (99.5% purity), were obtained from Fluka, Buchs, Switzerland. Water was ultrapure quality. Polyethylene glycol (PEG 3350), polyvinyl alcohol (10 kDa; Product

number

8136),

polyvinyl

pyrrolidone

(PVP

10

and

PVP

360),

Polyoxyethylenlaurylether (Brij 35 and Brij 58) were obtained from Sigma-Aldrich Chemie GmbH, Steinheim, Germnay. Tween 20, Pluronic F68, Nonidet P40, Tween 20 and Triton X100 were from AppliChem GmbH, Darmstadt, Germany. Sodium polyphosphate (NaPP) and sodium fluoride (NaF) were obtained from Riedel-de Haën, Seelze, Germany. All other chemicals were of analytical grade. 2.6.3.2 Generation of biomass

Saccharomyces cerevisiae wild strain FY 86, haploid, was obtained from Dr. V. Ganeva (Sofia University, Bulgaria). The strain was maintained on agar plates made from yeast extract 10 g/l, soy peptone 20 g/l and agar 20 g/l with D-glucose 20 g/l as additional carbon source. Yeast cells were grown on YPD medium [1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) glucose] utilising 300 ml cotton-plugged-conical flasks on a rotary shaker at 30°C. The culture volume was 100 ml and the shaker speed 150 rpm. Growth was monitored turbidimetrically at 600 nm. After reaching exponential phase (OD600 = 1.9), cells were collected by centrifugation at 2000g, washed twice with 10 mM phosphate buffer (pH 7.6) and re-suspended to give 7.5 × 108 cells·ml−1 (24 mg cell dry wt per ml) (Ganeva et al. 2004). 2.6.3.3 Physiochemical characterization of particles 2.6.3.4 Contact angle measurements Contact angles were measured as per the sessile drop method utilizing a commercial goniometric system (OCA 20, Data Physics instruments GmbH, Filderstadt, Germany). Three diagnostic liquids e.g. α-bromonaphtalene, formamide, and water were employed (Bos et al. 1999). Details of the experimental procedure, as applied for biomass and crushed chromatographic beads, were published elsewhere (Vennapusa et al. 2008).

183

Results 2.6.3.5 Zeta potential determinations Particle zeta-potential values were measured with a Zetasizer Nano ZS from Malvern Instruments (Worcestershire, United Kingdom). Fragmented Sepharose particles were utilized instead of Streamline beads due to their lower density and to avoid sedimentation during measurements. Before performing the measurement, particles were contacted with 20 mM sodium phosphate buffer at pH 7.6 for 2 h and further diluted to appropriate particle count (~200 particles total count). Zeta potential measurements were also performed on particles which were contacted with 1% solution of PVP 360 and then extensively washed with phosphate buffer. Zeta potentials were calculated from the electrophoretic mobility data according to the Smoluchowski’s equation (Ottewill and Shaw 1972). All the measurements were performed in triplicate. 2.6.3.6 Deposition of yeast cells (Partition experiments) Deposition of yeast cells onto chromatographic beads was studied by partition experiments. These experiments were performed in glass flasks (4 cm height, 1.5 cm diameter) with plastic caps. Vacuum dewatered chromatographic beads (0.5 g) were contacted with a cell suspension (2.0 ml; 0.03% dry weight) under gentle orbital stirring. The optical density of the cell suspension remaining in the supernatant was evaluated by absorbance at 600 nm. Cell number was calculated according to the following expression:

y = 0.3363x 2 + 0.1155 x

Equation 1

where y is the concentration of yeast of cells (% w/v wet basis) and x is the OD@600 nm. Cell suspensions having an optical density higher than 1.0 were diluted before photometry. Samples were taken after 3 h to evaluate total cell deposition (Fernandez-Lahore et al. 2000). Results were expressed as a Cell Partition Index (CPI) which was calculated according to:

CPI =

Cf Ci

Equation 2

where Cf is the final concentration of cells (t = 3h) and Ci is the initial concentration of the yeast cells (t = 0) in the supernatant.

184

Results 2.6.3.7 Protein adsorption in the presence of cells Partition tests were also employed to evaluate protein binding in the presence of cells. Bovine serum albumin (BSA) was utilized as a model protein when anion-exchangers were under study. Total BSA load was 20-30 mg per vial. Experiments were performed in 20 mM phosphate buffer (pH 7.6) and in the same buffer containing suspended yeast cells (5 % wet weight). Protein binding was evaluated after 3 h and the results expressed as an “affinity number”, which is defined as follows:

Affinity number = 1-

Cf

Equation 3

Ci

where Cf is the final concentration of protein (t = 3h) and Ci is the initial concentration of protein (t = 0) in the supernatant. 2.6.3.8 Biomass deposition experiments Biomass deposition experiments were performed employing an ÄKTA Explorer 100 system (GE Healthcare, Munich, Germany), under optimized conditions, as previously described (Tari et al. 2008). Streamline materials were utilized as collector beads. Several chromatographic support types were tested including anion-exchangers, cation-exchangers, hydrophobic interaction, and pseudo-affinity materials. Buffer compositions were adjusted according to the corresponding chromatographic mode: a) 20mM phosphate buffer, pH 7.6, for ion exchange chromatography (Buffer A), b) Buffer A containing 1.2 M ammonium sulphate for hydrophobic interaction chromatography (Buffer B), and c) Buffer A containing 0.25 M sodium chloride and 1 mM imidazol for Chelating systems. Particle breakthrough curves were obtained by monitoring the effluent suspensions at 600 nm. On the basis of such data, the biomass deposition parameter (α) was calculated as previously described (Redman et al. 2004). 2.6.3.9 Estimation of viscosity The viscosity of PVP 360 aqueous solutions at 25°C was estimated according to a previously published correlation (Yeh et al. 1998):

μ = 0.89 ×10 −3 e 0.875C

Equation 4

185

Results where μ is the viscosity of the polymer solution [Pa·s] and C is the concentration of PVP 360 [wt. %].

186

Results

2.6.4 Results and Discussions 2.6.4.1 The potential of additives to influence cell attachment Certain chemical components may have the ability to modify the interactions between microbial cells and chromatographic beads –in aqueous media- by promoting changes in the free interfacial forces / free energy between bodies: a) Ionic compounds can alter microbial deposition and transport through surface charge modification (Brown and Jaffe 2001). Polyphosphates are highly negative charged chemicals which were shown to decrease microbial adhesion to soils and synthetic membranes. These compounds find applications as microbial dispersants and to stabilise suspensions of mineral particles. Sodium polyphosphate [NaPP / n = 17] compounds can reduce the zeta-potential of microbial particles (Papo et al. 2002; Sharma et al. 1985). b) Polyvinylpyrrolidone [PVP] is a non-ionic polymer which has been shown to adsorb onto oxide-surfaces through an acid-base interaction i.e. surface hydroxyl groups, acting as Bronsted acids, can interact with PVP segments which are considered Lewis base in aqueous media (Pattanaik and Bhaumik 2000). PVP was also shown to interact with dye-affinity chromatography supports (Galaev et al. 1994). PVP 360 was utilized in this study. c) Some studies have demonstrated that Polyethylene glycol [PEG] is preferentially excluded from macromolecular surfaces. This might elicit an energetically favourable sharing of the co-solvent hydration shells surrounding the biological particle and the chromatography media, thus increasing the partition coefficients (Gagnon et al. 1996). PEG 3350 was utilized in this study. d) Poly (vinyl alcohol) [PVA] is a polymer having anionic character. This compound has been reported to bind to controlled-porosity glass beads and to reduce the zeta potential of such particles (Wisniewska et al. 2007). PVA adsorption would increase with pH due to the presence of non-hydrolysed acetate groups i.e. the polymer gain negative charge [89% hydrolysis in this work]. PVA adsorption depends on

187

Results electrostatic forces, hydrogen bonding, and conformational state. PVA having a MW of 10 kDa was utilized in this work. e) Adsorbed polymer / surfactants on the solid surface can modify both physical surface properties and the interaction between interacting bodies. Therefore, other non-ionic surfactants have been included: Polyoxyethylene cetyl ether [Brij 58], Polyoxyethylene

sorbitan

monolaureate

[Tween

20],

Polyoxyethylene-

polyoxypropylene block copolymer [Pluronic F68], and Ethylphenolpoly(ethylene glycolether)n [Nonidet P40].

2.6.4.2 Screening additives with partitioning tests 2.6.4.3 The effect of additives on cell deposition A preliminary exploration (screening) on the potential effect(s) of a number of different additives on yeast cell deposition onto commercial chromatographic beads was performed utilizing simple partition tests (Fernandez-Lahore et al. 2000; Lin et al. 2001). Partitioning tests were carefully optimized to accommodate a variety of yeast-adsorbent interaction pairs. Under such experimental conditions, yeast cell deposition onto non-functionalized agarose beads was less than 10% as judged by the cell partition index (CPI ≥ 0.90). Standard experiments were performed in 20 mM phosphate buffer (≈ 4.0 mS/cm) and therefore a certain degree of deposition onto cation-exchanger materials (CPI ≥ 0.7 for SPStreamline) was observed due to electrical double layer compression effects (Tari et al. 2008). Contact time was fixed to 3 h so as to evaluate the combined effects of the fast and the slow phases of cell deposition (Fernandez-Lahore et al. 2000). Anion-exchangers are known to strongly interact with microbial cells, mainly due to charge-mediated (electrostatic) effects (Lin et al. 2006). Indeed, partition tests run with the weak anion exchanger DEAE-Streamline and the strong (composite) anion exchanger QStreamline XL showed high cell deposition i.e. a CPI equal to 0.28 and 0.14, respectively. Subsequent incorporation of non-ionic polymers / surfactants to the liquid phase has reduced cell deposition in an extent which depends on the additive and the solid phase under consideration. Table 1 depict the result of screening tests performed for various adsorbents and additives. When DEAE beads were employed as the solid phase, the basal condition for cell deposition (i.e. CPI = 0.28 in buffer) was improved: CPI increased to 0.54 188

Results in the presence of Tween 20, to 0.43 with added Pluronic F68, to 0.60 with Brij 58, and 0.34 with PEG. Particularly effective in inhibiting cell deposition onto DEAE beads was PVP 360; in this case the CPI raised to 0.81. However, the same additive failed to avoid cell attachment onto the Q-XL material. This could be explained by the presence of external Dextran chains in the structure of the composite adsorbent. On the contrary, polymeric sodium phosphate (NaPP), an ionic agent, showed almost no effect in preventing cell deposition onto DEAE beads but inhibited such phenomena onto Q-XL beads. It could be hypothesized that NaPP may interact primarily with yeast particles, rendering them more negative when DEAE beads are present. Thus, cell-to-adsorbent interactions remain the same. However, the presence of Q-XL beads may trigger the interaction of the (positively) strongly charged Dextran chains with the (negatively) charged polyphosphate. This would shield adsorbent charges thus reducing interaction with suspended cells. The distinct behaviour of the two anion-exchangers rules out a predominant role for the increased conductivity of the liquid phase in the presence of NaPP (≈ 9.1 mS/cm). Further studies performed with the DEAE / yeast system in the presence of PVP 360 showed that CPI is fairly proportional to the concentration of the additive in the liquid phase. A PVP concentration of 1% (w/v) resulted in maximum inhibition of cell deposition. Kinetic studies also revealed that PVP 360 seems to interfere with cell deposition mechanisms at early stages i.e. the polymer might inhibit cell-to-support interaction (data not shown). This phenomenon is discussed in more detail in the sections below. Partition experiments were also performed with hydrophobic interaction supports (PhenylStreamline) in 20 mM phosphate buffer (pH 7.6) containing 0.75 M ammonium sulphate. The CPI for the control situation i.e. buffer without additive(s) was 0.75 which correlates well with previous reported values (Fernandez-Lahore et al. 2000). Addition of non-ionic polymer / surfactants failed to improve the baseline situation e.g. CPI fell within the range 0.69-0.83. Therefore, only a marginal effect –if any- was observed for HIC systems. The presence of high concentrations of ammonium sulphate is HIC systems might interfere with the potential action of the additives utilised in this work. Chelating Streamline was utilized to evaluate cell deposition as well. Zn (II) ions were immobilized within the IDA groups present in the matrix. Partition experiments were performed in 20 mM phosphate buffer (pH 7.6) containing 250 mM sodium chloride and 1 189

Results mM imidazol. Baseline CPI was 0.90 which is similar to values previously reported in the literature for IMAC (Fernandez-Lahore et al. 2000). For Chelating systems the addition of non-ionic polymers / surfactants has had a slightly deleterious effect since CPI values tended to be lower (0.69 – 0.87). This can be explained considering the effect of considerable amounts of sodium chloride in the liquid phase and / or a possible bridging effect exerted by the polymers.

190

0.28

0.75

0.90

DEAE

Phenyl

Chelating

n.d: not determined

0.14

None

Q-XL

↓

Bead type

0.94

0.70

0.54

n.d.

Tween 20

0.91

0.80

0.43

n.d.

Pluronic F68

191

n.d.

0.79

0.34

n.d.

PEG 3350

0.74

0.69

0.81

0.34

PVP 360

Additive type

0.69

n.d.

0.35

n.d.

PVA

0.87*

0.83

0.60

0.40

Brij 58

n.d.

n.d.

0.30

0.87

NAPP

Table 1: Cell depositions of intact yeast cells onto Streamline beads as observed by partition experiments. Additives were present at a final concentration of 1% (w/v). Experiments were run in buffers having a typical composition, depending on the chromatographic mode involved (see text). Contact time was 3 h. Control CPI (agarose matrix) was ≥ 0.90. The Chelating material was loaded with Zn (II) ions. CPI values were within ± 10%.

Results

Results 2.6.4.4 The effect of additives on protein binding A useful additive has to inhibit cell deposition onto a defined type of chromatographic support without interfering with protein (bioproduct) binding. This situation would lead to a decreased biomass attachment without compromising adsorbent capacity for the targeted species. Therefore, partitioning studies were also performed to assess protein sequestration from a cell suspension with and without a selected number of promising additives. Partition studies were performed with DEAE-Streamline. In buffer, this material showed a high affinity number (0.85); the introduction of cells into the system translated in a ≤ 5% reduction in the affinity (binding) for the model protein. It should be recall that this support type has also a strong tendency to capture cells (Table 1). The addition of chemicals to the solution phase, showed no dramatic effect on the (equilibrium) capacity of the adsorbent (Table 2). Among the additives tested, PVP 360 showed a maximum protection against cell deposition (CPI 0.81 vs. 0.28) while protein binding remained unaltered (Affinity number 0.80 vs. 0.83). Therefore, PVP was clearly acting without interfering with the chargemediated attraction between BSA and the adsorbent beads. Partition studies were performed also with Q-Streamline XL. This material suffers from an extremely high deposition of cells, probably due to the presence of densely charged Dextran chains within its composite structure. As sodium polyphosphate was found to be effective in inhibiting cell interactions with Q-XL, protein-binding capacities were checked with buffers containing this chemical. Unfortunately, NaPP was found to interfere with BSA binding as reflected by affinity numbers falling from 0.92 (cells, no chemical) to 0.13 (cells plus additives). It follows that NaPP most probably interacts with the adsorbent by masking positively charged sites, therefore inhibiting both cell and protein uptake.

192

Additive

None Tween 20 PVP 360 Brij 58 None NaPP

Adsorbent

DEAE

Q-XL

193

0.95 0.13

0.85 0.91 0.81 0.80

Buffer

0.92 0.13

0.83 0.92 0.80 0.74

Cell suspension

Affinity number (-)

0.14 0.87

0.28 0.54 0.81 0.60

CPI (-)

Table 2: Protein binding studies on Streamline materials in 20 mM phosphate buffer at pH 7.6, as judged by the affinity numbers. Chemical additives were present to a final concentration of 1 % (w/v). Experiments run in the presence of suspended yeast cells contained 5% biomass (wet weight). Affinity number values were within ± 5%. CPI values were within ± 10%.

Results

Results 2.6.4.5 Biomass deposition experiments Data gathered employing a range of adsorbents and chemical additives showed the beneficial effect of PVP 360 on preventing cell deposition onto the weak anion-exchanger. In order to confirm such data under dynamic conditions, biomass (yeast cells) deposition experiments were run. DEAE-Streamline beads were utilized as collectors. Figure 1 depicts the cell effluent profiles recorded as a function of PVP 360 concentration, within the range 0.01 to 1.0 %. Cell deposition behavior was characterized by a decreased interaction with the adsorbent in the presence of increased concentrations of the additive in the flowing phase. The observed lower deposition of cells was reflected by the “attachment efficiency” parameter (α); α values are shown in Table 3. The baseline condition i.e. no additive translated into a α value equal to 0.683 which has decreased by a factor of ≈ 8 due to the presence of the additive. The α value obtained after running cell deposition experiments in the presence of 50 mM sodium chloride, a condition known to improve system hydrodynamics with DEAE adsorbents, was 0.213. Combination of 1% PVP 360 and 50 mM sodium chloride resulted in an minimum value for the attachment coefficient (α = 0.052). However, the incorporation of charge-screening ions into the chromatographic mobile phase may also lead to a decreased capacity for the targeted proteins.

194

Results

Figure 1: Biomass deposition experiments onto DEAE-beads. Intact yeast cells were utilised as model biomass. Solution chemistry was provided by a 20 mM phosphate buffer pH 7.6, which contained PVP 360 at various concentrations: (z) Control (no additive), (z) 0.01% , (z) 0.05%, (z) 0.1%, (z) 0.5 %, (z) 1 % PVP 360.

Previous work has demonstrated that, as far as a threshold value for the attachment coefficient e.g. α ≤ 0.15 is not reached, the level of biomass deposition remains low enough so as to allow for proper bed fluidization and product capture. Therefore, the addition of PVP within the range 0.2 to 0.5 % would be sufficient to prevent biomass interference with EBA operation. On the basis of the gathered experimental evidence, anion-exchangers would operate –in the presence of moderate amounts of PVP 360- similarly to cationexchangers. The later are materials which operate without major limitations with a variety of feedstock compositions (Tari et al. 2008).

195

d

c

b

a

0 0 0.01 0.05 0.1 0.5 1 1 0

Naked beadsa

Naked beadsa

Coated beadsc None

None None None None None Saltb

None Saltb

Other additive

196

DEAE-Streamline beads were utilized. Sodium chloride at a final concentration of 50 mM was employed. By treatment with PVP 360 and extensive washing with buffer solution. PVP 360 added to the flowing phase at a fixed concentration.

PVP 360d (%)

Adsorbent type

0.514

0.134 0.256 0.382 0.528 0.614 0.731

0.017 0.280

C / C0 (-)

0.112

0.337 0.228 0.161 0.107 0.082 0.052

0.683 0.213

α (-)

Table 3: Lumped attachment parameter (α) calculated from yeast cell deposition onto DEAE beads. Runs were performed in 20 mM phosphate buffer at pH 7.6. PVP 360 was employed as an additive.

Results

0.89 0.92 0.97 1.37 2.13

Viscositya (mPa·s) 9.5 x 10-11 9.8 x 10-11 1.0 x 10-10 1.4 x 10-10 2.2 x 10-10

6.0 x 10-14 5.8 x 10-14 5.6 x 10-14 3.9 x 10-14 2.5 x 10-14

b

197

Hydrodynamic dragc (N)

Diffusion coefficientb (m2·s-1)

Liquid phase. Calculated according to (Yeh et al. 1998) For yeast cells. Calculated according to (Brown 2007) c For yeast cells. Calculated according to (Johnson et al. 2007) d For adsorbent beads. Calculated according to (Brown 2007)

a

0.01 0.05 0.1 0.5 1

Concentration PVP (% w/v) 4.9 x 10-3 4.7 x 10-3 4.5 x 10-3 3.2 x 10-3 2.0 x 10-3

Settling velocityd (m·s-1)

Table 4: The effect of PVP 360 concentration on mobile phase viscosity, on cell diffusion coefficient, on the hydrodynamic drag exerted on yeast particles, and on the adsorbent bead settling velocity.

Results

Results A potential drawback on the utilisation of PVP 360 in fluidised bed systems is the increase of viscosity due to the presence of such polymer in solution. In turn, viscosity can affect process characteristics (Table 4). For example, mass transfer properties of cells can be altered via a reduced diffusion coefficient. Moreover, cell transport might be affected due to increased hydrodynamic drag exerted on them. Bed hydrodynamics can be also compromised by a reduced adsorbent particle settling velocity, which may promoted bead elutriation. Therefore, it is advisable to keep the concentration of this additive to a minimum compatible with its function as “cell-deposition-preventing” agent. To get a better insight on the mode of action of PVP 360 as an additive preventing cell deposition, DEAE beads were contacted with the additive (4 CV) and subsequently washed with phosphate buffer (20 CV). Interestingly, cell deposition experiments performed on these beads –but in the absence of any additive in the solution chemistry- also demonstrated much less cell deposition than with untreated beads (α = 0.112) see table 3. It can be concluded that PVP 360 might have been retained on the adsorbent surface by a combination of physicochemical forces. This is in agreement with previous work with hydrophilic polymers (Dainiak et al. 2002; Viloria-Cols et al. 2004). However, protein binding sites remain available within the adsorbent structure. Due to its hydrodynamic radius of gyration (~ 19 nm) (Armstrong et al. 2004) is unlikely that PVP 360 would have complete access to the pores existing in the chromatographic support (pore radius = 29 nm) (Jungbauer 2005). Moreover, these experiments have demonstrated a beneficial effect of the additive on cell deposition by decoupling cell attachment from viscosity-related effects in the presence of PVP. Other experimental findings employing biomass deposition experiments have generally confirmed results obtained with partition tests (data not shown). These findings can be summarised as follows: a) Biomass deposition experiments serve to confirm that other additives like Tween 20 or Brij 58 were ineffective in inhibiting cell interaction with DEAE beads; b) The utilization of such chemical additives failed to inhibit cell attachment to Chelating materials; c) Introduction of PVP 360 and Brij 58 in hydrophobic interaction systems was limited due to a lack of solubility in ammonium sulphate containing mobile phases; d) PVP 360 was unable to prevent cell deposition onto Q-XL in agreement with partition experiments.

198

Results 2.6.4.6 Interfacial free energy of interaction between bodies The extended DLVO theory can be applied to understand the interaction of two bodies in aqueous media, on the basis of colloid chemistry principles. In this case, the XDLVO approach was employed to calculate energy vs. distance profiles between naked or PVPcovered interaction particles e.g. yeast cells and adsorbent beads. Performing such calculations require the input of experimentally determined parameters like contact angle values with three diagnostic liquids and zeta potentials. Details on the mentioned approach have been published elsewhere (Vennapusa et al. 2008). Contact angle on hydrated PVP 360 layers were taken from published work (Faibish et al. 2002) and employed to calculate surface energy parameters. Calculations resulted in the following values:

γLW 35.8 mJ·m-2, γ+ 2.2 mJ·m-2, γ- 34.2 mJ·m-2, and γAB 17.3 mJ·m-2 These values were assumed to be the ones corresponding to either yeast cells or adsorbent beads when PVP 360 is present on their surface. Utilizing this information, the free (interfacial) energy of interaction was calculated –at the closest distance of approximation i.e. 1.57 Å and in aqueous media- for four different cases (Table 5): a) A naked adsorbent DEAE-bead interacting with a naked yeast cell b) A polymer coated bead interacting with a naked cell c) A PVP coated bead interacting with a polymer coated cell d) Two polymer coated yeast cells Additionally, zeta potential determinations were performed for polymer coated DEAE beads and intact yeast cells in 20 mM phosphate buffer (pH 7.6). These measurements have shown that particle zeta potential values can be reduced in the presence of the additive (Table 5). Similar observations have been made for silica-based particles (Goncharuk et al. 2001).

199

Results Table 5: The interfacial free energy of interaction between bodies in aqueous media, at closest distance of approximation: a) intact yeast cells and DEAE-beads, and b) aggregation between two yeast cell particles. Calculations were performed assuming interaction and aggregation with and without chemical additive at pH 7.6 in 20 mM phosphate buffers.

a b

System

ΔGLW [mJ·m-2]

ΔGAB [mJ·m-2]

Zeta potential [mV]

Naked beada - Cell

- 1.4

+ 28.7

+15 / -18

Coated beadb - Cell

- 1.6

+ 20.0

+ 5 / -18

Coated bead – Coated Cell

- 3.4

+ 12.0

+ 5 / -15

Two covered cells

- 7.0

+ 11.4

-15 / -15

DEAE Streamline PVP 360 treated adsorbent

The interaction systems mentioned above can be better understood by interfacial free energy (U) as a function of distance (H) profiles. Figure 2 depict such energy / force curves where it can be observed that: a) Interaction between naked DEAE beads and intact yeast cells is characterized by a strong interaction, as reported previously. Charge effects i.e. Coulomb-type attraction is dominating. A secondary energy pocket would exist at 5nm; pocket depth is -230 kT. This is in agreement with biomass deposition experiments presented before (Figure 1 and Table 3; α = 0.683). b) Interaction between a polymer coated adsorbent bead and a naked cell resulted in a secondary energy pocket of a moderate depth (-100 kT). The reduction in interaction energy / forces can be explained on the basis of modifications observed in the zeta potential values, particularly for the coated adsorbent bead. This is in agreement with cell deposition experiment utilizing pre-treated DEAE beads as collectors (Table 3; α = 0.112).

200

Results c) The assumption of having PVP 360 molecules covering the cell surface would result in a situation characterized by increased LW attraction, decreased AB repulsion, and almost unaffected EL attraction (Figure 2; U ≈ -160 kT). This behaviour couldn’t be verified by biomass deposition experiments. Although a slight decrease in the zeta potential value for cells in the presence of PVP was noticed, evidence more likely supports the idea that this additive preferentially interacts with the adsorbent beads. d) The presence of PVP 360 does not severely compromise colloidal stability of cell particles in suspension and therefore no aggregation is expected to occur (Figure 2; U ≈ - 20 kT).

Figure 2: Energy (U) vs. distance (H) profiles calculated for the interaction between a DEAE-bead and a yeast cell, in 20 mM phosphate buffer at pH 7.6. Calculations were performed assuming 4 hypothetical cases (refer to text), as follows: (—) DEAE/Cell, (—) [DEAE]pvp/[Cell]pvp, (—) [DEAE]pvp/[Cell], (—) [Cell]pvp/ [Cell]pvp.

201

Results

2.6.5 Conclusions Several chemical additives were evaluated for their capacity to reduce cell deposition onto a variety of chromatographic adsorbents. A simple albeit effective method to reduce biomass deposition onto DEAE adsorbent beads is presented. Addition of PVP 360, a pharmaceutical grade polymer, seems to preferentially interact with the adsorbent under process-like conditions. Covered adsorbent beads retain capacity for proteins but substantially reduce the interaction with suspended cells. The preceding hypothesis is supported by biomass deposition experiments and XDLVO calculations. The utilization of safe additives may find practical application to improve the sorption performance of direct contact methods in the downstream processing of bioproducts.

2.6.6 Acknowledgements RRVP great fully acknowledges the doctoral fellowship from Jacobs university.

202

References

2.6.7 References Anspach FB, Curbelo D, Hartmann R, Garke G, Deckwer WD. 1999. Expanded-bed chromatography in primary protein purification. J Chromatogr A 865(1-2):129-44. Armstrong JK, Wenby RB, Meiselman HJ, Fisher TC. 2004. The Hydrodynamic Radii of Macromolecules and Their Effect on Red Blood Cell Aggregation. Biophys J 87(6):4259-4270. Bos R, Van der Mei HC, Busscher HJ. 1999. Physico-chemistry of initial microbial adhesive interactions--its mechanisms and methods for study. FEMS Microbiol Rev 23(2):179-230. Brown DG. 2007. Adaptable method for estimation of parameters describing bacteria transport through porous media from column effluent data: Optimization based on data quality and quantity. Colloids Surf A Physicochem Eng Asp 296(1-3):19-28. Brown DG, Jaffe PR. 2001. Effects of Nonionic Surfactants on Bacterial Transport through Porous Media. Environ Sci Technol 35(19):3877-3883. Curbelo DR, Garke G, Guilarte RC, Anspach FB, Deckwer WD. 2003. Cost Comparison of Protein Capture from Cultivation Broths by Expanded and Packed Bed Adsorption. Eng Life Sci 3(10):406-415. Dainiak MB, Galaev IY, Mattiasson B. 2002. Polyelectrolyte-Coated Ion Exchangers for Cell-Resistant Expanded Bed Adsorption. Biotechnol Prog 18(4):815-820. Desilets CP, Rounds MA, Regnier FE. 1991. Semipermeable-surface reversed-phase media for high-performance liquid chromatography. J Chromatogr A 544:25-39. Faibish RS, Yoshida W, Cohen Y. 2002. Contact Angle Study on Polymer-Grafted Silicon Wafers. J Colloid Interface Sci 256(2):341-350. Fernandez-Lahore HM, Geilenkirchen S, Boldt K, Nagel A, Kula MR, Thommes J. 2000. The influence of cell adsorbent interactions on protein adsorption in expanded beds. J Chromatogr A 873(2):195-208. Fernandez-Lahore HM, Kleef R, Kula M, Thommes J. 1999. The influence of complex biological feedstock on the fluidization and bed stability in expanded bed adsorption. Biotechnol Bioeng 64(4):484-96. Fernandez-Lahore HM, Lin DQ, Hubbuch JJ, Kula MR, Thommes J. 2001. The Use of IonSelective Electrodes for Evaluating Residence Time Distributions in Expanded Bed Adsorption Systems. Biotechnol Prog 17(6):1128-1136.

203

References Feuser J, Walter J, Kula MR, Thommes J. 1999. Cell/adsorbent interactions in expanded bed adsorption of proteins. Bioseparation 8(1-5):99-109. Gagnon P, Godfrey B, Ladd D. 1996. Method for obtaining unique selectivities in ionexchange chromatography by addition of organic polymers to the mobile phase. J Chromatogr A 743(1):51-55. Galaev IY, Garg N, Mattiasson B. 1994. Interaction of Cibacron Blue with polymers: implications

for

polymer-shielded

dye-affinity

chromatography

of

phosphofructokinase from baker's yeast. J Chromatogr A 684(1):45-54. Ganeva V, Galutzov B, Teissie J. 2004. Flow process for electroextraction of intracellular enzymes from the fission yeast, Schizosaccharomyces pombe. Biotechnol Lett 26(11):933-7. Goncharuk EV, Pakhovchishin SV, Zarko VI, Gun'ko VM. 2001. Properties of Aqueous Suspensions of Highly Dispersed Silica in the Presence of Polyvinylpyrrolidone. Colloid J 63:283-289. Johnson WP, Li X, Assemi S. 2007. Deposition and re-entrainment dynamics of microbes and non-biological colloids during non-perturbed transport in porous media in the presence of an energy barrier to deposition. Adv Water Res 30(6-7):1432-1454. Jungbauer A. 2005. Chromatographic media for bioseparation. J Chromatogr A 1065(1):312. Lin

DQ,

Fernandez-Lahore

HM,

Kula

MR,

Thommes

J.

2001.

Minimising

biomass/adsorbent interactions in expanded bed adsorption processes: a methodological design approach. Bioseparation 10(1-3):7-19. Lin DQ, Zhong LN, Yao SJ. 2006. Zeta potential as a diagnostic tool to evaluate the biomass electrostatic adhesion during ion-exchange expanded bed application. Biotechnol Bioeng 95(1):185-91. Ng MYT, Tan WS, Abdullah N, Ling TC, Tey BT. 2007. Direct purification of recombinant hepatitis B core antigen from two different pre-conditioned unclarified Escherichia coli feedstocks via expanded bed adsorption chromatography. J Chromatogr A 1172(1):47-56. Ottewill RH, Shaw JN. 1972. Electrophoretic studies on polystyrene lattices. J Electroanal Chem 37:133-142. Papo A, Piani L, Ricceri R. 2002. Sodium tripolyphosphate and polyphosphate as dispersing agents for kaolin suspensions: rheological characterization. Colloids Surf A Physicochem Eng Asp 201(1-3):219-230. 204

References Pattanaik M, Bhaumik SK. 2000. Adsorption behaviour of polyvinyl pyrrolidone on oxide surfaces. Mater Lett 44(6):352-360. Petro M, Berek D. 1993. Polymers immobilized on silica gels as stationary phases for liquid chromatography. Chromatographia 37(9):549-561. Poulin F, Jacquemart R, De Crescenzo G, Jolicoeur M, Legros R. 2008. A Study of the Interaction of HEK-293 Cells with Streamline Chelating Adsorbent in Expanded Bed Operation. Biotechnol Progr 24(1):279-282. Redman JA, Walker SL, Elimelech M. 2004. Bacterial Adhesion and Transport in Porous Media: Role of the Secondary Energy Minimum. Environ Sci Technol 38(6):17771785. Santarelli X, Muller D, Jozefonvicz J. 1988. Dextran-coated silica packings for highperformance size-exclusion chromatography of proteins. J Chromatogr A 443:5562. Schomburg G. 1991. Polymer coating of surfaces in column liquid chromatography and capillary electrophoresis. Trends Anal Chem 10(5):163-169. Sharma MM, Chang YI, Yen TF. 1985. Reversible and irreversible surface charge modification of bacteria for facilitating transport through porous media. Colloids Surf 16(2):193-206. Silvino Dos Santos E, Guirardello R, Teixeira Franco T. 2002. Preparative chromatography of xylanase using expanded bed adsorption. J. Chromatogr. A 944(1-2):217-224. Smith MP, Bulmer MA, Hjorth R, Titchener-Hooker NJ. 2002. Hydrophobic interaction ligand selection and scale-up of an expanded bed separation of an intracellular enzyme from Saccharomyces cerevisiae. J Chromatogr A 968(1-2):121-128. Tari C, Vennapusa RR, Cabrera RB, Fernandez-Lahore M. 2008. Colloid deposition experiments as a diagnostic tool for biomass attachment onto bioproduct adsorbent surfaces. J Chem Technol Biotechnol 83:183-191. Vennapusa RR, Hunegnaw SM, Cabrera RB, Fernandez-Lahore M. 2008. Assessing adsorbent-biomass interactions during expanded bed adsorption onto ion exchangers utilizing surface energetics. J Chromatogr A 1181(1-2):9-20. Viloria-Cols ME, Hatti-Kaul R, Mattiasson B. 2004. Agarose-coated anion exchanger prevents cell-adsorbent interactions. J Chromatogr A 1043(2):195-200. Wisniewska M, Chibowski S, Urban T. 2007. The temperature influence on the adsorption and electrokinetical properties in the nonionic polymer/controlled porosity glass (CPG) system. Mater Chem Phys 103(2-3):216-221. 205

References Yeh HM, Cheng TW, Wu HH. 1998. Membrane ultrafiltration in hollow-fiber module with the consideration of pressure declination along the fibers. Sep Purif Technol 13(3):171-180.

206

General Conclusions and Remarks

3.0 General conclusions and remarks Expanded bed adsorption (EBA) is an interesting integrated bioprocess technology unit operation where solid liquid separation, partial purification and concentration can be simultaneously achieved. This integrated unit operation loses its significance due to interaction or aggregation of biological particles onto the process surfaces during direct sequestration of bioproducts. Many authors have repeatedly addressed this interference of biomass during primary unit operations of downstream processing such as EBA. Undoubtedly, biomass effects are detrimental to appropriate sorption performance in EBA. Information on the underlying mechanisms which govern biomass deposition onto chromatographic beads, under real process conditions, is still scarce. To address the significant challenge of biomass interference on EBA performance, the current thesis work puts emphasis on having a fundamental understanding of the feedstock behavior during primary-sequestration unit operations in downstream processing. Principles of colloid chemistry were applied to understand the fouling behavior of the chromatographic supports. The physicochemical properties of yeast cells (model type of biomass) and chromatographic supports (several types) were evaluated by contact angles and zeta potentials measurements. From these experimental determinations, deposition onto the process surface was predicted using the XDLVO theory. Calculations were confirmed by independent experiments like biomass deposition in granular beads and laser diffraction spectroscopy. The XDLVO theory and biomass deposition experiments were able to explain the attachment behavior of cells as a function of varying solution chemistry, size of the biological particle, and functionalisation type of the process surface. In this regard, the approach developed in this work can be anticipated as a universal approach for understanding the biomass adhesion onto process surfaces. Furthermore, the tools presented here are useful in guiding process and material design and development. This piece of work has arrived to remarkable conclusions on the behavior of particulate feedstock component deposition onto different chromatographic supports of varied chemistries, ranging from ion-exchangers and hydrophobic interaction supports to metalion chelating surfaces. Besides the utilization of intact yeast cells, other systems like yeast cell debris and E.coli homogenate were explored with special focus on the ion exchanger type of chromatographic beads.

207

General Conclusions and Remarks The following partial conclusions can be mentioned: 1) Ion exchanges surfaces to biomass The extent of interaction of yeast cells (biomass) with anion- and cation- exchangers can be explained on the basis of the calculated secondary energy minima e.g. reversible adhesion can be predicted (Figure 1). The degree of interaction varied with the solution chemistry e.g. upon changes in buffer pH and conductivity. A positive correlation was found between energy minima and cell deposition, as evaluated by the so called cell transmission index (CTI). Coulomb-type interactions, which are related to the measured zeta potential of the interacting bodies, were confirmed to be dominant. The XDLVO approach also gave us a clear idea about influence of biomass particle size on the extent of fouling. The total forces acting during biomass adhesion are greatly altered with the size of the cell or cell fragments. Secondary interfacial energy minima values were experimentally validated via biocolloid deposition experiments (BDE). BDE is simple, straightforward and automatable diagnostic tool, which was developed during the current work. This technique allowed the calculation of a lumped parameter [α, attachment deposition coefficient] reflecting biomass interaction and aggregation phenomena. However, cell-to-cell aggregation is less likely to happen under the conditions prevailing in an ion-exchange process. This can be explained due to dominant electrostatic repulsion between (negatively charged) cells in low-conductivity mobile phases. The conclusion drawn with regard to deposition of biomass onto ion-exchangers from the fundamental understanding gathered during the current work is in full agreement with the known EBA operational constraints. 2) Hydrophobic interaction surfaces to biomass As hydrophobic interactions are expected to occur under high salt concentrations, a marginal contribution of charge-mediated effects is anticipated under such chromatographic mode of operation. The consideration of the XDLVO theory was very well justified in the scenario of hydrophobic interaction chromatography (HIC).

In the later case, the

interaction of biomass with the chromatographic beads could be explained due to Lifshitz-

208

General Conclusions and Remarks Van der Waals (LW) and acid-base (AB) interaction forces. Biomass deposition onto HIC supports is correlated to the development of a (reversible) energy secondary minimum, which can be observed to arise from LW and AB forces. Calculations indicated that moderate interactions between yeast cells and adsorbent beads can develop, especially in presence of higher salt concentrations at pH 7. It was also found that cell-to-cell aggregation is taking place in the context of hydrophobic interaction conditions i.e. at high salt concentrations. Because of the high salt concentration, the repulsive electrostatic forces are reduced allowing yeast cells to interact with each other. Again, LW and AB forces were relevant to aggregation phenomena. Buffer pH and conductivity were found to influence cell-to-bead interaction and cell-to-cell aggregation. In both cases, predictions based on the XDLVO approach were validated by biomass deposition experiments, laser diffraction spectroscopy, and confocal microscopy. 3) Chelating beads to biomass Immobilized ion-metal affinity (“Chelating beads”) chromatography (IMAC) is run at moderate salt concentrations (viz 250-750 mM sodium chloride). The interaction of biomass with IMAC-Cu2+ was observed to be related to the development of a (reversible) secondary energy minimum. An influence of buffer pH and conductivity was observed within normal operational windows. From XDLVO calculations it can be concluded that favorable interaction of yeast cells with Chelating beads takes place at pH ≥ 8. However, biomass deposition experiments failed to confirm such prediction i.e. a decrease in deposition coefficient at pH 8 was observed. This anomalous behavior can be explained considering that Cu2+ can be actually sequestrated from the Chelating beads by yeast cells, a fact usually exploited for biosorption of metal ions from wastewaters. Cell-to-cell aggregation behavior also observed

at IMAC process buffer conditions. The

aggregation phenomena especially with IMAC Cu2+ system were also discussed. The extent of cell- cell aggregation was found to be less at IMAC buffer conditions when you compare with that of the hydrophobic interaction buffer conditions.

209

U (kT)

General Conclusions and Remarks

Separation distance (nm)

Secondary energy minimum pocket

Figure 1: Secondary energy minimum where reversible cell adhesion can occur.

4) Effect of chemical additives on biomass deposition for various process surfaces The better understanding of underlying mechanisms operating during biomass attachment onto chromatographic beads may be helpful in guiding / explaining the application of chemical additives to prevent biomass interference during EBA. A portfolio of chemicals was introduced in an attempt to alter biomass deposition. Extensive testing has resulted in the utilization of a safe polymeric additive to drastically reduce yeast cell deposition on anion-exchangers without altering the protein binding capacity. Anion exchangers are known to strongly suffer from biomass interference. However, process performance of fluidized anion exchangers in the presence of yeast cells can be restored by addition of PVP 360 at concentrations ranging from 0.4 to 1.0 % (w/v). The role of various additives on different modes of chromatography was also discussed in detail but no chemical additive was found to inhibit biomass deposition onto hydrophobic and chelating materials.

210

General Conclusions and Remarks

Figure 2: General correlation between α and U.

Force vs. distance profiles were calculated for a large combination of model biomass and adsorbent bead types. The energy minima values (absolute value / U) obtained for each of the analyzed cases were correlated with the corresponding deposition coefficient values (Figure 2). A positive correlation was obtained. It was concluded that total interaction energy U ≤ -25 to -50 kT and biomass deposition parameter α ≤ 0.15 would be a safe region for EBA operation e.g. biomass interference can be neglected. Deviation in the correlation was found in such cases where moderate or high cell-to-cell aggregation occurs, for example in hydrophobic interaction systems.

211

General Conclusions and Remarks 5) Remarks

1) The work highlighted the use of yeast as model biomass onto various process surfaces. The tools developed for understanding of yeast interaction and aggregation onto the process surfaces can be extended to any different model biomass or any kind of process surface. From the preceding results and discussions, it allowed us to draw conclusion that mechanistic tools developed here can be universally applied. 2) Local level understanding of biomass interaction can lead to development and design of an optimized primary unit operation (like EBA). For example, the amount of hydrodynamic force required to prevent the interaction or modification of the process surfaces with polymeric brushes (length of brushes) to prevent interaction. 3) The data provided from this work can act as first step for global modeling of EBA. 4) The knowledge obtained from the physicochemical properties of biomass and absorbent surfaces would allow easy access to process predictions without any trial and error experiments in laboratory. Hence the developed approach here can be used to design EBA process where reduced time and effort is required. A fundamental knowledge about the roles of key variables affecting interactions /aggregation described herein is of great importance, because their correct manipulation can help to prevent or at least mitigate fouling. 6) Recommendations for further work

1) The current thesis work explored interaction of yeast as a model organism onto various adsorbent surfaces. There is still room to understand the behavior of several other feedstocks like yeast and bacterial debris, mammalian cells and plant cells which are commonly utilized in biotechnology industry as a host system for biopharmaceuticals production. It is very important to understand the compatibility of various kind of biomass onto process surfaces, which could allow drawing some general conclusion from where a global model for EBA process design can be proposed.

212

General Conclusions and Remarks 2) The biomass deposition transport phenomenon has to be better modeled where the three parameters - collision efficiency (cell to process attachment), blocking (amount of process surface blocked) and ripening (cell-to-cell aggregation), can be quantitatively obtained. Further quantitative straining models have to be implemented to completely rule out the role of physical attrition during these biomass deposition experiments. 3) XDVLO predictions made in the current work can be further directly verified experimentally with force/ distance curves utilizing atomic force microscopy. This would further broaden understanding and signifies the importance of the various forces to be taken into consideration. 4) Age of biomass culture and its influence on interaction and aggregation especially with the hydrophobic materials can be tested. Preliminary studies with the different age of the culture showed the differences in the deposition phenomenon especially with hydrophobic material type. 5) HIC/ EBA process limitations are not described in the literature (with any type of biomass) which is actually prone to have deleterious effects by biomass interference (observations from current research work). Hence there is room to explore more in this direction and define the solution/ operational windows for HIC materials in context of EBA. 6) PVP 360 additive which was found to inhibit the interaction of yeast with DEAE beads in this work can be further tested for its compatibility under real EBA operations. 7) Hydrodynamic shear and its impact on the secondary minimum can be explored in detail.

213

Appendix

4.0 Appendix Biomass deposition experiments, which were performed during the current thesis work, are geared towards in-depth understanding of the fate and transport of biomass through the packed bed chromatographic adsorbents. These understandings could be helpful in preventing the attachment or predicting the transport behavior of biomass during bioprocess scenario where direct sequestration of bioproducts expected in presence of biological particles like expanded bed adsorption. The physicochemical parameter α can be obtained from such experiments relates to the attachment efficiency of colloidal particles (yeast) onto the collector surfaces (process surface). Obtaining such quantitative information is possible through the application of colloid filtration theory. The physicochemical parameter determined from these experiments can at times lead to false quantative information which may be due to the physical filtration means. This indicates the cell effluent profiles go down due to the attrition effects which have been shown previously in many studies (Tufenkji et al. 2004). Among many models developed to predict the potential for straining, Bradford (Bradford et al. 2004) and Sakthivadivel (Sakthivadivel 1966; Sakthivadivel 1969) developed a model to predict the potential of straining based on the system geometry. According to this straining could have a significant influence when the ratio of the particle diameter to the median grain diameter (dp/dc) is greater than 0.05. In the current study dp/dc ≈ 0.04, which means that, physical filtration effects can be neglected. However there are some studies showing that straining observed when dp/dc values were as low as 0.002 (Tufenkji et al. 2004). To prove that in the current thesis work straining can be neglected, biomass deposition experiments were performed with colloids (yeast) utilized in the study with the non-interacting adsorbent beads (Streamline SP). These experiments were performed in dilute buffers conditions. Figure 1 depicts the biomass deposition experiments performed with the yeast and Streamline SP. At very low ionic strength of the buffer, the electrostatic double layer repulsion between the particles and beads packed in the column is substantial such that particle deposition (physicochemical phenomenon) can be neglected. Thus, in this type of experiment, any particle removal in the packed bed is attributed to the influence of a

214

Appendix physical mechanism such as straining. Hence it is expected that cell effluent concentrations should be similar to the concentrations of cells before in contact with the collector. The experimental evidence from the figure 1 suggest that there is a complete breakthrough of the curve ( C/Co ≈ 1 ) indicating that straining does not play a role in the removal of particles in porous media. This also could be reflected in the α ≤ 0.018 the minimum alpha which represents absence of any type of interactions.

Figure 1: Biomass deposition experiments between intact yeast to cat-ion exchangers

(Streamline SP) at very dilute buffer conditions (0.66 mS.cm-1, pH 7.6). From this experimental observation now it can be clearly stated that differences in the physicochemical parameter α reported in this thesis work are mainly due to influence of physicochemical forces.

215

References

References Bradford SA, Bettahar M, Simunek J, Van Genuchten MT. 2004. Straining and Attachment of Colloids in Physically Heterogeneous Porous Media. Vadose Zone J 3(2):384394. Sakthivadivel R. 1966. Theory and mechanism of filtration of non-colloidal fines through a porous medium. Berkeley: Hydraulic Engineering Laboratory, University of California. Report nr Rep HEL 15-7. Sakthivadivel R. 1969. Clogging of a granular porous medium by sediment. Berkeley: Hydraulic Engineering Laboratory, University of California. Report nr Rep HEL 15-5. Tufenkji N, Miller GF, Ryan JN, Harvey RW, Elimelech M. 2004. Transport of Cryptosporidium Oocysts in Porous Media: Role of Straining and Physicochemical Filtration. Environ Sci Technol 38(22):5932-5938.

216