High Throughput Solubility Analysis; a Tailored Approach to Supporting Drug Discovery Efforts Charles E. Taylor, Robyn A. Rourick and John P. Walsh Kalypsys, Inc. Pharmaceutical Sciences Department 10420 Wateridge Circle, San Diego, CA 92121
Objectives • Expanded assay to include multiple buffer solutions being used in‐house • Fully automated solubility assay → Robotic plate movement → Automated plate reading • Server based data handling → Queuing system integrated with proprietary “Request System” → Curve fitting tools → Automated report generation and archival
Optimization
Stock Plate
0.45 µM
UV Plates
96 Well Shallow Well
Acetonitrile
Process Plates
40µL
160µL
Buffer Solubility Comparison
Solvents
200µL
Verapamil Solubility
600
96 Well
96 Well
UV Plate
UV Plate
500 400 300 200 100
Sa lin e
-100 -200
260nm to 500nm Figure 1: Illustration showing the effect of shake time on the sample results. 4.5 hours was chosen because it provides most of the compounds enough time to 4.5 Hrs reach equilibrium. The compounds used in the illustration were: clozapine, t a m o x i f e n , k e t o c o n a z o l e , chlorpromazine, chloramphenicol, 7.00 14.00 21.00 28.00 35.00 42.00 49.00 56.00 63.00 70.00 furosemide, 2‐naphthoic acid and 4,5‐DPI
Effect of Shake Time in Sample Values 3.5 3
AU of Samples
2.5 2 1.5 1 0.5 0 0.00
Hours of Shaking Conc. (µM)
500 500 200 200
50
Organic Buffer (µL) 285 285 380 380
12
12
10 mM stock (µL)
15
15
8
8
Transfer (µL) from Col 3& 4
Transfer (µL) from Col 5& 6
Transfer (µL) from Col 7& 8
12.5 12.5 3.13 3.13
0
0
Figure 3: Chart illustrating the different solubility values obtained for each of the buffers tested. Ketoconazole is represented by the blue bars, and Verapamil is represented by the yellow bars. The validated phosphate buffer replicates, which were used as controls, are boxed in red.
15 100
15
15
15
15
15
15
15
100
100 100
100 100
Table 1: Liquid aliquots and serial dilution volumes used in constructing the calibration concentrations used to determine the unknown sample concentrations. Organic buffer refers to the 20% Acetonitrile composition in the buffer being used to construct the calibration curve. Acetonitrile helps to maintain the higher concentrations of the compound in the calibration curve.
100
200
300
400
500
600
Concentration [uM] 3. Informatics Tools • Queuing system • Curate results • Automated PDF report generation and archival
Figure 6: User curate ability in Solubility Manager
Figure 7: Incoming sample queue. Setup
2. Fully Automated Assay • Validate the use of a liquid handler with specific automation components:
Verapamil
A. Robotic Arm B. Shaker
Verapamil
C. Vacuum Block D. UV Plate Reader
Figure 8: Sample results page (from left to right); pass/fail indicator, tabulated results, triplicate spectra, calibration spectra, calibration curve, structure, compound ID, plate barcode, and well location.
B D
0
Figure 10: Bar chart showing the historical use of the solubility assay. The effect of introducing a fully automated solution provides that many FTE hours will be saved, and the use of the assay will increase with the throughput.
Buffer Systems
285 285 285 285 285 285 285 285
DMSO(µL)
50
S D (p M H EM 7. 4) (-p he D no M U EM AB lr ed / O (-p ) A he B no Pl at lr e ed 1 )+ 5% R R PM PM FB I( I( S -p -p he he no no lr lr ed ed ) )+ 5% U AB FB /O S A H B BS Pl S at + e 2 2 C 0m M Y et P M 45 ab H 0 ep ol As ic es S s a ta y bi Bu lit y f fe As r sa y Bu U AB ffe r /O A B AD Pl M B at ag e K ic in 3 B a se uf fe B rf uf or fe r N eu tro ph M ils ag ic B uf U fe AB r /O 2 A B Pl at e 4
0
UV Plate Read
0
5
Ketoconazole Solubility
700
Transfer Volume
0
10
7‐Oct‐07
Table 2: Summary of the Pass/Fail results for the various buffers in the system. The automation capable category considers the success with which the buffer was handled by the Tecan liquid handler and/or shaker. A High Standard Deviation of the samples is indicative of a ruggedness problem for inter‐ or intra‐day reproducibility. A failure in the baseline absorbance category would occur if the buffer system showed a chromophore of significant enough intensity to affect the current quantitation limits. Intertwined calibration absorbance curves are often a result of a poorly behaving UV solvent with compound.
0.5
15
7‐Aug‐07
Filter @
1
Manual 1 BioA 1 Profiler 1 Profiler 2 Profiler 3 Profiler 4 Profiler 5 Profiler 6
7‐Jun‐07
Process Step
1.5
20
7‐Apr‐07
@ 200RPM
Verapamil
25
7‐Feb‐07
Legend
Concentration [uM]
30
7‐Dec‐06
4.5 Hrs
600
7‐Oct‐06
Dilution
500
7‐Aug‐06
Serial
400
7‐Jun‐06
(Calibration)
300
7‐Apr‐06
(Samples)
200
7‐Feb‐06
w/ 20% Acetonitrile
100
7‐Dec‐05
Deepwell Plate
0
35
7‐Oct‐05
Filterplate
0
40
7‐Aug‐05
Phosphate Buffer
0.2
45
7‐Jun‐05
96 Well
0.4
Manual 1 BioA 1 Profiler 1 Profiler 2 Profiler 3 Profiler 4 Profiler 5 Profiler 6
7‐Apr‐05
96 Well
Intertwined Baseline Automation High Std. Dev. Calibration Buffer Capable Of Samples Absorbance Absorbance Saline Pass Pass Pass Pass PBS (pH 7.4) Pass Pass Pass Pass DMEM (‐phenol red) Pass Fail Fail Fail DMEM (‐phenol red) + 5% FBS Fail Pass Fail Fail RPMI (‐phenol red) Pass Pass Fail Fail RPMI (‐phenol red) + 5% FBS Fail Pass Fail Fail HBSS + 20mM Hepes Pass Pass Pass Pass CYP450 Assay Buffer Pass Pass Pass Pass Met. Stab. Assay Buffer Pass Pass Pass Pass ADB Kinase Buffer Pass Pass Fail Pass Magic Buffer for Neutrophils Pass Pass Pass Pass Magic Buffer 2 Pass Pass Pass Pass
AU
46µL
0.6
AU
10 µL
0.8
• Must be automation capable
DMSO
Phosphate
1
Solubility Assay Metrics
7‐Feb‐05
• Test multiple buffers used in various groups
170 uL 10mM DMSO Stock
Shake
Ketoconazole
Figures 5a and 5b: Graphs illustrating six replicate calibration curves for the respective compounds, which were constructed by the fully automated method. Also in the graphs are a manually prepared calibration curve and a s e m i ‐ a u t o m a t e d calibration curve. This graphs illustrate the capability of the fully automated method in production.
Compounds
96 Well
Buffer
Output and Metrics
1. Buffer Expansion Work
PB
Purpose: Solubility is a widely utilized assay throughout the Drug Discovery projects at Kalypsys. To increase the throughput, reduce the resource cost and maintain the credibility of the current assay, we aim to deploy a completely automated modified shake‐flask solubility assay that is capable of analyzing various buffer types. Methods: Using a modified shake flask method, a medium throughput plate‐ based method was developed and qualified against literature values. This method employed a filter plate as the shake vessel; allowing for ease of post‐agitation precipitate filtration. Calibration gradients in an organic mixture of the buffer allow for quantification of the sample solubility. UV plates were used to read both the sample and calibration solution absorbances; the latter provides for a linear correlation of the UV absorbance to concentration. Initially, this procedure was performed manually with multichannel pipettes with the user taking the plates to various instruments. The next evolution automated the liquid handling on a disposable tip liquid‐handler. In this configuration, the assay was expanded to include analysis in multiple buffer types; all buffers used in the company were investigated for solubility analysis compatibility. The next design iteration was to increase the assay throughput; achieved by migrating the assay to an integrated platform and upgrading the data handling tools for the assay. A liquid handler with the accessories capable of processing the solubility assay was installed; this system allowed for a theoretical throughput of 84 compounds in 5 hours. With the data quantity, a server based data handling package was needed for the purpose of: a sample queuing system, requesting test articles from compound management, parsing raw data files, curve fitting, outlier sample exclusion and automated report generation. Results: The assay was successfully migrated to the fully automated platform, with the capability of analyzing solubility in multiple buffer types. The data handling tools were successfully implemented. Each method developed was qualified by performing a parallel analysis in the previous method(s); the assay was deployed only after qualification. Conclusions: The deployment of this assay, with its versatility and throughput has greatly impacted the Discovery efforts at Kalypsys.
Experimental
Concentration [uM]
Introduction/Abstract
10 minutes
10 minutes
Sample 20 minutes Plate Prep
5 minutes
N/A
Cal Plate 3 hours (28 Cmpds)
40 minutes
N/A
Sample Analysis
20 minutes
10 minutes
N/A
Data Analysis
1 hour
15 minutes
15 minutes
10 minutes
4 hours 40 minutes
1 hour 20 minutes
25 minutes
Figure 11: FTE time dedicated to the assay at the various levels of development for a typical run size of 28 compounds. The data analysis time for the manual method is higher as that method was deployed before the Informatics Tools were deployed.
Acknowledgements The authors would like to thank the following individuals for their time and effort in bringing this assay to deployment and optimizing it for higher throughput: Mike Calcagno, Qiner Yang, Frank Balistrieri , Chris Peterson and Katie Belsky.
C
Figure 4: Bed layout of the robotic deck that is used to process the fully automated solubility assay.
Figure 9: From the results displayed in Figure 7, automatically generated PDF reports are created and archived in a Solubility Report Library on the company intranet. The results are also maintained in an Oracle® table, which can be data mined or displayed in SAR tables.
Document Number
Description
AC_M T_Solubility_v1 AC_HT_Solubility_v1 AC_MT_Solubility_Expansion_v1 AC‐MethDev‐2006‐012v1
Semi‐Auto and Manual Solubility Assay Fully Automated Solubility Assay Multiple Buffer System Validation Fully Automated Solubility Assay Development