Isolation Mass Spec

Methods 26 (2002) 260–269 www.academicpress.com

Isolation and mass spectrometry of transcription factor complexes G. Sebastiaan Winkler,a,1 Lynne Lacomis,b,1 John Philip,b Hediye Erdjument-Bromage,b Jesper Q. Svejstrup,a,* and Paul Tempstb,* a

Cancer Research UK London Research Institute, Clare Hall Laboratories, Blanche Lane, South Mimms, EN6 3LD, Hertfordshire, UK b Molecular Biology Program, Memorial Sloan–Kettering Cancer Center, 1275 York Avenue, New York, NY 10021, USA Accepted 1 February 2002

Abstract Protocols are described that enable the isolation of novel proteins associated with a known protein and the subsequent identification of these proteins by mass spectrometry. We review the basics of nanosample handling and of two complementary approaches to mass analysis, and provide protocols for the entire process. The protein isolation procedure is rapid and based on two high-affinity chromatography steps. The method does not require previous knowledge of complex composition or activity and permits subsequent biochemical characterization of the isolated factor. As an example, we provide the procedures used to isolate and analyze yeast Elongator, a histone acetyltransferase complex important for transcript elongation, which led to the identification of three novel subunits. Ó 2002 Elsevier Science (USA). All rights reserved.

1. Introduction Proteomics is the newest chapter in biological systems analysis. A more focused embodiment, termed targeted proteomics, calls for the examination of subsets of the proteome, e.g., those proteins that either are specifically modified, or bind to a particular DNA sequence, or exist as members of higher-order complexes, or any combination thereof. We have sought to apply this concept to the study of gene expression. The genome of eukaryotic organisms contains a plethora of factors important for the regulation of transcription. Genetic and biochemical analysis of gene transcription have identified numerous regulatory proteins, including chromatin-modifying enzymes. In addition, homology searching in the recently fully sequenced genomes reveals the presence of many additional, yet uncharacterized factors that are likely also important for transcriptional regulation. Because highmolecular-weight multisubunit complexes have frequently been identified by biochemical approaches, it

*

Corresponding authors. E-mail addresses: [email protected] [email protected] (P. Tempst). 1 Authors contributed equally to this review.

(J.Q.

Svejstrup),

can be expected that factors identified by genetic means or by sequence homology are also functional in the context of multiprotein assemblies. Here we describe a generic method that can be used to purify proteins and their associating partners and their subsequent identification by mass spectrometry. The complete procedure does not require previous knowledge of complex composition or function, and is fast, reliable, and suitable for proteome analysis. As an example, we provide the protocol used to isolate yeast Elongator, a histone acetyltransferase complex important for transcript elongation, which led to the identification of three novel subunits by mass spectrometry.

2. Methods 2.1. Affinity purification of protein complexes The procedure presented herein is based on the high affinity of histidine stretches to Ni-agarose and a hemagglutinin (HA) immunoaffinity procedure that allows elution by competition with excess peptide. Both tags have been used individually in both yeast and mammalian systems, facilitating the isolation of multisubunit

1046-2023/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 4 6 - 2 0 2 3 ( 0 2 ) 0 0 0 3 0 - 0

G. Sebastiaan Winkler et al. / Methods 26 (2002) 260–269

complexes (see, for example, [1–4]). However, the combination of both tags and the employment of competitor peptide that allows relatively high yields from the antiHA immunoaffinity step improve the use of these tags for the purification of rare complexes [5,6]. The protocol is rapid and permits the biochemical characterization of the isolated factor. 2.1.1. Expression of tagged Elp1 in yeast We constructed a tagged version of the yeast ELP1 gene, encoding the largest subunit of Elongator. The tagged gene was expressed at normal levels by replacing the endogenous chromosomal copy with the tagged sequence by homologous recombination in yeast. This approach makes it possible to avoid the creation of nonphysiological complexes with altered stoichiometry as might be caused by protein overexpression. Furthermore, careful phenotypic analysis was carried out to ensure that the sequence encoding the affinity tag did not interfere with gene/protein function. To facilitate carboxyl-terminal genomic tagging, we have developed a generic vector, pSE-304-HisHA (Fig. 1), in which multiple unique restriction enzyme sites precede the tag and a transcription termination signal [7]. 2.1.2. Preparation of immunoaffinity reagents The anti-HA immunoaffinity procedure uses the mouse 12CA5 monoclonal antibody. These antibodies can be obtained commercially (Roche), and the 12CA5 hybridoma is also widely distributed among research laboratories. We used crude (concentrated) cell culture supernatant or ascites fluid directly to prepare the immunoaffinity resin.

261

We routinely prepared anti-HA immunoaffinity resin by binding 1–3 mg 12CA5 antibody per milliliter of protein A–Sepharose resin (Amersham–Pharmacia Biotech) equilibrated in phosphate-buffered saline (PBS) for 30 min at room temperature with continuous mixing. As the elution conditions do not interfere with the antibody–protein A binding, the antibodies were not covalently crosslinked to the resin. This allows the recovery of protein A–agarose resin by removing antibody–peptide complexes with 0.1 M glycine (pH 2.5). 2.1.3. Purification of double-tagged Elp1 protein from yeast cells Protease-deficient yeast cells BJ2168 (MATa, prc1407, prb1-1122, pep4-3, leu2, trp1, ura352, gal2) [8] expressing a HisHA-tagged version of the ELP1 gene were grown to late-log phase and lysed with glass beads, and whole-cell extracts were prepared as described [7,9]. We found that the yield and purity of the isolated material were greatly improved by starting with chromatography on a conventional resin, such as Bio-Rex 70 (Bio-Rad). This resin was chosen because most DNA-related factors bind to it, but other resins, such as heparin–Sepharose (Pharmacia–Amersham), could also be used. The Bio-Rex fraction containing HisHAtagged Elp1 was then incubated with anti-HA immunoaffinity resin (Fig. 2A). In some instances, soluble whole-cell extract was incubated directly with the immunoaffinity resin. 1. Incubate portion of the Bio-Rex 70 fraction containing the majority of Elp1 protein (50 mL, 300 mg total protein) in buffer A (40 mM Hepes–KOH, pH 7.6, 1

Fig. 1. Plasmid pSE.HisHA-304. Indicated are the bacterial origin of replication and the ampicillin resistance gene, the TRP1 marker, and the promoter-less HisHA tag followed by a transcription termination signal derived from the ADH1 gene. Also shown are the nucleotide and amino acid sequences of the polylinker fused to the decahistidine stretch and HA epitope tag.

262

G. Sebastiaan Winkler et al. / Methods 26 (2002) 260–269

Fig. 2. Affinity purification of the Elongator histone acetyltransferase complex. (A) Schematic diagram of the HisHA-tagged Elp1 protein and the purification strategy. (B) SDS–PAGE analysis of protein fractions from the anti-HA immunoaffinity column. Top: Immunoblot. The HisHA-tagged Elp1 protein was detected using rat monoclonal antibody 3F10 recognizing the HA epitope (Roche). Bottom: Protein silver staining. Indicated on the left are positions of the size markers. (C) Analysis of protein fractions from the Ni-NTA affinity column by protein silver staining. Indicated on the left are positions of the size markers.

mM EDTA, 1 mM dithiothreitol, 20% (v/v) glycerol) containing 600 mM potassium acetate several hours to overnight at 4 °C with 0.8 mL anti-HA immunoaffinity resin with continuous, gentle mixing. 2. Collect the resin by gravity flow in a disposable polypropylene column holder (Econo-Pac, 1:5
12 cm, Bio-Rad). Reapply the flow-through twice onto the resin.

3. Wash the immunoaffinity resin twice with 10 mL buffer A containing 600 mM potassium acetate. 4. Subsequently equilibrate the column in buffer E (40 mM Hepes–KOH, 1 mM 2-mercaptoethanol, 20% (v/v) glycerol) containing 300 mM potassium acetate. 5. Finally, elute bound protein with excess peptide. Gently mix the resin with 1 mL prewarmed buffer

G. Sebastiaan Winkler et al. / Methods 26 (2002) 260–269

E containing 300 mM potassium acetate and 1 mg/ mL HA peptide (KKKRILKMYPYDVPDYARIL) and submerge the column holder in a water bath at 30 °C for 15 min. Occasionally, resin and buffer were mixed by gentle tapping. Collect the eluate by gravity flow and immediately place on ice. Repeat this step twice. Protein fractions were analyzed by SDS–PAGE and stained with silver nitrate (Fig. 2B). In addition to Elp1 and two known Elongator subunits, Elp2 and Elp3, three additional associating proteins with apparent molecular masses of 50, 35, and 30 kDa were identified. These proteins do not bind to anti-HA immuno-affinity resin in the absence of HisHA-tagged Elp1. Thus, virtually homogeneous Elongator complex was obtained in a rapid two-step procedure. However, a second high-affinity step using the decahistidine stretch was used to verify that these proteins are indeed associating with Elp1. This second affinity step can also be used to concentrate the protein fractions and is required when crude extract is incubated directly with anti-HA immunoaffinity resin, or when the target complex is present at relatively low levels in the extract. Moreover, the Ni-agarose efficiently removes the excess HA-peptide used for elution from the anti-HA resin, which might interfere with mass spectrometric analysis. 1. Pool the eluted fractions from the anti-HA immunoaffinity column and incubate several hours to overnight with 0.4 mL Ni-NTA agarose (Qiagen) at 4 °C with continuous mixing. 2. Collect the Ni-agarose resin by gravity flow in a disposable polypropylene column holder (Polyprep, 0:8
4 cm, Bio-Rad). Reapply the flow-through twice onto the resin. 3. Wash the Ni-agarose resin once with 1 mL buffer E containing 300 mM potassium acetate, and twice with 1 mL buffer E containing 300 mM potassium acetate and 10 mM imidazole. 4. Finally, elute bound protein with three washes of 0.4 ml buffer E containing 300 mM potassium acetate and 300 mM imidazole. All procedures were carried out at 4 °C. Protein fractions were analyzed by SDS–PAGE and stained with silver nitrate (Fig. 2C). The associating proteins with apparent molecular masses of 50, 35, and 30 kDa coeluted from the Ni-NTA agarose column. These proteins were prepared for identification by mass spectrometry and shown to be bona fide subunits of Elongator [7]. 2.2. Identification of proteins by mass spectrometry Proteins isolated in the manner described in the previous section usually yield a small amount of protein, often less than 1 pmol per band. Whereas chemical

263

microsequencing has long been the technique of choice for protein identification [10], femtomole-level analysis can only be done using a mass spectrometric approach. ‘‘Mass fingerprinting’’ uses a small set of proteolytic fragment masses—five or six, but accurate to within 30– 50 ppm—to establish identity [11,12]. Identification is most easily done by matrix-assisted laser desorption/ ionization (MALDI) reflectron time-of-flight (reTOF) mass spectrometry (MS). Such a data set does not, however, allow querying an expressed sequence tag (EST) database, nor is it usually enough to cope with mixed or contaminated proteins. In such instances, a multimode MS approach, particularly including electrospray ionization (ESI) tandem MS, is preferred [13]. Tandem MS enables protein identification on the basis of fragmentation data from a single, derivative peptide [14,15]. MS-based protein identification is multistep and all steps are critical to the successful outcome of the analysis. In general, mass spectrometers are not very tolerant of particulates, salts, detergents, buffer components, and any or all ionizable molecules other than peptides (this in case of peptide analysis, of course). Furthermore, it is a concentration-sensitive technique, requiring the sample to be presented for analysis in the smallest possible volume. As a rule, sample handling and adsorptive losses to tubes and pipet tips should be kept to a minimum, chemical modifications avoided, and no extraneous proteins introduced. Using the purified Elongator complex as an example, we consider these steps in turn. 2.2.1. In-gel tryptic digestion The purified Elongator-associated proteins were resolved by 10% SDS–PAGE and stained with Coomassie Blue R–250. Tryptic digests were carried out using the following protocol (Coomassie-stained gel piece 6 10
3
1 mm). 1. Destain (plus remove SDS) with 50% methanol (v/v), vortex, place in 37 °C bath for 15 min, spin down, and remove supernatant. Repeat three times. 2. If still blue, wash with 50% acetonitrile in 0.2 M ammonium bicarbonate. Note: All color and SDS must be removed from the gel pieces or the residual stain will cause peptide losses during later RP microtip step. 3. Rinse the gel four times with Milli-Q water. 4. Using a designated (and clean) Petri dish, tweezer, and scalpel, dice the gel to  1-mm3 pieces. Note: It is important that the gel does not become too dry during this procedure; otherwise it will be difficult to gather the pieces since they become hardened and sticky. On the other hand, the gel should not be too wet because this will delay drying down and promote losses (see next step). 5. Speed-Vac the sample to dryness for 15 min.

264

G. Sebastiaan Winkler et al. / Methods 26 (2002) 260–269

6. Once the gel pieces are semidry to bone dry, add 0:2 lg trypsin (Promega; modified trypsin, porcine) in 5 lL digest solution: 0.02% (w/v) Zwittergent 3– 16 (Calbiochem) in 0.1 M ammonium bicarbonate (NH4 HCO3 ). 7. Allow sufficient time for the enzyme solution to be entirely adsorbed by the dry gel. 8. Add sufficient digest solution (Zwittergent included) to reswell the gel pieces to the original volume and then completely submerge them. However, the final volume of supernatant should ideally not exceed 50 lL. Note: The precise (0.02%) Zwittergent 3–16 concentration and supernatant volume ð650 lLÞ are critical for the success of the microtip cleanup procedure. 9. Incubate for 2 h at 37 °C. 10. Sonicate (in bath) for 5 min; spin down for 1 min in Eppendorf centrifuge. 11. Draw up the entire supernatant (<50 lL) and immediately deposit on a 2-lL bed volume RP microtip for cleanup (see text). This will then provide desalted, concentrated peptide pools for MALDI re-TOF MS and CF NanoES-MS/MS analysis. The entire procedure should ideally be done inside a ‘‘clean box,’’ for instance, a PCR AirClean Systems workstation, and using dedicated handling tools and plasticware. It is especially important that all dye and SDS be removed from the gel pieces as we have observed these to adversely affect sample cleanup (microtip) afterward, resulting in low recoveries. Note that, after the initial drying and reswelling in trypsin containing buffer, no further drying steps are included anywhere in the procedure, all the way to the introduction into the mass spectrometers. We strongly recommend the inclusion of 0.02% Zwittergent 3–16 (a zwitterionic detergent) in the digest buffers for its unique ability to prevent adsorptive losses of even lowfemtomole amounts of peptides [16]. Zwittergent is conveniently, and completely, removed by passage over a reversed-phase microtip, a step already required for peptide desalting and concentration. However, care should be taken not to saturate the RP tip. As a rule, a 2-lL bed volume of Poros R2 beads can easily bind all Zwittergent from 50 lL of a 0.02% solution (or 1 lL of 1% Zwit). As such, there are limitations on what volume of gel pieces can be processed. Combining several bands from parallel lanes in a gel may result in volumes exceeding 50–100 mm3 , requiring 100–200 lL of digest buffer. Aside from the dilution factor and losses, the total amount of Zwittergent will hereby exceed microtip capacity, impeding sample preparation. 2.2.2. Modified silver stain and bleach Gel bands stained with Coomassie R-250, or its colloidal form (Pierce), are ideal for MS-based protein

identification. The Coomassie stain detection limit of an average band, <1 cm wide on a 0.5- to 1-mm-thick gel, is about 25–50 ng. For proteins of molecular weights under 50 kDa this amount represents about 1 pmol or more; in excess of what is ideally needed for simple mass spectrometric identification (0.5 pmol). In these cases, silver staining will be necessary and is recommended. However, some restrictions apply. The proteins under study should not be chemically modified, except perhaps to reduce and alkylate cysteine residues, which may be helpful in some cases for the digest to proceed optimally. Inadvertent covalent modifications will result in changes in peptide molecular mass, precluding ‘‘peptide fingerprinting’’-type database searches (see below). Treating proteins with an excess of aldehydes, a procedure commonly used in most commercial silver stains, may result in Schiff base formation with primary amines such as lysine side chains, interfering with subsequent tryptic digestion (at the C terminus of Lys and Arg) and accurate mass measurement. A modified silver stain has been developed [17] to minimize such effects. Second, it has been noted that silver staining, including the modified version, has adverse effects on overall peptide recovery after digestion. This effect progressively worsens with the duration of exposure (i.e., development time), as shown by comparative, quantitative analysis (L.L., H.E.B., and P.T., unpublished results). It can be remedied, to some extent, by bleaching the silver [18]. 1. Fix gel in 40% ethanol/10% acetic acid in water (v/v), overnight. 2. Wash for 10 min in 50% methanol (v/v). 3. Wash with water for 10 min to remove residual acid. 4. Sensitize gel by a 1-min incubation in 0.02% (w/v) sodium thiosulfate (Sigma, Catalog No. S8503). 5. Rinse with two changes of Milli-Q water for 1 min each. 6. Submerge gel and incubate in 0.1% (w/v) silver nitrate (Aldrich Chem. Co., Catalog No. 20,505-2) for 20 min at 4 °C (e.g., cold room). 7. Discard silver nitrate and rinse gel twice with Milli-Q water for 1 min each. 8. Develop gel in 0.04% formaldehyde [dilute 108 lL of commercially available 37% formaldehyde (Sigma, Catalog No. F-1268) in 100 mL of 2% (w/v) sodium carbonate (Fisher Sci., Catalog No. S263-3)] under vigorous shaking. Note: After the developer turns yellow, it should be immediately discarded and replaced with fresh solution. 9. Terminate the development by discarding the reagent and storing the gel in 1% acetic acid at 4 °C until further analysis. Note: Storage in glycerol solution is not recommended, as it interferes with proteolytic digests.

G. Sebastiaan Winkler et al. / Methods 26 (2002) 260–269

265

Table 1 Mass spectrometric protein identification: Sloan–Kettering System 1. Protein(s) in gel (see text) 2. Tryptic digest (see text) 3. Peptide pools, 3–4 lL volume (see text) 4. MALDI-reTOF MS (take 0.5 lL of step 3) 4.1. Extract mass (m=z) list; designate major, medium, minor peaks 4.2. Subtract background peaks (trypsin autolytic, keratin,. . .): go to step 5.1 5. MALDI List(s) and SEARCH (‘‘Peptide Fingerprinting’’) 5.1. Master list 5.2. Subtracted list(s) 5.2.1. Sub-list 1 5.2.2. . Sub-list 2 .. 5.2.n. Sub-list n; until no more unaccounted peaks 5.3. SEARCH 5.3.1. Programs: PeptideSearch (EMBL, Protana); MASCOT (Matrix Sci.) 5.3.2. Parameters: 30–50 ppm accuracy; 1–2 missed cleavage sites 5.3.3. Database(s): NR (NCBI)—integral, or species-separated go to step 6 6. Identification 6.1. Yes (top match clearly separated from rest): go to step 7 6.2. Maybe (top match not much separated from lesser matches, but separated from noise): go to step 8 6.3. No (no top match; some matches can be separated from noise, or not at all): 6.3.1. Perform ‘‘Functional’’ inspection of list by (i) domain expert or (ii) use of PINdb 6.3.2. Candidate proteins? (TFs, coactivators, repressors, other nuclear proteins, etc.): go to step 8 6.3.3. Still no candidates: go to step 9 7. Reinspect MALDI spectra 7.1. All peaks accounted for? Matched peaks: major/medium/minor (see step 4.1)? If Yes: STOP If NO: go to step 7.2 7.2. Subtract matched peaks from master list (step 5.1); create new Sub-list 1; 2; 3; . . . ; n: go to step 5.2 and repeat search 8. Confirm fingerprint ID by ESI-MS/MS 8.1. Take MALDI master step 5.1 or subtracted (step 5.2): List; convert major m/z from MHþ to MH2þ and MH3þ ; select 8.2. CF NanoESI-MS/MS (1; 2; 3; . . . ; peptides); take 1.5 lL of step 3 8.3. Confirmed? 8.3.1. Yes: go to step 7 8.3.2. No: go to step 6.3 9. CF NanoESI-MS (and MS/MS) (take 1.5 lL of step 3) 9.1. ESI-MS: find major precursor ions; compare with master list (step 5.1) or subtracted list (step 5.2); select precursor ions for MS/MS 9.2. MS/MS 9.3. SEARCH (with fragment ions) 9.3.1. Programs: SequenceTag (EMBL; Protana); PepFrag (PROWL, Proteometrics) 9.3.2. Input: m=z (MALDI), y 00 -ions 9.3.3. Databases: NR/species (NCBI); dbEST (NCBI) 9.4. Identification 9.4.1. Yes (single ID; clear y 00 -ion series); Confirm with 2nd (and 3rd. . . MS/MS): go to step 9.1 pick precursor ion predicted to match putatively identified protein 9.4.1.1. Yes: go to step 7 9.4.1.2. No: go to step 9.1 and start over 9.4.2. No (no signal; noninterpretable; or many ‘‘IDs’’): go to step 9.1 and pick other precursor ion

10. Once bands/spots of interest have been excised from the gel, prepare destain (‘‘bleach’’) solutions (A and B) fresh each time, as follows: A. 30 mM Potassium ferricyanide ½K3 FeðCNÞ6  (Sigma): Weigh out 0.05 g and dissolve in 5 mL Milli-Q water.

B. 100 mM Sodium thiosulfate pentahydrate ðNa2 S2 O3  5H2 OÞ (Sigma): Weigh out 0.0124 g and dissolve in 5 mL Milli-Q water. 11. Mix equal (500-lL) volumes of solutions A and B, cover each gel piece with destain solution, and vortex lightly until stain disappears.

266

G. Sebastiaan Winkler et al. / Methods 26 (2002) 260–269

12. Rinse approximately five times with water. 13. Cover each gel piece with 0.2 M NH4 HCO3 in 50% acetonitrile (v/v). 14. Incubate at 37 °C for 15 min. 15. Rinse approximately five times with water. 16. Cut (dice) gel band and prepare as usual for digest (see above). It is recommended that this procedure be followed closely, as even small deviations may result in later failure to identify the proteins. As a rule, gels should only be lightly stained with silver, which will pick up protein bands in the very low nanogram range. Anything below that may not be enough for analysis anyway. Proteins of apparent molecular mass over 100 kDa should be visible either by Coomassie or after very brief silver development to ensure successful identification. When not sure, Coomassie staining can be done first followed by silver staining. 2.2.3. Peptide sample preparation The digest mixture resulting from in-gel proteolysis cannot be directly introduced into a mass spectrometer as it is too dilute and contains many constituents interfering with analysis. Sample cleanup and concentration is now almost universally done using reversedphase microtips (i.e., with 1- to 2-lL bed volumes), either homemade devices [19] or commercially available Zip-Tips (Millipore). As a result, the peptide mixtures are then presented for analysis while dissolved in 0.1% formic acid/30% acetonitrile (in water); the solvent required for elution from the reversed-phase beads. Organic solvents promote easy desolvation and formation of gaseous ions, a prerequisite for mass analysis during electrospray ionization. Acid promotes solubility; and the low pH ensures almost universal protonation of all peptides, enabling operation of any type of mass spectrometer in the positive ion mode at all times. In general, we elute peptides from the RP tip in two batches of 3–4 lL each; the shorter and hydrophilic ones at 16% acetonitrile and the longer and hydrophobic ones at 30%. We refer to these successive eluates as the ‘‘16% pool’’ and ‘‘30% pool.’’ As with all other forms of reversed-phase packings and chromatography, SDS has a very detrimental effect on analyte separation and recovery, and should therefore be fully removed prior to loading of the tip. Further details about the procedure are beyond the scope of this report and can be found in an earlier publication by some of the authors [19]. 2.2.4. MALDI-TOF mass spectrometry A homogeneous protein, with archived sequence, contained within a single gel band in P0:25- to 0.5pmol amounts, and prepared for mass spectrometric analysis as described in the previous sections can be readily identified by simple peptide mass fingerprinting

as outlined in Table 1 (steps 4–7). For MALDI-TOF MS, the sample is introduced as a solid crystal, usually in a 0.5-lL volume or less. This technique is relatively simple, very accurate (in reflector mode), and very sensitive. For example, the 50-kDa Elongator-complex protein that eluted from the Ni-NTA column yielded a distinct series of tryptic peptide ions in the combined (16% + 30% pools) spectra (Fig. 3). Twelve of the fifteen selected, major peaks had m=z values that matched, to within 40-ppm accuracy, the calculated monoisotopic masses of predicted, singly charged [MHþ ] tryptic peptides of the yeast hypothetical protein YRL101w (theoretical molecular mass of 51,156 Da), providing 26% sequence coverage, with stretches of double or triple overlap. By contrast, random matches (‘‘noise’’) with all other yeast proteins of molecular mass 6100 kDa, at 40-ppm accuracy and with at maximum one missed cleavage site, numbered 63 of 15 (with less than 8% sequence coverage and no overlaps). The three unaccounted m=z values could not be matched to any other yeast proteins and must therefore derive either from modified peptides or from a nonyeast source. In our experience, an identification as presented here is to be considered one of ‘‘high confidence’’ (see step 6.1 in Table 1). Experimental details of the analysis and database search can be found in Fig. 3. 2.2.5. Multimode mass spectrometry In many instances, initial MALDI-TOF analysis and peptide fingerprint searching will not yield a clear-cut result, in that two or more proteins appear as top candidates (Table 1, step 6.2); sometimes no real candidates emerge (Table 1, step 6.3). Several options are available at this point, as presented in Table 1; all require additional analysis using tandem mass spectrometry (also known as MS/MS). This could be either to confirm a lead (Table 1, step 8) or to make a completely independent attempt at identification (step 9). Leads to identifying a protein may be obtained from inconclusive ‘‘first-pass’’ peptide fingerprint searches, whereby the return list has been inspected for potential candidates on the basis of ascribed function (e.g., transcription) or cellular location (e.g., nucleus). Most commercially available tandem mass spectrometers use electrospray ionization (ESI), which involves introduction of the sample in liquid form and, ideally, at a very low flow (nanoliters/minute), coupled to triple-quadrupole and hybrid quadrupole-TOF mass analyzers. It enables protein identification on the basis of fragmentation (mass) data derived from a single tryptic peptide [14,15]. This peptide need not be purified before analysis as the MS/MS method allows selection of peptides of any particular mass out of complex mixtures. Subsequent fragmentation results from high-speed collisions with gas molecules, which induce cleavages across the amide backbone of a peptide and provide ion series

G. Sebastiaan Winkler et al. / Methods 26 (2002) 260–269

267

Fig. 3. Identification of Elongator ‘‘p50’’ by MALDI-reTOF MS and peptide mass fingerprint search. The gel-bound protein was digested and peptides were prepared for mass analysis as described in the text. MALDI-reTOF mass spectra of the (A) 16% peptide pool and (B) 30% pool. Note that the spectra shown here were the result of a duplicate analysis in the absence of calibrants. Major peaks corresponding to m=z (mass-to-charge) values that could be matched to predicted tryptic peptides from yeast protein YRL101w are labeled with closed circles; those that remained unaccounted for are indicated with open circles. Experimental: Each peptide pool was analyzed twice by MALDI-reTOF MS, in the presence and absence of peptide calibrants [19]. Aliquots (0.5 lL) were deposited on the probe surface, mixed with a-cyano-4-hydroxycinnamic acid solution (MALDI-Quality, Bruker-Daltonics, Billerica, MA) on the plate, and allowed to dry at room temperature; calibrants were diluted from concentrated stocks and mixed to yield 6.25 fmol of each per 0.2-lL volume of the same solvent prior to mixing with the analytes. Spectra were acquired on a REFLEX III (Bruker-Franzen, Bremen, Germany) instrument equipped with a 337-nm nitrogen laser, a gridless pulsed-extraction ion source, and a 2-GHz digitizer. The instrument was operated in reflector mode; 25-kV ion acceleration, 26.25-kV reflector, and )1.4-kV multiplier voltages were used. Ion extraction was done 200 ns after each laser irradiance by pulsing down the source extraction lens to 17.7 kV from its initial 25-kV level to give appropriate time-lag focus conditions at the detector. Spectra were obtained by averaging multiple signals; laser irradiance and number of acquisitions (typically 150) were operator adjusted to yield maximal peak deflections, derived from the digitizer as TOF data and displayed in real time as mass spectra using a SPARC Station 5 (Sun Microsystems; Mountain View, CA). After recalibration with internal standards, monoisotopic masses were assigned for all prominent peaks (marked with closed or open circles) over background, and a peptide mass list was generated. This list was taken to search the yeast segment of a protein nonredundant database (NR, National Center for Biotechnology Information, Bethesda, MD) using the PeptideSearch [11] algorithm. A molecular mass range up to 100 kDa was covered, with a mass accuracy restriction of 40 ppm or better and a maximum of one missed cleavage site allowed per peptide. The search program can be downloaded over the Internet from the following site: http:// www.narrador.emblheidelberg.de/GroupPages/PageLink/peptidesearchpage.htm/. Or the search can be done directly at the following location: http:// peptsearch.protana.com/FR_PeptideSearchForm.html.

corresponding to the amino acid sequence. The process is often referred to as ‘‘sequencing,’’ but it rarely ever yields bona fide sequence (e.g., 10–20 residues) that could be used for a BLAST search, and then only if ample quantities are available. Rather, the data are more often used to confirm a known or surmised structure. Identifications based solely on MS/MS data are certainly achievable (as will be shown here), but single-peptide matches, even when statistically sound, must be interpreted with caution; two seems a dependable minimum. A case in point was the analysis of the 30-kDa Elongator-complex component. MALDI-TOF analysis provided us with 15 major peptide ions (Fig. 4A) that were used to search the yeast database, resulting in a list of seven candidate proteins with matches (P4 of 15) above background (63 of 15). Three of those had predicted molecular masses between 29 and 31 kDa, namely, two ribosomal proteins and one unannotated, hypothetical protein (i.e., the putative product of an open reading frame). To find out which of the three proteins was present, a second aliquot of the peptide mixture was taken for continuous flow (CF) NanoESI-

MS analysis: first in single stage (Q1) scan mode (Fig. 4B), then for tandem MS of selected precursor ions (Fig. 4C). The MALDI-TOF results were first taken to select ions that are predicted to correspond to a specific tryptic peptide from each of the three proteins. MALDI ions are almost always singly protonated (m ¼ ½M þ Hþ , z ¼ 1) and must be converted to the predicted double (m ¼ ½M þ 2H2þ , z ¼ 2) or triple (m ¼ ½M þ 3H3þ , z ¼ 3) positive charge states, expected to result from electrospray ionization, before locating them on the ESI-MS (Q1) scan. For instance, for the hypothetical protein YMR312w, we focused on a MALDI ion of m=z ¼ 1086:53 (which should correspond to the sequence DVTGSLHVCR if the protein was indeed identified correctly), which should correspond to a precursor ion of m=z ¼ 543:8 in the Q1 scan (indicated as 543:82þ in Figs. 4B, C), which was then selected for tandem MS analysis. The result is shown in Fig. 4C, and indicates the presence of a y 00 -ion series (i.e., collisioninduced fragment ions that all have the C-terminal residue in common, but differ in length at their N termini). The mass differences between successive y 00 ions equals

268

G. Sebastiaan Winkler et al. / Methods 26 (2002) 260–269

Fig. 4. Analysis of Elongator ‘‘p30’’ protein composition by multimode mass spectrometry. The gel-bound protein was digested and peptides were prepared for mass analysis as described in the text. (A) MALDI-reTOF mass spectrum of the 30% peptide pool. Major peaks corresponding to m=z values that could be matched to predicted tryptic peptides from yeast protein YMR312w are marked with filled squares, those matching ribosomal protein S4A with closed circles, and those matching ribosomal protein S1B with closed triangles. The ion (m=z ¼ 1086:53) indicated with the arrow corresponds to the peptide ‘‘DVTGSLHVCR.’’ (B) Continuous-flow (CF) NanoESI single-quadrupole scan of the same 30% pool. (C) CF NanoESI MS/MS of a doubly charged precursor ion, labeled ‘‘543:82þ ’’ in (B) and (C) (and corresponding to the singly charged ion labeled ‘‘1086.53’’ in (A)). A limited y 00 -ion series is indicated, plus the corresponding peptide sequence that could be deduced. Note that the sequence reads backward. Experimental: MALDI-reTOF MS analysis was as described under Fig. 3. ESI-MS (and MS/MS) was done on an API 300 triple-quadrupole instrument (Applied Biosystems/MDS-SCIEX, Thornhill, Canada), modified with a continuous-flow NanoES source as described [33]. Needle voltage ranged from 600 to 1000 V; the voltages for the orifice and the curtain plate were set at 5 and 350 V, respectively. Q1 scans were collected using a 0.5-amu step size and a 3-ms dwell time over a mass range from 400 to 1400 amu. Scans were averaged for statistical analysis, and Q1 resolution was set such that the charge state of singly, doubly, and triply charged ions could be ascertained. For operation in the MS/MS mode, Q1 was set to transmit the complete isotopic envelope of the parent. All spectra were averaged with a 0.5-Da step size and a 3-ms dwell time for 5 min over the mass range of the singly charged m=z. Q3 resolution was set such that the charge state of the fragment ions could be distinguished. Collision energies, as well as CAD gas pressures, were optimized individually for each peptide so as to obtain the best MS/MS. Spectra were inspected for uninterrupted y 00 -ion series using the ‘‘find higher AAs’’ routine of the BioToolbox (PE-SCIEX) software. This limited information was taken (together with precursor ion mass) to search the NR database using the PepFrag protein identification program from the PROWL resource available over the Internet (http:// prowl.rockefeller.edu/PROWL/pepfragch.html). Any protein identification thus obtained was verified by comparing the computer-generated fragment ion series of the predicted tryptic peptide with the experimental MS/MS data.

the molecular mass of individual amino acids, enabling us to read limited sequence (indicated in Fig. 4C). Please note that because of the C terminal anchor point, the sequence reads backward (right to left). The outcome of the analysis did not leave any doubt about sequence and identity of peptide ‘‘543:82þ ,’’ thus confirming the tentative identification of yeast protein YMR312w. This routine was then repeated for each of the other two predicted proteins (ribosomal proteins S4A and S1B) as well, excluding these candidates.

3. Concluding remarks The procedure as outlined here is a generic strategy to isolate novel protein complexes and identify the individual components by mass spectrometry. This method has been successfully employed with several different target proteins. The HA and histidine tags need not be on the same protein, allowing the isolation of different complexes with common subunits. For proteins prepared in this manner, especially yeast proteins, mass spectro-

G. Sebastiaan Winkler et al. / Methods 26 (2002) 260–269

metric identification should be relatively straightforward in most cases, and with good success, provided some basic rules are followed. This comes down primarily to avoiding the three major ‘‘No’s’’ of protein and peptide nanosample handling: no introduction of keratin (e.g., fingerprints, hair) at any stage; no large volumes anywhere, beginning with loading of the gels (one lane only); no overstaining with silver and/or staining with ‘‘bad’’ silver. By doing so, we have identified hundreds of proteins from many complexes, without the aid of any robotic devices or automated data acquisition and analysis schemes. During the past 5 years, more than 10 transcription-related complexes were thus successfully dissected using these procedures, leading to the discovery or new functional assignments of 130 proteins (see Refs. [6,7,9,20–32]). This is in addition to the 17 single factors involved in transcriptional regulation (published elsewhere) and more than 100 proteins from various transcription complexes currently under study. As many complexes, or ‘‘functional modules,’’ of proteins that govern gene expression can be predicted to exist, experimental approaches of affinity capture and dissection by mass spectrometry, as described here, will likely become standard procedures in the transcription field in the near future. Acknowledgments Work in our laboratories was supported by grants from the Imperial Cancer Research Fund and the Human Frontier Science Program Project RG0193/97 (to J.Q.S.) and by NCI Core Grant P30 CA08748 (to P.T.). G.S.W. was supported by an EMBO Long Term Fellowship. References [1] J. Field, J. Nikawa, D. Broek, B. MacDonald, L. Rodgers, I.A. Wilson, R.A. Lerner, M. Wigler, Mol. Cell. Biol. 8 (1988) 2159– 2165. [2] R. Janknecht, G. de Martynoff, J. Lou, R.A. Hipskind, A. Nordheim, H.G. Stunnenberg, Proc. Natl. Acad. Sci. USA 88 (1991) 8972–8976. [3] J.Q. Svejstrup, W.J. Feaver, J. LaPointe, R.D. Kornberg, J. Biol. Chem. 269 (1994) 28044–28048. [4] G.S. Winkler, W. Vermeulen, F. Coin, J.M. Egly, J.H.J. Hoeijmakers, G. Weeda, J. Biol. Chem. 273 (1998) 1092–1098. [5] C.M. Green, H. Erdjument-Bromage, P. Tempst, N.F. Lowndes, Curr. Biol. 10 (2000) 39–42. [6] S. Chavez, T. Beilharz, A.G. Rondon, H. Erdjument-Bromage, P. Tempst, J.Q. Svejstrup, T. Lithgow, A. Aguilera, EMBO J. 19 (2000) 5824–5834. [7] G.S. Winkler, T.G. Petrakis, S. Ethelberg, M. ToKunaga, H. Erdjument-Bromage, P. Tempst, J.Q. Svejstrup, J. Biol. Chem. 276 (2001) 32743–32749.

269

[8] E.W. Jones, Genetics 85 (1977) 23–33. [9] G. Otero, J. Fellows, Y. Li, T. de Bizemont, A.M. Dirac, C.M. Gustafsson, H. Erdjument-Bromage, P. Tempst, J.Q. Svejstrup, Mol. Cell 3 (1999) 109–118. [10] H. Erdjument-Bromage, M. Lui, D.M. Sabatini, S.H. Snyder, P. Tempst, Protein Sci. 3 (1994) 2435–2446. [11] M. Mann, P. Højrup, P. Roepstorff, Biol. Mass Spectrom. 22 (1993) 338–345. [12] D.J.C. Pappin, P. Højrup, A.J. Bleasby, Curr. Biol. 3 (1993) 327– 332. [13] G. Neubauer, A. King, J. Rappsilber, C. Calvio, M. Watson, P. Ajuh, J. Sleeman, A. Lamond, M. Mann, Nat. Genet. 20 (1998) 46–50. [14] J.K. Eng, A.L. McCormack, J.R. Yates III, J. Am. Soc. Mass Spectrom. 5 (1994) 976–989. [15] M. Mann, M. Wilm, Anal. Chem. 66 (1994) 4390–4399. [16] M. Lui, P. Tempst, H. Erdjument-Bromage, Anal. Biochem. 241 (1996) 156–166. [17] A. Shevchenko, M. Wilm, O. Vorm, M. Mann, Anal. Chem. 68 (1996) 850–858. [18] F. Gharahdaghi, C.R. Weinberg, D.A. Meagher, B.S. Imai, S.M. Mische, Electrophoresis 20 (1999) 601–605. [19] H. Erdjument-Bromage, M. Lui, L. Lacomis, A. Grewal, R.S. Annan, D.E. MacNulty, S.A. Carr, P. Tempst, J. Chromatogr. 826 (1998) 167–181. [20] B.R. Cairns, Y. Lorch, Y. Li, M. Zhang, L. Lacomis, H. Erdjument-Bromage, P. Tempst, J. Du, B. Laurent, R.D. Kornberg, Cell 87 (1996) 1249–1260. [21] B.Ø. Wittschieben, G. Otero, T. de Bizemont, J. Fellows, H. Erdjument-Bromage, R. Ohba, Y. Li, C.D. Allis, P. Tempst, J.Q. Svejstrup, Mol. Cell 4 (1999) 123–128. [22] J. Fellows, H. Erdjument-Bromage, P. Tempst, J.Q. Svejstrup, J. Biol. Chem. 275 (2000) 12896–12899. [23] L.C. Myers, C.M. Gustafsson, D.A. Bushnell, M. Lui, H. Erdjument-Bromage, P. Tempst, R.D. Kornberg, Genes Dev. 12 (1998) 45–54. [24] C.M. Gustafsson, L.C. Myers, J. Beve, H. Sp ahr, M. Lui, H. Erdjument-Bromage, P. Tempst, R.D. Kornberg, J. Biol. Chem. 273 (1998) 30851–30854. [25] Y.W. Jiang, P. Veschambre, H. Erdjument-Bromage, P. Tempst, J.W. Conaway, R.C. Conaway, R.D. Kornberg, Proc. Natl. Acad. Sci. USA 95 (1998) 8538–8543. [26] C. Rachez, B.D. Lemon, Z. Suldan, V. Bromleigh, M. Gamble, A.M. N€a€ar, H. Erdjument-Bromage, P. Tempst, L.P. Freedman, Nature 398 (1999) 824–828. [27] Y. Zhang, R. Iratni, H. Erdjument-Bromage, P. Tempst, D. Reinberg, Cell 89 (1997) 357–364. [28] Y. Zhang, Z.-W. Sun, R. Iratni, H. Erdjument-Bromage, P. Tempst, M. Hampsey, D. Reinberg, Mol. Cell 1 (1998) 1021– 1031. [29] H.-H. Ng, Y. Zhang, B. Hendrich, C.A. Johnson, B.M. Turner, H. Erdjument-Bromage, P. Tempst, D. Reinberg, A. Bird, Nat. Genet. 23 (1999) 58–61. [30] Y. Zhang, H.-H. Ng, H. Erdjument-Bromage, P. Tempst, A. Bird, D. Reinberg, Genes Dev. 13 (1999) 1924–1935. [31] D.W. O’Neill, S.S. Schoetz, R.A. Lopez, M. Castle, L. Rabinowitz, E. Shor, D. Krawchuk, M.G. Goll, M. Renz, H.P. Seelig, S. Han, R.H. Seong, S.D. Park, T. Agalioti, N. Munshi, D. Thanos, H. Erdjument-Bromage, P. Tempst, A. Bank, Mol. Cell. Biol. 20 (2000) 7572–7582. [32] A.J. Saurin, Z. Shao, H. Erdjument-Bromage, P. Tempst, R.E. Kingston, Nature 412 (2001) 655–660. [33] S. Geromanos, G. Freckleton, P. Tempst, Anal. Chem. 72 (2000) 777–790.