Protein-electroblotting and -microsequencing strategies in generating protein data bases from two-dimensional gels

Proc. Nati. Acad. Sci. USA Vol. 86, pp. 7701-7705, October 1989 Biochemistry

Protein-electroblotting and -microsequencing strategies in generating protein data bases from two-dimensional gels (computerized protein data bases/human genome sequencing)

G. BAUW*, J. VAN DAMME*, M. PUYPE*, J. VANDEKERCKHOVE*, B. GESSERt, G. P. RATZt, J. B. LAURIDSENt, AND J. E. CELISt *Laboratorium voor Genetica, Rijksuniversiteit Gent, B-9000 Gent, Belgium; Aarhus University, DK-8000 Aarhus C, Denmark

tInstitute for Medical Biochemistry and Bioregulation Research Centre,

Communicated by M. Van Montagu, July 3, 1989 (received for review March 8, 1989)

ABSTRACT Coomassie blue-stained, heat-dried, and computer-imaged two-dimensional gels used to develop comprehensive human protein data bases served as the protein source to generate partial amino acid sequences. The protein spots were collected from multiple gels, rehydrated, concentrated by stacking into a new gel, electroblotted onto inert membranes, and in situ-digested with trypsin. Peptides eluting from the membranes were separated by HPLC and sequenced. Using this procedure, it was possible to generate partial sequences from 13 human proteins recorded in the amnion cell protein data base. Eight of these sequences matched those of proteins stored in data bases, demonstrating that a systematic analysis of proteins by computerized two-dimensional gel electrophoresis can be directly linked to protein microsequencing methods. The latter technique offers a unique opportunity to link information contained in protein data bases derived from the analysis of two-dimensional gels with forthcoming DNA sequence data on the human genome.

Here we describe in detail a modified version of the blotting/microsequencing procedure that allows sequence analysis of protein spots recovered from Coomassie bluestained, heat-dried 2D gels. The devised protein recovery procedure can be used to concentrate minor protein spots collected from several stained gels. In addition, by making use ofthe information stored in the comprehensive human 2D gel protein data bases (5, 6), it was possible to analyze proteins of interest by selecting tissues or cell types where a particular protein was expressed in higher amounts. The method was used to generate partial amino acid sequences of 13 human proteins: 6 cell growth/transformation-sensitive markers, 1 epithelial-specific protein, and 6 polypeptides whose relative degree of expression is not affected significantly by the growth stage of the cell (ref. 6 and references therein). The identity of most of these proteins could be determined because sequences generated from them matched those stored in protein data bases.

Two-dimensional (2D) gel electrophoresis is generally considered the method with the highest resolution for protein separation at the microgram level (1-3). It is, therefore, regularly used to study phenotypically dependent alterations of protein expression in total cellular extracts or enriched cell fractions. The complex protein patterns that may often display up to 2000 spots can be analyzed by computerimaging and the information stored in comprehensive data bases (4-9). This allows a further detailed quantitative comparison of a large number of gels and a more thorough search for (a) protein(s) whose expression is typically associated with variations in the phenotype, with differentiation, cell cycle, cell lineage, neoplastic transformation, genetic diseases, etc. (see, for instance, refs. 4 and 9). Identified marker proteins can then be further characterized by comigration experiments with known proteins or mixtures of proteins derived from isolated cell organelles (nuclei, mitochondria, Golgi, vacuoles, membranes, extracellular spaces, etc.). Alternatively, immunological cross-reactivity with specific antibodies may serve for protein identification. Systematic 2D gel protein analysis has now gained another dimension with the possibility of sequencing (major) protein spots after elution (10, 11) or electroblotting onto inert membranes (12-18). The generated NH2-terminal sequences are generally of sufficient length for a search of protein identity or similarity or for the generation of specific DNA probes for cloning purposes. Proteins that are NH2terminally blocked (either naturally or artifactually) are cleaved in situ and sequences are obtained from the generated peptides (16, 19, 20).

MATERIALS AND METHODS Cultured Cells and Tissues. Human MOLT-4 cells were grown in suspension in Dulbecco's modified Eagle's medium containing 10% (vol/vol) fetal calf serum and antibiotics (penicillin at 100 units/ml and streptomycin at 50 ,ug/ml). Fetal human tissues dissected from a 4-month normal human male fetus were used in some cases as source of proteins. These experiments have been approved by the Ethical Scientific Committee of the Aarhus Amtskommune. 2D Gel Electrophoresis. The procedures for computerized 2D gel electrophoresis have been described elsewhere (6, 21). Some of the protein preparations were partially enriched by ammonium sulfate fractionation and ion-exchange chromatography (22) or by cell organelle fractionation (B.G. and J.E.C., unpublished data), prior to 2D gel electrophoresis (23) (for details, see Table 1). Protein Recovery from Dried 2D Gels. Protein spots from Coomassie blue-stained and heat-dried gels were excised with a minimum of polyacrylamide and submerged in 50 mM boric acid (adjusted to pH 8.0 with NaOH) containing 0.1% SDS (buffer A). The buffer volume (1-2 ml) was not critical as protein losses due to elution were found to be minimal in this particular buffer system. After 2 hr of rehydration, the swollen gel pieces were taken up with tweezers and placed in a gel slot of a new slab gel. This gel was cast using spacers that were 0.5 mm thicker than those of the original 2D gel so as to facilitate loading of the pieces. Here, we used 1mm-thick 2D gels and 1.5-mm-thick "concentrating" gels. The stacking gel (5% polyacrylamide) extends 2 cm beneath the bottom of the slot, which is 6 mm broad and between 1.5

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Abbreviations: IEF, isoelectric focusing; PVDF; polyvinylidene

difluoride; NEPHGE, nonequilibrium pH gel electrophoresis; 2D, two-dimensional. 7701

7702

cm and 3 cm deep (depending on the number of gel pieces collected in the slot). The separation gel was as described by Laemmli (24). The gel pieces (up to 15 pieces could be combined in a single slot) were then overlayed with gel sample buffer [1% SDS/10% (vol/vol) glycerol/0.1% bromophenol blue/50 mM dithiothreitol/13 mM Tris HCl, pH 6.8) and occasionally trapped air bubbles were removed. Gel electrophoresis was carried out until the dye reached the bottom of the separation gel. Protein Electrotransfer. Proteins were electroblotted onto glassfiber sheets coated with poly(4-vinyl-N-methylpyridinium iodide) or polyvinylidene difluoride (PVDF) membranes as described (15) using 50 mM Tris/50 mM boric acid (pH 8.3) as transfer buffer. The transfer was carried out for at least 8 hr at 35 V by using a Bio-Rad Transblot apparatus. Gels containing >12.5% polyacrylamide were equilibrated for 2 hr in buffer A (see above) before transfer to minimize gel distortion during the blotting process. PVDF-membrane- and glassfiber-membrane-bound proteins were visualized by staining with Amido black and fluorescamine, respectively (12, 15). Membrane in Situ Protease Cleavage. Proteolytic digestion was carried out as described (19). Briefly, the membrane piece carrying the protein was excised, cut into pieces of approximately 3 by 3 mm, and collected in an Eppendorf tube. They were then immersed in 200-500 Al of a 0.2% polyvinylpyrrolidone (30 kDa) solution in methanol. After 30 min, the quenching mixture was diluted with an equal volume of distilled water and further incubated for 5-10 min. The supernatant was then discarded and the membrane pieces were washed four times with 200-500 ,ul water and once with 500 ttl 0.1 M Tris HCl (pH 8.5, buffer B). The buffer was removed and replaced by a volume of the same buffer enough IEF a,¢|;;asS~~~~~~~~~~~~~~~~~~~o

Cl)

iF

W,_f*m$'. :

Proc. Natl. Acad. Sci. USA 86 (1989)

Biochemistry: Bauw et al.

_ .t

....

i5 c

to submerge the membrane pieces (between 100 and 150 ,ul). To this was added 1 ,ul of a freshly prepared solution of trypsin at 1 mg/ml in buffer B. The digestion proceeded for 4 hr at 37°C. The supernatant was then transferred into a second Eppendorf tube and the membrane pieces were further washed once with 100,ul of 80% (vol/vol) formic acid and four times with 100 ,ul of distilled water. All washing solutions were added to the digestion mixture in the second Eppendorf tube. In some cases, the peptide solution was stored at -20°C until HPLC analysis. In situ digestion on coated glassfiber sheets was carried out as for PVDF membranes, except that quenching of remaining protein-absorbing sites was done with 0.2% polyvinylpyrrolidone/50% (vol/vol) methanol/50% (vol/vol) H20. Peptide Separation by Reversed-Phase HPLC. The combined washing solutions (±600 ,l) were loaded on a C4 reversed-phase column (0.46 x 25 cm; Vydac Separations Group) and the peptides were eluted with a linearly increasing gradient of acetonitrile in 0.1% trifluoroacetic acid. The column was equilibrated in 0.1% trifluoroacetic acid; the gradient was started 5 min after injection and reached 70% (vol/vol) acetonitrile after an additional 70 min. Eluting peptides were detected by UV absorbance at 214 nm (absorbance units full scale: 0.1 or 0.2) and collected by hand in Eppendorf tubes. A Waters-Millipore HPLC apparatus consisting of two pumps (model 510), a gradient controller, and a model 481 variable wavelength detector was used for these separations. The major peptides were further dried in a SpeedVac (Savant) concentrator and stored at -20°C prior to sequence analysis. Amino Acid Sequencing. Peptides were selected for amino acid sequence analysis on the basis of peak height and peak resolution and were redissolved in 30 ,l4 of 0. 1% trifluoracetic

_

_;.

t. o ... ..

110

-68

...

-55

_

o

x

...

..

o,

_

v

I

-

:

9

-43

-36

-30

FIG. 1. Fraction of a synthetic image of an IEF fluorogram of [35S]methionine-labeled proteins from amnion cells (for the complete 2D gel pattern, see ref. 5). Proteins that have been microsequenced are indicated with their corresponding number in the comprehensive human amnion protein data base (5). MW, Mr.

Proc. Natl. Acad. Sci. USA 86 (1989)

Biochemistry: Bauw et al. acid/30% acetonitrile before loading on a precycled Polybrene-coated glass filter. The sequence analysis was carried out with a gas-phase sequencer (model 470A, Applied Biosystems) equipped with an on-line phenylthiohydantoin amino acid derivative analyzer (model 120A) and run with the sequencing program recommended by the manufacturer. Computer Search for Identity or Similarity. The amino acid sequence comparisons were carried out using the FASTDB computer program of Intelligenetics. Two protein data bases were screened, the Protein Identification Resources (release 18) from the National Biochemical Research Foundation and the Swiss-Prot data base (release 9) from EMBL. RESULTS Fig. 1 shows the imaged 2D gel isoelectric focusing (IEF) protein pattern of a [35S]methionine-labeled total extract of transformed human amnion cells (AMA). Proteins selected

7703

for microsequencing included six proliferation-sensitive and/ or transformation-sensitive polypeptides (IEFs 6318, 7205, 8214, 8505, and 9109 and NEPHGE 3004) (7), an epithelial marker (IEF 9105), and six polypeptides (IEFs 8502, 8704, 9105, 9205, 9209, and 9806) whose rate of synthesis is not affected significantly by changes in growth rate and/or transformation (NEPHGE is nonequilibrium pH gel electrophoresis). To facilitate microsequencing, these proteins were cut out from Coomassie blue-stained gels of (i) partially purified protein fractions of human MOLT-4 cells (IEFs 6318, 7205, 8502, 8505, 8704, 9205, 9209, 9109, and 9806 and NEPHGE 3004; Table 1) or (it) total protein extracts of fetal human tissues (IEFs 5206, 8214, and 9105; see Table 1). With the exception of NEPHGE 3004, all other microsequenced proteins are indicated with their corresponding number in the master AMA protein data base (Fig. 1) (5). Gel pieces excised from 3 to 15 dry gels depending on their abundance were re-eluted and concentrated in a one-

Table 1. Partial amino acid sequences generated from proteins isolated from two-dimensional gels Ref. number in Molecular AMA protein mass, Protein Sequence Source kDa data base* IP?PEAVKPDD?(D)E?APAKIP MOLT-4*t 110.9 IEF 9806

Residues sequenced

Ref(s).

?VPPMANNPSYQGI ?TDAPQ(P)(K) IEF 8704

93.6

MOLT 4t

IEF 8505

56.3

MOLT-4t

EIEDPEDRKPED SGTSEFLNK (F)AFQAEV GLFDEYGSK

IKPHLMSQELPED?(D)KQPVK

Endoplasmin

3-PHase or PDI

LITLEEEMTK IEF 8502

52.8

MOLT-4t

QLAPI?DKLGETYKD(H)EN TIFTGHTAVVEDVS?(H)LL?E

IEF 6318

37.3

MOLT-4t

IEF 7205

36.7

MOLT-4t

IEF 5206

35.5

Fetal human lung§

DFSIHR (T)PSSDVLVFDYTKHPSK MTDQEAIQDL ADKDYHFKVDNDENEHQLSL ?FAFVQYVNE(R) SAAEMYGS?FDLDYDFQ(R) IVADKDYSVTANSK YLMAEK L?TDGDKAFVDFLSDEIKEE

IEF 9205

34.3

MOLT-40

IEF 8214

33.0

Fetal human lung§

IEF 9209

31.9

MOLT-4

IEF 9109

31.1

MOLT-4

IEF 9105

31.0

Fetal human skin§

EV(S)FQ(S)TGER QVYEEEYGSSLEDDVVG GTVTDFPGFDER VLTEIIASR

Human B23 Xl B23 homology hnRNP protein C LDH H chain

Lipocortin V

?GTDEEKFITIFGT(R)

168-176 75-81 395-403 351-370 317-326 402-419

52-61 24-46 51-61 100-117 77-90 172-177

126-143 6-18 108-117

25, 26

27, 28

29, 30 31 32

33

187-201

?YNHIK ?FGDLR

IQADGLV?GS(S)K YSEKEDKYEEEIK EENVGLHQTLDQTLNELN?I QTF?EAMA?L?TL(S)E ENLTL?TA?NA?(E)(E)GGE?PQEPQ

TM ,3 chain homology

177-189 228-247

34

VFYLK

SAYQEAMDISK 35 20-24 Human cyclophilin VSFEL 118-123 TE?LDG 77-88 SIYGEKFEDENF ,B-PHase, f subunit of prolyl-4-hydroxylase; PDI, protein disuffide isomerase; Xl, Xenopus laevis; hnRNP, heterogeneous nuclear ribonucleoprotein; LDH, lactate dehydrogenase; H, heavy; TM, horse platelet tropomyosin. *From Celis et al. (5). tIsolated from gels of partially purified cyclin/proliferating cell nuclear antigen preparations from human MOLT-4 cells. Steps in purification involved 40-80% ammonium sulfate fractionation, DEAE-Sephacel chromatography (fraction eluted with 0.3 M KCI), and HPLC (TSK DEAE-SPW) chromatography (23). The fraction eluted with 0.45-0.8 M sodium acetate was applied to the gel. tIsolated from gels of 0.6 M NaCl extracts of nuclear pellets from human MOLT-4 cells. The extracts were further purified by hydroxyapatite. Proteins eluting with 0.3 M potassium phosphate (pH 7) were applied to the gel (IEF or NEPHGE) (B.G. and J.E.C., unpublished data). §Cut from gels of total extracts. Question marks in the sequences indicate the positions where residues could not be identified unambiguously; residues in parentheses are the most probable assignment.

NEPHGE 3004

18.2

MOLT-4

7704

Biochemistry: Bauw et al.

dimensional gel using the protein-stacking properties of the discontinuous Laemmli gel system (24). Proteins were then electroblotted on membranes and digested in situ with trypsin. The peptides released to the supernatant were further separated by reversed-phase HPLC and sequenced. Compared to similar previously published procedures (15, 16, 20), this approach shows many improvements. (i) Gels can be handled by conventional methods involving Coomassie blue-staining (dried, stored, and/or scanned for synthetic imaging) and no special precautions are necessary to guarantee subsequent amino acid sequence determination. The standard use of Coomassie blue-stained, dried gels as a protein source increase considerably the versatility of the method. Indeed, it is now possible to use gels that are routinely generated during the development of the comprehensive human protein data bases (5, 6). Furthermore, the information from these data bases allowed us to select gels from tissues or cells in which the protein of interest is most abundantly present (here we used fetal human tissues as a better source of some proteins; Table 1). In this study we often used gels that have been stored in a dried state for >6 months at room temperature. As will become evident from the sequence analyses (Table 1) and much to our surprise, we never noticed any deamidation at amide residues or oxidation of methionine residues. (ii) This protein recovery technique is easily used to concentrate spots from multiple gels to study less abundant polypeptides. (iii) The internal sequencing strategy employed here also avoids situations where proteins are NH2-terminally blocked either as a result of co-II or post-translational modification or due to spurious artifactual reactions with components of the gel matrix and/or chemicals used in the procedure. (iv) The membrane in situ digestion procedure used in this study is similar to that originally described by Aebersold et al. (16) but has been adapted to fit with the PVDF- or polybase-coated glassfiber blotting procedure (19). Compared to other in situ cleavage methods, this approach combines several advantages that are illustrated for proteins IEFs 9806, 8502, and 8214 in Fig. 2. First, the number of peptides released from the membrane into the supernatant is generally much smaller than expected from the number of potential cleavage sites. As a consequence, peptide HPLC chromatograms are extremely simple and most peptides are obtained in pure form (compare the peptide patterns of IEF 9806, a 110-kDa protein, and IEF 8502, a 53-kDa protein). Such a situation is not encountered when proteins are exhaustively digested in the gel matrix (36). In this case, the complexity of the peptide pattern is generally directly related to the size of the digested protein so that large proteins are difficult to analyze. In addition, peptide chromatograms from such digests are seldom free of artifactual peaks due to gel contaminants or protein dyes. Our modified in situ cleavage procedure was also found to be advantageous compared to the gel in situ partial digest protocol used to obtain internal protein sequences (20). Indeed, the latter is difficult to reproduce and the large fragments obtained are often derived from the same region in the sequence. As a result, fragments are recovered in low yield and often display the same NH2-terminal sequence. Another major advantage of this procedure resides in the unusually high sequencing efficiencies encountered with peptides released from membrane-bound proteins (initial sequencing yields often exceed 80%). The latter point is illustrated in Fig. 2D showing the traces of the phenylthiohydantoin amino acid derivative chromatograms of cycles 1-12 of a peptide from protein IEF 8214 (Fig. 2C). Clearly,