Microsequence analysis of peptides and proteins

ANALYTICAL

BIOCHEMISTRY

170, 19-30 (1988)

Microsequence

Analysis of Peptides and Proteins

VIII. Improved Electroblotting of Proteins onto Membranes and Derivatized Glass-Fiber Sheets

QIN-YUXUANDJOHNE.SHIVELY Division of Immunology, Beckman Research Institute of the City ofHope, Duarte, California 91010 Received August 13, 1987 We have quantitatively examined the various parameters affecting the electrotransfer and sequence analysis of proteins from sodium dodecyl sulfate (SDS) gels to derivatized glass fiber paper or to polyvinyldifluoride (PVDF) membranes. Transfer yields in the range of 90-95% can be obtained for proteins in the molecular weight range of lo-90 kDa for transfer from 12% SDS gels to glass fiber paper derivatized with either QAPS (N-trimethoxysilylpropyl-N,N,N-trimethylammonium chloride) or APS (aminopropyhriethoxysilane). In order to achieve these yields, it was necessary to modify the conditions described by R. Aebersold et al. (.I. Biol. Chem. 261, 4229-4238, 1986). We activated the glass fiber paper with dilute ammonia water and derivatized the activated glass fiber paper with QAPS and APS in anhydrous solvents which were allowed to slowly absorb moisture during the derivatization process. The transfer yield varied with transfer time versus molecular weight of the protein for a given percentage gel. Shorter transfer times and higher yields were obtained for higher molecular weight proteins on 8% gels. Lower molecular weight protein gave higher yields from 12% gels under similar transfer conditions. Sequencing yields of the transferred proteins were in the range of 40-80%, but a number of background peaks were observed on HPLC analysis of the phenylthiohydantoin amino acid derivatives. Transfer yields in the range of 85-95% were observed for similar experiments with PVDF membranes. In order to achieve these yields, it was necessary to modify the conditions described by P. Matsudaira (J. Biol. Chem. 262, 10035-10038, 1987). A lower voltage and longer transfer times gave higher transfer yields. In order to achieve consistently high transfer yields, it was also necessary to precoat the PVDF membranes with Polybrene. The PVDF membranes were cut into approximately l-mm-wide strips and inserted into a continuous flow reactor (J. E. Shively, P. Miller, and M. Ronk, Anal. Biochem. 163, 517-525, 1987) for sequence analysis. Overall yields of samples loaded onto gels, electrotransferred to Polybrenecoated PVDF membranes, and sequenced ranged from 50-60% for &lactoglobin (IO-50 pmol loaded onto SDS gels) to 20-302 for bovine serum albumin and soybean trypsin inhibitor (50 pmol loaded onto SDS gels). A comparison of the two methods shows clear advantages for the PVDF membranes over the derivatized glass fiber paper, including the ability to directly sequence the Coomassie blue-stained PVDF membranes, and the lower backgrounds observed on subsequent sequence analysis. Q 1988 Academic &ss, IIIC.

phoresis is a common analytical method for resolving complex mixtures of proteins at the

Microsequence analysis of proteins can be performed in the low picomole range, but the isolation of proteins in this range with suitable purity for microsequence analysis is a continuing problem. Since SDS’ gel electro-

difluoride; SDS, sodium dodecyl sulfate; CFR, continuous flow reactor; DTT, dithiothreitol; PTH, phenylthiohydantoin; DPTU, N,N-diphenylthiourea; DEA, diethylamine; BSA, bovine serum albumin; STI, soybean trypsin inhibitor; DTE, dithioerythritol; PITC, phenylisothiocyanate; THF, tetrahydrofuran.

’ Abbreviations used: APS, aminopropyltriethoxysilane; QAPS, N-trimethoxysilylpropyl-N, N,Ntrimethylammonium chloride; PVDF, polyvinyl19

0003-2697188 $3.00

CapyrishtQ 1988 by Academic Rw, Inc. All rights of reprcduction in any form rmmd.

20

XU AND SHIVELY

0. l- to lo-pg level, it has been a long sought after goal to sequence samples isolated from SDS gels. Samples trapped in polyacrylamide or agarose cannot be sequenced because of the incompatibility of these substrates with the organic reagents and solvents used in microsequence analysis. The first approach was to electroelute samples from the gel into miniature dialysis chambers (1,2). This method suffered from low sample recovery and high backgrounds and was difficult to reproduce from one lab to another. Samples stained with Coomassie blue and sequenced gave interfering background peaks on the HPLC analysis of the amino acid PTH derivatives. Two groups have explored methods of electroblotting samples from gels to positively charged glass fiber paper (3,4). The advantage of using glass fiber paper as the transfer substrate is its chemical inertness and physical stability during Edman chemistry. Since proteins with bound SDS bear a negative charge, the transfer to positively charged glass fiber paper is a good strategy. Vandekerckhove et al. (3) precoated glass fiber paper with Polybrene, the polycationic polymer routinely used as a carrier in microsequence analysis. The technique gave sequencing yields of about 30-50% for samples in the range of 0. l- 1.O nmol. Drawbacks included low repetitive yields and the inability to stain the transferred proteins with Coomassie blue. Some problems with the differential transfer of different molecular weight proteins was observed. Aebersold et al. (4) covalently derivatized glass fiber with positively charged silylating agents. Although the results claimed for this method are exceptional, 60-75% initial yields and 92-96% repetitive yields for as little as 24 pmol of standard proteins, these results have not been reproduced in other laboratories. In general, this method has been limited to samples > 100 pmol with initial yields in the range of 20-30%, thus suggesting some difficulties in standardizing the chemistry and transfer conditions. Aebersold

et al. (4) were unable to stain proteins with Coomassie blue before transfer to positively charged derivatized glass fiber paper. This problem was partially solved by the use of the fluorescent dye 3,3’-dipentyloxacarbocyanine iodide. They also explored the transfer of proteins at high pH after removal of SDS to acid-etched glass fiber paper. This method appears to have a number of inherent drawbacks, including the problem of removing SDS from the protein, and has not been further used. An alternative approach is to transfer samples to hydrophobic membranes. Nitrocellulose has been widely used for the Western blot technique (5), but is not compatible with the solvents used in microsequence analysis. Recently, Matsudaira (6) has demonstrated that polyvinyldifluoride (PVDF) gives high yields on samples which are electrotransferred and subjected to microsequence analysis. A major advantage of the method is the ability to stain transferred samples with Coomassie blue without encountering artifact peaks on the HPLC analysis of the PTH amino acid derivatives. However, this study suffers from a lack of quantitative data on protein transfer from gels to PVDF and the need to optimize parameters which affect the yields of sample transferred. In one case the author reports a 108% initial sequencing yield for a sample loaded onto an SDS gel and electrotransferred to PVDF, when, in fact, it is difficult to obtain initial sequencing yields of >70-80% for samples which are applied directly to a sequencer. This report describes modifications of the electrotransfer methodology for glass fiber derivatization, resulting in higher transfer and sequence yields. The parameters atfecting sample transfer to derivatized glass fiber paper and to PVDF membranes were thoroughly examined and optimized. Electrotransfer of samples from SDS gels to PVDF was quantitated and conditions optimized for proteins of different molecular weights. The subsequent sequence analysis of samples on PVDF was adapted to a continuous flow

MICROSEQUENCE

AND ELECTROBLOT

reactor (CFR) (7), resulting in a further optimization of the methodology. MATERIALS

AND METHODS

Materials. Glass fiber paper was Whatman GF/F. PVDF (Immobilon) was 15 X 15cm sheets from Millipore. Polybrene was obtained from Aldrich, APS and QAPS from Petrarch. Methods. Glass fiber paper was treated with a dilute ammonia (pH 10.2) overnight, washed five times sequentially with water, methanol, and THF, and air dried. For comparison of activation methods, some paper was treated with concentrated nitric acid overnight and washed as described above. The activated paper was treated with APS/ THF ( l/9) or QAPS/THF ( I /9) for 2-3 h at room temperature, loosely covered with aluminum foil. During the treatment there is a slow absorption of water vapor and polymerization of the silyl derivative. The paper was washed five times sequentially with THF and methanol and air dried. The derivatized papers were stored 1 day at room temperature, and thereafter at -20°C. SDS gels were prepared according to O’Farrell(8) using either 8 or 12% gels. The stock solutions (4X) for the upper gel buffer was 0.5 M Tris/HCl, 0.4% SDS, pH 6.8, and for the lower gel buffer 1.5 M Tris/HCl, 0.4% SDS, pH 8.8. The running buffer was 0.025 M T&-O. 192 M glycine, 0.1% SDS, pH 8.3. Minigels ( 10 X 10 cm) were polymerized overnight and prerun with 0.05% 2-mercaptoethanol or DTT for 20 min at 3 mA. Samples were loaded at 7 mA for 30 min, and electrophoresis continued at 15 mA for 60 min. After electrophoresis of the samples, the gels were transferred to a Bio-Rad electroblot apparatus and electroblotted essentially according to Towbin et al. (5). Samples migrate toward the anode due to the presence of bound SDS. Electrotransfer was performed for variable amounts of time at 25-30 V, 300 mA, constant current. In the case of derivatized glass fiber paper, two sets of paper were

ANALYSIS

21

applied. The voltage settings were 25-30 V at 300 mA. In the case of electrotransfer of samples onto PVDF, two layers of membranes were applied to the apparatus. The voltage settings were 25-30 V at 300 mA. Prior to electrotransfer the PVDF membranes were treated with Polybrene (10 mg/ml in methanol/ water, I/ 1) for l-2 s, air dried, and rinsed with methanol. The PVDF membranes must be prewetted with methanol prior to electrotransfer. Protein standards (phosphorylase A, bovine serum albumin, ovalbumin, fl-lactoglobulin, and soybean trypsin inhibitor) were obtained from Sigma and radioiodinated with 125I according to Hunter and Greenwood (9). The radiolabeled proteins were separated from free iodine by centrifugal gel permeation chromatography. Radiolabeled samples run on SDS gels initially exhibited no free iodine. The specific activities of the iodinated proteins were in the range of 0.02-0.03 pCi/pg. The radiolabeled samples were mixed with unlabeled samples to give 100,000 cpm per 100 pm01 of protein. Quantitation of samples electrotransferred was performed by counting the gels and membranes and comparing these results to equivalent samples run on SDS gels, but not electrotransferred. Samples transferred to derivatized glass fiber paper were detected by autoradiography, cut out, and counted in a Beckman y-counter. The gels were dried and counted. For sequence analysis, the samples were cut into a l.O-cm disk and applied to a microsequencer equipped with a Teflon cartridge (10). Prior to sequence analysis, the samples were washed with 0.1-0.2 ml of ethyl acetate. PTH amino acids were separated and quantitated according to Hawke et al. (1 1). For samples transferred to PVDF, the membranes were rinsed with water for 2-3 min, stained for 1 min with 0.2% Coomassie blue R-250 (methanol/water/acetic acid, 50/40/ lo), destained for lo- 15 min with methanol/water acetic acid (45/48/7), rinsed

22

XU AND SHIVELY

with water for 1 min, and air dried. The bands were cut into thin strips (3-5 X 0.5-1.0 mm) and applied to a microsequencer equipped with a continuous-flow reactor (7). The microsequencer is similar to that described by Hawke et al. (11) and was connected on-line to a Hewlett-Packard 1035 HPLC. The PTH amino acid separations were performed on a Beckman Ultrasphere C 18 column (0.2 X 250 mm) at a flow rate of 0.25 ml/min using an ammonium acetate-acetonitrile/methanol solvent system as described by Shively et al. (7). The derivatives were detected at 269 nm on a Model SPD-6A Shimadzu detector. PVDF strips were inserted into the CFR so that the inlet and outlet lines were not obstructed. One to five strips can be inserted into a single CFR. The strips were washed with 0.1-0.2 ml of ethyl acetate prior to sequence analysis. The data were collected and analyzed on a Perkin-Elmer LIM system.

A ,”

I

100 f f b E B ‘c a

90 -

,

80. 70.,.,T. 60 20

30

40

50

60

70

80

B 100

5P 8o 1 60 H 8 a

40 20

RESULTS

Derivatized glass fiber paper. Glass fiber paper (GF/F) was activated by treatment with either dilute ammonia water or concentrated nitric acid and then fiber derivatized with either APS or QAPS to impart a positive charge to the paper. A time course study of the transfer yields of radiolabeled proteins from 12% SDS gels to ammonia activated, APS- or QAPS-derivatized glass fiber paper is shown in Fig. 1. At 50 V (700 mA) the optimal transfer time for proteins in the molecular weight range 2 l-68 kDa was 40-70 min. For these proteins 85-95% of the protein was transferred to APS-derivatized glass fiber paper, but for phosphorylase A (95 kDa) the transfer was only 45% complete at 70 min (Fig. 1B). Similarly, the optimal transfer time for proteins in the range of 2 l-68 kDa to QAPSderivatized paper was 40-70 min, but phosphorylase was transferred with 90% efficiency at 70 min (Fig. 1A). In this respect QAPS-derivatized glass fiber paper has slightly better results for high

0

20

60

80

rimLaIn)

PIG. 1. Electrotransfer rates of radiolabeled proteins onto derivatized glass fiber paper. The samples were 5 pg each of soybean trypsin inhibitor (Cl, 21,000), ovalbumin (0,44,000), bovine serum albumin (m, 68,000), and phosphorylase A (A, 97,000). (A) Electrotransfer of samples to ammonia-activated, QAPS-derivatized glass fiber paper from a 12% gel. (B) Electrotransfer of samples from ammonia-activated, APS-derivatized glass fiber paper from a 12% gel. (C) Electrotmnsfer of samples from ammonia-activated, QAPSderivatized glass fiber paper from an 8% gel.

molecular proteins when transferring from 12% gels. In the case of 8% gels for either QAPS- or APS-derivatized glass paper, transfer efficiency for proteins in the range of 2 l-44 kDa is optimal in 30 min, but falls off at longer

MICROSEQUENCE

AND ELECTROBLOT

23

ANALYSIS

increasing transfer times, it was clear that a larger percentage of the sample was on the second layer. It was assumed that sample losses indicated electrotransfer through both layers into the surrounding buffer. The capacity of the derivatized glass fiber paper was not exceeded even at the 1-nmol range. The increase in sample bandwidth on transfer from the gel to the derivatized glass fiber paper was two- to fourfold, indicating significant band diffusion. Due to the high background obtained with Coomassie blue stain-

B

cycle

1

ski

cycle2

St*

rtd

cycle 3

I OTE

604 20

.

,

I

40

60

.

1 60

:

1 100

C

1 0

3

6

cycle 7

20

5fd

9 12 15 0

3

6

9 12 15

0

1,c 3

6

9

12 15

Cycle 9

Cycle 8

100

&eeular6G&t

in ICE

FIG. 2. Correlation ofelectrotransfer yield with molecular weight. Data are taken from Fig. 1. Times are (0) 70, (A) 45, and (m) 30 min. (A) APS-detivatized glass fiber paper, 12% gel. (B) QAPS-detivatized glass fiber paper, 12% gel. (C) QAPS-derivatized glass fiber paper, 8% gel.

times. For proteins in the range of 68-95 kDa, the optimal transfer time is 20-40 min, 80-92% efficiency. The results are replotted in Fig. 2 to demonstrate the dramatic effect of molecular weight on the efficiency of transfer to either APS- or QAPS-derivatized glass fiber paper. In all cases, efficient transfer of the samples required two layers of derivatized glass fiber paper. In cases where decreasing transfer yields were observed for

1y #JJJ E

0

3

6

9

G

12 15

0

3

6

9 l2

15 0

3

6

9

12 16

Time hid

FIG. 3. Sequence analysis of soybean trypsin inhibitor after electrotransfer to QAPS-derivatized glass fiber paper. The sample (50 pmol) was electrotransferred from a 12% gel to QAPS glass fiber paper for 70 min and subjected to microsequence analysis. Cycles 1,2,3,7,8, and 9 are shown. The identified PTH derivative at each cycle is labeled. Forty percent of the sample was analyzed for each cycle. Quantitation (corrected for amount injected) is given in Table 1. Background peaks are identified as follows: DTE is present in the sequencer solvents and is labeled, the second two peaks on cycle 1 are due to Tris and glycine in the transfer buffer, and the two late peaks are due to DEA in the sequencer buffer and the formation of DPTU from PITC.

24

XU AND SHIVELY TABLE 1

MICROSEQUENCEANALYSIS OF SAMPLES ELECTROTRANSFERREDTO DERIVATIZED GLASS FIBER PAPER” Amino acid

APS (250 pm00

QAPS (150 pmol)

QAPS (50 pmol)

Asp Phe Val

170 117 138 99 106 62 95 40 52 44

57 47 42 39 31 24 20 11 22 18

39 34 31 29 21 13 39 12 16 13

LeU

Asp Asn Glu GUY Asn Pro

’ Soybean trypsin inhibitor (Mr = 21,000) was electrotransferred from 12% SDS gels in the amounts shown. The sample was transferred to glass fiber paper activated with dilute ammonia and derivatized with either aminopropyltriethoxysilane (APS) or N-trimethoxysilylpropyl-N,N,N-trimethylammonium chloride (QAPS). The sequencer yields at each cycle are shown. The initial yields for the samples are 68, 38, and 7896, respectively. The repetitive yields ranged from 92 to 93%.

ing, it was impossible to stain glass derivatized paper. The problem was less severe with Amido black staining, but the overall sensitivity was low. Samples had to be located either by autoradiography or by staining and marking guide strips on the original gel. Activation of the paper with concentrated nitric acid resulted in much lower transfer yields under similar conditions, 1O-20% maximum transfer (data not shown). The use of two transfer buffers was explored, Tris/glycine or N-ethylmorpholine/acetate. Both transfer buffers gave similar results. One of the samples, 50 pmol of soybean trypsin inhibitor, was subjected to microsequence analysis. The sample band was located by staining guide strips on the gel, cutting out the band into a 1-cm disk, and placing the disk into a microsequencer. The initial yields ranged from 30 to 60% with repetitive yields in the range of 90-93%. The data are summarized in Table 1 and repre-

sentative chromatograms are shown in Fig. 3. When Tris/glycine was used as the transfer buffer several characteristic background peaks were observed on cycle 1 (Fig. 3) and diminished on subsequent cycles. The amount of the background peaks was diminished but not eliminated by prewashing the disk with ethyl acetate or by using N-ethylmorpholine as the transfer buffer. PVDF membranes. Initial experiments included the electrotransfer of samples onto A 100

20

40

60

80

100

120

40

60

80

100

120

B

201 20

C

0

50

100 Time

150

(mln)

FIG. 4. Electrotransfer rates of radiolabeled samples onto PVDF membranes. The samples (m, fi-lactoglobulin; 0, ovalbumin; and Cl, BSA) were electrotransferred from 12% gels to Polybrene-coated PVDF. (A) 200 pmol. (B) 50 pmol. (C) 25 pmol.

MICROSEQUENCE A

AND ELECTROBLOT

ANALYSIS

25

B

1234

5

5 10 15 20 25 (mid

1

5

234

10

5

15

20

25 (mid

FIG. 5. Photograph of stained gels and membranes after electrotransfer of samples. The samples (100 pmol of BSA and &lactoglobulin) were electrotransferred from 12% gels to Polybrene-coated PVDF for the times shown and the gels and membranes stained with Coomassie blue. Membrane and gel strips were cut out for each time point and assembled as shown. (A) Staining of samples transferred to PVDF membranes for each time point. (B) Staining of samples left on gels for each time point.

nylon membranes. Although the transfer yields were good (80-90%), the nylon membranes were not stable to the sequencer chemistry. This was not a problem with PVDF membranes. However, electrotransfer of samples onto PVDF membranes gave transfer yields in the range of 30-50% at 50 V (500 mA) even at short transfer times (10 min). The problem was slightly improved by lowering the voltage settings to 25-30 V (300 mA), adding two layers of membranes, and prewetting the membranes with methanol or acetonitrile. The most efficient transfers to PVDF membranes were obtained by precoating the membranes with Polybrene. The time course for electrotransfer of /?-lactoglobulin, ovalbumin, and bovine serum albumin to Polybrene-coated PVDF membranes under these conditions is shown in Fig. 4. Over the range of 25-200 pmol of samples on 12% gels, the optimum transfer time was 100 min. The efficiency of transfer was 90-95% for @lactoglobulin and ovalbumin and 80-90% for bovine serum albumin. Longer transfer times led to a uniform decrease in recovery of the radiolabeled sample from the PVDF membrane. It was assumed

that this was due to electrotransfer of the sample through both sets of membranes. There was little detectable increase in bandwidth during the electrotransfer process. The samples could be stained with Coomassie blue directly on the membranes. In the absence of Polybrene no background staining was observed. In the presence of Polybrene a uniform background was observed, but did not interfere with detection of the samples. Typical staining results are shown in Fig. 5. Samples transferred to PVDF membranes were stained, cut into multiple l-l .5 X 5-mm strips, and inserted into a CFR. The strips were washed with ethyl acetate and subjected to microsequence analysis. Results for several proteins in the range of lo- 100 pmol are shown in Tables 2 and 3. The initial yields ranged from 50 to 60% for B-lactoglobulin and 20 to 40% for bovine serum albumin. Repetitive yields were in the range of 9 l-94%. Representative chromatograms for electrotransferred samples are shown in Figs. 6-8. Little or no background peaks were observed from the stained PVDF strips. The sensitivity of the method is good over the entire sample range analyzed. Results were

26

XU TABLE

AND

2

MICROSEQUENCE ANALYSIS OF ,&LACWGLOBULIN ELEC~ROTRANSFERREDTO PVDF MEMBRANES~ Amino acid

50 pm01

20 pm01

10 pm01

Lell

27

Be

18 31 11 7

12 8

6 5 6 4 4 3 4 3 4 5

Vd

Thr Gln Thr Met

13 6

5

LYS GUY

8 8 5 6

LA%

10

I

Asp Ile Gln

3

5 5 4 2 5 6 3 3 2

LYS

Val Ala

7 4 3

12

LeU

8 4 3 4 4 2 5

Ala

7

GUY

Thr Trp Tv Ser

5 6 4 4

SHIVELY

age and current settings, and a given percentage gel, the rate of transfer of proteins depends on molecular weight. For proteins in the molecular weight range lo-70 kDa, the transfer yields are high from 12% gels but rather low for 8% gels. Conversely, for proteins around 100 kDa, the transfer yields are low for 12% gels and higher for 8% gels. This generalization holds true when an optimized

TABLE

MICROSEQUENCEANALYSIS OF SOYBEAN TRYPSIN INHIBITOR AND BOVINE SERUM ALBUMIN ELECTROTRANSFERREDTO PVDF MEMBRANES” Amino acid

ST1 (50 pmol)

Asp Phe

11 13 17 11

Vd

Leu Asp Asn Glu GUY

Asn pro

’ @Lactoglobulin (Mr = 18,400) was electrotransferred from 12% SDS gels in the amounts shown. The sample was transferred to PVDF membranes as described under Materials and Methods. The transfer yield in each case was 90%. The sequencer yields at each cycle are shown. The initial yields (calculated from the amount applied to the SDS gel) for the samples are 54, 60, and 60%, respectively. The repetitive yields ranged from 9 1 to 95%.

Leu Glu Asn GUY GUY

Thr Tyr Tv

Ile L42.l

similar when Tris/glycine or N-ethylmorpholine was used as transfer buffer. DISCUSSION

General. These studies demonstrate the need to quantitate the amount of sample transferred from SDS gels to the glass derivatized paper. In this respect, the use of radiolabeled proteins was invaluable. Low sequencing yields may be due to low transfer or poor initial yields on the sequencer, two unrelated problems. For a given set of volt-

3

Ser Asp Ile Thr Ala

9 9 8 7 8 4

Amino acid Asp Thr His LYS Ser Glu Ile Ala His

BSA (50 pmol) 10 9 5 8 6 9 9

12

I

Phe

4 6 4 5 4 4 6 4 6 2 3 4 3 6

LYS

4 4 9 6

A%

Asp

I

LeU

I

GUY

3

Glu Glu His Phe

8 2

LYS GUY

6 4

LeU

I

Vd

9

7 I

LISoybean trypsin inhibitor (MI = 2 1,000) and bovine serum albumin (Mr = 68,000) were electrotransferred from 12% SDS gels in the amounts shown to PVDF membranes as described under Materials and Methods. The transfer yields were 80-90%. The sequencer yields at each cycle are shown. The initial yields (calculated from the amount applied to the SDS gel and the amount recovered on cycle 1 from sequence analysis) for the samples were 34 and 20% respectively. The repetitive yields ranged from 94 to 98%.

MICROSEQUENCE

AND ELECTROBLOT

27

ANALYSIS

std

MA

0

Y ,

PMV I

21.0

26.5

RA

I

155

DPTU

TIME

W

FKIL

1

320

375

(minutes)

FIG. 6. Sequence analysis of 6-lactoglobulin after electrotransfer to PVDF membranes. The sample (20 pmol) was electrotransferred to PVDF for 100 min from a 12% gel. The strips were cut out and applied to a microsequencer equipped with a CFR. About one-third of the sample was analyzed for each cycle. Cycles 1 2 3,4, and 5 are shown with the expected PTH derivatives shaded. PTH derivatives expected for each cycle are shaded. Quantitation is shown in Table 2. The major background peak is DPTU.

reader is cautioned to verify some of these results before assuming exactly identical transfer yields. Derivatized glass fiber paper. The results presented here demonstrate that transfer yields of 90-9576 can be obtained for APS- or

transfer time is chosen, but if one is willing to change transfer times according to the sample, molecular weight, and percentage gel used, transfer yields can be improved. It is likely that our results are typical of what can be expected for similar equipment, but the

0.0050

DPTU r

‘N STGQH 10.0

Y

PMV

I

RA

I

I

15.5

21.0 TIME

26.5

w

FKIL I

32.0

I

375

(minutes)

FIG. 7. Sequence analysis of &lactoglobulin after electrotransfer to PVDF membranes. The sample (10 pmol) was analyzed as described in Fig. 6. Cycles 2, 3, 4, 5, and 6 are shown with quantitation given in Table 2.

28

XU AND SHIVELY std

0.0026

D

4.0

E

N STGQH

R/i

Y

PMV

W

I

I

1

I

10.7

174

241

30.8

FKIL I

375

TIME (minutes) FIG. 8. !kquence analysis of BSA after electrotransfer to PVDF membranes. The sample (50 pmol) was analyzed as shown in Fig. 6. Cycles 2, 5, 8, 1 I, and 14 are shown with quantitation given in Table 3.

QAPS-derivatized glass fiber paper. These results agree with Aebersold et al. (4). The fact that many groups have had difficulties reproducing these yields is almost certainly due to irreproducible chemistry. Our attempts to work with either unactivated or nitric acid-activated glass fiber paper always resulted in low transfer yields. The activation step causes hydrolysis of the silica to present new silanol groups on the glass surface. If this step is not accomplished, very low levels of substitution occur in the silanization step. We used hydrolysis with dilute ammonia for activation, a step which was more reproducible than acid hydrolysis. We also found that polymerized APS or QAPS was more stable than the monomeric forms, which probably hydrolyze from the glass surface during Edman chemistry. Polymerization of these reagents depends on the controlled addition of water. Aebersold et al. (4) added water directly to the reagent, probably causing significant polymerization before covalent attachment to the glass surface. Our method allows the slow adsorption of moisture from the atmosphere, promoting a slower, more controlled polymerization after covalent attachment of the monomer to the surface.

In spite of these improvements, the use of derivatized glass fiber paper has inherent drawbacks, including the need for activation and chemical derivatization, the band spreading observed during the transfer step, the inability to stain the bands with Coomassie blue, and the background peaks observed on the first few cycles of Edman chemistry. Thus, the use of hydrophobic, chemically stable membranes such as PVDF is clearly preferable to the use of derivatized glass fiber paper. PI!DF membranes. Our initial experience with electrotransfer of samples to PVDF membranes was encouraging in view of the sharpness of the bands transferred to PVDF and the ability to stain the transferred bands with Coomassie blue with no interference on subsequent microsequence analysis. However, when the same voltage settings as described above for derivatized glass fiber paper were used, the transfer yields were low. We were able to optimize the transfer yields by lowering the voltage settings and coating the membranes with Polybrene. It was also clear that the membranes must be prewetted with methanol or acetonitrile prior to electrotransfer or the electroblotting buffer

MICROSEQUENCE

AND ELECTROBLOT

would not wet the hydrophobic membrane. Otherwise, the rates of transfer for proteins of different molecular sizes were quite similar to that observed for derivatized glass fiber paper. It was necessary to use a double layer of membranes to obtain yields in the range of 90%. This finding suggests that the movement of the sample through the membrane is always occurring. Thus, prolonging the transfer time is not a good idea. Optimal yields depend on careful analysis of transfer yields for a given set of conditions. We also investigated other membranes, including positively charged nylon membranes. These membranes were physically unstable, lasting only one or two cycles of Edman chemistry. We are not aware of the availability of other membranes of comparable stability to the PVDF membranes. Microsequence analysis of the small PVDF strips presents unique problems for a cartridge based microsequencer. Since the sample is likely to be present on two or more strips, it is necessary to lay the strips across the Zitex seal. If the membranes move, or reagent and solvent flow is not uniform across the membranes, poor or irreproducible chemistry is likely to occur. The glass fiber paper must be removed from the cartridge or it will adsorb the chemicals in preference to the hydrophobic membrane. These problems prompted us to insert the membrane strips into our previously described CFR. We found that the samples were easily inserted into the CFR and that the chemistry was reproducible. The backgrounds observed were uniformly low from cycle to cycle and sample to sample. The results shown in Figs. 6-8 are comparable to or superior to those observed for standards spotted directly onto Polybrene-coated glass fiber paper in a cartridge or onto Polybrene-coated silica in a CFR. The results are significantly better than those we obtained for samples electrotransferred to derivatized glass fiber paper. It is clear from this work that electrotransfer yields on PVDF and microsequence analysis of samples on PVDF are high

ANALYSIS

29

enough to allow the routine sequence analysis of lo-50 pmol of sample run on an SDS gel. This is a major improvement over the previous methods. The method is straightforward and requires no special chemistry for the PVDF membranes. The use of the CFR is optional, but in our opinion, well worth the effort. The adaptation of a microsequencer from a cartridge to a CFR is trivial and represents no major obstacle to implementing this methodology. We predict that the sequence analysis of samples from SDS gels will gradually gain in popularity, but will be still limited for several reasons. The first is that a large number of proteins (estimated as high as 50%) possess a blocked amino terminus. Although this problem can be overcome by protease mapping on the SDS gel, it requires having a pure protein band at the protease digestion step. The second refers to the problem of sample purity. Whole cell or tissue digests applied to SDS gels give 30-50 discrete bands which actually represent hundreds of unseparated proteins. Thus, this approach cannot reasonably lead to one-step purifications of proteins. It is also unlikely that complex mixtures can be resolved in two steps. Thus, we warn that in the case of complex mixtures, the method should be used with careful attention to obtaining other criteria of sample purity before deciding that the sequence obtained belongs to the protein of interest. Chief among these criteria must be quantitative measurements of the amount of sample electrotransferred and the initial yield of the amino-terminal residue at cycle 1. If quantitation is not performed, it may be possible to sequence a minor impurity in the presence of large amounts of the desired protein. ACKNOWLEDGMENTS We gratefully acknowledge the expert technical assis-

tance of Fun-Shan Chen and Jimmy Calaycay and funding from NIH Grants CA37808, HD 14900, DK33155, and HL 28481. A portion of this work was supported by the City of Hope Cancer Center grant from the National Cancer Institute.

30

XU AND SHIVELY

REFERENCES 1. Hunkapiller, M. W., Lujan, E., Ostrander, F., and Hood, L. E. (1983) in Methods in Enzymology (Hits, C. H. W., and Timasheff, S. N., Eds.), Vol. 91, pp. 227-236, Academic Press, New York. 2. Hunkapiller, M. W., and Lujan, E. (1986) in Methods of Protein Microcharacterization (Shively, J. E., Ed.), pp. 89-101, Humana Press, Clifton, NJ. 3. Vandekerckhove, J., Bauw, G., Puype, M., Van Damme, J., and Van Montagu, M. (1985) Eur. J. Biochem. 152,9-19. 4. Aebersold, R., Teplow, D. B., Hood, L. E., and Kent, S. B. (1986) J. Biol. Chem. 261, 4229-4238.

5. Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA X,4350-4354. 6. Matsudaira, P. (1987) J. Biol. Chem. 262, 1003510038. 7. Shively, J. E., Miller, P., and Ronk, M. (1987) Anal. Biochem. 163,5 17-525. 8. O’Farrell, P. H. (1975) J. Biol. Chem. 250, 4007-402 1. 9. Hunter, W. M., and Greenwood, F. C. (1964) Biothem. J. 43,91-98. 10. Hawke, D., Harris, D. C., and Shively, J. E. (1985) Anal. Biochem. 147,3 15-330. 11. Hawke, D., Yuan, P.-M., and Shively, J. E. (1982) Anal. Biochem. 120,302-3 11.