University of Virginia Health System

GUIDE TO PROTEIN SEQUENCING

ºAssistant Director, *Laboratory Specialist Advanced, #Director, Biomolecular Research Facility,
University of Virginia, Box 441, Charlottesville, VA 22908 (434) 924 2356

Introduction

The Biomolecular Research Facility operates an Applied Biosystems Procise model 494 sequencer for N-terminal sequencing. This instrument uses Edman chemistry to remove amino acids sequentially from the amino terminus of a polypeptide. The removed amino acids are identified by reverse phase chromatography. Under favourable conditions, the instrument can identify a stretch of fifty amino acids.

Chemistry

The Edman chemistry uses different reagents from earlier implementations e.g. gas phase sequencer (Hewick et al. 1981, Hunkapiller et al. 1983) but the principles remain the same. The base N-methylpiperidine makes the immobilised sample basic, after which the amino group of the terminal amino acid residue reacts with phenylisothiocyanate to produce an anilinothiazolinone amino acid. Liquid or gaseous trifluoroacetic acid cleaves the derivatized amino acid from the end of the peptide chain. Another reaction converts the anilinothiazolinone amino acid to a phenylthiohydantoin (PTH) amino acid which is then identified by reverse phase chromatography on a C18 column using an on-line HPLC. We use the current Applied Biosystems operating procedures and programs for both the sequencer and the on-line PTH-amino acid analyzer.

This process is repeated for each new terminal amino acid that results from a cycle of derivatization. sequencing, length of The process can not be continued indefinitely for two reasons: firstly the cycle of derivatization and hydrolysis is, on average, only 94% efficient and secondly there is acid cleavage of peptide bonds. After 20 cycles (amino acids), there is 0.9420 =29% of the initial quantity of amino acid produced by a derivatization and hydrolysis cycle. One factor decreasing efficiency is loss of sample from the support. For samples on PVDF, 1 to 10% of the sample may be lost off the support in each cycle Matsuidaira (1989b); this work used Immobilon-P and it is possible that later formulations of PVDF will lose less protein. It is also common to see major losses of sample at the end of a peptide immobilized on a Polybrene treated glass fiber filter, and hydrophobic peptides do not stick well to Polybrene.

Another factor reducing efficiency of the reaction occurs when a serine residue appears in the sequence. The hydroxyl group of serine may be esterified by acid, and when the amino group of the serine residue becomes N-terminal during the sequence, the esterified hydroxyl group may participate in an OThe acid used to cleave the derivatized amino acids from the end of the chain also cleaves peptide bonds within the chain. There is specific cleavage after aspartate residues in anhydrous acid (Brandt et al., 1980) as well as non-specific cleavage. Also Ne the yield of the new amino acid gets smaller relative to the background, it becomes impossible to determine which amino acid has been newly released from the original protein. Thus even if there is sufficient sample, 500 pmole or more, we cannot reliably interpret sequence data after about 50 amino acids or less, depending on the protein. Isoaspartate formation also reduces sequence yield. Formation of isoaspartate appears relatively slow, even under the alkaline conditions favouring its formation (Geiger and Clarke, 1987; Violand et al., 1990). However if it occurs, the sequence will stop at that point (Teshima et al., 1991).

Blocked proteins

The reaction described above requires a free -amino group -amino group on the terminal amino acid; if one is not present, we cannot derive any sequence data from the sample without further treatment. The most common moieties blocking the terminal amino group are either acetyl or other acyl groups, or pyroglutamate residues formed by cyclization of glutamine or glutamate residues. Other blocking groups have also been found (Wold 1981) and it is believed that other less well defined reactions occur in vitro that also block the terminal amino group.

Sample preparation

The three most common ways of supplying samples are: in a volatile solvent, in a buffer solution, in a gel.

Samples in volatile solvents are dried in 15 al aliquots on a glass fiber disk which has been coated with Polybrene which retains proteins on the disk by non-covalent interactions.

When a sample is in buffer, it is spun on to a disk of PVDF in a modified centrifugal filter (ProSpin cartridge) or the solution may be drawn through the PVDF by absorption forces (ProSorb). In this way, salts and detergents pass through the PVDF membrane while the protein remains on the PVDF which is loaded into the sequencer Sheer, 1990). Binding of samples is reported to be more efficient in non-acidic solution (Baumann, 1990).This technique should not be used for proteins which are to be manipulated further because of the difficulty of removing proteins from PVDF (see later).Amicon sell ion exchange membranes (Microcon-SCX) for the same purpose, using a centrifuge to pass the solution through the membrane; the peptides or proteins can be eluted from these membranes, unlike with PVDF where removal of adsorbed molecules is difficult.

Samples purified by gel electrophoresis are blotted to a sheet of PVDF, which is stained and the band of protein cut out and loaded into the sequencer. This technique is described later.

Another method of loading samples is to attach them covalently to a support. A commercially available support is the Sequelon family of modified PVDF from Millipore. In Applied Biosystems instruments, some peptides sequence poorly on these supports, while some give better data. We obtained better results when sequencing a synthetic peptide applied to a Polybrene treated glass fibre filter then the same peptide attached to a Sequelon membrane. However when identifying phosphorylation sites, it is necessary to covalently bind peptides to a PVDF based support to prevent sample losses from the solvents used.

Identification of amino acids

To identify the PTH amino acids produced by the sequencer, they are injected on a C18 column, which is equilibrated in acetate buffer at about pH 3.9 and eluted with a gradient of acetonitrile. The elution positions of arginine, histidine, and pyridylethyl-cysteine are sensitive to pH and buffer concentration. A mixture of PTH amino acid standards are injected at the beginning of each sequencer run for identification of amino acids and quantitation of the amino acids released from the protein being sequenced.

To see cysteine in a sequence, it must be alkylated, preferably before the peptide is loaded on the sequencer, to prevent destruction by acid. Applied Biosystems recommends alkylation with 4-vinylpyridine according to the procedure listed in Appendix A. Another procedure is that of Andrews and Dixon (1987). Alkylation with 4-vinylpyridine produces a pyridylethyl-cysteine residue which can be identified in analyses of both PTH-amino acids (from the sequencer) and PTC-amino acids, which are produced in our procedure for amino acid analysis. Another preferred reagent is N-isopropyliodoacetamide (Molecular Probes, Oregon) (Krutzsch and Inman, 1993) which gives carboxamidoacetamide-cysteine, whose PTH derivative is easily separated. Reaction of cysteine with iodoacetate produces carboxymethyl-cysteine (Gurd 1972) which may be confused with glutamine in an analysis of PTH-amino acids; note that iodoacetate may also alkylate methionine (Jones et al., 1994). Cysteine can be alkylated after electrophoresis on PVDF (Ploug et al., 1992).

Yields of serine , and threonine are low because these residues can be esterified by TFA, and the esters can dehydrate, giving dehydroalanine from serine. In our sequenator, DTT reacts with the dehydroalanine giving a peak that elutes between alanine and tyrosine. The corresponding products from threonine are seen with large amounts of threonine. The acyl group that esterifies the hydroxyl group may undergo a ON shift, acylating the terminal amino group, thereby blocking it and preventing further sequencing. This shift may explain the drop in yield seen after serine and threonine when sequencing isolated peptides (Aebersold, 1989). Prolonged exposure of a protein to formic acid appears to esterify serine and threonine (Tarr and Crabb, 1983) but aminoethanol appears to reverse the reaction.

Histidine and arginine may give lower yields, reportedly because of poor extraction from the solid supports for the sample.

Tryptophan is also prone to oxidation by sunlight and other agents (Pirie, 1971). Methionine may be oxidized in proteins but this does not appear to be a common problem (Brot and Weissbach, 1983).

Modified amino acids

Phosphorylated amino acids cannot be identified by standard sequencing procedures. Phosphoserine and phosphothreonine eliminate the phosphate group giving a low yield of a derivative; the released phosphate is poorly extracted from the sample support (Soderling and Walsh, 1982). Phosphotyrosine is insoluble under standard conditions and not easily identified by chromatography. However if the phosphoproteins are labelled with P32, the position of the labelled residue can be determined after digestion with a proteinase of known specificity (Shannon and Fox, 1995). Chemical conversion of phosphoserine is another method of identifying this residue (Meyer et al, 1986).

Heavily glycosylated proteins are reported not to give good data because the hydrophilic carbohydrate chains interfere with the non polar reagents and solvents used in the sequencing reaction.

In some proteins, palmitic acid (C16) is attached to cysteine, serine or threonine through a (thio)ester link (Sefton and Buss, 1987); myrisitic acid (C14) is attached through an amide link to amino terminal glycine residues. One method of identifying acylated cysteine is to deacylate with hydroxylamine and alkylate the cysteine (Bizzozero et al., 1990).

Carboxy-terminal sequencing

There are three strategies to obtain C-terminal sequence data.

A few laboratories offer C-terminal sequencing by chemical methods; hundreds of picomoles of sample are needed and 5 amino acids of sequence is a standard result. Proline may not be identified.

A traditional procedure is to cleave amino acids from the C-terminus using a broad specificity carboxypeptidase, usually carboxypeptidase Y or a mixture of carboxypeptidases A and B. By noting the time at which levels of different amino acids reach a maximum, one can deduce the sequence of 4 or 5 residues (Hayashi, 1977). Sequencing of cytochrome P-450 from the C-terminus has been unreliable, possibly because of the presence of proline (Black and Coon, 1986). An example of this procedure is in Asano et al. (1986).

More recently, Patterson et al. (1995) released amino acids using a differing concentrations of carboxypeptidase and measuring the mass of the remaining peptide by mass spectrometry; this method is limited by the ability of the mass spectrometer to distinguish the loss of different amino acids from the peptide.

A third technique for obtaining C-terminal sequence data is isolation of the C-terminal peptide and determine its sequence by standard N-terminal sequencing; if there is sufficient peptide, one may be able to determine the complete sequence of the C-terminal peptide, especially if the amino acid composition can be determined to see if all amino acids present according to amino acid analysis are also found on sequencing. To isolate the C-terminal peptide, Pierce and Clontech sell a column of immobilized anhydrotrypsin. After digesting the protein with trypsin, the peptides are applied to the column. The internal peptides, with a C-terminal lysine or arginine bind to the anhydrotrypsin but the C-terminal peptide washes through the column. This technique has been applied to other purifications (Hirabayashi and Kasai, 1992).

Techniques to isolate the C-terminal peptide using esterification of carboxyl groups, digestion of the protein and isolation of the C-terminal peptide, which lacks a free carboxy group, have been described by Furka et al. (1983) and Dopheide and Ward (1981). Ion exchange has also been used (Gorman and Shiell, 1993).

A technique that may involve smaller volumes of reagents is to digest a protein with endoproteinase lys-C and couple the peptides to a Sequelon-DITC membrane. All peptides except the C-terminal peptide are covalently bound to the membrane through the _-amino group and all peptides are bound through the N-terminal amino group. Treatment with acid cleaves the N-terminal amino acid from the peptide in a reaction like that found in Edman sequencing. The C-terminal amino acid is released from the membrane while all other amino acids remain bound through their C-terminal lysine (Dixon et al., 1993).

Sample Requirements

Samples must contain only one protein or peptide to avoid ambiguous data. If there is more than one polypeptide, an amino acid will be released from each polypeptide in every sequencing cycle. If the amounts of polypeptide are very different, the amounts of amino acid will be also, enabling each amino acid to be assinged to the correct sequence. If there are two protein present in similar amounts, the sequence data will be ambiguous. Because of differences in amino acid yields, and drops in yields after serine, threonine and proline, large differences in amounts of the proteins are needed to confidently determine the sequences.

Although the instrument can sequence 10 pmole or less protein, more sample is preferred. Larger amounts of sample increase the chance of identifying amino acids whose yield is poor, and obtaining a long sequence run. More importantly, with small quantities, samples are easily lost during handling, so that the amount seen for analysis is much less than the amount estimated. With a small quantity of sample, if there is no data, the question remains as to whether the sample is blocked and therefore unsequenceable, or whether there was insufficient sample.

Samples in volatile solvents suitable for direct application to the sample support should be in less than 200 µl, the volume of samples containing salts and other non-volatile material should be less than 1 ml; a PVDF strip up to 8 mm long can be loaded but further strips must be shorter to fit into the cartridge. A high concentration sample is preferable to a larger area of low concentration sample.

Substances which can interfere with sequencing reactions are listed in appendix D.

It is reported that dialysis may easily introduce contaminants, so dialysis should be done with thoroughly cleaned tubing and should be done in the presence of salt or acid to prevent contaminants adsorbing to the protein. We have seen samples that have been dialyzed and contain contaminants. Reportedly "molecular biology grade reagents ;contain insoluble material and UV-absorbing material (Matsuduira, 1989).

One of our preferred methods of final sample preparation is reverse phase chromatography using a gradient of water and acetonitrile with 0.1% trifluoroacetic acid. This procedure often gives a pure sample, and leaves the sample in a volatile solvent. Although samples may appear pure on electrophoresis or ion exchange chromatography, they often contain more than one peptide chain when analyzed on the sequencer. The data from such samples usually can not be interpreted reliably. Despite the resolution that reverse phase HPLC has for peptides, chromatography under one set of conditions may yield more than one peptide in a single peak so that rerunning under different conditions is desirable.

Not all of a sample applied to the sequencer will sequence. This is a standard observation. Hewlett-Packard report that treating a sample with 6M guanidine-HCl immediately before application to their instrument increased the initial yield from 52% (sample in 2% TFA) to 68% (Application Note 93-3). Lottspeich (1990) argues that incomplete reaction of applied protein is caused by interactions of the protein with the support, making portion of the protein unavailable to reactants.

Peptides can also be applied to PVDF; without further treatment, the loss of peptide from the PVDF is significant, but 100 µg of Polybrene applied to PVDF reduces sample wash out, at the expense of reduced recovery of charged amino acids (Werner et al., 1996). The addition of Polybrene reputedly makes sequencing samples on PVDF as efficient as sequencing samples on glass fibre disks.

Losses of Protein

;It is claimed that small amounts of protein are easily lost during drying (Grego et al., 1985; Wilson et al., 1986; Esch, 1984). One group reports that the amount of protein they see after extensive preparation is often less than 10% of the estimated value (Stone et al., 1989b). When isolating a tryptic peptide, Lu and Lai (1986) found that drying a tryptic peptide and dissolving it in 50% formic acid gave about 10% of the sequenceable protein seen when the peptide was collected in a polypropylene tube and applied directly to the sequencer; formic acid may have blocked the N-terminal thus reducing yields on the sequencer. Tempst et al. (1990) suggest that at concentrations below 0.1 µg/µl proteins may adsorb to polypropylene tubes. Gary Hathaway (unpublished results) as a rule of thumb assumes 10 ng of protein/cm2 is adsorbed to polypropylene; 1 ml in a 1 ml pipet tip is 2.9 cm2, in a 1.5 ml tube is 6 cm2. Nevertheless, polypropylene tubes are the preferred tubes to use because glass may adsorb proteins. Polystyrene, which is used in ELISA plates because of its protein adsorbing properties, should be avoided. One company reports that their low protein binding syringe filters adsorb 2 µg of protein/cm2, which shows how readily small quantities of protein may be lost.

On controlled pore glass, initial adsorption of protein is increased by the pI of a protein and the subsequent rate of loss is an inverse function of the molecular weight of the protein (Messing, 1969). Initial adsorption is 11% for cytochrome c with rates of loss of 5% for this protein, which was the worst of those studied. Acidic urea solutions removed protein from the glass.

Proteins differ in their adsorption to silica; the process can be saturated, appears insensitive to ionic strength and is affected by conformation of the protein (Morrissey and Stromberg, 1974). Repeated washings appear to remove some of the adsorbed protein (Bull, 1957). Adsorbance of up to 4 mg/mm2 of glass have been reported (Morrissey and Stromberg, 1974; Bull, 1957), which corresponds to 6 µg in a 10 x 50 mm tube.

Up to 50 ml of solution can be concentrated on small bore columns, at flow rates up to 1 ml/min Simpson et al. (1989a); recoveries are claimed to be 90%. This group suggests using 0.01% Tween 20 to reduce losses of peptides during handling.

Don Hunt's lab in Chemistry routinely handle very small amounts of peptides by collecting in pthalate or siliconized polypropylene tubes and freezing samples as soon as they are collected by placing them on dry ice. However acetonitrile dissolves material from siliconized tubes which may interfere with subsequent operations. They concentrate samples by drying in a Speed-Vac but do not go to dryness. Solutions used to try to redissolve peptides include 30% acetonitrile and 70% formic acid. Another group found that time and temperature do not have large influences on adsorption of a peptide to polypropylene tubes, but addition of TFA to 33% improves recovery of peptide to 90% from 50% (Erdjument-Bromage et al., 1993).

An illustration of protein losses from pipetting into successive tube is provided by Stevenson et al. (1994) who showed that the non-ionic detergent Nonidet P-40 reduced the loss.

One report lists recovery of 10-30% from Centricon-10 units (Jones et al., 1994) but >80% recovery when desalting with gel filtration columns. Amicon have since shown that treatment of Centricons with a number of solutions, including 5% Tween-20, 5% polyethylene glycol, 5% Triton X-100 and 5% SDS reduced losses (Amicon document 2301). Amicon have developed Centriplus concentrators for which they report 90% recovery from 15 ml of 1 µg/ml solution of BSA and 50% recovery from a 250 ng/ml solution.

Long, high speed centrifugation increases recovery of proteins during trichloroacetic acid precipitation; even at 0.1 µg/ml, some proteins give 80% recovery (Hwang and Chu, 1996).

We have observed complete loss of sample when digesting 20 pmole of BSA in a gel slice, but addition of Tween 20 or PVP-360 allowed recovery of peptides (Shannon, 1995, unpublished observations). However Tween 20 interferes with mass spectrometry, while PVP-360, which is compatible with mass spectrometry, elutes in the latter part of a reverse phase separation of peptides. Lottspeich at a meeting (1996) suggested PPG-4000 to reduce sample loss.

Supplying a sample for sequencing

The most convenient container for samples is usually a 1.7 ml microcentrifuge tube. Polypropylene is the preferred material for storing protein samples, and the 1.7 ml size allows ready retrieval of samples, whether liquid or pieces of PVDF. The tube must be labelled with unique information; "1" is not a unique identification. Tubes with frosted surfaces are preferred because ink easily rubs off smooth tubes, especially after removal from a freezer.

You also need to decide what information is useful. The ideal sequence run gives continuous data from one polypeptide, but not all runs are ideal. Many intact proteins give no data because the N-terminal is blocked; if there is sufficient protein, there will be a background of amino acid. Normally we stop the run as soon as it is clear that no data is being obtained. In some samples, there are two or more polypeptides, even though other analytical methods suggest that a sample is pure; sometimes we can decide which amino acid belongs to which sequence but often, we have at each cycle, two or more amino acids and we cannot decide which sequence each belongs to; in such cases, you must decide if this data is of any use. Sometimes there will be amino acids which are not seen, either because amounts are so low, or because of the presence of unalkylated cysteine or because of some modified amino acid which is not seen on our analytical system.

Results

The raw data from a sequencing run is a series of chromatograms of PTH-amino acids. Amino acids are identified by their elution times. The chromatogram also has peaks from the injection (these occur in the first 4 minutes of the chromatogram), and the derivatives PMTC, DPTU and DPU formed by reactions involving PITC. These derivatives are useful for aligning successive chromatograms. Ideally there is one, large, easily identified amino acid. In impure samples and late in a run from even a pure sample, there are many amino acid peaks. To identify the amino acid produced in a sequencing cycle, successive chromatograms are compared to identify which amino acid(s) increase in a cycle.

The chromatographic data collected by our data system is archived on floppy disks. We provide a sheet listing the amino acids released by each cycle of the sequencer, and our interpretation of the sequence. We also compare the sequence with the library of sequences from the National Biomedical Foundation if a comparison is desired. We compare the sequences with the fasta program of Pearson and Lipman (1988). At the University of Virginia we are fortunate to have one of the authors of this widely used sequence comparison program on campus and a well maintained library of sequences.

Blocked Proteins

If a protein yields no sequence data, it may be blocked, although sometimes samples do not contain adequate amounts of protein. It has been reported that the majority of soluble proteins have blocked N-termini (Brown and Roberts, 1976; Driessen, 1985). One group could sequence five of fourteen proteins from a total cell extract (Hanash et al., 1991). In both blocked and unblocked proteins, alanine, serine, methionine and pyroglutamate are overrepresented at the N-terminal. If a large amount of sample is present, it will generate some background amino acids from acid cleavage of internal peptide bonds. To confirm the presence of significant amounts of protein in a sample that did not give data, we can digest the sample with cyanogen bromide after it has been applied to the sequencer. This process takes a day, and does not give any interpretable sequence data, because there will be several sequences. It will however tell if protein was applied to the sequencer (Simpson and Nice, 1984).

If you do have a N-terminal blocked protein and you need the N-terminal sequence, you can try deblocking although success is not assured. Normally it is more productive to digest the protein and isolate and sequence petpides. One problem is that it is not easy to determine what is blocking the N-terminus. If the blocking group is a pyroglutamate residue, also known as 5-oxopyrrolidine-2-carboxylic or pyrrolidone carboxylic acid, it may be removed by treatment with pyroglutamate aminopeptidase as described by Podell and Abraham (1978). This reaction has also been used by Strydom et al. (1985) and Zalut et al. (1980). This reaction can be performed on samples that have been electroblotted on to PVDF (Moyer et al., 1990). If this reaction is conducted, a control is advisable to see if the reaction works. We have used [Phe]2-TRH, whose elution from a C18 column is slightly altered after deblocking with pyroglutamate aminopeptidase and an amino acid analysis of the peptide isolated from HPLC will show the loss of the pyroglutamate residue. Because of variability of the commercial preparation of the enzyme, Fischer and Park (1992) recommend determining the optimal activity for each batch of enzyme. These authors have successfully deblocked 100 pmol or less of protein. At high concentrations of enzyme, they observed cleavage at proline. Similarly Dilone et al. (1994) observed removal of Gly-Pro from a peptide by pyroglutamate aminopeptidase.

Cyclization of terminal glutamine residues occurs slowly, but is accelerated by phosphate and other buffers, extremes of pH and temperature (Khandke et al., 1989); in 0.1M acetic acid at -20°C, one peptide showed 15% conversion in 7 months. During long purifications, glutamine residues may cyclize to pyroglutamate (Blombäck, 1967). Enzymes which catalyze the cyclization reaction have been described (Fischer and Spiess, 1987; Busby et al., 1987). If treatment with pyroglutamate aminopeptidase has no effect, the blocking group may be an acyl group.

There are two strategies to obtain sequence data from proteins with N-acetyl groups. One is to digest the protein and sequence peptides; if the N-terminal peptide is isolated, it can be deblocked and sequenced by Edman sequencing or mass spectrometry. The second is to deblock the whole protein. For a review, seen Chin and Wold (1986). Digestion of proteins is described later, so here are described the two methods for deblocking polypeptides, namely enzymatic and chemical.

The commercially available enzymes for removing N-acetyl amino acids use peptides as substrates, but not large proteins. Pierce Chemical Company sells an acylamino acid releasing enzyme. This enzyme is said to work only on peptides less than 30 residues and releases acyl amino acids or acyl dipeptides ; the suggested protocol uses 100 nmole of protein. Some references to the procedure, including identification of the removed amino acid are Nakamura et al, 1974; Jones and Manning, 1985; Tsunasawa and Sakiyama 1984.

The chemical methods used acid. The most recent uses trifluoroaccetic acid and methanol, which reportedly makes this method successful on all amino acids, rather than just N-acetyl serine (although almost all data shown is for N-acetyl serine), and the random acid cleavage of the polypeptide chain is claimed to be small (Bergman et al., 1996). In my hands, this method did not deblock some acetylated peptides, apart from one with N-acetyl serine, and did not deblock cytochrome c, which has a terminal acetyl group. Other methods, which probably are only successful for N-acetyl serine, are heating the peptide with 1 M HCl at 110°C for 10 to 20 minutes which may remove an acetyl group (Fordyce et al., 1979; Chin and Wold, 1985) or add anhydrous TFA to a polypropylene tube and incubate at 40°C for 1 hour (LeGendre et al., 1993) or other acidic conditions (Ozols, 1989). Hulmes et al. claim deacetylation by exposure to trifluoroacetic acid vapour at room temperature for about a week; they report cleavage at serine residues also with efficiencies of about 25%. Similarly, Wellner et al. (1990) found that limited acid treatment did not deblock cytochrome c, but did deblock peptides with N-acetyl serine or N-acetyl threonine at the N-terminal.

A survey of acetylated proteins found that 43% of N-terminal residues are serine or threonine suggesting that limited acid treatment has significant chances of success (Persson et al., 1985). Examples of the determination of an acetyl blocking group are seen in Haniu et al., (1984); Asano et al., (1986).

Proteins with an N-formyl group at the N-terminus may be partially deblocked by heating for 2 h at 55°C in 25% trifluoroacetic acid; success is variable (Shively et al., 1982). Another procedure is to incubate the sample in a 1.5 ml tube with 30 µl of 0.6M HCl for 24 hours at 25°C as starting conditions (LeGendre et al., 1993). When we tried this procedure on a bacterially expressed protein on which N-formyl methionine was expected, there was only slight acid cleavage, rather than deblocking.

A recent review of obtaining sequence data from acetylated proteins used relatively large amounts of protein and found that the specificity of the enzyme will result in little or no data from some proteins (Krishna, 1992). Uses of this enzyme in real sequencing include Schinina et al. (1996). An interesting way to use this enzyme is described by Tsunasawa et al., (1990). The authors digested a protein, blocked the newly released -amino groups with PITC and oxidized to phenylcarbamyl groups, then removed the acetylated amino acid with the enzyme. The mixture of peptides was then applied to the sequencer and only the N-terminal peptide was sequenced. Krishna et al. (1991) obtained limited sequence data also using chemically blocking the N-termini of peptides. To obtain data, hundreds of pmole of protein are required; identification of the N-terminal amino acid is possible with large amounts of protein. A similar technique using ion exchange chromatography is described by Allen (1989). A procedure for isolating N-terminal peptides using adsorption of dinitrophenol derivatized amino groups has been reported (Kaplan and Oda, 1987). Another technique was described by Akiyama et al. (1994).

A systematic application of the above techniques to proteins blotted on PVDF was decribed by Hirano et al. (1993). The procedure appears relatively insensitive, and may introduce peptide bond cleavages from the attempted acid based deblocking steps which may confuse latter data, especially for large proteins.

Some intracellular proteins have myristic acid on their N-terminal glycine (Sefton and Buss, 1987). The residues which determine whether a terminal glycine is myristoylated are within ten residues of the N-terminal (Towler et al., 1988) and include a lysine at position 7 (Kaplan et al., 1988). A N-Acetyltransferase from yeast displays substrate specificity which is affected by parts of the protein distant from the N-terminal (Lee et al., 1990); the enzyme from rat liver is mainly affected by the first three amino acids (Yamada and Bradshaw, 1991). N-acetylation is believed to be a co-translational event (Yamada and Bradshaw, 1991).

Blocked N-terminal peptides can be selectively isolated from digests by cation exchange chromatography (Gorman and Shiell, 1993).

We suggest that if a protein is blocked, it be digested and the peptides be sequenced, unless obtaining the N-terminus is vital.

Sequencing Proteins From Gels

Many proteins are isolated by gel electrophoresis. To obtain sequence data, the two normal methods are to either transfer to PVDF and sequence, or to digest the protein and sequence peptides. Transferring the protein to PVDF and then attempting sequencing is the more simple approach but success is limited because of the frequency of blocked proteins. Digestion of protein from gels is described separately.

When preparing samples for electrophoresis, boiling may hydrolyze peptide bonds, especially Asp-Pro bonds (Geiger and Clarke, 1987); indeed boiling may be used for peptide mapping (Rittenhouse and Marcus, 1984) although boiling for 3 minutes has been considered to give negligible hydrolysis (Kowit and Maloney, 1982), or according to Rittenhouse and Marcus (1984), about 1%. Tubulin is a protein that is very susceptible to non-enzymatic hydrolysis of peptide bonds, namely Asp-Pro, Gly-Ser; in addition, heating may cause ß-elimination by cysteine, producing dehydroalanine, which may react with lysine, leading to cross links between peptides (Correia et al., 1993). The Asp-Pro hydrolysis appears acid catalyzed, because of the weak buffering of Tris at pH 6.8, and the drop in pH with temperature; phosphate pH 7 gives adequate pH control without raising the pH to where disulfide interchange can become a problem (Cannon-Carlson and Tang, 1997). Kubo (1995) noted that even at pH 6, inorganic cations do not cause hydrolysis, while Davagnino et al. (1995) preserved immunoglobulins by using a higher pH Tris buffer. Another danger of excessive heating is loss or oxidation of all reducing agent, with possible reformation of disulfide bonds.

Sample preparation of hydrophobic peptides at low temperatures for electrophoresis has been described recently and is claimed to increase the solubilisation of peptides and reduce streaking (Hennessey and Scarborough, 1989)

Harrington (1990) claim that using diacrylylpiperzaine as the cross-linker in gels increases the yield of sequenceable protein by two. They mention that bisacrylamide breaks down to formaldehyde which may react with amino groups and cause blockage of the N-terminal.

N-terminal blockage is a potential problem but there are no definitive reports on its prevalence; Reim and Speicher (1992) find 10-15% blocking during transfer, which is within the range of normal variability in yield. The following steps may be taken to reduce the chance of N-terminal blockage: use of reducing agent in the solution of protein, pre-electrophoresis of gels to remove potential reactants, degassing of gel solutions to reduce the amount of catalyst needed and hence the concentration of free radicals, use of reducing agent in the electrophoresis buffers, recrystallization of SDS, long polymerization of gels (up to 3 days) to reduce concentrations of potential reactants, using sucrose instead of urea in a sample buffer, heating samples to 37°C for 10 minutes; (Walsh et al., 1988; Moos et al., 1988; Speicher, 1989). It has been claimed that the most important reagent to reduce is ammonium persulphate (A. Louie, Applied Biosystems seminar, 1991). One possible reactant in gels which could lead to blocked N-termini is free acrylamide (Geisthardt and Kruppa, 1987). Data of Bosisio et al (1980) suggests that 50 mM acrylamide may be present in a typical gel. Moos et al. (1988) recommend conducting electrophoresis at near neutral pH, rather than the usual pH 8.5 and suggest that electrophoresis at lower than normal pH is the most important factor in preventing protein blockage. However it has since been reported that a low pH for electrophoresis improves sequencing yields of some proteins but decreases the sequencing yields of other proteins (Stone et al., 1989a);blocking during electrophoresis The low pH also causes some proteins to smear on gels. Lysine and terminal amino groups can be modified in gels by Schiff base formation, a process reversed at high pH (Kosiarz et al., 1978) which may be why it does not seem to be a problem in sequencing where the sample is exposed to base. Ward et al. (1990) suggest that electroblotting causes considerable loss of tryptophan.

Hemoglobin (5µg) in sample wells increases the amount of protein transferable out of a gel on to membranes, possibly by blocking reaction of acrylamide with proteins, thereby incorporating some protein into the gel (Gillespie and Gillespie, 1997). Overnight incubation dcreased protein yield rather than increasing it, and pre-electrophoresis had no benefit. This report agreed with an earlier finding that keratin, a common contaminat of gels, is partially solubilized by reducing agents, causing streaks and giving the appearance that 2-mercaptoethanol is a source of contaminats (Ochs, 1983).

Pre-running of gels may decrease resolution, but casting the stacking gel with resolving gel buffer and performing pre-running with stacking gel buffer in the upper tank maintains resolution (Dunbar and Wilson, 1994)

Another possible reaction that occurs in acrylamide gels is alkylation of cysteine by free acrylamide. This phenomenon has been observed by several laboratories, some of which have used the cysteinyl-S-propionamide adduct to identify cysteine (Ward et al., 1990, Ploug et al., 1989). However only some cysteine residues may react (Chiari et al., 1992). Controlled alkylation prior to electrophoresis is preferred.

A double, inverse gradient gel system may increase resolution of proteins of similar sizes (Zardoya et al., 1994). To separate large proteins by electrophoresis, FMC recommend their agarose gels, which can separate proteins up to 600,000 (FMC, 1991).

To separate small peptides as peptides, small as 1,000 daltons, Schägger and von Jagow (1987) suggest electrophoresis using Tricine buffers. Novex (Encinitas, CA), Schleicher and Schuell, Integrated Separation Systems (Natick, MA) and the Nest Group sell precast gels that are claimed to separate peptides as small as 1,500. The Nest Group claim that their gels have longer shelf life and better resolution, in part due to the use of high quality SDS than gels made in the normal laboratory. Integrated Separation Systems also sell a stain that is claimed to efficiently fix and stain small peptides, but the stain is not stated to be compatible with further analysis of the peptides. Hoefer list an electrophoresis system for peptides down to 1,500 in their 1992 catalogue.

Elution of proteins from gels

Below are methods for removing proteins from gels, but these methods have limited uses. Normally to obtain sequence data from an intact protein, transfer to PVDF is the method of choice. If the protein is blocked, the normal method is to digest in the gel and then elute, separate and analyze peptides.

There are devices for direct elution of protein from gels. ISCO, BioRad, Hoefer, Schleicher and Schuell, and Amicon make such apparatus, and another has been described by Hunkapiller et al. (1983). The Amicon unit claims recoveries of 80% to 95%, with unfixed proteins; they suggest staining with cold KCl, sodium acetate or Coomassie without acid. Promega claim that proteins stained during electrophoresis with their ChromaPhor Protein Visualization System yield 75% recovery. One report that stains with 4M sodium acetate finds 70% recoveries when 100 µg of protein is applied to a gel (Ohhashi et al., 1992).

One procedure for passive elution of proteins from gels is that of Ward et al. (1991), in which the gel slice is soaked in 3 x 1 ml of 20 mM Tris-HCl pH 7.4, 0.02% Tween 20 for 20 hours at 25oC. Coomassie ,effect on elution of proteins stained BSA was poorly eluted from a gel (1%) whereas 90% of unstained BSA was passively eluted after 70 hours (Ward et al., 1991). To remove Coomassie, staining was only for 15 minutes when possible, then the gel pieces were washed with 3% SDS, 50% 2-propanol to remove Coomassie, then washed with water for 24 hours. The SDS is to dissociate the Coomassie and the propanol keeps the protein fixed (Ward, personal communication).

Promega sell a Protein Recovery System to be used in conjunction with their Protein Visualization System which stains proteins during electrophoresis (Promega Technical Bulletin 100). Promega claims recovery of 50% or better for proteins with molecular weights of 31, 55, 97 Kd (Larson and Shultz, 1993). The Recovery System uses homogenization of the gel followed by extraction with 100 µl of 50 mM ammonium bicarbonate, 0.1% SDS. The supernatant from centrifuging the crushed gel contains the protein. After an essential concentration step, the SDS and dye are removed by acetone precipitation. The dye inhibits some proteinases. The system appears to be based on earlier reports of passive elution of proteins from gels (Hager and Burgess, 1980; Bray and Brownlee, 1973).

Another method of from gels is to soak the gel slice, containing protein visualized with 1 M KCl, in 1 ml of 70% formic acid for 10 minutes and then centrifuge the mixture on to a piece of PVDF. The ProSpin cartridges sold by Applied Biosystems appear ideal for this method. Recoveries of protein from electrophoresis step at 10-20%. The method has been used with proteins from 13,000 to 67,000 (Hermansen et al. 1992).

Another method for eluting proteins from gels is to add 200 µl of 50% trifluoroacetic or formic acid, crush the gel with a Kel-F pestle and sonicate; recoveries decrease with the protein size (Asquith et al., 1992)

Feick and Shiozawa (1990) report that elution of proteins from gels with formic acid:acetonitrile:2-propanol:water 50:25:15:10 v/v/v/ followed by gel filtration gives recoveries of over 60% in most cases.

Unstained proteins can be detected in gels by phase contrast (Johnson et al., 1990) but the commercial instrument has been withdrawn from the market.

Zinc or copper staining is claimed to be almost as sensitive as silver staining and reversible, so that the stain does not interfere with blotting or sequencing (Fernandez-Patron et al., 1995a; Ortiz et al., 1992). When treated with the right solutions, proteins can be digested and blotted with minimal loss compared to unstained proteins, even after storage for years. Other stains are the Rapid Reversible Stain of Diversified Biotech which is claimed not to affect biological properties of proteins and to be removable for elution of the protein from gels. Nile Red stains proteins in SDS gels, and with modifications, in IEF gels, without interfering with subsequent operations (Bermudez et al., 1994). Fluorescence detects 10 ng of protein (Alba et al., 1996). ESA (1997) sell Rev-Pro reversible stain for PVDF and nitrocellulose membranes; it is claimed to be more sensitive than Coomassie Blue, and compatible with sequencing.

Proteins can be removed from gel slices by homogenization and spinning in an Amicon Microcon concentrator fitted with a Micropure insert for filtration (Amicon document 1901; Sheer, 1994). In further studies, Amicon (Krowczynska et al., 1995) showed that unstained gels give the best recovery, or using KCl stained gels while Coomassie or zinc imidazole staining reduces recovery. The elution buffer is less important, with best results from 50 mM Tris-Cl, 0.1 mM EDTA, 150 mM NaCl or PBS.

An interesting method is reverse staining of proteins with zinc, followed by electroelution on to a small reverse phase column, from which the protein is eluted with a gradient of acetonitrile (Fernandez-Patron et al., 1995b). This method may separate chromatographically proteins which are not resolved by electrophoresis. However it requires construction of special apparatus and the best sensitivity reported was 50 pmole of protein.

Removal of gel matrix

Another method of recovering proteins that are purified by electrophoresis is to use FMC's agarose based ProSieve electrophoresis system, in which the agarose gel is melted to allow extraction of the protein. Recoveries of 90% are reported (Morgan et al., 1991). Bands are not as sharp as acrylamide gels. Epicentre Technologies (1202 Ann Street, Madison, WI 53713, 800-284-8474) sell GELase™ to digest agarose but we suggest that the absence of proteolytic enzymes be determined before using this product.

BioRad sells the reversible cross linker N,N'-bis-acrylylcystamine for acrylamide. Although this product has been available for several years, it has not been widely used. One reason is that achieving a completely soluble gel may require elevated temperatures during polymerization, purification of the reversible cross linker (Hansen, 1981) and the use of formamide to avoid non-reducible cross links (Hansen et al., 1980). This cross linker is less reactive than bis-acrylamide and may give gels that do not perform well (Gelfi and Righetti, 1981). Ghaffari et al. (1988) obtained satisfactory resolution, although less than from bis-acrylamide and 50-70% recovery of proteins at a 1 µg load.

Sample concentration from multiple gels

To analyze proteins separated on 2-D gels, Bauw et al. (1990) cut out bands from multiple gels and loaded up to 20 gel segments on top of another gel which concentrated the protein. The second gel has a custom made well which concentrates the sample, which can be gel pieces or dilute protein solution, horizontally to give a band 5 mm2 (Rasmussen et al., 1991).

Another technique needs only a different spacer to form a funnel web well. The data shows excellent recovery from multiple gel slices (Lombard-Platet and Jalinot, 1993). A variation runs the protein from multiple pieces of gel into an agarose gel, from which the protein is extracted by melting (Rider et al., 1995).

Another gel concentration technique uses a concentration gel inside a Pasteur pipet, leaving the protein in 1-2 µl of gel (Gevaert et al., 1996).

Proteins from multiple gels can be concentrated on one piece of PVDF by homogenizing the gel and spinning the extract on to PVDF (Warlow et al., 1995). The protein can then be sequenced directly or digested on PVDF.

Blotting to PVDF

The standard procedure now for sequence analysis of proteins separated by electrophoresis is to blot on to PVDF,( polyvinylidene difluoride) membranes (Immobilon-P from Millipore) as described by (Matsudaira, 1987; LeGendre and Matsudaira, 1988). Nitrocellulose cannot be used in the sequenator because of its susceptibility to the harsh conditions employed.

We suggest amount of sample loading 100 pmoles on a gel to increase the chances of having sufficient sequenceable protein present after losses during electrophoresis and transfer. One simple rule for proteins with molecular weights up to 30,000 is that if a stained blot can be photocopied, it can be sequenced. After blotting, it is wise to stain the gel to see if all protein transferred from the gel. Also a second piece of PVDF will indicate how much protein passes through the primary membrane. The amount of protein passing through PVDF is less with some brands (see Selection of PVDF) and may be reduced by presoaking the gel in transfer buffer (Mozdzanowski and Speicher, 1990).

It is reported that the amount of protein on PVDF can be quantitated, down to 0.5 µg, by eluting the bound Coomassie Blue (Kain and Henry, 1990). One laboratory has quantified immunostained and Coomassie stained protein on PVDF by wetting the membrane with methanol to make it transparent and then scanning (Parrado et al., 1993). quantitation of protein In our laboratory, Dr. E. Baramova loaded about 1.3 nmole of a protein on a gel and was able to sequence over 25 amino acids.

Amino acid analysis of proteins electroblotted to PVDF shows typical recoveries of 28%-47% (Tous et al., 1989, Nakagawa and Fukuda, 1989). Stone and Williams (1993) report an average blotting efficiency of 34% for proteins between 14,000 and 116,000. Henzel et al. (1994b) report that thin gels increase transfer efficiencies, especially for larger proteins; 0.5 mm thick gels give transfer efficiencies of 60-90%.

Transverse gradient gel electrophoresis can determine the optimal concentration of for electroblotting, by showing resolution and transfer efficiency on one gel (Smejkal and Gallagher, 1994). As a guide, these authors show complete transfer of proteins up to 45,000 with gels up to 12% acrylamide, up to 60,000 with a 7% gel and up to 120,000 with a 6% gel.

Although not tested with PVDF, ultrasound has been used to transfer proteins to nitrocellulose (Kost et al., 1994).

Novex in their catalogue suggest Tricine gels because they are more porous than the corresponding Tris containing gels and therefore may give better transfer.

A new approach to efficient transfer of proteins is the use of different buffers on each side of the PVDF transfer membrane (Laurière, 1995). On the gel side, high pH and SDS promotes elution of the protein from the gel, which on the other side of the PVDF, low pH and methanol promote binding to the PVDF. Under these conditions, large and small proteins transfer efficiently to the membrane.

Selection of PVDF

The PVDF that is best known is Immobilon-P from Millipore. PVDF is also available from Applied BioSystems (ProBlott), BioRad and Schleicher & Schuell (Westran); these companies claim that their products are superior to Immobilon in retaining proteins. Westran is PVDF on a polyester support. In support of the claims of Applied Biosystems, Promega in their Probe-Design kit for generating and separating peptides, state that peptides are more easily eluted from Immobilon and better retained on ProBlott. Millipore has released a second version of Immobilon, the Immobilon-PSQ membrane which they claim binds more proteins and gives much higher yields for . In 1992 Porton Instruments released Hyperbond™ membrane which is claimed to retain small peptides better than existing high density PVDF and to allow more efficient sequencing reactions with more data obtained as a result.

In our laboratory, we have compared three brands of PVDF. Ken Klotz, Department of Cell Biology, transferred 50 pmole of ß-lactoglobulin to the PVDF. ProBlott from Applied Biosystems gave twice as much signal as Immobilon and slightly higher repetitive yields; Westran also retained more sequenceable protein than Immobilon but less than ProBlott. The experiment described here used only one protein and one condition for transfer, (10 mM CAPS, pH 11, 10% methanol, 500 mA, 40 min from 1.5 mm thick 10% gel) so that the advantage of ProBlott may not always be as great under other conditions; also it should be remembered that many people have successfully obtained data with Immobilon. It is claimed that three brands of PVDF selection of PVDF perform similarly when excess SDS is removed but ProBlott and Bio-Rad PVDF are superior when excess SDS is present (Speicher et al., 1990). Later this lab showed that ProBlott and Bio-Rad's Transblot bind small proteins better than Immobilon-P, and that Immobilon-P is more sensitive to SDS (Mozdzanowski and Speicher, 1992a). PVDF which binds proteins more tightly gives sequencing performance as good as looser binding PVDF, and better yields of tryptophan (Reim and Speicher, 1992).

Jungblut et al. (1990) reported only small differences in efficiency of transfer of a range of proteins on to ProBlott™, Immobilon™, a polypropylene membrane (Selex-20) and a siliconized glass fibre (Glassybond). When the increased initial yields obtained with the siliconized glass fibre are considered (Eckerskorn and Lottspeich, 1990), the authors suggest the siliconized glass fibre is the best medium for electroblotting and sequencing of proteins. However this membrane is not in widespread use.

Another study of transfer membranes showed that ProBlott usually binds more protein that the other tested membranes. When the efficiency of transfer and sequencing are considered, ProBlott appears to usually be the best choice of membrane (Baker et al., 1991). Immobilon PSQ was not included in this study. A Millipore representative suggests that because both Immobilon PSQ and ProBlott have 0.1 µm pores, rather that 0.45 µm of the original Immobilon, they will bind protein in a similar manner. Micron Separations (MSI) sell PVDF-Plus, with 0.45 µm pores, which they say is comparable to Immobilon-P. LeGendre et al. (1993) report recoveries of 30-80% of protein on Immobilon-P and 100% on Immobilon-Psq and ProBlott.

Buffers for electroblotting

Some studies claim that there is not a systematic difference between CAPS and Tris-glycine buffers (Speicher, 1989; Stone et al., 1989a), although some proteins exhibited less efficient transfer out of gels, or less efficient binding on PVDF membranes in one of the buffers. More recently Mozdzanowski and Speicher (1990) claim that Tris-glycine and Tris-borate buffers (50 mM Tris, 50 mM boric acid, not titrated, pH ~8.3, 10% methanol, 225 mA, 2-4h) give 40-50% recovery of blotted proteins on the first PVDF membrane compared to about 30% recovery obtained with CAPS buffer and sodium borate. Tris buffers must be removed to prevent interference with sequencing. Baker et al. (1991) also suggest that Tris-borate pH 8.5 is often superior to CAPS. Speicher (1989) used 12.5 mM Tris, 96 mM glycine pH 8.3 for 2 to 4 hours as his routine conditions which he claimed gave quantitative transfer of most proteins, the exceptions being those in the top 15% of the gel (a range of acrylamide concentrations were used; the concentration used which gave the reduced recovery was not identified) where recovery was still 75% or more.

Bauw et al. (1990) report that Tris-borate (50 mM Tris, 50 mM borate, 8 h, 35 V) gives better recovery of proteins than Tris-glycine although transfer is slower.

Bhavsar et al. (1994) find that addition of BSA as a carrier protein improves efficiency of transfer. Addition of BSA would interfere with sequencing applications, but deoxycholate may improve transfer efficiency, or decrease it.

Transfer time for electroblotting

Speicher (1989) suggests that transfer times need to be empirically optimized but CAPS needs shorter times than Tris-glycine buffers. Later this author (Mozdzanowski and Speicher, 1990) suggests that 2 hours of blotting is sufficient and that longer transfer times do not remove proteins that remain in the gel, nor do they move proteins from the first PVDF membrane to the second, a finding confirmed by Jungblut et al. (1990). Mozdzanowski et al. (1992) showed that overtransfer does not occur i.e. once a protein has been bound to PVDF, it does not elute during the transfer.

A study by Baker et al. (1991) shows that a large protein does not transfer as fast as a small protein. With some membranes, there was loss of protein in less than an hour of transferring but no data after one hour was shown.

Effect of methanol and SDS on electroblotting

Gültekin and Heerman, 1988 report that in the transfer buffer prevents binding of proteins to PVDF; methanol is reported to reduce the elution of some proteins from gels (Eckerskorn et al., 1988). Mozdzanowski and Speicher (1990) suggest 10% methanol, because less methanol decreases binding to PVDF; he also reports that SDS usually decreases binding to PVDF. A few membrane proteins, transfer from gels were transferred more efficiently without methanol; apparently methanol was not needed to dissociate SDS from the protein. Mozdzanowski and Speicher, 1990) claim that most losses losses of protein associated with electrophoresis and blotting occur during transfer and are independent of the buffer used but are caused by high concentrations of (Speicher et al., 1990). To avoid the losses, they recommend the presoaking of the gel in transfer buffer. For some proteins below 20,000, presoaking the gel increases the amount of blotted protein by 70%, even though less protein is transferred out of the gel (Mozdzanowski et al., 1992). The amount of protein bound to PVDF is about 50% of that in the gel. Eckerskorn et al. (1988) also report that methanol decreases the movement of large proteins out of gels. Speicher explains that methanol is needed to dissociate SDS from the protein; too little methanol will cause incomplete SDS removal and hence poor binding to PVDF, whereas too much methanol will remove SDS from the protein in the gel. Because the charge on the SDS converts the electric field into movement of the protein, removal of SDS in the gel by high concentrations of methanol will result in inefficient transfer of the protein from the gel. Guidelines from Applied Biosystems are to use 20% methanol for transferring protein below 20,000, 10% methanol for proteins between 20,000 and 80,000 and no methanol for larger proteins (Applied Biosystems seminar, 1991). For large proteins, it may be necessary to add to the transfer buffer to help move the proteins out of the gel.

in the transfer buffer reduces the amount of protein bound by PVDF several fold, but addition of 0.1% NaCl partially restores binding efficiency (Tovey and Baldo, 1989).

Transfer of large proteins

Large proteins are the most difficult to transfer and sequence because of low yields on transfer and then relatively high backgrounds of amino acids during sequencing. Some factors which may increase the success of electroblotting are listed here. Besides reducing the methanol concentration, or even omitting methanol (Pluskal et al., 1986), another technique that may improve extraction of proteins from the gel is to presoak the gel in transfer buffer for 30 minutes (Mozdzanowski and Speicher, 1990). Increasing current density may increase efficiency of transfer of large proteins at the expense of smaller proteins (Jungblut et al., 1990). A blotting procedure varying the current is claimed to give good recoveries of both large and small proteins (Otter et al., 1987).

Blotting from a gel with a low concentration of may also help and is more successful than reducing methanol concentration in the transfer buffer (Mozdzanowski et al., 1992). Adding 0.05% to the transfer buffer may help.

At the Max Planck Institute in Munich, one recommended blotting procedure is to use 10 mM sodium borate (no adjustment of the pH) overnight in the cold room with cold circulating water at 200 mA.

Transfer of small peptides

Millipore report that proteins less than 20,000 transfer easily and are insensitive to methanol concentration. Because they can diffuse, equilibration of the gel in transfer buffer should be less than ten minutes. Porton Instruments claim that their Hyperbond ™ membrane is more efficient than existing PVDF membranes at binding small proteins. To enhance the binding of small peptides to PVDF, Millipore suggest using 20% methanol in the transfer buffer and reducing the electric field by 50%; a procedure is described by Otter et al. (1987). They also recommend their new product, Immobilon-PSQ.

Using semi-dry blotting apparatus

This type of apparatus is less efficient in transferring proteins than a tank transfer apparatus (J.W. Fox, unpublished observations; reported in poster of Mozdzanowski and Speicher, 1989)

A study using a Millipore semidry blotting apparatus, found that the amount of membrane proteins extracted from a gel was inversely related to the molecular weight of the protein (above 12,000). Addition of 0.1% SDS to the cathode buffer increased the transfer of larger proteins and 35% or 40% methanol in the anode buffer improved the retention of proteins less than 12,000 on the PVDF. However the high methanol concentrations decreased the amount of larger proteins bound to the PVDF. These modifications are not applicable to transfer in a tank type apparatus (Lissilour and Godinot, 1990). Parameters affecting transfer efficiency in a semi-dry apparatus have also been studied by Jungblut et al. (1990).

Other procedures for electroblotting

Millipore has a book of electroblotting protocols for different applications, Protein Blotting Protocols for the Immobilon -P Transfer Membrane.

One of the protocols that may be useful for some situations is for electroblotting of large proteins from non-denaturing gels; this procedure may be useful because of the difficulty of transferring large proteins. No methanol is used in the transfer buffer; data on the amount of protein transferred is not shown. A procedure for transfer of basic proteins from non-denaturing gels to nitrocellulose has also been published (Van-Seuningen and Davril, 1990)

Non-denaturing gels can be coupled with SDS-PAGE; two denaturing gels may be used to concentrate the sample (Trudel and Asselin, 1994).

Xu and Shively (1988) recommend transfer times of 100 minutes, at a lower voltage than specified by Matsuduira (1987) and report decreases in recovered protein if the electroblotting is continued. These authors also claim that coating the Immobilon-P membranes with Polybrene increases the yield of transferred protein, as reported by Matsuduira (1987) and Walsh et al. (1988), but no longer recommend the use of Polybrene (conference report).

Amount of PVDF used in electroblotting

PVDF is very hydrophobic and poorly wetted by the butyl chloride used in the sequencer to extract ATZ-amino acids from the PVDF which binds the protein being sequenced. It has been found that putting a large, or very small amount of PVDF in the sequencer is more likely to cause variable flow of butyl chloride over the PVDF membrane, causing variation in the amounts of amino acid seen (Speicher, 1989); this reported variation is consistent with our observations. The suggested ideal size for PVDF membrane is 4 x 9 mm, with two pieces this size; larger amounts can be loaded but are prone to erratic flows of reagents and hence erratic levels of amino acids. Another piece of data on the amount of PVDF is that of Stone et al. (1989a) who found that loading protein on a large amount of PVDF decreased the repetitive yield during sequencing.

Staining of electroblotted proteins

; Destaining is also reported to affect the yield of sequenceable protein (Speicher, 1989); destaining with 50% methanol/10% acetic acid was reported to reduce yields of sequenceable protein by 30% or more compared to destaining with 50% methanol alone. Phang et al. (1996) report that omission of acetic acid from staining and destaining solutions slows the processes but does not affect the final results. The staining procedure suggested is to stain with 0.5% Coomassie blue in 40% methanol and destain in 50% methanol (Speicher, 1989) (time unspecified but possibly five minutes and not more than half an hour). This author has since reported that Coomassie may reduce the yield of sequenceable protein, although he was not working with electroblotted proteins, but rather proteins adsorbed on PVDF; Ponceau S was suggested as a stain which causes less loss of protein (Purcell & Speicher, 1989). Choli et al. (1989) report that 40% of lactoglobulin spotted on to PVDF is sequenced, but only 24% of that which is stained with Coomassie; Ponceau S was not tested. Drying the blot after staining enhances visibility of proteins (Sanchez et al., 1992).

Another destaining procedure is to soak the gel in 5% acetic acid:1-butanol 9:1 biphasic mixture; this system is claimed to be more rapid than conventional destaining (Molnar et al., 1990).

Ponceau S is less sensitive than Coomassie. Staining is stronger at acidic pH and is reversible at basic pH (Aebersold, 1989). Schleicher & Schuell claim that destaining in 10% 2-propanol/10% acetic acid gives permanent staining (instructions for Westran).

Promega have three protein staining systems. They claim that their ChromaPhor Protein Visualization System, which stains proteins during electrophoresis, is compatible with blotting and sequencing but they do not show data on recoveries (Breitlow, 1992). This company also reports loss of small and hydrophilic proteins from PVDF during staining and destaining with Coomassie blue. The suggested conditions are to stain for 30 s in 40% methanol, 5% acetic acid, 0.02% Coomassie blue and destain for 1 min in 40% methanol, 5% acetic acid. They do not address the issue of loss of sequenceable protein caused by Coomassie blue or acetic acid (Shultz, 1992).

In addition to staining proteins after transfer, Thompson and Larson (1992) successfully transferred proteins stained with Coomassie Blue or ChromaPhor stain. By soaking the stained gel in 1% SDS, 50 mM Tris-Cl pH 7.5, these workers presumably coated the protein with SDS so that an electric field would move the protein out of the gel, even though it had been fixed. After soaking in SDS, the gel was soaked in transfer buffer for 15 minutes.

Copper phthalocyanine 3,4',4",4'"-tetrasulphonic acid has been proposed as more sensitive than Coommassie blue, easily removed with 0.5 M NaHCO3/20% ethanol (Bickar and Reid, 1992). Another chelating method uses a Ferrozine/ferrous complex or ferrocyanide/ferric complex for protein on PVDF or nitrocellulose (Patton et al., 1994). The iron based stains are as sensitive as gold stains, but are reversible and after reversal, cause no interference with sequencing or digestion.

An alternative to staining is to dry PVDF for 10 minutes after blotting, rewet in 20% methanol and look at it in front of white light or place in 20% methanol over a light box. Protein bands will appear more translucent than background (LeGendre and Matsudaira, 1989). By wetting the PVDF in 50% methanol and scanning at an unspecified, protein bands can be quantitated to 1 µg (Parrado et al., 1993).

Merewether et al. (1995) used Coomassie Brilliant Blue G stain instead of R, because there was only one, late eluting peak. Amido Black gave negative deflections in the baseline, interfering with low level peptide collection.

Blotting to other supports

A procedure to transfer proteins from polyacrylamide gels to glass fiber (Aebersold et al. 1986) does not appear to be used any more.

Alimi et al. (1993) transferred proteins to carboxymethylcellulose membranes purchased from Schleicher and Schuell (NA-49). Most of the proteins studied were basic but alcohol dehydrogenase, serum albumin, ovalbumin and catalase also transferred to the membranes with "high yields". A major advantage of CMC membranes is ease of eluting proteins, with 50 mM HCl

Proteins transferred to nitrocellulose cannot be loaded into a sequenator because the membrane disintegrates under the conditions used. Although some prefer to use nitrocellulose, (Aebersold et al., 1987; Aebersold, 1989), PVDF appears more popular.

When proteins are blotted on to nitrocellulose, up to 90% of small proteins can be eluted with acetonitrile solutions (Parekh, et al., 1985). Triton X-100, at 0.1% to 1% also may elute proteins; Polyethylene glycol under low salt conditions is a second choice. (Gelman Sciences, usenet message, 1994) . Lui et al. (1996) investigated elution of proteins from PVDF, and found that high temperatures or high TFA partially degrade proteins; alkaline solutions of acetonitrile are moderately successful; 1% piperidine in 40% acetonitrile is up to 90% effective. SDS and some other ionic detergents were ineffective, whereas several non-ionic detergents werw effective, and the favoured detergent was

3-16 (3%); adding organic solvent to detergents did not enhance elution. Zwittergent 3-16 gave greater recoveries (up to 90%) of peptides from digestion of proteins on nitrocellulose. Staining proteins reduces elution efficiency, so the authors suggest staining a narrow strip of nitrocellulose to locate the band of nitrocellulose to be eluted. Aebersold (1989) suggests that extensive drying of nitrocellulose causes irreversible binding of proteins to nitrocellulose and that desorption is favoured at basic pH. Another procedure for removing proteins from nitrocellulose is to dissolve the membrane in acetone and precipitate the proteins with ammonium bicarbonate (Anderson, 1985).

A theoretical study with observations of the binding of proteins to nitrocellulose shows that salt and 20% methanol promote binding of proteins (van Oss et al., 1987).

Electroblotting from IEF and 2-D gels

It has been reported that proteins can be blotted from isoelectric focusing gels after removal of ampholytes by perchloric acid (Hsieh et al., 1988). Proteins have also been blotted from immobilized pH gradients (Knierim et al., 1988). A recent two dimensional electrophoresis system, using acid urea gels reportedly can handle relatively large amounts of protein, and separation by hydrophobic properties can follow using a third electrophoresis (Vanfleteren, 1989). A two dimensional mapping system using acid urea minigels has been described earlier (Davie, 1985). Blotting from two dimensional gels on to derivatized glass fibre has been reported (Bauw et al., 1987).

Proteins separated on two-dimensional gels have been analyzed by Kennedy et al. (1988a, 1988b) by performing digestions of the protein in the gel, then separating fragments by electrophoresis, transferring to PVDF and sequencing. It is claimed that data was obtained from 1-10 µg of protein, which may require pooling of material from several gels. To analyze proteins separated on 2-D gels, Bauw et al. (1990) cut out bands from multiple gels and loaded up to 20 gel segments on top of another gel which concentrated the protein. Jahnen et al. (1990) also describe sequencing of proteins separated on 2-D gels. Immobilized pH gradients have been used; ten gels give sufficient amounts of the more abundant proteins to enable sequencing (Hanash et al., 1991).

Hochstrasser et al. (1988) claim that a modified 2-D electrophoresis system increases resolution over O'Farrel's original system.

Sequencing peptides produced by digestion of a protein.

Often a protein must be digested to peptides to obtain sequence data, most often because of a blocked N-terminal, and also because sequencing from the N-terminal may not give enough data. The steps of digestion and separation of peptides cause some sample loss, so more protein is needed than when sequencing the N-terminal of a protein.

Choice of digestion.

There are enzymatic and chemical methods for digesting proteins, but enzymatic digestions are more common. An ideal digestion cuts only at a specific amino acid, but cuts at all occurences of that amino acid. The number of digestion sites should not produce too many peptides because separation of peptides becomes too difficult. On the other can, too few digestions produces peptides too large for complete analysis.

The most common digestions are with trypsin and lysine specific proteinases, because these enzymes are reliable, specific and produce a suitable number of peptides. The next most common digestion is at aspartate or glutamate using endoproteinase Glu-C or endoproteinase Asp-N. Chymotrypsin is sometimes used, although it does not have a well defined specificity. Proteinases of broad speficity may generate too many peptides to separate, and the peptides may be too short to give useful data. Of the chemical cleavages, cyanogen bromide is the most common. All the chemical digestions are less efficient that a good enzymatic digest. However they do produce only a few peptides, which can ease the purification problem.

Sequencing a few large peptides may be more efficient than sequencing several smaller peptides. With a long peptide, there is a better chance of having a long sequence stretch for primer design, and there are less gaps between peptides. Also the actual sequencing can be faster because of the elimination of the startup steps of the sequencer operation.

Often standard analytical reagents are used for the digestions described above. If there is concern that the proteinase being used is contaminated with proteinases of different specificity, sequencing grade proteinases are available from Boehringer-Mannheim and Promega. It has been reported that some proteinases are inactive; one group prefers sequencing grade trypsin and endoproteinase Asp-N from Boehringer, endoproteinase Lys-C from Wako Chemicals, and chymotrypsin and subtilisin from virtually any supplier (Tempst et al., 1990). These enzymes may be stored frozen in 0.1 M ammonium bicarbonate. We have not noted inactive proteinases.

Digestion conditions

Digestions with some proteinases may be performed in 20% organic solvent, effect on digestion which appears to improve the specificity of thermolysin (Welinder, 1988). Fernandez et al. (1992) find that 10% acetonitrile with or without 1% reduced Triton X-100, use in digestions allows tryptic digestion of proteins. This group also showed that carboxymethylation of a protein allows more complete .

Overnight digestions are common, to ensure that a protein is fully digested. However, such times may be longer than necessary, although only if there is sufficient protein for multiple conditions can the necessary digestion time be investigated. When subtilisin was denatured, with a trypsin:substrate ratio of 1:100, 10 minutes was sufficient for digestion, and longer digestions gave a more complex peptide pattern, possibly due to the formation and then the action of pseudotrypsin (Christianson and Paech, 1994).

Use of controls during digestion of a protein

When performing an enzymatic digestion, it is very important to perform a control digestion containing no substrate; also include a positive control, like cytochrome c. We have more than once sequenced fragments of the proteinases used to perform a digestion. Using proteinases of known sequence -trypsin, endoproteinase Glu-C, chymotrypsin, thermolysin- makes finding autolytic fragments easier.

It is also vital to ensure that the protein of interest is digested. We have seen incubations of proteins with proteinases in which no significant digestion occurs. After performing a digestion, some of the digested protein should be run on a gel to see if there is intact protein left, or if the digestion products were separated by chromatography, some of the intact protein should be run to see that it is not present in the digestion mixture. Running a control digestion of the proteinase alone is not sufficient to be sure that there are no products from the proteinase.

To digest proteinase resistant proteins, Riviere et al. (1991) suggest either heating for 30 min at 50oC in 6 M guanidine-HCl, which can be diluted to 2 M; chymotrypsin, Achromobacter protease or subtilisin can be used for digestion but trypsin works poorly. Another procedure is heating at 37°C for 30 min in 8 M urea followed by digestion with Achromobacter protease or subtilisin.

Urea can form cyanates which will react with amino groups. Boehringer Mannheim recommend 20 mM methylamine to block this reaction. 0.1M ammonium bicarbonate also is adequate to prevent N-terminal modification (Stone and Williams, 1993: Fernandez et al., 1992) Proteolysis may be performed in detergents, although may cause problems because it must be removed prior to HPLC separation and cannot be tolerated by all proteinases (Stone et al., 1989c). Proteolysis in acetonitrile also worked well.

When preparing proteins for digestion, excess detergent should be removed to prevent inhibition of enzymatic digestion and interference with reverse phase separation of peptides (Shannon, unpublished observations, Stone et al., 1989b). However, low levels of Tween 20 have been suggested to reduce losses of peptide during digestions (Simpson et al., 1989a,b). This grouphas abandoned use of Tween 20 for digestions, possibly because it interferes with with mass spectrometry; they have not said anything about sample losses (Reid et al., 1995).

A complication to proteolytic digestions is the formation of new peptide bonds by the proteinase under sub-optimal pH. transpeptidation One study found 10% formation of a new peptide bond with trypsin, and 1% levels of products using endoproteinase Glu-C (Canova-Davis et al., 1991).

To separate peptides produced by a digestion of a protein, the most common method is reverse phase HPLC. When performing such separations, it is wise to chromatograph peptides in two solvent systems to reduce the chances of having two or more peptides in one peak. When collecting peptides, Aebersold (1989) recommends immediate freezing to avoid reduced yields after serine residues.

Moritz et al. (1996) state that broad peaks in the chromatogram of peptides from a digest are disulfide linked peptides, which can be avoided by reducing and alkylating the protein before digestion, which also gave more peptides, in greater quantity.

For digests of less than 50 pmole, Kenny et al. (1994) suggest that electrophoresis on 150 µm diameter capillaries gives better recovery than chromatography on 2.1 mm diameter columns. However use of electrophoresis for separation of peptides from digests has not been widely used.

Two dimensional electrophoresis can separate peptides also; its main use is separation of fragments which are not eluted from reverse phase columns (Nokihara et al., 1994).

One scheme to reduce sample handling when obtaining proteolytic fragments of a protein is to load the protein on a reverse phase column to concentrate and desalt it, and possibly alkylate it. Proteinase introduced into the column digests the protein and the fragments can then be eluted and separated. This procedure has been described for the Hewlett-Packard sequencer with its sample cartridge. 20% acetonitrile is essential to obtain digestion, causing hydrophilic peptides to be washed off the support with starting buffer. Some bonds are not cleaved with the protein on the column although they are cleaved in solution (Burkhardt, 1993). A similar technique is used by the laboratory of Don Hunt to generate peptides for mass spectrometry.

An example of successfully obtaining data from a protein is work done by Joel Hockensmith and Larry Mesner in the Department of Biochemistry. Intact ATPase, molecular weight 90,000 was blotted on PVDF but neither the intact protein nor proteolysis products gave sequence data. 500 pmole of protein was isolated, digested with CNBr and peptides separated on Tricine gels and blotted on to PVDF. Approximately 20 pmole of peptide was sequenced off PVDF. Two samples contained more than one sequence and one had only one peptide which was sequenced for over 25 amino acids.

Jones et al. (1994) optimized digestion of one protein by altering temperature and urea concentration, thereby increasing yields of peptides and reducing non-specific digestion. Another digestion with endoproteinase Glu-C gave either incomplete digestion, or artifacts of non-specific cleavage and transpeptidation (Hara et al. 1996).

Michrom BioResources sell small columns which trap protein while reduction, alkylation and digestion are performed (Baldwin et al., 1995). The procedure is confusing as written, and recoveries are not given. Using a conventional C18 cartridge for alkylation of lysozyme followed by elution gave disappointing yields in my hands.

Immobilized proteinasescan be used for protein digestion. Davis et al. (1995) have described the production and use of capillary columns of trypsin for preparation of samples for mass spectrometry. Raising the temperature of digestion to 37°C increases autolysis, as does the presence of acetonitrile. Ronnenberg et al. (1994) immobilized proteinases on soft gels with diisocyanate chemistry, which has less bleeding than CNBr; the sensitivity of the technique was not demonstrated, but an absence of autolysis products was claimed.

Digestion of proteins separated by electrophoresis

After isolating a protein by electrophoresis, there are several options for digesting the protein to obtain internal sequence data. The gel may be dissolved, as described above; this option seems rarely used. Electroelution or passive elution can remove the protein from the gel; these techniques can work, and have been used successfully here. Digestion in the gel or after transferring to PVDF are commonly used, and in the hands of experienced laboratories, are equally efficient (Erdjument-Bromage et al., 1995); transfer to nitrocellulose was less efficient.

For both in gel digestion and on PVDF digestions, electrophoresis was performed using high purity reagents (Bio-Rad), and the gels, 1.5 mm thick, aged for 24-48 hours prior to use. Fresh buffers for the gel are suggested. Samples were heated at 37°C for 15 minutes in sample buffer. A control piece of gel or PVDF is advised to identify artifacts from reagents. Using chromatographically repurified Coomassie blue (Sigma or Aldrich) reduces the number of artifact peaks seen during the peptide separation; dye from other suppliers may have a higher dye content but is not recommended. Coomassie Blue R-250 is about 50% pure, with about 45% of the impurities being less polar compounds, and the remainder tri and tetra-sulfonated molecules, blue-green colour, that inhibit binding by Coomassie Blue (Kundu et al., 1996; the authors describe a purification scheme.

Filtering through Nalgene disposable filterware may extract something which inhibits the enzyme by 50%. If trypsin is used, which is common, modified trypsin from Promega or Boehringer will give less autolysis products.

In all cases, a high protein concentration is necessary for successful analysis, which means 2 µg or more of the protein of interest per lane (0.05 µg/mm3). Williams and Stone (1995) found that the concentration of protein in the gel is more important than the total amount, mentioning one protein which gave data when redigested at a lower amount but higher concentration than an unsuccessful first digestion.

In gel digestions give greater yields of peptides than digestion on PVDF, especially of large or glycosylated peptides (Merewether et al., 1995). In contrast to digestions in solution, methionine appeared to be oxidised. On the other hand, Mørtz et al. (1994) found that in gel digestions gave large unidentified peaks and were not reproducible and digestions on PVDF gave slightly greater yields than nitrocellulose.

Moritz et al. (1996) state that broad peptide peaks in a digest are disulfide linked peptides, which can be avoided by reducing and alkylating the protein before digestion.

in gel digestion

The bands were cut in small pieces and 1 ml of 50% methanol added and the supernatant discarded after 20 minutes at room temperature; the destaining was repeated. The gel pieces were dried for 2 minutes in a Speed-Vac, then add 0.5 µg of trypsin in 200 µl of 0.1M NH4CO3 0.1% Tween 20 and incubate for 20 hours at 30°C. Extract peptides twice with 0.1% TFA, 50% acetonitrile (volume not specified) for 30 minutes at 4°C and concentrate supernatants in Speed-Vac to < 100 µl. Lysyl endopeptidase has been used for this procedure.

The overall yield of this process has been estimated at 12% (Williams and Stone, 1996).

digestion on PVDF

The procedure here is from Fernandez et al. (1994). Blotting uses 10 mM Tris 100 mM glycine pH 8, 10% methanol to high retention membranes, namely ProBlott, Trans-Blot or Immobilon Psq. Transfer in a tank rather than semi-dry apparatus is believed to give higher yields. After cutting the PVDF in small pieces, add 50 µl of 1% reduced Triton X-100, 10% acetonitrile, 0.1M Tris pH 8. After 30 minutes at room temperature, add 0.2 µg of trypsin and incubate 24 hours at 37°C, then vortex, sonicate for 5 minutes, spin and remove supernatant. Add another 50 µl of above buffer and repeat, then use 100 µl of 0.1% TFA and pool supernatants. The suggested enzyme:substrate ratio is 1:10 but 5-fold variations work.

If MALDI-TOF data is to be obtained, octyl or decyl glucopyranoside or decyl or dodecyl maltopyranoside should be used to eliminate detergent clusters during mass spectrometry (Kirchner et al., 1996); 1% octylglucopyranoside in 25mM ammonium bicarbonate 10% acetonitrile allows MALDI-TOF with saturated cyanohydroxycinnamic acid in 50% acetonitrile 0.1% TFA ((Gharahdaghi et al., 1996).

Sutton et al. destain Coomassie stained PVDF with 70% acetonitrile and digest in 4 µl of 25 mM ammonium bicarbonate, 1% octyl glucoside, 10% methanol, 20 µg/ml modified trypsin; extract peptides with 10 µl of formic acid:ethanol =1:1 for MALDI-TOF. Henzel et al. (1994) find that Coomassie does not interfere with alkylation, but does interfere with mass spectrometry or capillary LC. They remove Coomassie with the following, mentioning that direct extraction with methanol can cause significant protein loss: wet with 1-2 µl methanol, add 100 µl water, mix, add 400 µl methanol, mix; add 100 µl chloroform, mix, then remove solvents. To remove peptides> 3,000, incubate blot with 20 µl DMSO and shake for 30 minutes.

Iwamatsu and Yoshida-Kubomura (1996) exploit the retention of hydrophobic digestion products on PVDF by performing sequential digestions on electroblotted proteins. Their suggested order of digestion is lysyl peptidase, endoproteinase Asp-N, trypsin. The reasoning is that hydrophobic peptides not released by the first digestion may be digested and released by a later digestion.

Lui et al. (1996) claim that only 50% of peptides are recovered from a endoproteinase Lys-C digest, and 70% from a tryptic digest on membranes (PVDF or nitrocellulose not specified), better recoveries using Zwittergent 3-16 when digesting on either nitrocellulose or PVDF; recoveries were greater off nitrocellulose than PVDF. Thus this grouup recommends use of Zwittergent 3-16 when digesting proteins on nitrocellulose or PVDF membranes.

Digestion of proteins in gels

This process has been long used for peptide mapping (Cleveland et al., 1977).

To digest proteins in a gel with cyanogen bromide, the gel slices were dried, then treated with 0.5 ml of cyanogen bromide in 70% formic acid (protein:CNBr 1:20 or 100 by weight), then the slices were dried again (Jahnen et al., 1990). The resultant peptides could be separated by extraction with 1% trifluoroacetic acid followed by reverse phase separation (Jahnen et al., 1990). To reduce peaks from Coomassie Blue staining, staining was only for 15 minutes or less, and then Coomassie was removed by treating the gel pieces with 2 1 ml aliquots of 3% SDS 50% n-propanol for 3 hours or until the gel was clear followed by 24 hours of washing with water. Extraction of Coomassie Blue increased the yield of peptides and eliminated the Coomassie Blue peaks on chromatography (Ward et al., 1991). The extraction may also remove which can interfere with reverse phase separations of peptides (Bosserhoff et al., 1989).

Limited acid hydrolysis of proteins in gels, after removal of Ponceau S or Coomassie blue stain, appears successful in generating internal sequence data and avoids problems with enzymatic reactions (Vanfleteren et al., 1992).

Stone and Williams destain proteins by adding 500 µl of cold 95% acetone at 4°C for 30 min, and washing twice with 0.1M ammonium bicarbonate, presumably to remove SDS.

A procedure claimed to be more efficient than earlier procedures specifies destaining the gel, incubating in 0.2M ammonium bicarbonate, 50% acetonitrile, and then, unlike other procedures, completely drying the gel, which is then rehydrated with ammonium bicarbonate, 0.02% Tween 20 and trypsin (Hellman et al., 1995). The acetonitrile removes some Coomassie. Avoiding the use of Coomassie may be more efficient, but most people are used to using it, and the procedure works adequately when Coomassie stains is used. The proteins are alkyated before digestion; incubate in sample buffer containing 10 mM DTT, then add 20 mM iodoacetate (N-isopropyliodoacetamide not tested)(Hellman, ABRF posting, March 1997). The author (personal communication) suggests that by completely drying the gel, proteinase is better drawn into the gel to digest all of the substrate. Reportedly some samples that did not give data with other in gel digestions gave peptides with this protocol. The rest of the procedure is common to other methods This procedure can be used on dried gels. Hellman (e-mail, 1997) says that the Tween has been omitted from this procedure successfully, which was used by many for the 1997 ABRF in gel digestion survey. Moritz et al. (1994) also omit Tween in their later procedure for in gel digestions.

A variation is to digest the protein and eluate with SDS and endoproteinase lys C, remove the SDS and then peform a complete digestion (Hwang et al., 1996). A 60% yield is claimed, with success from 10 pmole protein; the authors report the Rosenfeld's procedure does not always work, and is sensitive to the amount of proteinase.

Another variation is a stress on the importance of reducing and alkylating proteins before digestion to prevent disulfide bonds forming (Jenö et al., 1995), a point made by Moritz et al. (1996), whose procedure for alkylation in a gel is described in appendix A; this report also shows more efficient digestion after alkylating the heavily cross-linked BSA. The authors suggest that reduction and alkylation prior to electrophoresis may not work efficiently in mixtures but do not offer direct evidence. Instead they digest in the gel prior to electrophoresis and find for efficient production of peptides it is necessary to add SDS which is removed by precipitation with guanidinium chloride.

One group report more autolysis during in gel digestions than during digestions on PVDF, even using autolysis trypsin (Rasmussen et al., 1994).

Digestion of electroblotted proteins

Digestion of proteins on PVDF is common. However the digestion may give different results from digestions in solution because of failure to extract hydrophobic proteins with the common aqueous digestion and extraction systems or failure to cleave in hydrophobic regions of the protein (Rasmussen et al., 1991). However digestion on PVDF gives less autolysis products than in gel digestions, even with modified trypsin (Rasmussen et al., 1994). Henzel et al. (1994) find that alkylation of proteins on PVDF is unaffected by Coomassie, but Coomassie must be removed prior to capillary LC and electrospray mass spectrometry. Methanol extraction removes proteins, but the chloroform/methanol extraction of Wessel and Flügge (1984) is suitable. After reduction and alkylatin, the blot is incubated with PVP-360 and trypsin; hydrophobic peptides are extracted with DMSO.

Applied Biosystems (Yuen et al. 1988, 1989) has described procedures for performing amino acid analyses, alkylation, and digestions of proteins that have been electroblotted on to PVDF. We prefer not to perform alkylations in the sequencer as described by Yuen et al. (1989) because we have found that the sequencer is contaminated by the reagents, so that extensive washing is needed before the sequencer can be used again. A method for with 4-vinylpyridine prior to electrophoresis described by Tempst et al. (1990) is to reduce the protein in Laemmli sample buffer, with 2-mercaptoethanol reduced to 0.5%, heated to 60°C for 10 min then 37°C for 20 min; then a 20% solution of 4-vinylpyridine in ethanol is added to give a final concentration of 1.5%. After 30 min at room temperature in the dark, the sample is loaded on to a gel. Drs. P. Glee and K. Hazen in the Department of Pathology have successfully used this procedure enabling identification of a cysteine during sequencing. Without alkylation, digestion, incomplete of proteins may be less complete. Cysteine can be alkylated after electrophoresis on PVDF (Ploug et al., 1992, Iwamatsu, 1992) prior to digestion of the protein.

Tempst et al. (1990) found that endoproteinase Glu-C was not successful for in situ digestion of proteins. In agreement with the report of Bauw et al. (1990), on nitrocellulose, not all possible proteolytic cleavages of substrates occurred. This group advises keeping the reaction volume below 25 µl. Up to 25 mm2 of nitrocellulose can be handled, with enzyme concentrations of 0.04 µg/µl or more. The membrane should never be allowed to dry.

Another method of obtaining sequence data from a cyanogen bromide digestion is to electroblot the protein on to PVDF and digest with 150 µl of 0.15 M CNBr in 70% formic acid, then elute the fragments with 75 µl of 2% SDS, 1% Triton X-100, 50 mM Tris pH 9.5 for 90 min at room temperature . The fragments are then separated by electrophoresis, blotted and sequenced. The yield of sequenceable fragments was 2-5% of the material loaded on the gel. (Scott et al. 1988). It has been reported that CNBr digestion on PVDF leaves about 19% of protein, an amount increased to 73% by reaction with PITC (Wadsworth et al., 1992).

BNPS-skatole can digest proteins at on PVDF; the fragments can be separated by running a second gel, or presumably by chromatography (Crimmins et al., 1990).

A modified procedure for digesting protein blotted on to nitrocellulose has been described by Tempst et al. (1990). In studies on digestion of proteins on Fischer et al. (1991) used TES buffer pH 8 with 5% acetonitrile, 1 µg of enzyme/100 pmole of substrate (less enzyme gives less digestion, more gives autolysis products) for 8 hours after which no further digestion occurred. Yields of sequenceable peptides were 10-30%; again, not all possible peptides were seen. Proteinases that were used successfully were endoproteinase Glu-C, endoproteinase Asp-N, endoproteinase Lys-C and trypsin. Luo et al. (1991) have also described digestion of proteins blotted on to nitrocellulose.

To digest blotted proteins, Wong et al. (1992) applied 5 µl of methanol to PVDF, then added cyanogen bromide in 0.1 M HCl and incubated for 3 h. at 45°C. Before extracting electroblotted and digested proteins, Wong et al. (1992) used a modified extraction (Wessel and Flügge, 1984) to remove . The methanol wetted blots were placed in 100 µl of water, 400 µl of methanol added, the tube was vortexed for 2 min and 100 µl of chloroform added. After another 2 min, the liquid was removed prior to extraction with DMSO. Methanol or acetone have also been used to remove Coomassie from PVDF. Digestion was enhanced by alkylation of cysteine with iodoacetate (Henzel et al., 1994). To digest with endoproteinase Lys-C, this groups dried off all but 20-50 µl of DMSO, added 0.1 M ammonium bicarbonate to give 10% DMSO, then digested with enzyme at a ratio of 1:20 substrate for 17 h at 37°C. Complete removal of DMSO causes difficulty in dissolving the peptides . DMSO did not alter the pattern of digestion, while a mixture of SDS, proteolysis and DMSO enhanced proteolysis. Similarly Fischer (1993) increased yields of hydrophobic peptides by digesting with trypsin in the presence of 50% DMSO. Another procedure to help extract peptides is the use of 10% trifluoroacetic acid (Rasmussen et al., 1994).

Millipore have developed Immobilon-CD membrane for digestion of electroblotted proteins. The surface is cationic and peptides can be eluted readily after digestion. Common dyes cannot be used, but Promega's ChromaPhor is suitable. Millipore now supply a stain for the membrane. Immobilon-CD appears to bind proteins as efficiently as unmodified PVDF. Patterson et al. (1992) eluted over 70% of bound proteins with 4 M guanidine-HCl and 0.1% Triton X-100, about 70% of CNBr peptides with 70% formic acid and 50% of peptic fragments using 2 M acetic acid. Enzymatic digestions can be performed on the membrane or after eluting the proteins. However it appears that digestion requires an enzyme active at either low pH or in guanidine, precluding the use of trypsin.

Cyanogen bromide may give a different digestion pattern when the substrate is on PVDF rather than in solution (Zhang et al., 1994).

Elution of electroblotted proteins from PVDF

Yuen et al. (1988, 1989) also reported that 25% of the protein applied to the gel can be extracted from the membrane. The most efficient solvent for lactoglobulin was 70% 2-propanol, 5% TFA. Szewczyk and Summers (1988) report higher recoveries of protein when mixtures of Triton X-100 and SDS are used to extract proteins from the membranes. Simpson et al. (1989b) reported recoveries of 57-78% using 2% SDS, 1% Triton X-100, 0.1% DTT in 50 mM Tris-Cl pH 9. Detergent can be removed by a precipitation technique described above (p.2, Stone et al. (1989)) or extraction with heptane: isoamyl alcohol 4:1 (Bosserhoff et al., 1989). Detergent can also be removed by "inverse gradient" reverse phase chromatography as described by Simpson et al. (1987, 1989a,b), the protein is digested and the digestion products separated by microbore HPLC (Simpson et al., 1989a,b). SDS may also be removed by precipitation with guanidine-HCl, although recoveries of peptides are variable, especially for large or hydrophobic peptides (Riviere et al., 1991)

Montelaro (1987) has reported extracting 75% of the protein A bound to Immobilon-P membranes with 40% acetonitrile in 0.1 M ammonium acetate pH 8.9 at 37°C for 3 hr. An application of removal of protein from PVDF was reported by Stone et al. (1989a) who applied a protein to PVDF in 70% formic acid and then performed a cyanogen bromide digestion by adding 10 µl of 70 mg/ml cyanogen bromide and incubating in the dark for 24 hours. To obtain the peptides, the supernatant was removed, then the PVDF was incubated in 200 µl of 40% acetonitrile for 3 h at 37°C. This supernatant was removed and the PVDF was incubated with 200 µl of 40% acetonitrile, 0.05% trifluoroacetic acid for 20 min at 50oC. The three supernatants were combined with 60 µl of water and dried. Recoveries in this procedure are reported to be 15-100% and protein dependent (Stone et al., 1989b). Another procedure uses 20% acetonitrile in the digestion mixture to enhance digestion of the protein and elution of fragments (Choli et al., 1989).

Although the procedure of Szewczyk and Summers (1988) gives higher recoveries, losses during detergent removal make this latter process less efficient.

Wong et al. (1992) have extracted proteins from PVDF with 2 consecutive 2 hour extractions with 200 µl of DMSO with shaking. This procedure is more efficient in extracting proteins from Immobilon-P than from ProBlott, with efficiencies of 32% of BSA and 70% of carbonic anhydrase extracted from ProBlott.

Another technique is to digest a protein in the presence of SDS and separate the peptides by reverse phase HPLC, using a DEAE pre-column to remove the SDS . This technique has been successfully used by Kawasaki et al (1990). Using a DEAE column ahead of the reverse phase column can disturb the baseline early in the gradient (Shannon, 1992, unpublished observations). Columns designed specifically for the removal of SDS are now available.

SDS can be quantitated by its interaction with methylene blue (Hayashi, 1975).

As described above, proteins can be electroblotted to nitrocellulose, eluted, digested and sequenced, or digested on nitrocellulose and the peptides eluted. Aebersold (1989) describes a procedure for such digestions.

One method of generating and isolating fragments is to use Promega's Probe-Design Peptide Separation System. The protein of interest is separated on a gel, transferred to an Immobilon PVDF membrane, digested with cyanogen bromide and the peptides are resolved on a gel system designed for peptides, which are then transferred to ProBlott PVDF membrane for sequencing. The necessary materials are supplied by the maker.

A different method for eluting proteins is to displace them with another protein. Chertov et al. (1992) eluted a hormone with a solution of 0.1 M ammonium bicarbonate pH 8, 25% acetonitrile, 0.5 mg/ml bovine serum albumin. This technique gives another separation problem if pure protein is needed.

While digesting proteins on PVDF, Zhang et al. (1994) observed that 30% acetonitrile with 2.5% TFA was more efficient than 60% acetonitrile-2.5% TFA in extracting hydrophobic peptides from PVDF and that 60% acetonitrile was more efficient for hydrophobic peptides; both solutions also contained 30 mM 4-hydroxycinnamic acid for mass spectrometry and peptide extraction. This group also observed that vacuum drying of a CNBr digestion on PVDF prevented extraction of peptides, but after drying in an open vial, peptides could be extracted with the solution mentioned above.

Another study showed that for the short peptide angiotensinogen, 2% formic or acetic acid removed 70% of bound peptide (Krishnamurty et al., 1994).

After digestion, Mørtz et al. (1994) found the best procedure for eluting peptides from PVDF was 20 µl of 50% TFA for 10', and 2 x 50 µl of 0.1% TFA for 15'. Iwamatsu and Yoshida-Kubomura (1996) mention that more peptides are extracted with 20% acetonitrile than 10%.

Cyanogen Bromide

Cyanogen bromide cleaves on the carboxy side of methionine residues. Because there are usually only two methionine residues per 100 residues, several peptides, often large, are generated. Peptides produced by cyanogen bromide digestion have homoserine at their carboxy terminal, unless the peptide comes from the carboxy terminal of the intact protein. Thus the absence of homoserine from a peptide suggests that it is found at the C-terminal of a protein. Homoserine can be readily identified in an amino acid analysis, but is hard to distinguish from serine when PTH amino acids from the sequencer are analyzed. A cyanogen bromide digestion is simpler to perform than an enzymatic digestion. Performing the reaction in 70% formic acid usually avoids the protein solubility problems that often occur with reactions that are performed in aqueous buffers. The reaction is carried out according to the procedure of Gross (1967). Although a ratio of cyanogen bromide to methionine is often quoted, achieving a precise ratio is difficult because the amount of methionine in 1 mg or less of protein requires an amount of cyanogen bromide that cannot be weighed easily. We have found that the excess of cyanogen bromide produced by adding the smallest amount of cyanogen bromide that can be taken from a large piece does not produce harm. Strydom et al. (1985) also used a large excess of cyanogen bromide. Other examples are Asano et al. (1986). Handle the cyanogen bromide in the hood; if you wish to weigh it, put the cyanogen bromide in a preweighed, sealable container for weighing. After digestion has taken place, the formic acid is removed on a Speed-Vac, equipped with an acid trap to reduce the amount of formic acid reaching the pump and to reduce the amount of cyanogen bromide which escapes. Dilute the sample 10 x with water first, and repeat the drying to reverse possible esterification of serine and threonine residues with formic acid; this process is repeated. Some references are: Reiman et al. (1984), Asano et al. (1986), Pastuszyn et al. (1987).

Tarr and Crabb (1983) report that exposure to formic acid appears to esterify serine and threonine residues and that aminoethanol reverses the esterification; however standard procedures do not use aminoethanol. Goodlett et al. (1990) confirmed formylation of serine and threonine by mass spectrometry and claimed that the formylation did not affect Edman sequencing but did not offer quantitative evidence. There are reports that Met-Cys, Met-Thr and Met-Ser bonds are not efficiently cleaved (Doyen and Lapresle, 1979; Schroeder et al., 1969). With a Met-Cys sequence, methionine is converted to homoserine (loss of 30 mass units) with variable cleavage of the peptide bond. The latter authors achieved more efficient cleavage of Met-Thr and Met-Ser bonds by using 70% trifluoroacetic acid although the reaction proceeded more slowly than in formic acid. These authors also find that 4 hours at 50°C caused random cleavage in catalase. Shively et al. (1982) state that use of trifluoroacetic acid for cyanogen bromide digestions avoids blocking of terminal amino groups and Goodlett et al. (1990) showed that use of trifluoroacetic acid instead of formic acid reduced esterification of serine and threonine. Guandine hydrochloride may also be used and may give greater yields of peptides, and avoids possible reduction of disulfide bonds by formic acid, which was not directly shown (Villa et al., 1989). Although pyridine formate buffers can block the N-terminus of peptides (Shively et al., 1982; Levy et al., 1981), formic acid is still commonly used for cyanogen bromide digestions.

A study of reaction conditions found that 1 mg/ml CNBr in formic acid for 12-18 hours, gave formylation, oxidation of tryptophan; acetic acid and 0.1M hydrochloric acid gave more acid hydrolysis of peptide bonds; 70% trifluoroacetic acid with 10 mg/ml CNBr for 6 hours gave little acid hydrolysis, no adducts and reduced oxidation of tryptophan (Andrews et al., 1992). When digesting ovalbumin, the cleanest appearing peaks on chromatography were from cyanogen bromide in 70% TFA; 70% formic acid was next best, and 6M guanidine with 3M HCl gave the smallest and broadest peaks (Shannon, 1996, unpublished observations); the yield of peptide was about 25%.

Methionine sulfoxide and sulfone are unaffected by cyanogen bromide. The cyanogen bromide digestion converts tryptophan to monohydroxy tryptophan (+16 mass units).

An alternate procedure is to use cyanogen bromide and potassium iodide to cleave at both methionine and tryptophan residues (Huang and Huang, 1994). Yields of 80-100% are claimed. Another procedure is to use a mixture of formic and heptafluorbutyric acids, with reported yields of 80% (Ozols et al., 1977).

One reason for incomplete cleavage of proteins by cyanogen bromide is the formation of an unstable intermediate, which will give rise to some of the expected products (Zhang et al., 1996).

Trypsin and Endoproteinase Lys-C

Trypsin cleaves on the carboxy side of lysine and arginine residues, although cleavage at Arg-Pro and Lys-Pro bonds is slow. The bond RXS(P) is not cleaved (Cohen et al., 1991). The specificity of trypsin, its ability to work under denaturing conditions, and its low cost make tryptic digestions a favored digestion. Before performing a tryptic digest, it may be worth seeing how many arginine and lysine residues are present, and whether it might be better to produce fewer, larger fragments by digesting at lysine only with endoproteinase lys-C, or arginine only by citraconylation of lysine residues before digestion.

Trypsin is reported to be active in 6 M urea and 1 mg/ml SDS; our experience suggests that it is not active in 3 M guanidine hydrochloride. Aebersold (1989) reports that trypsin is more active in bicarbonate buffer than Tris-Cl and that trypsin is not inhibited by up to 40% acetonitrile. Suzuki and Terada (1988) reported that 0.05% SDS in 0.2M ammonium acetate pH 8 stimulates hydrolysis of BSA, but this concentration of SDS inhibits activity against a . No other proteins were tested to determine the specificity of the effect of SDS. Stone and Williams (1993) find that trypsin will digest suspensions of proteins unlike endoproteinase lys-C.

Prolonged incubation of trypsin appears to give rise to pseudo-trypsin (C-trypsin) which can hydrolyze bonds adjacent to aromatic residues (Keil-Dlouhá et al., 1971). About one third of trypsin can autolyse to this form whose activity towards chymotrypsin substrates is low but apparently sufficient to give chymotrypsin like digestion (Smith and Shaw, 1969). Promega Corporation and Boehringer Mannheim sell trypsin which is claimed to be resistant to autolytic digestion because it has been alkylated. Methylation of trypsin makes the enzyme more resistant to breakdown and stabilizes the activity, but does not prevent all autolysis (Rice et al., 1977). Although calcium also stabilizes trypsin, reductive methylation further stabilizes it.

Cleavage with endoproteinase lys-C, which cleaves on the carboxy side of lysine (Jekel et al., 1983), may be performed at pH 8 in 0.1 M NH4CO3 and 5 M urea (Reiman et al. (1984). Steffens et al. (1982) used the same buffer but with 0.1% sodium dodecyl sulfate, proteinase:substrate 1:100 by weight, for 2 hours at 37°C, then added another batch of proteinase and repeated the incubation.

Kawasaki and Suzuki (1990) report that some bonds are cleaved more slowly in the presence of SDS but digestion still occurs even at 1% SDS. Later these workers reported that the digestion pattern in 0.05% SDS is the same as without SDS, but some bonds are not cleaved at higher concentrations, unless more proteinase is used; digestion will occur in up to 2% SDS (Kawasaki and Suzuki, 1992).

Achromobacter protease I is also reported to cleave only on the carboxy terminal side of lysine. Brenner et al. (1990) report that Achromobacter protease and endoproteinase Lys-C have different activity at some lysine residues, that endoproteinase Lys-C produces less nonspecific cleavages but requires more enzyme and longer digestion times. With both enzymes, most non-specific cleavages are on the C-terminal side of arginine

Submaxillaris protease (Schenkein et al., 1977) and clostripain or endoproteinase Arg-C are claimed to digest specifically at arginine. However, with a preparation of submaxillaris protease from Pierce, we found that digestion was not specific, and we have had no other experience with these enzymes. One report showed useful digetss by this enzyme, but not at arginine residues (Proudfoot et al., 1995). This enzyme has been used by Simpson et al. (1987). Boehringer Mannheim suggest denaturing the substrate in 5M urea but conducting the digestion in 0.1M urea at pH 8, or using 1 mg/ml SDS. Our mass spectrometry lab has also found this enzyme ineffective (M. Kinter, personal communication, 1997), as did Iwamatsu and Yoshida-Kubomura (1996).

Instead we suggest modification of lysine residues with citraconic anhydride before digestion with trypsin (see Reiman et al., (1984); Lau et al., (1985); Atassi and Habeeb, (1972)).

Annan and Biemann (1993) peracetylate protein prior to digestion an