Electrophoresis 1994, IS, 1535-1540 Focusing in extremely alkaline pH gradients 1535

Alessandra B o d Pier Giorgio Righetti' Giuseppe Vecchio' Steen Severinsen3

'Chair of Biochemistry, Faculty of Science, University of Calabria, Arcavata di Rende (Cosenza) 'Istituto di Chimica degli Ormoni, CNR, Milano 3 N ~ ~ ~ Nordisk, Bagsvaerd

Focusing of alkaline proteases (subtilisins) in pH 10-12 immobilized gradients

Isoelectric focusing in very alkaline immobilized pH gradients (IPG) was adopted for checking the purity and assessing the p l value of two strongly alka- line proteases: Savinase and Durazym. The first enzyme (known to be the most alkaline) contains 5 Asp, 5 Glu, 7 His, 7 Tyr, 5 Lys, no Cys and 8 Arg resi- dues and should have a theoretical p l of 9.7. Yet, when focused in a pH 9-11 IPG interval, it was lost in the cathodic compartment. After repeated attempts at creating even more alkaline pH intervals, a pH 10-12 IPG range was finally optimized and proved successful in focusing both enzymes midway between the two electrodic compartments. The p lof Savinase was measured as 11.15 * 0.15; that of Durazym as 10.95 & 0.20 and that of the plmarker cytochrome c as 10.6 f 0.17. Both enzymes (and a number of minor components in each preparation) were proven to be active by an in situ zymogram consisting of a caseinlagar overlay. The discrepancy between theoretical and experimental p l values could not be fully reconciled: when correcting for pK values of amino acids in proteins at lO"C, instead of the tabulated values at 25"C, the p l should increase to a value of 10. Differential UV spectra showed that ca. ' I z Tyr are buried in the protein interior and are thus unable to contribute to sur- face charge. This further increases the plvalue by 0.3 pH units to a p l of 10.3, still quite removed from the experimentally assessed plvalue (in the gel, at l 0 T ) of 11.15.

1 Introduction

Immobilized pH gradients (IPG) in alkaline pH intervals (pH 10-11) were made possible by the introduction, in 1987, of a novel acrylamido buffer having a pK of 10.3 (at 10°C; N,N-diethylaminopropylacrylamide) [l]. In such pH ranges, cytochrome c (a protein marker usually added in conventional focusing to monitor the alkaline part of the pH gradient, and routinely lost in the cathodic compartment) was focused midway between the electrodes and was found to have a p l of 10.45. Lyso- zyme gave a major component with p l 10.55 and ribo- nuclease gave a band with p l 10.12. The same alkaline IPG interval was subsequently used for focusing elastase and trypsin from porcine pancreas [2]. Elastase gave a sharp array of three bands, with the following pls (at 10 'C): 10.60 (major component), 10.53 (intermediate species) and 10.45 (minor isoform). The enzyme activity in these bands was ascertained by a zymogram con- sisting of an overlay of a cellulose acetate strip impreg- nated with a fluorogenic substrate. Trypsin was resolved into two almost equally abundant bands with pls of 10.70 and 10.53, respectively. However, this early tech- nique of alkaline IPGs was never quite popular, since results were not well reproducible, due to the fact that CO, was constantly adsorbed by the gel layer and inter- fered with protein focusing. In addition, due to the elec- troosmotic flow, thinning the alkaline gel extreme, the gel had to be reswollen in a gradient of a viscous solu- tion (e.g., sorbitol) able to slow down the liquid trans- port process during focusing [3].

We have now reevaluated focusing under extremely alka- line conditions, due to two favorable new aspects: (i) A

Correspondence: Prof. P. G. Righetti, Via Celoria 2, Milano 20133, Italy

Abbreviation: AAEE, N-acryloylaminoethoxyethanol

special ,,submarine" chamber has been produced, allowing for focusing under a thin veil of light paraffin oil, able to prevent CO, absorption by the open-face gel layer [4]. (ii) A novel, quaternary acrylamido titrant (with pK > 13) was described [5 ] , allowing, in principle, the extension of the pH gradient well above pH 11. We have applied this novel technology to the analysis and p l assessment of two strongly alkaline proteases, Savinase and Durazym, which so far had baffled any attempt at focusing under steady-state conditions.

2 Materials and methods

2.1 Materials

All IPG experiments were performed in the LKB 2117 Multiphor I1 horizontal electrophoresis system together with the LKB 2297 Macrodrive 5 power supply and Mul- titemp I1 thermostat. IPG gel casting was carried out by using the LKB 2117-903 2-D gradient and Immobiline gel kit. Acrylamide, N,M-methylenebisacrylamide (Bis), N,N,A",N-tetramethylethylenediamine (TEMED), Repel- Silane, GelBond PAG film, agarose and Pharmalytes were purchased from Pharmacia-LKB Biotechnology (Uppsala, Sweden). N-Acryloylaminoethoxyethanol (AAEE) was synthesized as described in [6]. The ten- acrylamido-buffer kit (p l select) was purchased from Fluka (Buchs, Switzerland). Guanidine HCl was from Serva (Heidelberg, Germany). Cytochrome c and vita- mine-free casein were from Sigma (St. Louis, MO, USA). Pure preparations of Savinase and Durazym (both of them alkaline proteases) were available from Novo Nor- disk (Bagsvaerd, Denmark). The former enzyme was purified from Bacillus lentus, whereas the latter was a modified enzyme produced by recombinant-DNA tech- niques.

0 VCH Verlagsgesellschaft mbH, 69451 Weinheim, 1994 0173-0835/94/1212-1535 $5.00+.25/0

1536 A. Boss] et a / . Electrophoresis 1994, 15, 1535-1540

2.2 Analytical IPGs

The technique is described in detail in Righetti [5]. Briefly, the following analytical IPG gels were made: in pH 9.0-11.0, pH 9.2-11.2, pH 9.5-11.5 and pH 10.0-12.0 ranges. All these pH ranges were calculated and opti- mized with the computer program of Giaffreda etal. [7] (available from Hoefer Scientific, San Francisco, CA, USA). They were prepared as 5%T, 4%c matrices, in the case of poly(acrylamide), and as 6%T, 4%c in the case of poly(AAEE). All gels had a size of 10 X 11 cm and were 0.5 mm thick. After casting and polymerization, the gels are extensively washed in distilled water, dried and then reconstituted to their original volume by reswelling in a cassette in 10% glycerol (pH ranges of 9-11, 9.2-11.2 and 9.5-11.5) or 30% sorbitol (pH 10-12 range) and 0.6% LKB Ampholine pH 9-11 (as seen from refrac- tive lines and ninhydrin stain, they occupy only the lower half of the gel in the pH 10-12 interval). In addi- tion, all gels were run under light paraffin oil, so as to prevent CO, adsorption (in the commercially available tray). Protein samples (generally 1-2 mg/mL) were usu- ally loaded (in 30-50 pL volume) in plastic troughs at the anodic gel side after a brief profocusing step (1 h) at low voltage gradients (500 V). Analytical IPGs were run at 10°C, 1500 V for a maximum of 4h. Staining was in Coomassie Brilliant Blue R-250 in 10% acetic acid, 30% ethanol in presence of 0.1% copper sulfate [8].

2.3 Enzyme visualization

For an assessment of the enzyme activity of the focused bands, an insitu visualization was applied, using an overlay technique developed by Arvidson and Wadstrom [9]. After focusing in IPGs, the gel was immersed in 0.4 M Tris-HCI buffer (pH 7.6) for 5 min. After this equilibra- tion step, the gel was covered with a second gel (same size, same thickness), pasted to a film of GelBond, con- sisting of 1.5% agarose containing 1% casein, in pres- ence of 1 mM CaCl,, at pH 7.6. The sandwich was then incubated in a humid chamber at 30°C for 20 min. Upon staining both gels in Coomassie Blue, the enzyme activity in the overlay was detected by white bands on a blue background, as opposed to blue zones on a white background in the corresponding protein staining in the IPG gel.

2.4 Differential UV spectrophotometry

To measure Tyr exposure, UV spectra were recorded with a Jasco Uvidec 610 spectrophotometer, equipped with a dedicated software for recalculating the spectra in terms of a second derivative. The instrumental conditions were the same as described by Balestrieri etal. [lo]. For deter- mining the percentage of surface-located Tyr vs. those buried inside the protein, we used the following equa- tion, according to Ragone etal. [ll]:

respectively, as obtained from the second derivative spectra, and r, is the a/b value of an equimolar mixture of aromatic amino acids in ethylene glycol solvent, which presents the same characteristics of the interior of a protein matrix. The calculation of the 2r, value was done according to the following equation [l l] :

(2) r, = (AX + B)/(Cx + 1)

where A, B and C are constants determined in ethylene glycol and x is the molar ratio of Tyr vs. Trp for Savinase based on amino acid analysis data.

3 Results

Some preliminary experiments in IPG gels in the pH 9-11 interval failed to properly focus both Savinase and Durazym. The main enzyme bands were apparently lost in the cathodic compartment and only minor, acidic components were properly focused in the gel, notwith- standing the fact that the theoretical p l of the most alka- line species (Savinase) was calculated to be 9.7. Although no attempts had ever been reported at pre- paring immobilized gradients above pH 11, we could optimize a pH recipe spanning from pH 9.2 to 11.2 (see Fig. 1). This pH interval was reasonably linear and was obtained by mixing 11.5 mM of pK 9.59 and 8.6 mM of pK 4.6 Immobilines in the acidic chamber. The cathodic chamber (pH 11.2 extreme) contained 11.8 mM of pK 10.3 Immobiline. As seen in Fig. 2A, a good focusing pattern was obtained for both enzymes and for cyto- chrome c, used as p l marker. Analogous results were ob- tained by using a poly(AAEE) matrix (Fig. 2B), although the train of bands was focused more closely to the cathode, due to a matrix effect which weakens all Immo- biline buffers [12].

It was nevertheless difficult to assess with precision the p l of the main bands of both Savinase and Durazym, as both enzymes focused close to the cathodic gel extremity. For a better evaluation of the p l values of both proteins, an even higher pH interval was calculated with the aid of the computer program of Giaffreda etal. [7]. Figure 3A shows the pH profile and deviation from linearity of an IPG interval spanning the pH 9.5-11.5 range. This gradient had the following composition: P H ( n ) E ( " )

11.0

10.5

10.0

9.5

050

0 5 0

were a is the degree of Tyr exposure to the solvent, r, and r,. are the values of the ratio alb (see Fin. 6 ) for

9.0

F r a C t i o n Number

Figure I . Profile of a pH 9.2-11.2 IPG. U, pH gradient; -0-, devia- tion from linearity. The gradient was calculated and optimized with

native" and unfolded (in 6 M guanidine chloride)protein, the program of Giaffreda etal. [7]

Electrophoresis 1994, IS, 1535-1540 Facusing in extremely alkaline p H gradients 1537

Figure 2. Focusing of alkaline proteins in the pH 9.2-11 2 IPG of Fig. 1. Samples: (l), (2) Savinase; (3) cytochrome c; (41, (5) Durazym. About 50 pL of sample (1-2 mg/mL) were applied in plastic troughs on the gel surface at the anodic side, after 1 h prefocusing. (A) 5%T, 4%C poly(acry1amide); (B) 6%T, 4OhC poly(AAEE). Running time: 4h at 1500 V and 10°C. '-Submarine" gel, under a layer of light paratfin oil. Staining with Coomassie Brilliant Blue in copper sul- fate. Here and in all subsequent figures the cathode is uppermost.

9.0

0 0 50 100 15 0 2 0 0 5 0 100 150 2 0 0

Fract ion Number Fraction Number

Figure 3. (A) Profile of a pH 9.5-11.5 IPG. -&, pH gradient; -S, deviation from linearity. The average buffering power of the gradient is 6 mequiv. L-' pH-'. The gradient was calculated and optimized with the program of Giaffreda etal. 171. (B) Focusing of fl), (2) Savinase, (3) cytochrome c and (4)-(6) Durazym in the same pH interval (in a poly- acrylamide gel). All other conditions as in Fig. 2.

Figure 4. (A) Profile of a pH 10-12 IPG. -0-, pH gradient; -C-, deviation from linearity. The average buffering power of the gradient is 8 mequiv. L-' pH-'. The gradient was calculated and optimized with the program of Giaffreda etal. [7]. (B) Focusing of (l), (2) Savinase, (3) cytochrome c and (41, (5) Durazym in the same pH interval (in a poly- acrylarnide gel). All other conditions as in Fig. 2.

Elecfrophoresis 1994. IS, 1535-1540 1538 A. Bossi er a!.

acidic chamber: 6.0 mM o f pK 9.59, 6.3 mM of pK 10.3, 6.7 mM of pK > 13 and 16 mM of pK 4.6 Immobilines; basic chamber: 9.2 mM of pK 10.3 and 2.4 mM of pK> 13. While in all recipes only the Immobiline pK 4.6 was adopted as an acidic titrant, any other acidic Immobiline (pK 1, pK 3.1, pK 3.6) could have been used interchange- ably. In the employed gradient, both enzymes were seen to focus at a good distance from the cathodic gel extremity, allowing for a proper p l assessment (see Fig. 3B).

In order to evaluate how far in the alkaline region the IPG technique could be extended, attempts were made at creating an extremely alkaline range, encompassing the pH 10-12 interval. In principle, such a gradient should not even perform well, due to the strong elec- troosmotic flow created by the net positive charge of the matrix, forcing a strong solvent flow directed towards the anodic gel extremity. The best gradient we could generate, with existing Immobiline buffers, is shown in Fig. 4A. The recipe contained 10.8 mM of pK 10.3 Immo- biline with 7.5 mM of p K 4.6 Immobilines at the acidic extremity and just 10 mM of pK > 13 at the cathodic gel end (for computational purposes, a pK value of 13.5 was assumed). Note that, in fact, this gradient has a marked sigmoidal profile. This is due to the deficiency of Immo- biline buffers in this pH region and to the fact that, in the cathodic gel half, the pK > 13 Immobiline acts essen- tially as a titrant, the buffering power being provided (in the pH 11-12 region) mostly by bulk water. By chemical laws, such a gradient could not be quite linear [131. Nevertheless, such a gradient performed well, and allowed us to focus both enzymes in the central part o f the gel, at an adequate distance from both the anodic and cathodic gel extremities (Fig. 4B). It was mandatory, under such extreme alkaline conditions, to reswell the gel in 30% sorbitol, so as to reduce the strong electroos- motic flow and cathodic gel thinning. Two enzymes with the following p l values (main band) were observed: p l 11.15 f 0.15 ( n = 4) for Savinase and p l 10.95 rf: 0.20 (n = 4) for Durazym. It might be asked if, at such pH extremes, both species still retain enzyme activity. We thus applied in situ enzyme visualization by preparing a casein-agarose gel overlay. As shown in Fig. 5A, all bands showed enzyme activity, producing white proteo-

B A

lytic zones against a blue background (Fig. 5B shows the corresponding positive, blue protein staining against a white background).

4 Discussion

4.1 On p l assessments of subtilisins

Subtilisins are alkaline serine proteases produced by a wide variety of Bacillus species. In view o f their indus- trial application for detergent and food processing, subti- lisins have been extensively investigated as promising targets for protein engineering. For example, Takagi et al. [14] produced the mutant subtilisin E, cloned in E. coli, with a higher activity, by optimizing the amino acid adjacent to the catalytic triad (Ser2z,, His,, and Asp,,). Another subtilisin, Aqualysin I, produced by Thermus aqunticus YT-1, has an optimum temperature of 80°C for caseinolytic activity [15]. In view of the fact that the market of subtilisins, as additives for household laundry detergents, commands the largest share of the world- wide market of industrial enzymes (estimated at $200 million in 1991), it is easy to understand the pressure in industry at producing newer mutants with improved per- formance. Savinase also belongs to the class of subtili- sins, and it is a serine endopeptidase with an extended binding cleft comprising at least eight binding subsites [16]. It is a relatively small protease, composed of 269 amino acids. Its main commercial use is also as an addi- tive in washing powders; thus, this enzyme has to with- stand the relatively harsh conditions that occur in laun- dering. Savinase contains the following charged resi- dues: 5 Asp, 5 Glu, 7 His, 7 Tyr, 5 Lys, no Cys and 8 Arg residues; its theoretical p l value is thus 9.7. Our experi- mental p l value therefore seems to be off by more than one pH unit. How can we reconcile such a huge discrep- ancy? First of all, it should be noted that all amino acid pK values, as given in the literature, are measured at 25 "C, whereas all pK values of Immobilines have been assessed at 10 "C, the operating temperature during focusing conditions. As shown by us, the pK values of basic Immobilines can increase by as much as 0.5 pH units (and more for the highest pK values) by lowering the temperature from 25°C to 10°C [17]. Moreover, in

Figure 5. Focusing and enzyme visualiza- tion of alkaline proteases in the pH 10-12 IPG of Fig. 4. (A) Casein-agarose overlay of gel (B), after 4 h of focusing at 1500 V. The contact time was 20 min at 30°C, followed by Coomassie Biue staining. (B) Protein staining. All other conditions as in Fig. 2.

Ekctropkoresis 1994, IS, 1535-1540 Focusing in extremely alkaline pH gradients 1539

220 (4 340

0.25

Q) 0 C 0

0 v) 0.0 P U

+

- 0.25c

220

0

conventional isoelectric focusing, Fredriksson [18] has shown that by lowering the temperature from 25’C to 4°C (the standard focusing temperature) the p l shift of some standdrd alkaline proteins (pl 10.0) is as high as +0.7 pH units. However, in our particular case, the only shift in pKvalue that will affect the p l of Savinase is that of Lys (and to a minor extent of Tyr), i.e. of the only two amino acids which will buffer in proximity to the protein pl. When correcting these two pK values according to Fredriksson [18], the new p l (calculated at 10°C) shifts from 9.7 to 10.0, still quite remote from our value of 11.15. We have thus tried to assess the environment of Tyr residues in the native enzyme structure. This can be obtained by differential UV spectrophotometry [lo, 111. As shown in Fig. 6, it appears that ca. 50°/o of the Tyr residues are buried in the protein core, the other half being exposed to solvent. If we thus subtract from the net surface charge the contribution of Tyr residues, the new p l value reaches 10.3, still quite far from our experimental data.

In recent literature [19, 201, excellent agreement was found among theoretical pls predicted from a protein amino acid sequence and experimental pls as derived from an IPG gel. The agreement between the two sets of data was, in the case of a set of 29 proteins, as good as k 0.01 to 0.02 pH units, quite remarkable, indeed. How- ever, two major points should be highlighted: (i) these data refer to randomized structures denatured in 9.8 M urea; and (ii) the agreement was only proven for acidic proteins, the correlation ending at pH 7.0. Thus, it remains to be seen if such a good correlation actually exists, even at moderately to very alkaline p l values, in view of the many existing difficulties (e.g., uncertainties

Figure 6. Absorption (dotted line) and second derivative (solid line) spectra of Savinase (I mg/mL) solubilized (A) in water and (B) in 6 M guanidine HCl After the derivative operation each spec- trum presents two minima, positioned at 284 and 291 nm, and two maxima at 288 and 295 nm, the peak-to-peak distances a and b are evident and are the arithmetic sum of d2A/dA2. The value of r,, was found to be 0.813 and that of r, was 1.236 (average values of 3 sets of data). The ratio Tyr/Trp is 2.333. The calculated

on the true pK values of alkaline species, especially when incorporated into the gel matrix; lack of precise data on the true incorporation of such alkaline species as the quaternary base, which moreover is not a pure acryl- amide, but a rnethacrylamide derivative). Another factor which might contribute to this p l discrepancy is the pres- ence of 30% sorbitol in the matrix. According to Gel- sema etal. [21-231, the addition of up to 60% sucrose or glycerol to the focusing media produces relatively small errors in p l assessments, of the order of t 0.1 pH units. However, this has been explored only in the interval 3 < p l < 9 and might not apply to values up to pH 12. Addi- tionally, note that a good part of the disagreement between the two plvalues could come from the fact that the protein is focused in the native state, so that fixed pK values, as attributed to free amino acids, might not apply any longer. In fact, while we could probe the Tyr environment, and correct for its contribution to total sur- face charge, we could not assess the state of all other charged residues.

3. (nm) = 0.5.

4.2 On the use of extremely alkaline pH gradients

Nothwithstanding the above shortcoming on p l assess- ments, our pH 10-12 gradients work well, yielding highly reproducible data. There are several reasons for this. First of all, the Altland technique [4] of working with a “submarine” gel is a must. Light parafin oil completely prevents CO, absorption and allows reaching true steady-state conditions. In addition, when running such alkaline gels, the matrix should be prepared and run immediately because, on storage, hydrolysis of amido bonds would continuously change the pH gra- dient in the matrix. A good alternative is, of course, to

1540 A. Bossi et d. Elecfrophoresis 1994, 15, 1535-1540

use the novel poly(AAEE) matrix, which is 500 times more resistant to alkaline hydrolysis. An additional improvement came from the commercial introduction of the pK > 13 Immobiline (MAPTAC, methacrylamidopro- pyltrimethyl ammonium chloride) by Fluka. Originally, a different compound, namely N,N,N-triethylaminoethyla- crylamide (QAE-acrylamide) was offered; however, this compound resulted in erratic incorporation and pro- moted gel detachment from the supporting GelBond PAG film. With the present chemical, reproducible gra- dients and good mechanical stability of gel matrices is routinely obtained. This might sound odd, since a metha- crylamide derivative should have a lower incorporation efficiency than an underivatized acrylamide, due to the presence of a methyl group on the double bond. Yet, by monitoring the focusing position of Savinase and Durazym in progressively more alkaline matrices (from pH 9.2-11.2 up to pH 10-12), we know that these pro- teins focus in identical pH regions, as shown in the rele- vant figures (however, data on absolute incorporation levels are at present lacking). Another advantage of MAPTAC is that the nitrogen carries three methyl sub- stituents, whereas QAE-acrylamide contains three N-ethyl groups. The latter compounds would thus be more hydrophobic and could favor hydrophobic interac- tion with proteins.

In conclusion, we feel that the present development has much to offer in the analysis of extremely alkaline pro- teins. It is well known that histones could never be ana- lyzed, neither by zone electrophoresis in alkaline pH media nor by steady-state focusing conditions in carrier ampholyte-generated pH gradients or IPGs. The standard technique for histone analysis has been zone or disc electrophoresis under acidic conditions (e.g., the cationic pH 4.3 disc system of Reisfeld etal. [24]). In zone electrophoresis, histones have been separated by the acid-urea-detergent gels of Zweidler and Cohen [25], containing 5 % free acetic acid as buffering ion. The present technique now offers a valid alternative for steady-state analysis of histones (work in progress).

Supported in part by grants from the Radius Project in Bio- technology (ESA, Paris), by Progetto Strategic0 Comitato Chimica (CNR, Roma) and by Minister0 delll’llniversita e Ricerca Scientifica to PGR.

5 References

[l] Gelfi, C., Bossi, M. L., Bjellqvist, B., Righetti, P. G., .I. Biochem.

121 Sinha, P. K., Righetti, P. G., J. Biochem. Biophys. Methods 1987, 15,

[3] Mosher, R. A., Bier, M., Righetti, P. G . , Electrophoresis 1986, 7,

[4] Altland, K., Banzhoff, A,, Electrophoresis 1986, 7, 529-553. [5] Righetti, P. G., Immobilized p H Gradients: Theory and Methodo-

[6] Chiari, M., Micheletti, C., Nesi, M., Fazio, M., Righetti, P. G.,

[7] Giaffreda, E., Tonani, C., Righetti, P. G., J. Chromatogr. 1993, 630,

[8] Righetti, P. G., Drysdale, J. W., J. Chromatogr. 1974, 98, 271-321. [9] Arvidson, S., Wadstrom, T., Biochim. Biophys. Acta 1973,310,418-

[lo] Balestrieri, C., Colonna, G., Giovane, A., Irace, G., Servillo, L.,

[ l l ] Ragone, R., Colonna, G., Balestrieri, C., Servillo, L., Irace, G. ,

[12] Bossi, A., Righetti, P. G., Chiari, M., Electrophoresis 1994, IS,

1131 Gianazza, E., Dossi: G., Celentano, F., Righetti, P. G.; J. Biochem.

[14] Takagi, H., Morinaga, Y., Ikemura, H., Inouye, M., J. Biol. Chem.

[IS] Graycar, T. P., Bott, R. R., Caldwell, R. M., Dauberman, J. L., Lad, P. J., Power, S. D., Sagar, H., Silva, R. A., Weiss, G . L., Wood- house, L. R., Estell, D. A,, Ann. N . I: Acad. Sci. 1992, 672, 71-79.

[16] Gron, A., Meldal, M., Breddam, K., Biochemistry 1992, 31, 41-48. [17] Bjellqvist, B., Ek, K., Righetti, P. G., Gianazza, E., Gorg, A, , Wes-

termeier, R., Postel, W., J. Biochem. Biophys. Methods 1982, 6,

Biophys. Methods 1987, 15, 41-48.

199-206.

59-66.

logy, Elsevier, Amsterdam 1990.

Electrophoresis 1994, 15, 177-186.

3 13-327.

420.

Tota, B., Comp. Anal. Biochem. 1978, 106, 49-54.

Biochemistry 1984, 23, 1871-1875.

1112-1117.

Biophys. Methods 1983, 8, 109-133.

1988, 263, 19592-19596.

317-339. [l8] Fredriksson, S., J . Chromatogr. 1977, 135, 441-446. [19] Bjellqvist, B., Hughes, G . J., Pasquali, C., Paquet, N., Ravier, F.,

Sanchez, J. C., Frutiger, S . , Hochstrasser, D., Electrophoresis 1993,

[20] Bjellqvist, B., Basse, B., Olsen, E., Celis, J. E., Electrophoresis

[21] Gelsema, J. W., De Ligny, C. L., J. Chromatogr. 1977, 130, 41-50. 1221 Gelsema, J. W., De Ligny, C. L., Van Der Veen, J. Chromatogr.

[23] Gelsema, J. W., De Ligny, C. L., Van Der Veen, J. Chromarogra.

[24] Reisfeld, R. A,, Lewis, U. J., Williams, D. E., Nature 1962, 195,

[25] Zweidler, R., Cohen, L. H., Fed. Proc. Fed. A m . Soc. Exptl. Biol.

14, 1023-1031.

1994, 15, 529-539.

1977, 140, 149-155.

1978, 154, 161-174.

281-283.

1972, 31, 926-927.

Received May 16, 1994