High-performance hydrophobic interaction chromatography of proteins

ANAL\‘TI(‘AI

RIOCHEMISTRY

High-Performance

137,

464-477

(1984)

Hydrophobic

J. L. FAUSNAIJGH,

interaction

E. PFANNKOCH,

Received

S.

October

Chromatography

GUPTA,

AND

F.

of Proteins’

E. REGNIER

3, 1983

A new, weakly hydrophobic, high-performance liquid chromatography column has been developed for the separation of native proteins based on their relative hydrophobicities. Starting with a covalently bound, hydrophilic polyamine matrix. packing materials were synthesized through acylation with anhydrides and acid chlorides of increasing chain length to obtain increasingly hydrophobic surfaces. Proteins in aqueous buffers were induced to bind hydrophobically to the columns by the use of high salt concentrations in the mobile phase. Elution was achieved by decreasing the ionic strength of the solvent in a linear gradient. A mixture of cytochrome c, conalbumin, and &glucosidase was used as a standard to test the resolving power of newly synthesized columns. On a 4-cm butyrate column. baseline resolution was achieved in 20 min with a gradient of 3.0 p sodium sulfate in 0.1 M potassium phosphate buffer. pH 7.0, to water. The static loading capacity for each column was determined using a hemoglobin binding assay. Capacities normally ranged between I50 and 180 mg of hemoglobin per gram of support. Since proteins are not denatured in hydrophobic interaction chromatography. enzymes eluted from the column retained enzymatic activity. Samples of a-amylase and &glucosidase ranging in size from 10 to 200 peg were recovered from the butyrate column with greater than 9% enzymatic activity in all cases. In a single trial. the enzyme citrate synthase was recovered from the benzoate column with 92%) retention of enzymatic activity.

In general, most hydrophobic amino acids in proteins are located in the relatively inaccessible interior of the three-dimensional structure of the protein. However. some hydrophobic patches may be found at the surface of the protein. This fact has been exploited ( 1-5) for the chromatographic separation of proteins in a process known as “hydrophobic interaction chromatography” (HIC).’ In HIC, proteins are induced to adsorb to weakly hydrophobic matrices by high-ionic-strength

mobile phases and are selectively desorbed during a descending salt gradient. The high salt concentration enhances nonpolar interactions and at the same time quenches electrostatic effects (6). The descending salt gradient weakens the hydrophobic interaction and allows the protein to elute from the column. HIC on soft-gel supports has been used in the purification of many proteins. Column packing materials have been prepared by coupling hydrocarbon chains of 4-8 carbon atoms to carbohydrate matrices through ether or isourea linkages. The principal difference between these two coupling techniques is that the ether linkage is neutral while the isourea bonding produces a weakly cationic species. Affinity of proteins for these weakly hydrophobic materials has been shown to be proportional to the chain length of the bonded phase (2,7). According to Melander and Horvath (S), the type of salt used can have a pro-

I This is Journal Paper Number 9633 of the Purdue University Agricultural Experiment Station. ’ Abbreviations used: HIC. hydrophobic interaction chromatography; IEC, ion-exchange chromatography; RPC. reverse-phase chromatography; CMETS. 2-(carbomethoxy)ethyltrichlorosilane; TEPA. tetraethylenepantamine: BSA. bovine serum albumin: OVA, ovalbumin: n-AMY. n-amylase: @-GLU. &lucosidase; LYZ. lysozyme; CYT C. cytochrome c: CON. conalbumin; p. ionic strength, = ‘/2 z:nl,=y; HP-HIC, high-performance hydrophobic

0003-2697/84

interact&n

chromatography.

$3.00

Copyright 1~ 1984 by Acadcmlc Press, Inc. All rights of reprnduct~on ,n any form reserved.

464

HGH-PERFORMANCE

HYDROPHOBIC

found effect on hydrophobic interactions. This property of a salt is quantified by its molal surface tension increment, which is a measure of the increase in surface tension caused by the addition of the salt to the mobile phase. Apparently, high salt concentrations are not detrimental to most proteins. Pahlman et uf. (9) have shown in circular dichroism studies that neither 3 M NaCl nor 1 M Na2SOJ alter the conformation of bovine serum albumin (BSA) or ovalbumin (OVA). Fujita et ~(1.(IO) was able to recover as much as 75%) of the initial enzymatic activity of several yeast enzymes after chromatographing in 3 M (NH4)$04. lJse of n-alkyl- and phenyl-derivatized carbohydrate supports with high-saltconcentration mobile phases in protein purification is now common. In addition to descending salt gradients, glycols, urea, and sucrose have been used to elute tightly bound proteins from HIC columns (4,6,10. I 1). Hydrophobic adsorption of proteins at high salt concentrations has been observed with several HPLC supports. Pfannkoch et ul. ( 12) observed that all commercially available highperformance size-exclusion chromatography columns for water-soluble polymers have a weakly hydrophobic character at high salt concentration. This has been further documented by Kato and Hashimoto (13.14) for proteins with ihe TSK SW-series columns. The feasability of high-performance hydrophobic interaction chromatography (HP-HIC) was demonstrated by Chang et al. in 1976 (15) with heptanol and polyethylene oxide-derivatized silica. The elution behavior of proteins on these weakly hydrophobic columns was identical to that described on carbohydrategel analogs. This paper describes the synthesis and chromatographic behavior of several HP-HIC columns. The support materials were synthesized starting with a polyamine bonded-phase silica previously described in the preparation of weak cation-exchange supports ( 16). HIC supports of increasing alkyl chain length prepared in this study were characterized with

INTERACTION

465

CHROMATOGRAPHY

regard to protein loading capacity. MATERIALS

retention,

AND

resolution,

and

METHODS

Materials. Vydac 101TPB5 was obtained from The Separations Group (Hisperia. Calif.). 2-(Carbomethoxy)ethyltrichlorosilane (CMETS) was purchased from Petrarch Systems (Levittown, Pa.). Tetraethylenepentamine (TEPA), butyric anhydride. benzoic anhydride. Valery1 chloride. benzyl alcohol. and phenylethyl alcohol were obtained from Aldrich (Milwaukee, Wise.). Trifluoroacetic anhydride was obtained from Sigma Chemical Company (St. Louis, MO.). Acetic anhydride was purchased from J. T. Baker Chemical Company (Phillipsburg. N. J.). Propionic anhydride was purchased from Matheson, Coleman and Bell (Norwood, Ohio). Hexanoic anhydride was obtained from Eastman Organic Chemicals (Rochester. N. Y.). All proteins were obtained from Sigma Chemical Company (St. Louis, MO.). except for &glucosidase (&GLU) which was purchased from Worthington Biochemical Corporation (Freehold, N. J.). All buffers and solvents were AR grade. Synthesis c$packitg muterid. HIC supports (Fig. I) were synthesized in a three-step coating process. In the initial reaction, .5+m macroporous silica wassilylated with CMETS. TEPA was subsequently coupled to the support at multiple sitesby nucleophilic displacement of the methoxy group from the organosilane as previously described ( 16,17). The resulting derivatized material had 1.53pmol of amine per gram of support. In the final step of the synthesis, 700 mg of support material was acylated with 3.9 mmol of the anhydride in a reaction mixture of 5.6 ml of dioxane and 3.5 ml of pyridine. In synthesizing the Valery1 support material, 700 mg of the TEPA-derivatized silica was acylated with 3.9 mmol of Valery1 chloride in 9.4 ml of N,N-dimethylformamide. TEPA. although drawn asa linear polymer in Fig. 1, is probably a mixture of linear and branched polymers. This is signif-

466

FAUSNAl.JGH

ET Al

Columns were tested on a Micromeritics 7000B liquid chromatograph with a variable wavelength Chromonitor 785 uv detector, a Linear 260 chart recorder. and a Rheodyne 7 120 injector valve. Ilemoglohi~~ hindiqq USSUJ~. A static hemoglobin binding assay was adapted from a previously published ion-exchange capacity assay( 15) to measure the protein loading capacity of these new HIC columns. A 2.5% (w/v) solution of hemoglobin was prepared in 0.1 M potassium phosphate buffer, pH 7.0, containing 5.0 p sodium sulfate. After the hemoglobin had dissolved, the solution was centrifuged to remove any particulate matter. One hundred milligrams of support material was weighed into a test tube. Approximately 2 ml of the hemoglobin solution was added to the test tube and gently vortexed. After centrifuging. the supernatant was aspirated away and the support material was washedwith 5.0 y sodium sulfate in 0.1 M potassium phosphate buffer, pH 7.0, until the supernatant was clear. The hemoglobin was desorbed and collected in volumetric flasks by repetitive washing and centrifugation with 0.1 M potassium phosphate buffer, pH 7.0, until the supernatant was again clear. From the absorption of the supernatant at 410 nm, the milligrams of hemoglobin bound per gram of support was calculated assumingan O.D. of 1.O for a 0.5-mg/ ml hemoglobin solution. C‘hronzatqyuphp. Proteins were eluted from the columns in a 20-min linear gradient from high sodium sulfate concentration in 0. I M potassiumphosphate buffer at pH 7.0 to water. No influence on the chromatographic profile or effluent pH was seenwhen water or buffered solution was used to elute the proteins. All separations were conducted at room temperature and a flow rate of 1 ml/min. Rrcovrr~~ qf enzyme uctivitJp. Protein sampleswere eluted at 1 ml/min in a 20-min linear gradient from 3.0 p sodium sulfate in 0.1 M potassium phosphate buffer, pH 7.0, to water. Control samples were injected into empty capillary tubing. cu-amylase(a-AMY) was asL’qllipmcnt.

?

step I

-(CII~SICH~CH~COCH~

step 2

(CMETSI

Ik

H2N-(CH2CH2NH13~CHpCH2NH2

-

CH3CH&OikH&H3

ITEPA)

101,

step

00 3

(prop~cmc

anhydrIde)

1

0’ l’)S,-0,

PH

CH3

FIG. 1. Schematic representation of the synthesis of the covalently bound HIC packing material.

icant in that the branched polymer will result in residual tertiary or sterically hindered secondary amines in even the most exhaustively acylated coating. Therefore, HIC packing materials derived from this synthetic route could have properties more similar to the conventional isourea-type HIC materials than the ether-bonded materials.

HIGH-PERFORMANCE

HYDROPHOBIC

INTERACTION

sayed with sta.rch and 3,5-dinitrosalicylic acid as the color reagent ( 18). P-GLU was assayed with p-nitrophenyl-P-D-glucopyranoside ( 19). RESULTS

Charactrrization

of the Packing Material

Several characteristics of the support materials synthesized by the procedure outlined above are presented in Table 1. The cationic character of the support materials may be measured by ion pairing with picric acid (20). This picric acid assayis capable of quantitating primary, secondary, and tertiary amines but not quaternary amines or amides. The ionpairing capacity of the TEPA-bonded phase support was reduced from 153 pmol/g of support to less than 30 pmol/g by acylation in all cases.This data indicates that 80 to 90% of the amines in the covalently bonded TEPA layer were derivatized and that acyl ligand density was 1.4 to 1.6 pmol/m’. From the ionpairing capacity, 0.3 pmol of residual amine/ m’ was present on the acylated packing material. Therefore, the HIC columns have a

TABLE COL~JMN

weak anion-exchange character. Most ion-exchange (IEC) and reverse-phase (RPC) columns have 2 to 4 pmol of stationary phase groups/m’ (2 1). Attempts to separateproteins on the acetate column in the anion-exchange mode with 50 mM potassium phosphate buffer, pH 7, were unsuccessful.Although the picric acid assaydemonstrated that there were residual amine groups in the packing, the density of cationic groups in the packing material was low and they appeared to be inaccessible to proteins. Therefore, there are no detectable electrostatic contributions to the binding of proteins to these packing materials. Protein loading capacity of the support materials was measuredwith a hemoglobin binding assay. When the support materials were completely saturated with hemoglobin in a static batch assay, loading capacities were in the range 150- 180 mg/g of support. These values are comparable to those for cation- and anion-exchange materials prepared with the same silica support matrix. The color of the support material ranged from gray to light brown after the assay. indicating that some hemoglobin was irreversibly bound.

I

CHARACTERISTICS Retention CYT

Column Tritluoroacetate Acetate Propionate Butyrate Valery! Hexanoate Benzoate

Ion-pairing capacity (woN2) 6.66 17.08 29.67 21.39 14.49 10.52 25.36

Hemoglobin binding capacity (w/g) 211 NA’ 147 185 I82 135 162

467

CHROMATOGRAPHY

and resolution

C

of proteins

CON

p-GLU

_~ tit (min) 1.0” 0.8 0.7 0.8 0.6 1.4 2.2

y The trifluoroacetate chromatogram was recorded with an initial ionic strength were recorded with an initial ionic strength of 3.0 NaZSO,. ’ NA = Not available. ’ ‘2, Protein irr:versibly retained.

IR R, 5.2 3.7 5.3 7.7 6.0

(min) 8.0 5.3 6.0 8.1 1.7 zc Ix.

of 3.5 Na2S04.

R, 1.9 4.2 5.2 5.7 5.1

fK (min) 11.8 11.3 12.8 15.0 17.7 24.2 24.3

All other chromatograms

468

FALJSNALIGH

The HIC columns gave very reproducible chromatograms. Chromatograms recorded before and after 48 h of continuous use at pH 2.0 showed no significant deviations. All columns remained functional with only slight deterioration during the 6 months of this study. Issocratic~ Chromatography

The nature of the HIC interaction wastested using two small aromatic alcohols, benzyl alcohol and phenylethyl alcohol. As the ionic strength was increased, the retention time of these probes increased, indicating a hydrophobic interaction with the support material. The same effect was seenwith standard proteins (Fig. 2). All proteins tested except lysozyme (LYZ) eluted in the void volume at low ionic strength. From these retention curves, an estimate can be made of the native or surface hydrophobicity of the various proteins. Cytochrome L’(CYT C) appeared to be a very hydrophilic protein and, therefore, required high ionic strengths to induce adsorp-

IO

F I AL.

tion to the support material. Of the proteins tested, LYZ appeared to be the most hydrophobic. Although proteins may be eluted from the HIC columns isocratically. the peakswere very broad and the resolutionswere very poor. Only proteins of widely differing retention properties may be resolved by isocratic elution. It became apparent early in these studies that gradient elution would produce superior protein resolution.

For each of the columns. a mixture of CYT C. conalbumin (CON), and /I-GLU wasloaded onto the column in 3.0 p sodium sulfate in 0.1 M potassium phosphate buffer, pH 7.0. Elution was achieved in a 20-min descending salt gradient. The retention times (fR) and resolution (R,) of these proteins are presented in Table 1. CYT C was only retained on the more hydrophobic hexanoate and benzoate columns. The butyrate column (Fig. 3) gave the best resolution of this mixture with values

20

30 N%S04,

40

p

FIG. 2. The retention of several proteins and small hydrophobic alcohols was measured isocratically as a function of the ionic strength of sodium sulfate on the tiifluoroacetate column, The proteins were lysoayme. 0; conalbumin, 0: ovalbumin, A; bovine serum albumin. A; myoglobin. 0: and cytochrome c. n . The small hydrophobic alcohols were phenylethyl alcohol. V: and benzyl alcohol, 0.

HIGH-PERFORMANCE CO1

HYDROPHOBIC

INTERACTION

CHROMATOGRAPHY

469

ml”

:hrome C

7

J---4

/ I

I

I

5

IO

15

TIME

(min)

FIG. 3. A 20-~1l injection of a mixture of cytochrome e (3.5 mg/ml). conalbumin (2.5 mg/ml). and /$glucosidasc (10 mg/ml) was separated in a 20-min linear gradient from 3.0 p sodium sulfate in 0.1 M potassium phosphate butfer. pH 7.0. to water on a 0.41 K 4-cm butyrate column.

of 7.7 between CYT C and CON and 5.7 between CON and p-GLU. The peaks eluting from the butyrate column were very sharp compared to the other columns. The more hydrophobic columns gave very broad peaks, although the separation between the individual proteins was quite good. The initial ionic strength had a significant effect on proiein chromatographic behavior (Fig. 4). Proteins eluting very early or very late in the gradient tended to have sharper peaks than those eluting in the middle of the gradient. For example. the peak width of CON was 1.2 ml when it eluted either early or late in the gradient, but it was 1.X ml when it eluted in the middle of the gradient. There are two possible explanations for this phenomenon. The first is that resolution might be greater at intermediate ionic strengths and, therefore, broad peaks are the result of subfractionation of a heterogeneous protein sample. Such a phenomenon has been ob-

FIG. 4. The effect of initial ionic strength on protein retention was analyred with a 20.~1 injection ofa mixture of cytochrome e (2.5 mg/ml). conalbumin (1.5 mg/ml), and &glucosidase ( IO mg/ml) on a 0.41 k 5-cm acetate column. The major peahs of cytochromc c. conalbumin. and &glucosidase. in that order. were separated in a 20. min linear gradient to water at a flow rate of I ml/min. In chromatogram A. the initial ionic strength of sodium sulfate was 3.5, in B it was 3.0, and in C it was 7.5 in 0. I M potassium phosphate buffer, pH 7.0

470

FAUSNAUGH

served with IEC columns (22). The second explanation is that proteins eluting at intermediate ionic strengths undergo more adsorption-desorption stepsduring elution with a concomitant increase in kinetic contributions to band-spreading. Support for this explanation was seen when the gradient time was decreasedfrom 20 to 5 min. Less bandspreading of intermediate peaks was observed relative to the samepeaks eluting in a 20-min gradient (unpublished data). Further work is needed to clarify this phenomenon. The initial ionic strength also had a significant effect on the selectivity of the HIC columns (Fig. 4). The most tightly retained species was affected the least by decreasing the initial ionic strength of sodium sulfate. The retention of P-GLU was reduced by 30% when the initial sodium sulfate ionic strength was decreased from 3.5 to 2.5. In contrast, the retention of CON was reduced by 80% for the same change in initial ionic strength. By changing the initial loading conditions, the resolution between any two proteins could be dramatically altered. Unlike RPC (23) protein retention on HIC columns is a function of bonded-phase chain length (Fig. 5). (A column derivatized with acetic anhydride is designated as having a chain length of one.) This behavior of the HPHIC column is analogous to the gel-type HIC columns. The increase in protein retention with increasing chain length indicates that these support materials have increasing hydrophobic surfaces. With increasing chain length, the retention of fl-GLU varied from 11.3 to 24.2 min in going from the acetate to the hexanoate column. In contrast, the retention of CON varied from 5.3 min to infinity for the same series of columns. Thus, alkyl chain length provides another means of varying retention and selectivity in the HIC mode.

Recovery of Enzyme Activity Recovery of CY-AMY and /3-GLU from the butyrate column is shown in Table 2. Sample sizesranging from 10 to 200 pugwere gradient

ET

AL.

CHAIN

LENGTH

FIG. 5. The effect of chain length on protein retention was measured for conalbumin (0) and fl-glucosidase (0). A 204 injection of each protein at a concentration of 2.5 mg/ml of conalbumin and 10 mg/ml of n’-glucosidase was chromatographed on each of the columns in a 20min linear gradient from 3.0 fi sodium sulfate in 0.1 M potassium phosphate buffer. pH 7.0. to water.

eluted in 20 min from the column, collected, and assayedfor enzyme activity. Enzyme recovery in all casesexceeded 90% with a-AMY and 95% with P-GLU. In a single trial, citrate synthase was recovered from an HIC column with greater than 90% yield. This data indicates that the adsorption-desorption process in HIC does not irreversibly alter the threedimensional structure of these proteins. This makes HIC potentially useful asa preparative technique. DISCUSSION

AND

CONCLUSIONS

Seven columns of increasing hydrophobicity were synthesized and evaluated. Acyl groups on the support surface ranged from a methyl group to a five-carbon chain. An additional column containing a phenyl group was synthesized to examine aromatic hydrophobicity. The acyl ligands were attached to the surface through an amide linkage. The less

HIGH-PERFORMANCE

HYDROPHOBIC

INTERACTION TABLE

471

CHROMATOGRAPHY

2

RECOVERY OF ENZYME ACTIVITY cu-Am$ase

/3-Glucosidase Units recovered (XlOF)

Units recovered (X10 ‘) Amount protein injected (pg) 10 25 50

100 200

Sample

Control

Percentage recovery

Sample

Control

Percentage recovery

7.8 23. I 45.6 87.5 135.0

7.5 23.2 49.4 95.0 139.0

103 99.6 93.3 92.1 97.1

I.0 2.5 6.0 11.4 22. I

1.0 2.6 4.9 Il.4 25.5

100 96 122 100 98

hydrophobic supports gave fairly sharp peaks, but the proteins were only weakly retained even at high ionic strength. This resulted in decreasedresolution. In contrast, the very hydrophobic supports such asthe hexanoate and the benzoate columns retained proteins very strongly even at low ionic strength, but the peaks were very broad. This again resulted in decreasedre:solution. The addition of propanol to the buffers did not affect the peak width, but did decreasethe retention time (unpublished data). Overall, the intermediate hydrophobic butyrate column gave the best resolution of the proteins tested. The chrornatographic mode on theseweakly hydrophobic columns has been shown to be an induced hydrophobic interaction between the solute and the support as seen by the increased retention of proteins and small hydrophobic alcohols with increasing mobilephase ionic strength (24). There was also an increase in retention time with increasing chain length. This is unlike RPC in which there is no relationship between retention time and alkyl chain length. The ligand density and spatial orientation of the functional groups on the HIC columns differ significantly from the RPC columns. The ligand density on HIC columns is 1.4 to 1.6 pmol/m2 as opposed to 4 pmol/m’ on the RPC columns. Both HIC and RPC separate proteins on the basisofiheir hydrophobicity. RPC utilizes the intrinsic hydrophobicity of the primary

sequence, while HIC columns only interact with accessible hydrophobic groups in the three-dimensional structure of the protein. Since the number of these residues is smaller than the total number of hydrophobic residues in a protein, the number of amino acid residues that will interact with an HIC column is relatively small. Jennissen (25) has shown with phosphorylase kinase that this number is four or five. The number of hydrophobic residueson the surface of a protein may not reflect the hydrophobicity of the amino acid sequence of the protein. For example, LYZ was found to be a very hydrophobic protein relative to other protein standards by HIC, in contrast to RPC where it is relatively hydrophilic (26). The initial mobile-phase ionic strength had a sizable effect on the retention of proteins in HIC. A decrease in the initial ionic strength had a greater effect on weakly retained proteins than on the more strongly retained species. The pH of the buffer may also alter the chromatographic profile. This has already been seen in nonideal size-exclusion chromatography ( 13). where high salt was usedto induce a hydrophobic interaction with the surface of the column. By varying chain length, initial ionic strength. and pH, it should be possible to manipulate protein retention so that resolution can be maximized. HIC appears to separate proteins with respect to their surface hydrophobicity. There-

472

I-Al.ISNAUGH

fore. unlike RPC, proteins eluting from the column should retain their biological activity. This wasfound to be the casefor both (u-AMY and fi-GLU. Since these HIC columns exhibit high static loading capacities and biologically active enzymes are recovered, this mode of chromatography should be adaptable to preparative separations. The columns synthesized in this study have exhibited many advantages with regard to the separation of proteins. Despite the high salt concentrations required in this new mode of HPLC, proteins are recovered with high enzymatic activity. The retention of proteins is easily manipulated and the columns have high static loading capacities. Further studies on mobile-phasevariables such aspH, salt effects, and temperature are needed. ACKNOWLEDGMENTS The authors Mr. R. Drager Fen L. Chang This work was

wish to thank Mr. W. Kopaciewicz and for their helpful discussions, and Dr. Lifor performing the citrate synthase assay. supported by NIH Grant GM 25431.

E-f

Ho&tee. B. J. H. ( 1975) Hl~~c~hc,th Bu~/)I‘\ K(,\ ~‘rttwn~rtr. 63, 6 I X-674. 7. Arfmann. H. A.. and Shaltiel. S. (1976) Ixr .I Bmkon. 70 .- 36’1-274. x. Melander. W.. and Horvath, C. ( 1977) .-ln+l /1roc~/rc~rn. Biotplry~.

183,

200-Z

I 5.

Pahlman.

IO. I I. 13.

13.

S., Rossengren. J.. and Hjcrten. S. (1977) ./ Chrotnaict,yr. 131, 99-I OX. Fujita. T. d (I/. ( 1980) J. Rioc~lnv?!. 87, 89- 100. Le Peuch. C.. and Balny. C. (1978) I*FBS Lerl. 87, 232-234. Pfannkoch. E.. Lu, K. C.. Regnier. F. E.. and Barth, H. G. (1980) J. ~%rcvm/ct,qr. Sb. 18, 430-441, Kato, Y.. and Hashimoto. T. ( 1983) J. Ili,q/r Rmhrt. Chrottmtqyr.

~‘ottvtttwt.

14. Kato. Y.. and Hashimoto, (%romu/~~,yr.

15. Chang. 16. 17. IX.

19. 20. ?I. 2’. 23.

~‘ottmrrtt

6, 45-46.

T. ( 1983) J. Il;,$!

Ke.\o/~c.

6, 324-325.

K. M., and Regnier. F. E. 120, 32 l-333. Gupta. S.. Pfannkoch. E.. and Regnier. F. E. ( 1983) .,I& Uioc~hctn. 128, 196-20 I. Gupta. S., and Regnier. F. E. U. S. patent application 44 I .057. Bemfeld. P. ( 195 I ) it?Advances in Enzymology (Nerd. F. E., cd.). Vol. 12. pp. 379-428. Interscience. New York. Bahl. Om. P., and Agrawal. K. M. L. (1968) J B& Chw~. 243, 9X- 102. Alpert. A. J.. and Regnier. F. E. ( 1979) J Chmtttuftyr. 185, 375-392. Berendsen, G. E.. Pikaart, K. A., and de Galan. L. ( 1980) J. I,iq. ~‘/rrott~ur~t,~r. 3, I4371464. Vanacek. G.. and Regnier. F. E. (1980) At&. Biochwt. 109, 345-353. Pearson. J. D.. and Regnier. F. E. (1983) /. Liq. C‘lrr,j( 1976)

REFERENCES I. Er-el, Z.. Zaidenzaig. Y., and Shaltiel. S. (1972) Biochcw Bioph?:\. Re.s. C’onzmw~. 49, 383-390. 2. Shaltiel. S.. and Er-el, Z. (1973) Proc,. Null. .~uu/. .S?i. c:x‘l 70, 77X-78 I. 3. Hofstee, B. H. J. ( 1973) Bio&nr. Rnqhjx Res. C‘0tntn1rti. 53. 1 137-l 144. 4. Porath. J.. Sundberg, L., Fornstedt. N.. and Olsson, P. ( 1973) Nuiwc (London) 245. 465-466. 5. Hjerten, S. (1973) J. C~rnvxuto~~r. 87, 325-331.

AL.

tttufoq

S. H.. Gooding. .I. Chnttnu/~t,yr.

6, 497-S

IO.

24. Morris. C. J. 0. R. ( 1976) 7’ret~ds Riochcvn. SC;. 1. N207-N’OX. 75. Jennisen. H. P. ( 1976) ffop~l’c~-.S~~~~/~~r:s /: Plry,Gtl ~%W~?. 357, 1201-1203. 26. O’Hare. M. J.. and Nice. E. C. (1979) J. C’hrotttu/ct,qr 171, 209%226.