Water Solubilization of Membrane Proteins

0 downloads 0 Views 835KB Size Report
A reagent and a derivatization procedure have been developed ... and this reagent was used to monitor the protein deri- .... Pyridine, obtained from J. T. Baker.
Vol. 263, No. 17, Iseue of June 15, pp. 7989-7996 1988 Printed in ~ . S . A .

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Water Solubilization of Membrane Proteins EXTENSIVE DERIVATIZATION WITH A NOVEL POLAR DERIVATIZING REAGENT* (Received for publication, December 11, 1987)

Robert C. Morton$ and GerhardE. Gerbers From the Department of Biochemistry, McMaster University, Hamilton, Ontario L8N 325,Canada

insufficient to maintain the solubility of fragments due to the A reagent and a derivatization procedure have been developed which result in trimesylationallof-OH and presence of extended hydrophobic domains and the relative -NH2 groups on proteins: serine, threonine, tyrosine, paucity of amino groups which are derivatized (12). The and lysine side chains are all completely derivatized. derivatization of additional functional groups or the attachThe parametersaffecting the kinetics of trimesylation ment of a more polar reagent would be expected to result in of serine, threonine, and tyrosine peptideswere stud- water solubility of even the hydrophobic proteins. ied using dinitrophenylated peptides as model systems. A reagent and a derivatization procedure have been develConditions for solubilization of proteins in anhydrous oped which result in the attachment of trimesylate (1,3,5organic solvents for the derivatization are described; benzenetricarboxylate) groups to all hydroxyl andamino removal of blocking groups results in the polar, highly groups on proteins: serine, threonine, tyrosine, and lysine side water-soluble protein derivative which behaves as a chains are all completely derivatized. The parameters affectmonomer during gel permeation chromatography in ing the kinetics of trimesylation of serine, threonine, and simple aqueous buffers even in the absence of detertyrosine residues were studied using dinitrophenylated pepgents. Synthesis of the tritiated reagent is described, and this reagent was used to monitor the protein deri- tides as model systems. Conditions for the solubilization of proteins in anhydrous organic solvents, for the derivatization vatization. procedure, and thesubsequent removal of the blocking groups from the reagent are described. The procedure results in the attachment of two carboxylate ions at each site of derivatiThe determination of the amino acid sequence of intrinsic zation and results in polar, highly water-soluble derivatives membrane proteins has progressed very much more slowly even in the case of the very hydrophobic protein bacteriothan that for water soluble proteins. The protein fragmenta- rhodopsin; this behaves as a monomer during gel permeation tion, purification of the fragments, and theactual sequencing chromatography in simple aqueous buffers even in theabsence of the fragments are all plagued by the problems associated of detergents. Synthesis of the tritiated reagent is described, with the proteins’ lack of water solubility. Several chemical and this reagent was used to monitor the protein derivatizafragmentation methods can be carried out in solvents which tion. It is expected that this method should be generally solubilize membrane proteins, at least partially (1). The use applicable for the conversion of membrane proteins intowater of organic solvents for isolation of fragments during gel per- soluble ones: this should facilitate the enzymatic fragmentameation in reverse phase chromatography has greatly facili- tion and isolation of the fragments as well as improve their tated the isolation of hydrophobic fragments (2-5). Further- behavior during automated Edmandegradation in both spinmore, detergents (6, 7) have been used to increase the solu- ning cup and gas-liquid phase sequencers. bility of hydrophobic fragments during fragmentation and MATERIALS ANDMETHODS sequencing. These methods notwithstanding, sequence studies of intrinsic membrane proteins often end upbeing difficult, Chemicals that were purchased from Sigma included trimesic acid especially in the hydrophobic core (4,8).The development of (1,3,5-benzenetricarboxylicacid) (I), 1,l’-carbonyldiimidazole(CDI),’ DNA sequencing has resulted in the determination of the dicyclohexylcarbodiimide,4-pyrrolidinopyridine (PPY), 1-fluoro-2,4dinitrobenzene, L-proline, and the hydrochloride salts of L-serinamamino acid sequence of many membrane proteinsasthis ide, L-threoninamide, L-glycyltyrosinamide,L-glycinamide, and purmethodology obviously circumvents the problems cited above ple membrane of Halobacterium halobium. Tetrabutylammonium hy(9). However, protein sequencing is still required to confirm droxide (B-NOH, 1.54 M aqueous solution) and trimethylsilylethanol the deduced sequence, to identify sites of posttranslational (Me&iCH,CHz-OH)were obtained from Aldrich. Triethylamine modification (lo), and to obtain direct experimental topolog- (Et3N) and trifluoroacetic acid were purchased from Pierce Chemical Co.; the Et3N was distilled from ninhydrin under nitrogen to remove ical data. In the case of water-soluble proteins, succinylation of lysine any primary orsecondary amine contaminants. Storage of the distillate was at 4 “C under nitrogen, over Fisher 4-A molecular sieve, and amino groups is usually sufficient to keep the denatured protected from light. Sephadex G-150, LH-20, and LH-60 were obprotein or its fragments adequately water-soluble for struc- tained from Pharmacia LKB Biotechnology, Inc. All other chemicals tural studies (11); in the case of very hydrophobic intrinsic were reagent grade. membrane proteins,this degree of modification is usually HPLC-grade methanol, tetrahydrofuran, dimethylformamide

* This work wassupported by Medical Research Council of Canada Grant MA6488 (to G. E. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solelyto indicate this fact. $ Recipient of a Medical Research Council of Canada Studentship. I To whom correspondence should be addressed.

(DMF), dimethyl sulfoxide, benzene, petroleum ether, and diethThe abbreviations used are: CDI, carbonyldiimidazole; B-N, tetrabutylammonium; DMF, dimethylformamide; EtSN, triethylamine; tmse, Me&CH2CH2, trimethylsilylethyl; TM acid, trimesic acid (1,3,5-benzenetricarboxylicacid); PPY, 4-pyrrolidinopyridine; HPLC, high pressure liquid chromatography; DNP, 2,4-dinitrophenol; CCL,carbon tetrachloride; (tmse),TM, di-protectedtrimesic acid.

7989

7990

Water Solubilization of Proteins Membrane

yl ether were purchased from Caledon. Anhydrous DMF was made (1.2 equivalents). After 3 h at room temperature, the reaction was by a modified procedure of Thomas and Rochow (13); DMF was diluted with 2 volumes of aqueous 10 mM sodium phosphate, pH 7. initially dried by the addition and subsequent removal under reduced Following the removal of the ethanol under reduced pressure, the pressure of 0.1 volume of benzene followed by storage over 4-A excess l-fluoro-2,4-dinitrobenzene was selectively extracted into diemolecular sieve and under nitrogen. Dry benzene was obtained by thyl ether. Subsequent acidification of the aqueous phase to pHl, by distillation from calcium hydride and was stored under nitrogen and titration with 6 N HCl, permitted the extraction of DNP-proline into over molecular sieve ( 4 4 . Pyridine, obtained from J. T. Baker ethyl acetate. The extracts were washed with water and then evapoChemical Co., was incubated overnight with KOH pellets, decanted, rated under reduced pressure. Residual water was removed by drying and then distilled from barium oxide under nitrogeq. The anhydrous the DNP-proline several times from a mixture of anhydrous benzene pyridine was then stored under nitrogen and over 4-A molecular sieve and DMF. DNP-proline was stored as a 0.25 M solution in anhydrous in amber bottles. Carbon tetrachloride (CCla),purchased from Fisher, DMF over a molecular sieve (4-A) under nitrogen. was distilled from phosphorous pentoxide under nitrogen. All other DNP-Peptide Synthesis-(a) DNP-prolylserinamide andDNPsolvents were reagent grade. The petroleum ether fraction used was prolylthreoninamide: DNP-proline was activated by the addition of the one boiling between 30 "-60 "C. 1.1 equivalents of CDI in dry tetrahydrofuran as a 60 mM solution High Pressure Liquid Chromatography-The HPLC system (Mil- and incubated at room temperature for 1 h under nitrogen. To the lipore-Waters) consisted of two "60 pumps, a 720 system controller, reactive imidazolide derivative, 3 equivalents of the hydrochloride an M 730 data module, a WISP 710B automatic injector, and a 440 salt of serinamide or threoninamide was added as a freshly prepared absorbance detector with the appropriate filter for 365 or 280 nm 1.54 M aqueous solution in the presence of 1 equivalent of BhNOH. detection. The column routinely used for the analysis of DNP( b ) DNP-prolylglycinamide and DNP-prolylglycyltyrosinamide: peptides was a 4.6 X 250-mm pBondapak ODS comprised of 10 p~ The formation of DNP-prolylimidazolide was carried out under the CISbonded particles that had been end-capped after the column was same conditions as outlined above except that the solvent was dry packed. The elution buffers were Caledon methanol (HPLC grade) DMF and the concentration was 80 mM. At the appropriate time, 3 and British Drug House sodium acetate dissolved in glass distilled equivalents of the hydrochloride salt of glycinamide or glycyltyrosiwater; the pH was adjusted with British Drug House glacial acetic namide was added as an anhydrous 80 mM solution in DMF, in the acid. presence of 2 equivalents of Et3N. Synthesis of (trnse)zTM Acid (ZV)-AIl steps inthe synthesis of the After 12 h, each peptide reaction was diluted with 0.5 volume of (tmse),TM acid were performed in a closed vesselunder nitrogen and water and the desired DNP-peptideextracted into ethylacetate. at room temperature. The reactive triimidazolide derivative of tri- Isolation and purification of the peptide products was accomplished mesic acid (11)was generated by the addition of 4.8 mmol of trimesic by preparative HPLC on a reverse phase pBondapak CIS column acid (I) in 6 ml of anhydrous DMF, in approximately 1-ml aliquots, which was equilibrated and run isocratically with 1%aqueous acetic to a screw cap tube containing 2.33 g of CDI (3 equivalents) in 4 ml acid containing the following specified percentages of methanol: of dry DMF. During the addition, the mixture was continuously DNP-prolylglycinamide and DNP-prolylserinamide required 10% vortexed. After 1 h, 2 equivalents of Me3SiCH2CH20Hwere added to methanol, while DNP-prolylthreoninamide and DNP-prolylglycyltythe activated trimesic acid (11), and the subsequent esterification rosinamide required 15 and 20% methanol, respectively. The DNPreaction was allowed to proceed overnight. Following a dilution with peptides were dried several times from anhydrous pyridine and then an equal volume of dry DMF, the remaining imidazolides were hydro- stored as a 25 mM solution in dry pyridine over molecular sieve and lyzed by the addition of 2 ml of 2.5 M aqueous NaOH. Under these under nitrogen. basic conditions, three petroleum ether extractionsafforded the comFormation of Tetrabutylummonium Salt-Tetrabutylammonium plete removal of the tri-estercontaminant. The aqueous solution was trifluoracetate (BQN-trifluoroacetate) was made by the addition of then acidified with HC1 topH 1 allowing the desired product, 1.1 equivalents of trifluoroacetic acid to 1.0 equivalent of tetrabutyl(trnse),TM acid, to be selectively extracted into diethyl ether/petro- ammonium hydroxide. The excess acid was removed under reduced leum ether (1:9, v/v). The mono-ester side product remained in the pressure and the resulting hydrophobic salt was repeatedly (3 times) aqueous phase and could only be extracted with 100% diethyl ether. dissolved in benzene, and the solvent was removed under reduced The extracts containing (tmse);rM acid were pooled, washed with pressure. The resulting residue was then stored as a 1 M solution in 0.1 N HC1, and thenonce with 4 M NaCl. The solvent was evaporated absolute ethanol under nitrogen. under a stream of nitrogen, and theproduct was repeatedly (3 times) Standard Trimesylation-The DNP-peptides were trimesylated dissolved in dry benzene and the solvent removed under reduced under nitrogen as a 500 pM solution in dry pyridine in the presence pressure. The dried (tmse)2TM acid was stored as a 0.2 M solution in of 100 mM activated reagent ((tmse)2TM anhydride), 100 mM PPY, dry benzene, over molecular sieve (4-A) and under nitrogen. and 20 mM BmN-trifluoroacetate. The addition of the DNP-peptide Activation of (tmse),TM Acid-The reactive anhydride derivative of choice initiated the reaction, and atspecific times an aliquot of the of (tmse),TM acid was generated as a 0.2 M solution at room temper- reaction was quenched in 10% Hz0 and 0.1% EhN in DMF. The ature by the addition of 1.0 equivalent of dicyclohexylcarbodiimide extent of peptide acylation was determined by HPLC analysis of a to 2.25 equivalents of (tmse)2TMacid in dry CCL. After 1 h, this mixture was diluted to 50 mM with dry CClc and thedicyclohexylurea portion of the quenched reaction which was applied to a pBondapak precipitate was removed by filtration through glass wool. The CC4 Cle reverse phase column. The HPLC conditions are those specified was removed under reduced pressure, and theanhydride residue was in Fig. 3 and gave the following retention timesfor the DNP-peptides redissolved in anhydrous benzene. This afforded a stock solution of and their corresponding derivatives: 5.0 and 15.9 min for the serine the activated reagent from which appropriate aliquots were used for dipeptide; 6.4 and 16.2 min for the threonine dipeptide; and 7.8 and trimesylation, after the removal of the benzene under reduced pres- 17.5 min for the tyrosine tripeptide. Following 100% DNP-peptide derivatization, the excess reagent sure. was deactivated by the addition of Me3SiCH&H20H to a final Synthesis of PHITrimesic Acid-The trimesic acid (660 mg) was placed into the bottom of an 18 X 150-mm Pyrex test tube, and the concentration of 0.1 M while remaining under nitrogen and at room tube was then restricted. The tritiated water (500 pl containing 2.5 temperature. The alcoholic quench proceeded for 0.5 h, after which Ci) was then added and rinsed in with 50 plof 12 N HCl. The contents time the deblocking of the carboxyl groups was achieved by the of the tube were then frozen in liquid nitrogen, the tube was evacu- addition of B*NF, in sufficient quantity to have at least 2 equivalents ated, purged with nitrogen, and then sealed. The sealed tube was for each 2-(trimethylsily1)ethylgroup (14). This deprotection reaction placed in a pressure device containing water to equalize the pressure was essentially instantaneous. Preparation of Apobacterinrhodopsin-The purple membrane of H . and heated at 275 "C for 31 days. After cooling, the tube was opened, the ['H].HzO removed by evaporation, and the dry residue recrystal- halobium cells was suspended (2 mg/ml) in 0.5 ml of dimethyl sulflized twlce from 15 ml of boiling water. The specific activity of the oxide to which 0.5 ml of 4.0 M NaCl containing 1 M NHzOH.HC1, product was 5 mCi/mmol, and it was found to co-elute with authentic pH 7, was added. The suspension was stirred for 1 h at 37 "C to ensure complete disappearance of the purple color (15, 16). The trimesic acid on HPLC analysis. DNP-Proline Synthesis-During the synthesis and subsequent apomembrane was diluted 8-fold with H20, pelleted by a 100,000 X g storage of DNP-proline, as with the other DNP-peptides, care was centrifugation for 1 h, resuspended in 1 ml of distilled HzO, and taken to protect these compounds from prolonged exposure to light. lyophilized. The residue was redissolved in 88% formic acid and I-Proline (150 pmol) was dissolved in ethanol/water (3:1, v/v, 2 ml) applied to a Sephadex LH-20 column (1 X 40 cm) that was equiliin the presence of 2 equivalents of Et3N; the dinitrophenylation brated and run in formic acid/ethanol (30:70) to achieve the delipireaction was initiated by the addition of l-fluoro-2,4-dinitrobenzene dation of the apobacteriorhodopsin. Fractions (0.5 ml) were collected

Water Solubilizationof Membrane Proteins under nitrogen, and the void volume peak was pooled and stored at -20 “C. Solubilization of Apobacteriorhodopsin in Anhydrous PyridineThedelipidatedproteinwasdriedunderreducedpressure in the presence of BQN-trifluoroacetate (100 pmol/mg of protein) which wasaddedas a 1 M solution in absolute ethanol. The protein/salt 200 pl of dry pyridine which was residuewastwicedissolvedin

7991

cally impossible; 2) the stability of aromatic esters is much greater due to theelectronic effect of the aromatic ring; 3) the carboxylate groups on the ring further increase its electronic density and hence stabilize the ester linkage; 4) the steric hindrance of the bulky group would be expected to hinder hydrolysis of the ester. removed under reduced pressure. The resulting residue was redisSeveral methods of activation of this reagent were studied solved in50 pl of pyridine. and thefollowing complications had to be considered. 1) The multifunctional nature of the reagent, upon activation of one RESULTS ANDDISCUSSION carboxyl group resulted in extensive cross-linking of the reConversion of membrane protein into truly water-soluble agent with other reagents as well as with the protein; the ones by means of extensive derivatization of all possible extra carboxyl groups thus needed to be protected. 2) The functional groups with a polar reagent has the potential of same factors responsible for the increased stability of aromatic making these proteins more amenable to sequencing by conesters mentioned above also result in lower reactivity of the ventional methods. It was therefore our objective to develop activated groups. The aromatic ring thus needs to be initially methods to derivatize the hydroxyl groups of serine, threoactivated to increase the reactivity of the acylating group; this nine, and tyrosine in addition to the amino group of lysine with a polar reagent. This paper describes the development activating group should be later changed to an inactivating of suitable reaction conditions for such derivatization, the one to stabilize the desired product. Initially, activation of the trimesic acid by forming its development of a suitable reagent, and the chromatographic triimidazolide was investigated. This reagent resulted in rapid behavior of the protein derivatives. acylation kinetics and theactivation of all carboxyl groups as The complete derivatization of hydroxyl groups, particularly the secondary hydroxyl groups of threonine, obviously their imidazolides protected them from cross-linking with requires reaction of proteins under anhydrous conditions. The each other. However, it was found that during aqueous first requirement thus was an aprotic solvent system capable quenching of the activated groups, extensive protein crossof dissolving proteins under anhydrous conditions. Associa- linking was obtained, as indicated by gel permeation chrotion of proteins with the very hydrophobic tetrabutylammon- matography (data not shown). Thus, although the tri-activated reagent had the requisite reactivity it did not produce a ium counter-ions resulted in their solubility in anhydrous pyridine, provided that thefollowing two-step procedure was useful derivative. Another activation procedure investigated was the generafollowed the lipid-free protein was first dried in the presence of BQN-trifluoroacetateandthen dissolved in absolute tion of the (tmse)zTMacid imidazolide (111)with carbonyldiethanol/toluene (1:2, v/v) and redried; only then was the imidazole in benzene. This reagent, although unreactive in protein residue readily soluble in dry pyridine. The presence pyridine even in the presence of equimolar amounts of the of the tetrabutylammonium counter-iongreatly enhances the acylation catalysts 4-dimethylaminopyridine (19, 20) or 4protein’s solubility inanhydrous solvents. This procedure pyrrolidinopyridine (19), was found to be fairly reactive in ensures that both formic acid and water are completely re- DMF in the presence of 1 M Et3N. However, due to reagent moved from the protein as assessed by reagent reactivity and instability and problems of protein aggregation in DMF, this [3H]Hz0 removal (data not shown). This procedure renders form of activated reagent was eventually abandoned. both water-soluble proteins such as bovine serum albumin The approach finally taken and described in this paper is and membrane proteins soluble in anhydrous pyridine as the shown in Fig. 1. The trimesic acid was activated to the tritetrabutylammonium salt, ina form suitable for acylation. imidazolide (11) in DMF using 3.0 equivalents of CDI. AddiOther solvents were analyzed for protein solubilization but tion of 2.2 equivalents of trimethylsilylethanol resulted in the were not found to be useful: DMF, although comparable to formation of mostly the (tmse)zTM imidazolide (ZZI), alpyridine for the initial solubilization of proteins, was poor for though some (tmse)TMand ( t m ~ e ) ~ T products M are also reagent stability and permits protein aggregation (17); di- generated. To purify the (tmse)zTM acid (ZV) the activating methyl sulfoxide, the solvent of choice for complete protein group is hydrolyzed off by an aqueous quench, and the redenaturation (17), generally reduced the rate of acylation of quired (tmsefZTMacid is then easily obtained by a selective the reagent (data not shown). extraction procedure. The purified (tmse)zTM acid was subOnce the protein is dissolved under anhydrous conditions, jected to NMR analysis, and itconfirmed the presence of two it can be acylated by a variety of acylating agents. A commonly trimethylsilylethyl groups for every trimesic acid molecule used reagent for derivatizing lysine is succinic anhydride, and (data not shown). The activation of the (tmse)zTMacid to its this was expected to acylate all the hydroxyl groups of the symmetric anhydride was achieved by the addtion of cycloprotein under these conditions. To study the reaction quantitatively, DNP-peptides were used and the reactions moni- hexylcarbodiimide,and itsreactivity was enhanced by a factor tored by HPLC. It was found that the peptides were readily of 1000 by the addition of the acylation catalyst PPY (19). The time course of acylation using (tmse)zTM anhydride acylated by succinic anhydride. The standard acylation conditions result inthe succinylation of the serine and threonine prepared as described under “Materials and Methods” was residues of apobacteriorhodopsin as judged by the extent of investigated using HPLC analysis of DNP-Pro-X-NHz pep[‘4C]succinic anhydride incorporation (18). However, it was tides (where X = Ser, Thr, or Gly-Tyr). The standardreaction found that the succinate esters were unstable under aqueous conditions, described under “Materials and Methods,” reconditions; these esters were cleaved spontaneously even at sulted in complete and essentially instantaneous acylation of the tyrosine and serine peptides, while the threonine peptide neutral pH. A large number of possible reagents was therefore consid- required 4 min for complete acylation (Fig. 2). The presence ered to avoid the drawbacks of succinylation. The reagent of 20 mM Bu4N-trifluoroacetate, which is necessary for the chosen was trimesic acid (1,3,5-benzenetricarboxylicacid) on solubilization of proteins, was observed to consistently result the basis of the following considerations: 1) the cyclization in a small decrease in the rate of acylation (closed symbols). responsible for the hydrolysis of the succinate ester is steri- This decrease correlates with the decrease in the rate observed

7992

Water Solubilization of Membrane Proteins 0

II

‘3% + 3 Im-H

3 Im-C- Im __L_,

DMF Im

I

II

Im

1

2 Me3 SiCH2CH2 OH

FIG. 1. Synthesis of (tmse)zTM acid. In this scheme, Zm represents imidazolyl and Me3SiCH,CH2 denotes a 2(trirnethylsily1)ethylgroup. Me3SiCHfH2-0,cB0

I

1 . DMF f“-----

2. NaOH(aq)

&,P I

I

Me3SiCH $H

+ 2 Im-H

C

Im

0 I

m

0

2

4

6

8

TIME (min) FIG. 2. Time course of DNP-peptide trimesylation. DNPprolylthreoninamide was trimesylated at a concentration of 0.5 m M with 100 m M (trnse)*TM anhydride in the presence (circks) and absence (triangles) of 100 mM PPY and in the presence (cbsed symbols) or absence (open symbols) of 20 mM BQN-trifluoroacetate. At the indicated times, the reaction was quenched with aqueous DMF (10% HQO)and analyzed for the extent of trimesylation by HPLC using similar conditions to those specified in Fig. 3.

when the polarity of the pyridine is altered by the addition of BQNCl (data not shown). The synthesis of [3H]trimesicacid was performed to provide radioactive reagent for the direct evaluation of protein derivatization and of the stability of the derivatives. It had been shown that the aromatic protons of benzoic acid can be exchanged with water at 275 “C under acid catalysis (21, 22); this method was therefore used to tritiate thetrimesic acid as described under “Materials and Methods.” Recrystallization of the product from a minimum volume of boiling water afforded the tritiated trimesic acid; analysis by HPLC indicated a single peak of radioactivity co-eluting with authentic trimesic acid. The synthesis andacylation of model peptides was routinely monitored by reverse phase HPLC. The use of DNP-peptide derivatives permitted detection of all peptide-containing products formed, and theintegration of chromatograms monitored at 365 nm was used as thebasis of all quantitation.The purity of each of the synthetic peptides was assessed by HPLC as shown in Fig. 3A for the DNP-Pro-Ser-NH2; the peptides had all been purified by preparative HPLC and were thus better than 99% pure asjudged by their elution profile monitored a t 365 nm. The characterization of the acylated peptide and its subsequent deprotection were performed with the reactive imidazolide reagent (Fig. 1; IZI) in DMF; however, identical peptide products were obtained with the symmetric anhydride and PPY in pyridine. Acylation by (tmsehTM imidazolide in DMF in the presence of 1 M Et3N resulted in the complete disappearance of starting material and the concomitant formation of a single very hydrophobic product (Fig. 3B). Although the results are shown only for the serine peptide, comparable results were observed for all three model peptides using identical procedures and the same analytical system. That the new hydrophobic product was trimesylated was demonstrated in two ways. Firstly, the hydrophobic peptide product was very sensitive to B a N F as shown in Fig. 3C;

Water Solubilization of Membrane Proteins

I

0

20

10

30

TIME (min)

BI

i

?

1 10

20

30

TIME ( m i d

n

f

VI (D

2

Y

g o

a

D CT

8m a

0.

1

0

10

20

1

30

TIME ( m i d

FIG. 3. Tritiated trimesylation of DNP-prolylserinamide. The representative HPLC chromatograms in A, B, and C were obtained by injecting samples containing 10 nmol of DNP-prolylseri-

7993

since this is a highly selective reagent for the removal of the trimethylsilylethyl blocking group and a brief reaction with this reagent converted the hydrophobic product to a polar one, it was concluded that the blocking groups were in fact attached to the peptide. The results show that under the conditions used the conversion is essentially complete, and no protected starting material remains under the conditions used for Fig. 3. The use of only 0.5 equivalent of the B b N F in this reaction resulted in partial deprotection, generating an intermediate product having a retention time of 8.6 min; the addition of more B-NF resulted in the complete conversion of this product to one that eluted at 5.3 min. We conclude thatthe intermediatehad one of the trimethylsilylethyl groups still remaining andthatthe hydrophobic product eluted at 15.9 min isthe desire product, i.e. the peptide acylated by the (tmse),TM acid. Secondly, to confirm the trimesylation of the serine peptide the tritiated(tmse),TM imidazolide wasused for the acylation (Fig. 3B). In this case, after completion of the acylation the excess reagent was quenched by the addition of trimethylsilylethanol. The figure indicates that in the case of the reagent blank (open circles) this results in the complete conversion of the (tmse)nTMimidazolide to the (trnseLTM acid. In thepresence of peptide, a single new peak of radioactivity is observed (closed circles) to coincide with the new peptide product formed. The amount of radioactivity in this peak confirms the incorporation of 1 mol of trimesic acid/mol of DNP chromophore into thisproduct. These resultsthus demonstrate that thehydrophobic product obtained contains one molecule of trimesic acid and two molecules of the trimethylsilylethyl substituentsper molecule of peptide and is the desired trimesylated peptide. The deprotection of the radioactive trimesylated peptide is shown in Fig. 3C. The entire quenched reaction mix depicted in B was subjected t o B h N F treatment to remove the protecting groups. In thecase of the reagent blank (open circles), the (tmse)sTM acid disappeared completely and asingle product was eluted in the position of TM acid. In the peptide case (closed circles), the hydrophobic product completely disappeared and a new polar product was observed in addition to the expected TM acid. The amount of radioactivity found in this peak corresponded exactly to the attachment of a single trimesyl group. Similar results were obtained with the threonine and tyrosine containing peptides. These resultsshow that the BhNF treatment results in the complete removal of the trimethylsilylethyl groups and produces the desired peptide product acylated by TM acid. The reagent thus acylates peptide hydroxyl groups and produces namide at different stages during the trimesylation reaction onto an equilibrated pBondapak CIS reverse phase column developed as indicated in A . Tritiated (tmse),TM imidazolide (100 mM) was utilized as the acylation reagent in the presence of 1 M EtsN in anhydrous DMF. The retention time of the pure DNP-dipeptide prior to acylation is shown in A . The results in B were obtained when the DNPdipeptide was injected after complete trimesylation and following reagent inactivation by excess Me3SiCHtCH20H. Fractions were collected a t 0.8-min intervals, diluted with Amersham aqueous counting scintillant, and the radioactivity determined using a Beckman LS-7800 scintillation counter; the results shown were obtained when the reaction was performed in the presence (0)and in the absence (0)of the DNP-prolylserinamide. The chromatogram in C was obtained after the acylated DNP-dipeptide was deblocked by excess BbNF. As in E,fractions were collected at 0.8-min intervals, and the radioactivity in each was determined. In B and C, i, ii, iii, and iv indicate the relative positions at which TM, (tmse)TM, (tmse),TM, and (tmse)aTM acid, respectively, are eluted under these HPLC conditions.

Water Solubilization of Proteins Membrane

7994

two carboxylate ions where there was only one hydroxyl group previously. The radioactive (tmse)2TManhydride was used to derivatize delipidated apobacteriorhodopsin and the time course of acylation monitored as described in thelegend to Fig. 4. The acylation of the protein proceeds very rapidly at first and then slows down somewhat, finally reaching completion at about 1.5 h. This change in rates reflects the different reactivities of the functional groups on the protein and was observed in the case of the model peptides as well. Thus, lysine and tyrosine residues would be expected to be fully modified at the first time point (5 min); the second stage, completed by about 15 min, reflects acylation of the serine side chains; the third, much slower rate is due to acylation of the threonine residues as expected on the basis of the rates observed with the model peptide which required 4 min for completion. The additional time required in the case of the protein is likely due to theincreased steric hindrance inherent in the molecule. The extent of modification was based on the totalamount of radioactivity incorporated for a known amount of protein and shows that complete modification of the protein is obtained. The trimesylated apobacteriorhodopsin reaction was quenched with Me3SiCH2CH20H (0.6M) and then applied to a Sephadex LH-60 column (1x 26 cm) that was equilibrated and run in DMF. The acylated protein eluted near the void volume, and, after concentrating the appropriate fractions, the protein product was deprotected under standard conditions. The deprotected protein was concentrated under reduced pressure, dissolved in formic acid/ethanol (5:95, v / ~ ) , and applied to a Sephadex LH-20 column that was equilibrated and run in formic acid/ethanol (5:95, v/v). The fractions elutedat thevoid volume were pooled and concentrated I

I

I

1

1

0

-

-

75

100

125

150

EFFLUENT VOLUME (ml)

-

-

50

FIG. 5. Elution profile of trimesyl-apobacteriorhodopsin from a Sephadex 6-160 column. Apobacteriorhodopsin (200 pg) that had undergone trimesylation using the standardconditions specified in Fig. 4 was isolated from the excess (tmseI2TM anhydridethat had been quenched with excess Me&iCH2CH20H by utilizing a Sephadex LH-60 column (1 X 26 cm) that was equilibrated and run in DMF. The acylated protein eluting near the voidvolumewas pooled, deprotected with excess BQNF. The trimesyl-apobacteriorhodopsin was isolated by dissolving it in formic acid and applying it to a Sephadex LH-20 column (1 X 25 cm) that was equilibrated and eluted with 5% formic acid in ethanol. The protein near the void volume wasconcentrated inthe presence of aqueous Na2C03to adjust the pH to8. The protein concentratewas then applied to a Sephadex G-150 column (1X 120 cm) that was equilibrated and eluted with 50 mM NH4HC03, pH 8.5, and 0.02% NaN3. The absorbance of the effluent was monitored a t 280 nm by an Isco absorbance monitor. V, and V, represent the void volume and the totalvolume of the Sephadex G-150 column, respectively.

I

0

25

2oo

t

-

-

40 .-

20

-

-

-

0. 0.0

1

I

I

1.o

I

2.0

I

1

3.0

TIME (h) FIG. 4. Time course of apobacteriorhodopsintrimesylation. Apobacteriorhodopsin was solubilized in anhydrous pyridine (see “Materials and Methods”) and trimesylated under the standardconditions using 100 mM tritiated (trnse)nTM anhydride with a specific activity of 0.7 mCi/mmol and in the presence of 100 mM PPY.At the indicated times, an aliquot of the reaction was removed and diluted into 200 pl of 0.2 M Me3SiCH2CH20Hin pyridine, and after 15 min the excess alcohol was removedunder reduced pressure. Deprotection was achieved by the addition of 2 equivalents of BQNF. The derivatized protein was subsequently diluted to 0.5 ml with DMF and applied to a Sephadex LH-60 column (1 X 26 cm) that was equilibrated and eluted in DMF. Aliquots of the 0.5-ml fractions collected were diluted with Amersham aqueous counting scintillant and the radioactivity determined using a Beckman LS-7800 scintillation counter. Radioactivity eluting near the void volume was determined and plotted as percentage of theoretical yield normalized to the amount of protein recovered as established by amino acid analysis.

K av FIG. 6. Determination of apparent molecular weight of trimesylated apobacteriorhodopsin by gel permeation chromatography on Sephadex G-160. The column (1 X 118 cm) was equilibrated and eluted with 50 mM NH.HCO3, pH 8.5, containing 0.02% sodium azide. Thestandard proteins used are glucose 6phosphate dehydrogenase, M , = 105,000; bovine serum albumin, M , = 66,000; chicken ovalbumin, M , = 45,000; carbonic anhydrase, M,= 29,000; whale myoglobin, M , = 17,000;and cytochrome c, M , = 12,500. The elution behaviour of trimesylated apobacteriorhodopsin is indicated by the open circle.

Water Solubilization of Membrane Proteins in the presence of Na2C03. Thissecond column afforded the removal of the hydrophobic tetrabutylammonium counter-ion that becomes associated with the protein after deprotection. The trimesylated protein residue was then dissolved in HZ0 and subjected to gel permeation chromatography on a standardized column of Sephadex G-150 equilibrated with a simple aqueous buffer (Fig. 5). As shown in Fig. 6, the protein eluted at a position corresponding to a molecular weight of 35,000 which is exactly what was expected on the basis of the total calculated molecular weight of the fully derivatized protein. This extensively derivatized protein not only elutes in its monomeric form from the Sephadex G-150 column but was also shown to run as a single discrete band on an acidic (pH 2) lithium dodecyl sulfate-polyacrylamide system (23) (data not shown). The trimesylated proteinwas observed to have a greater mobility than the bromphenol blue dye in a standard Laemmli gel, pH 6.8 (24),presumably due to itslarge negative charge density. At pH 2, all of the carboxylate ions on the protein should be fully protonated, hence reducing its mobility. The described methods show that complete trimesylation of all serine, threonine, tyrosine, and lysine side chains of a protein is possible; the resulting derivative has good water solubility even when the protein itself is an extremely hydrophobic membrane protein. The solubilization of membrane proteins in the absence of detergents or chaotropic agents by this derivatization provides a new tool for structural studies of hydrophobic membrane proteins. In particular, this method promises to be useful for enzymatic removal of pyroglutamyl residues from the NH2 terminus (data not shown). The susceptibility of these derivatives to other proteases as well as their behavior in a spinning cup sequencer are being determined to assess the general applicability of the approach. REFERENCES 1. Mahoney, W. C., and Hermodson, M. A. (1979) Biochemistry 1 8 , 3810-3814

7995

2. Takagaki, Y., Gerber, G. E., Nihei, K., and Khorana, H.G. (1980) J. Biol. Chern. 255,1536-1541 3. Gerber, G.E., Anderegg,R. J., Herlihy, W.C., Gray, C. P., Biemann, K., and Khorana, H. G. (1979) Proc. Natl. Acad. Sci. U. S. A . 76,227-231 4. Gerber, G. E., and Khorana, H. G. (1982) Methods Enzymol. 88, 56-74 5. Regnier, F. E.(1983) Methods EnzymoZ. 9 1 , 137-190 6. Hunkapiller, M.W., and Hood, L. E. (1978) Biochemistry 1 7 , 2124-2133 7. Bailey, G. S., Gilbert, D., Hill, D. F., and Petersen, G. B. (1977) J. Biol. Chem. 253,2218-2225 8. Ovchinnikov, Y.A., Abdulaev, N. G., Feigina, M. Y., Kiselev, A. V., and Lobanov, N. A. (1979) FEBS Lett. 100, 219-224 9. Ozols, J. (1986) J. Biol. Chem. 2 6 1 , 3965-3979 10. Carrington, D. M., Auffret, A., and Hanke, D. E. (1985) Nature 3 13,64-67 11. Gounaris, A. D., and Perlmann, G. E. (1967) J. Bwl. Chem. 2 4 2 , 2739-2745 12. Ozols, J., Heinemann, F. S., and Johnson, E. F. (1985) J. Biol. Chem. 260,5427-5434 13. Thomas, A. B., and Rochow, E. G. (1957) J. Am. Chem. Soc. 79, 1843-1848 14. Lipshutz, B. H., and Pegram, J. J. (1980) Tetrahedron Lett. 2 1 , 3343-3346 15. Gerber, G. E., Gray, C. P., Wildenauer, D., and Khorana, H. G. (1977) Proc. Natl. Acad. Sci. U. S. A . 74,5426-5430 16. Oesterhelt, D., Meentzen, M., and Schuhmann, L. (1973) Eur. J. Bwchem. 40,453-463 17. Llinas, M., DeMarco,. A.,. and Lecomte, J. T. J. (1980) . . Biochemistry 19,1140-1145 18. Morton. R. C.. and Gerber. G. E. (1981) Can.Fed.Biol.Sci. (Abstr. 469) ' 19. Holfe, G., Steglich, W., and Vorbruggen, H. (1978) Angew. Chem. Znt. Ed. Engl. 17, 569-583 20. Guibe-Jampel, E., LeCorre, G., and Wakselman, M. (1979) Tetrahedron Lett. 1 3 , 1157-1160 21. Werstiuk, N. H., and Kadai, T. (1973) Can. J. Chem. 5 1 , 14851486 22. Werstiuk, N. H., and Kadai, T.(1974) Can. J. Chem. 5 2 , 21692171 23. Jones, G.D., Wilson, M. T., and Darley-Usmar, V. M. (1981) Biochem. J. 1 9 3 , 1013-1015 24. Laemmli, U. K. (1970) Nature 227,680-685 ~I

~~