functional expression and site-directed mutagenesis of a synthetic

THE JOURNAL OF BIOLWCAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 269, No. 15, Issue of April 15, pp. 11178-11185, 1994 Printed in U.S.A.

Functional Expression and Site-directed Mutagenesis of a Synthetic Gene for a-Bungarotoxin*

(Received for publication, December 21, 1993)

Julie A. RosenthalS, Sigmund H. HSU, Dinesh Schneider, Lisa N. Gentile+, Norma J. Messier, Charles A. Vaslet, and Edward Hawrot9 From the Section of Molecular and Biochemical Pharmacology, Division of Biology and Medicine, Brown University, Providence, Rhode Island 02912

In order to explore the structure-function relationships of the curaremimetic a-neurotoxins we have constructed and cloned a synthetic gene for Bungarus mul- ticincfus a-bungarotoxin which is expressed in Escherichia coli. The recombinant a-bungarotoxin is expressed as a fusion protein with a-bungarotoxin linked to the COOH-terminal end of the T7 Gene 9-encoded coat protein. After treatment of the fusion protein with Fac- tor Xa protease, a recombinant a-bungarotoxin is released that co-migrates with authentic a-bungarotoxin upon reverse-phase high performance liquid chromatography and ion-exchange chromatography. Final yields of active recombinant a-bungarotoxin were about 0.4 mgfliter of starting bacterial culture. The recombinant a-bungarotoxin contains 10 additional residues linked to the NH,-terminal Ile of the a-bungarotoxin sequence due apparently to the inaccessibility of the engineered cleavage site to Factor Xa. Nevertheless, the recombinant a-bungarotoxin is capable of binding to the nicotinic acetylcholine receptor with an apparent affinity that is only decreased “1.7-fold from that of authentic a-bungarotoxin. Alanine substitution of a residue, Asps0, highly conserved among a-neurotoxins and previously suggested to play a key role in receptor recognition, resulted in a recombinant a-bungarotoxin whose receptor binding activity is indistinguishable from authentic a-bungarotoxin.

The snake venom-derived a-neurotoxins comprise a family of over 70 polypeptides of known primary amino acid sequence (Endo and Tamiya, 1987). The cy-neurotoxins are traditionally classified as either “short,” with 60-62 residues, or “long,” with 70-74 residues, and produce their physiological effect, skeletal muscle paralysis, by binding to the postsynaptic nicotinic acetylcholine receptor (nAChR)’ thereby preventing the normal

* This work was supported in part by National Science Foundation Grant BNS 90-21227, National Institutes of Health Grant GM32629, and a Howard Hughes Medical Institute Medical Student Research Training Fellowship (to S. H. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “udvertisemenP in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Training Grant GM07601 and was done in partial fulfillment of the $Supported in part by National Institutes of Health Predoctoral

requirements for a Ph.D. degree from Brown University (to J. A. R. and L. N. G.).

and Biochemical Pharmacology, Div. of Biology and Medicine, Brown 6 To whom correspondence should be addressed: Section of Molecular

University, Providence, RI 02912. Tel.: 401-863-1283; Fax: 401-863- 1595.

The abbreviations used are: nAChR, nicotinic acetylcholine receptor; Bgtx, a-bungarotoxin; Dm, dithiothreitol; IC,,, concentration resulting in a 50% reduction of the binding of radiolabeled authentic Bgtx to the nAChR; MES, 2-(N-morpho1ino)ethanesulfonic acid; MeSH,

activation of this receptor by the neurotransmitter, acetylcholine. These curaremimetic a-neurotoxins, together with the K-neurotoxins, which bind to a neuronal form of the nAChR (Chiappinelli, 1993; Oswald et al., 1991), and the more dis- tantly related cardiotoxins constitute a large family of snake venom toxins with a conserved protein fold motif. These toxins share five invariant or conserved residues (Tyr-PheZ4, Gly43, Pro4’, Asp-Gld3, and A d 6 ) and four core disulfide bridges which together are presumed to be important in the formation of the characteristic tertiary structure consisting of three large main-chain loops (Endo and Tamiya, 1987; Menez et al., 1992; Chiappinelli, 1993).

One of the most intensely studied of the long a-neurotoxins is a-bungarotoxin (Bgtx) which is obtained from the venom of the banded krait, Bungarus muZticinctus. As a result of its ex- tremely high affinity for the nAChR, Bgtx has provided a criti- cally important biochemical tool for both the purification and the detailed functional and structural analysis of the nAChR (Stroud et al., 1990; Changeux et al., 1992; Chiappinelli, 1993). The x-ray crystal structure of Bgtx has been reported (Agard and Stroud, 1982; Love and Stroud, 1986) and refinements to the structure have been made based on two-dimensional NMR experiments (Basus et al., 1988; Kosen et al., 1988). Bgtx is 74 amino acids in length (molecular weight of about 8,000) and contains a fifth disulfide bond near the tip of loop 2. Bgtx, apart from the COOH-terminal tail, is a relatively flat molecule with all three main-chain loops roughlyin one plane (e.g. see Fig. 6) covering an area of about 40 x 30-A with a depth of about 20 A (Love and Stroud, 1986). The two strands of the middle polypeptide loop together with one strand from the third (COOH-terminal) loop form a triple-stranded, anti-parallel p sheet which is the major secondary structure found in the molecule and which is a common feature throughout this family of proteins.

There are 9 residues that are conserved in the short and long a-neurotoxins but not in the functionally unrelated cardiotoxins (Endo and Tamiya, 1987). These amino acids are believed to contribute either directly or indirectly to the high specificity and affinity of the neurotoxins for the nAChR. It has been proposed that two of these conserved residues, Asp3’ and in Bgtx, form an ion pair near the tip of loop 2 that mimics the structure of acetylcholine thus contributing to toxin specificity (Tsernoglou et al., 1978). Sequence comparisons within this family of toxins and chemical modification studies have not, however, been able to determine conclusively the identity of all of the functionally relevant residues in the neurotoxins (Endo

2-mercaptoethanol; NOE, nuclear Overhauser effect; PBS, phosphate- buffered saline; PCR, polymerase chain reaction; rBgtx, recombinant Bgtx (BGTXlM3 construct); D30A, rBgtx with Ala substituted for Asp3’ (BGTXlM3-D3OA construct); RP-HPLC, reverse-phase high performance liquid chromatography; FPLC, fast protein liquid chromatography.

11178

Site-directed Mutagenesis of a-Bungarotoxin 11179

and Tamiya, 1987). More recently, several groups have taken advantage of heterologous expression systems in Escherichia coli to produce recombinant forms of neurotoxins and other ion channel toxins that would be amenable to site-directed mutagenesis studies of structure-function relationships (Howell and Blumenthal 1989; Ducancel et al . , 1989; Boyot et al . , 1990; Park et al . , 1991; Fiordalisi et al . , 1991, 1992; Gallagher and Blumenthal, 1992; Herve et al . , 1992; Pillet et al . , 1993). In one such study of a number of site-directed mutations introduced into the short a-neurotoxin, erabutoxin a, it was shown that substitution of a His for the conserved Asp residue corresponding to Asp3' in Bgtx produces a 46-fold reduction in the apparent Kd for the nAChR without producing a significant effect on secondary structure (Pillet et al . , 1993).

A second emerging approach towards determining those toxin residues important for receptor recognition involves the direct structural elucidation of high affinity complexes formed between Bgtx and peptide fragments that correspond to the cognate sequence in the nAChR forming the major determinant of the toxin binding site (Wilson et al . , 1985). We recently reported an NMR-derived solution structure for the stoichiometric complex formed between Bgtx and a dodecapeptide (a185- 196) derived from the primary sequence of the a-subunit from the nAChR of Torpedo californica (Basus et al . , 1993). In this complex, 10 Bgtx residues, including Asp3', were shown on the basis of intermolecular NOES to participate in the "contact zone" between the receptor peptide and Bgtx. Further NMR analysis and structural refinement of this complex as well as of those formed with longer peptide fragments of the a-subunit would benefit from the availability of a recombinant Bgtx which could be labeled metabolically with 15N in a simple heterologous expression system. In this paper, we report the design and construction of a synthetic gene for Bgtx. We show that this synthetic gene can be expressed at high levels in E. coli as part of a fusion protein. Proteolytic cleavage of the isolated fusion protein results in a recombinant form of Bgtx that co-migrates with authentic Bgtx upon RP-HPLC and ion- exchange chromatography. We also utilize a "charged to alanine" scanning mutagenesis strategy (Gibbs and Zoller, 1991; Bass et al . , 1991; Wells 1991) to explore the role of the charged side chain of Asp3' in directing receptor recognition.

MATERIALS AND METHODS Construction of the Synthetic Gene for a-Bungurotoxin-A synthetic

gene for Bgtx was designed by back-translating the primary amino acid sequence of native Bgtx and then optimizing the codon usage for expression in E. coli. To obtain optimal expression of the rBgtx we used the Gene 9-fusion protein strategy first described by Howell and Blu- menthal (1989). Plasmid pSR9 (Protein Polymer Technologies, Inc.), kindly provided by K. M. Blumenthal, is a derivative of pBR322 containing a copy of the structural gene for the Gene 9 coat protein of T7 bacteriophage. Expression of Gene 9-fusion proteins are readily induc- ible in the E. coli host strain BL21(DE3) (Studier and Moffatt, 1986) which contains the structural gene for T7 bacteriophage RNA polymerase integrated by A into the host chromosome of BL21, and under lac repressor control through the lacUV5 promoter.

To create the desired synthetic gene a set of six overlapping oligonucleotides, A to F, ranging from 28 to 93 bases long were synthesized (Applied Biosystems 380A DNA synthesizer) and purified by polyacrylamidehrea gel electrophoresis followed by elution in 0.5 M ammonium acetate. Oligonucleotide A, a 28-mer (5"CATGCGTAACGT- GTCGACATCGAGGGTC), corresponds to the junction between the COOH-terminal polylinker of Gene 9 and the Factor Xa protease recognition site, IEGR, at the NH,-terminal end of the Bgtx synthetic gene construct (see Fig. 4 for amino acid sequence in this region). Oligonucleotide B, an 83-mer (5"GGCGGGCAGGTAACAGCGGAGA- TCGGGGAGGTAGCGGTGGTGTGGCAAACAATACGACCCTCGATm WACG'ITACGCATG), has a 28-base complementary overlap with oligonucleotide A at its 3'-end. Oligonucleotides A and B include the polylinker SalI site (GTCGAC) as well as 12 extra flanking bases to facilitate restriction endonuclease digestion of PCR products. Oligo-

nucleotide C, a 71-mer (5'-G'ITACCTGCCCGCCGGGTGAAAACCTGT- GCTACCGTAAAATGTGGTGCGACGCTTTCTGCTCCTCCCGTGG), has a 14-base complementary overlap with oligonucleotide B. Oligo- nucleotide D, an 81-mer (5'-CAGCAGGTAAC'ITCITCGTACGG?Tm- TGGACGGGCAGGTAGCAGCGCAACCCAGTTCAACAACTTTACCA- CGGGAGGAG), has a 12-base complementary overlap with oligonucleotide C. Oligonucleotide E, a 75-mer (5'-AG'ITACCTGCTGCTCCACC- GACAAATGCAACCCGCACCCGAAACGTCAGCCAGGTTAGMS- TTATGCGTAACGT), has a 12-base complementary overlap with oligonucleotide D. Oligonucleotide F, a 28-mer (5'-ACG'ITACGCATMGCT TCTAACCTGGCT), is fully complementary with the 3'-end of oligonucleotide D. Oligonucleotides E and F contain the HindIII site (AAGC'IT) distal to the translational stop codon at the COOH-terminal end of the coding region (74 amino acids) for Bgtx. Both oligonucleotides E and F contain 11 extra bases flanking the HindIII site to facilitate restriction endonuclease digestion of PCR products. Oligonucleotides A and F also served as convenient flanking primers for PCR amplification of the synthetic gene.

Equimolar mixtures of oligonucleotides (A, B, C and D, E, F) were annealed in two half-reactions at 80 "C for 10 min, then cooled to 23 "C overnight. The two half-reactions were then combined, heated to 25 "C for 20 min, and cooled to 23 "C overnight. The single-stranded gaps between the overlapped oligonucleotides were filled in by Taq polymerase (Promega) followed by T4 DNA ligase (Life Technologies Inc.). The final construct was amplified using standard PCR procedures with oligonucleotides A and F as flanking primers. The PCR product was isolated by agarose electrophoresis and then digested with SalI and HindIII according to the manufacturer's specifications (Life Technolo- gies Inc.). The doubly digested fragment was ligated into a SaWHin- dIII-digested pSR9 plasmid at 15 "C. HBlOl competent cells (Life Tech- nologies Inc.) were transformed with the ligation mixture and single colonies were screened for insert by PCR using oligonucleotides A and F. One of the isolates, designated pBGTX1, was sequenced by double- strand DNA cycle sequencing (Life Technologies Inc.) and confirmed by M13 single-strand DNA sequencing. The synthetic gene for Bgtx in pBGTXl was found to contain a 10-base repeat located within the overlap region between oligonucleotides D and E. The altered sequence was corrected by recombinant PCR using two complementary oligonucleotides (23 and 24 bases in length) containing the correct sequence. The nucleotide sequence of the corrected construct, pBGTXlM3, was confirmed by double-strand DNA cycle sequencing. Isolated plasmid (Promega, MiniPrep kit) was used to transform the expression strain BL2UDE3) (Novagen) with pBGTXlM3. Except for expression studies in BL21(DE3), the pBGTX class of plasmids were standardly main- tained in the HBlOl host strain for all genetic manipulations and for sequencing purposes.

Construction of theAsp3'+Ala Mutant of a-Bungarotoxin-The D30A mutant was produced from pBGTXlM3 by recombinant PCR using two overlapping internal primers containing the desired mutation, a single base change (GAC to GCC). Primer D30A1 (5'-GTGGTGC[w- C]GClTl'C-TGC[&C]TCCCGTGG-3') and primer D30A2 (5"CCAC- GGGAGC'JGCAGnGCGGCGCACCAC-3') also encode a silent mutation a t t h e S34 codon (double underline region) to introduce a new diagnostic restriction site (PstI) that could be used for identification purposes. Each internal primer was added to the appropriate flanking primer to generate half the mutant by PCR. The two half-reactions were annealed, and the full product was amplified by PCR as with pBGTXlM3. The construct was digested by Sal1 and HindIII and in- serted into pSR9 as described above to form pBGTXlM3-D30A. The sequence of this mutant construct was confirmed by double-strand DNA cycle sequencing.

Fusion Protein Expression-BL2l(DE3) cells were grown at 37 "C to late log phase (A, = 0.7) in Luna broth and induced with 0.5 m~ isopropyl-P-o-thiogalactoside (Life Technologies Inc., ultra-pure grade) for 3 h. Cells were centrifuged, washed in cold PBS (40 ml; all volumes based per liter of starting culture), and resuspended in 10 ml of 10 m~ Tris, pH 8.0, containing 50 m~ NaCl, 2 m~ Na,EDTA, 1 m~ phenyl- methylsulfonyl fluoride, 1 p~ pepstatin, 0.5 p~ leupeptin, 0.2 pg/ml aprotinin, 1 mg/ml lysozyme, 2 m~ D m , and 50 m~ MeSH. The sus- pension was passed through a French Pressure cell (SLM Instruments) a t 20,000 p.s.i. DNase I (Life Technologies Inc.) was added to 1 pg/ml as well as MgCl, and MnCl, to make 10 and 1 mM final concentrations, respectively. After 20 min the extract was clarified by centrifugation (87,000 x g) for 1 h at 4 "C. The supernatant, containing 1&15 mg/ml protein, was collected and after being brought to 0.6 M NaCl by addition of 5 M NaCl, was passed over a Whatman DEAE-52 cellulose column equilibrated with 50 m~ Tris, pH 7.0, containing 0.6 M NaCl to remove nucleic acids. This step was used in place of streptomycin sulfate pre-

11180 Site-directed Mutagenesis of cu-Bungarotoxin cipitation (Howell and Blumenthal, 1989) which we observed to co- precipitate the Rgtx-fusion protein. Saturated ammonium sulfate was added a t 4 "C to the nonadsorbed protein fraction to a final concentration of 50% and the precipitated protein was collected by centrifugation a t 8.000 x g for 10 min. ppically, we obtained about 150 mg of ammonium sulfate-precipitated proteiditer of starting culture. In this fraction, -50% of the total protein appeared to be the desired fusion protein as indicated by SDS-polyacrylamide gel electrophoresis.

Protein Disulfide Refolding-In order to obtain refolded Bgtx by simple dialysis, ammonium sulfate-precipitated fusion protein was resuspended in 50 mu Tris. pH 7, containing 5 mu MeSH and 2 mu Na,E:DTA a t a concentration of 1-4 mg/ml protein and dialyzed against the same buffer a t 4 "C. Following the third buffer change, the dialysis was allowed to proceed overnight and no further MeSH was added. This procedure led to functionally active fusion protrin as measured by competition binding assays but a t about half the final yield achievable with the procedure described below utilizing mixed-disulfide formation (Seely and Young, 1991).

To facilitate a more efficient reformation of the correct disulfide pairs via a mixed-disulfide intermediate, the ammonium sulfate-precipitated fusion protein was resuspended in 50 mu Tris, pH 8.2, containing 100 mu NaCl and 5 mu DTT. to a final protein concentration of 1 mgYml. After 60 min a t 4 "C, mixed disulfides were formed by adding solid I.-cystine-HCI to 50 mu followed by the slow addition, with stimng, of NaOH (10 s) to adjust the pH to 12. After a further incubation for 10 min a t 4 "C, the reaction mixture was diluted 6-fold into 50 my Tris, pH 8.2, 100 mu NaCI, 20 mu MeSH and the pH was adjusted to 8.2 with HCI. The reaction mixture was purged with N, and incubated a t 4 "C overnight.

P urification ofRecornhinant Bgfx-The refolded fusion protein could be purified by chromatography on Pharmacia Q-Sepharose Fast-Flow equilibrated in 50 m.w Tris, pH 7.0, containing 50 mx! NaCI. A linear gradient of NaCl (50 mu to 0.6 u in 10 column volumes) was used to elute the remainder of the adsorbed fusion protein. The major fraction eluted from the column a t a NaCl concentration of about 0.3 M similar to rrsults reported by Howell and Blumenthal ( 1989). This fraction was either dialyzed against 20 mu Tris, pH 8.0, containing 100 mu NaCl or the buffer was changed using a Millipore Centriprep filtration unit. Preliminary studies in which we monitored Factor Xa cleavage offusion protein, by the appearance of material co-migrating with Bgtx upon reverse-phase HPLC, suggested an optimal enzymatic digest time of 16 h. After digestion with Factor Xa protease (New England Biolabs; a t 1% enzyme/substrate a t 23 "C for 16 h in the presence of 2 mu CaCI,), the reaction mixture was applied to a Pharmacia Mono-Q column and the material not retained by the column, containing the released recombinant Bgtx which is a basic protein, was collected. This fraction was then purified by reverse-phase HPLC using a C, column.

Refolded fusion protein could also be purified in a batch procedure using an anion-exchange resin (Macroprep Q, Rio-Rad). The eluted fusion protein was concentrated by an Amicon pressure flow cell (YM30 membrane 50 p.s.i). The fusion protein was dialyzed against 20 mu Tris, pH 8.0, containing 100 mu NaCl and treated with Factor Xa protease as described above. After a 16-h digest, MES salt was added to the enzyme digest to adjust the pH to 5.5, and the recombinant toxin was isolated by FPLC cation-exchange chromatography (S-Sepharose, Pharmacia LKI3 Biotechnology Inc.). The recombinant toxin was eluted in 50 mu NaOAc, pH 5.5. with a 5-15? gradient of 1 u NaC1. The protein fraction eluting at about 132 mu NaCl was lyophilized and then desalted using a PD-10 gel-filtration column (Pharmacia) pre-equilibrated with 0.5% NH,HCO,, pH 7.5.

The Ellman's reaction (Habeeb, 1972) was used to determine the amount of free sulfhydryl contained in 0.05 pmol of purified rBgtx. Under the condition used, we could readily detect 4 nmol of MeSH. We found less than 0.1 mol of free -SH per mol of rBgtx.

Characterization ofRecornhinant Bgtx-Protein concentrations were determined using a modified Lowry assay (Bio-Rad, immunoglobulin G as standard) or Bradford assay (Bio-Rad, bovine serum albumin as standard) when reducing agents were present. SDS-polyacrylamide gel electrophoresis was performed with 7.5% homogeneous gels or with high density gels (20% acrylamide. 30% ethylene glycol) using the pHast System (Pharmacia). Purified fractions were also analyzed for homogeneity by HPLC reverse-phase chromatography (C, column, Vy- dacl. The column was equilibrated in 0.lor trifluoroacetic acid and elution was performed using a trifluoroacetic acid/acetonitrile/water gradient of 20-60% acetonitrile. The final purified fractions were further analyzed by electrospray mass spectroscopy (Yale Cancer Center Mass Spectroscopy Facility). The electrospray source (Anal-flica, Branford, CTI was used in conjunction with a triple quadrupole instrument from

plasmid pSR9. The DNA sequence for the synthetic gene encoding FIG. 1. Construction of the Gene 9-a-bungamtoxin fusion in

Rgtx was derived from back-translation of the primary amino acid sequence. Codons were optimized for use in E. coli RL211DE3) cells. A Factor Xa recognition site (IEGR) was engineered flanking the 5'-end of the Bgtx sequence. The entire region was ligated into the pSR9 plasmid after HindIII-Sal1 endonuclease digestion.

BG Biotech (Manchester, United Kingdom). The sample concentration was 0.1 pg/pl in 5 0 4 methanoVwater. NH,-terminal sequencing was performed using standard procedures by the W. M. Keck Foundation Biotechnology Resource Laboratory a t Yale University School of Medi- cine.

The binding activity of recombinant toxin was determined in a competition binding assay using '2'I-labeled Bgtx and acetylcholine receptor-enriched membrane preparations. The plasma membrane fraction was obtained from frozen electric organs of T . cnlifornica (Pacific Bi- omarine, Venice, CA) essentially as described by Czajkowski et al. (1989). For use in binding assays, Torpedo membranes were diluted to 10 pg/ml in PBS. Flat-bottomed 96-well microtiter plates (Nunc, Immu- nosorb) were incubated with 0.1 ml of diluted membranes and centrifuged a t 1.000 x g. The wells were washed three times with 100 pl of PBS and quenched with 0.2 ml of 2% bovine serum albumin (Roehringer Mannheim, Fraction V) in PBS for 1 h a t room temperature. Wells were washed twice with 200 pl of 0.2% bovine serum albumin in PBS and 0.1 ml of prepared samples or authentic Bgtx standards was added. Snake venom-derived Bgtx was obtained from Miami Serpentarium (Miami. FL). The serially diluted samples and standards were prepared in 0.2% bovine serum albumin to a volume of 0.125 ml and added to 0.125 ml of '"Mabeled Bgtx (Amersham, 3000 Ci/mmol) diluted to contain 100,000 cpdwell. The final concentration of "'I-labeled Bgtx in the assay is about 3 nu. Following a 2-h incubation a t room temperature, the wells were washed 4 times with 0.24 bovine serum albumin in PRS. The remaining '2SI-labeled toxin bound to receptor was removed by the addition of 100 pl of 2.59 SDS, 0.25 s NaOH, and wiping the wells twice with cotton swabs, and the bound radioactivity was determined in a y-counter. Nonspecific binding in the presence of a large excess of unlabeled Bgtx was usually less then 10% of the total binding and the binding data were not corrected for this low level of nonspecific binding. The competition curves and resulting IC, values to fit the data were generated by nonlinear curve-fitting using the commercial program ENZFIT (Elsevier, The Netherlands) modified to analyze single-site competition data.

RESULTS

Bacterial Expression Construct-We undertook to express Bgtx in E. coli as a fusion protein appended to the COOH- terminal portion of the T7 Gene 9 coat protein. This expression system has the advantage that it avoids the formation of in- soluble inclusion bodies commonly formed in many other bacterial overexpression systems. The Gene 9 protein is a highly acidic protein which can be overexpressed to high levels in E. coli and which can be readily isolated and purified by anion- exchange chromatography (Howell and Blumenthal, 1989). The Gene 9 protein offers the further advantage that it is free of cysteine residues that could otherwise interfere with the reformation of the five disulfides in Bgtx. Fig. 1 describes the major features of the Gene 9-fusion protein expression system. The specific design and construction of the synthetic gene for Bgtx and its attachment to Gene 9 is detailed under "Materials and


Methods." The final construct, pBGTXlM3, was designed to have the peptide sequence IEGR, placed immediately NH,- terminal to the Bgtx sequence. The "restriction" protease, Fac- tor Xa, has been shown in many other fusion-based expression systems to cleave the peptide bond on the COOH-terminal side of the Arg residue within the sequence IEGR. Control experiments with authentic Bgtx indicated that Factor Xa did not produce any cleavage within the 74-amino acid sequence of Bgtx (data not shown).

Production of Recombinant Toxinsane of the major ques- tions in the feasibility of the production of recombinant toxin was the efficiency and accuracy of the refolding of cu-bungarotoxin's five disulfide bonds. To determine if refolding of Bgtx from the unfolded state was possible, authentic Bgtx was denatured in either urea or guanidine in the presence of 50 m~ DIT and allowed to renature by dialysis against buffers containing either MeSH or DIT at 1-5 m ~ . In preliminary studies we determined that dialysis against MeSH or DIT led to a recovery of from 7 to 17% of the initial activity as measured by a binding displacement assay using Torpedo membranes enriched in nAChRs (data not shown). In contrast, recoveries following treatment with a redox mixture of glutathione or with air oxidation were 51-2%. Recoveries of activity following simple dialysis to remove denaturant in the absence of a reducing agent were typically less than 1%. The demonstration that reduced and denatured native Bgtx could be refolded with a yield of up to 17% made us more confident that a recombi- nantly expressed Bgtx could be successfully refolded.

Both the total crude extract and the soluble, high-speed centrifugation supernatants from induced cultures show expression levels of the Bgtx-containing fusion protein similar to those previously described for other Gene 9-fusion proteins containing somewhat shorter polypeptide appendages (Howell and Blumenthal, 1989; Park et al., 1991). As judged by Coomassie staining of SDS-polyacrylamide gels and protein determinations of the various fractions, greater than 90% of the recombinant fusion protein was recovered in the supernatant following centrifugation of extracts at approximately 100,000 x g. The Bgtx-containing fusion protein was purified further by ammonium sulfate precipitation and chromatography on Q-Sepharose. About half of the total fusion protein, as judged by SDS-polyacrylamide gels, did not adsorb to the column under these conditions or when re-applied to a re-equilibrated column. Although the run-through fraction had detectable nAChR binding activity and appeared to contain intact fusion protein as assessed by SDS-gel electrophoresis, there was no increase in specific binding activity after Factor Xa digestion of this material. No further characterization of this material was attempted. From a 1-liter culture grown in a Fernbach shaker flask, we typically recovered approximately 35 mg of highly enriched fusion protein following salt elution from the anion- exchange column. The Bgtx-containing fusion protein represents the majority of the total protein in this fraction as judged by Coomassie staining of SDS-polyacrylamide gels (data not shown). The rBgtx-fusion protein at this stage of purification has demonstrable nAChR binding activity as determined by competition of 1251-Bgt~ binding. In these assays, the rBgtx- fusion protein exhibited an IC,, of 2.1 x lo-, M in comparison to authentic Bgtx with an average IC,, of about 1.5 x lo-@ M under the assay conditions used. The slope of the binding displacement curves suggested that the active fusion protein was apparently homogenous in its binding activity (data not shown). The 1,000-fold reduction in binding affinity of the recombinant Bgtx-containing fusion protein is greater than the 70-100-fold reduction in the affinity reported for a recombinant erabutoxin- fusion protein (Ducancel et al., 1989). The greater reduction in activity seen with the Bgtx-Gene 9 fusion protein may be due to

I * ~ . ' . ' . " ' . I 19 20 21 22 23 24 25

L . ~ . ' " . ' . ' . l 19 20 21 22 23 24 25

I . l . 1 . ' . ' . ' . I

19 20 21 22 23 24 25

Time (min) FIG. 2. Reverse-phase HPLC elution profiles of native and rB-

gtx. Samples (10 pg) were injected onto a C, column equilibrated with 0.1% trifluoroacetic acid and eluted with a linear gradient of 2040% acetonitrile beginning at 7 min aRer sample injection. A, rBgtx purified by Q-Sepharose FPLC following Factor Xa cleavage, as described under "Materials and Methods." B , authentic Bgtx. C, authentic Bgtx and rBgtx were combined (5 pg of each) and co-injected.

electrostatic interactions between the numerous basic residues in Bgtx and the highly negatively charged Gene 9 protein.

The rBgtx-fusion protein was treated with Factor Xa to release the rBgtx. Initially, we further purified the rBgtx, which is a basic protein, by applying the Factor Xa reaction mixture to a Mono-Q column and collecting the material that did not ab- sorb to the column. The rBgtx from the "run-through" fraction was then purified further by RP-HPLC with authentic Bgtx as a standard. The results shown in Fig. 2 demonstrate that the rBgtx eluted at a position in the acetonitrile gradient nearly identical to that of authentic Bgtx. Upon RP-HPLC, the authentic Bgtx, whose mass and purity have been verified by

11182 Site-directed Mutagenesis of a-Bungarotoxin

8K -

RG. 3. Coomassie stain of a high density SDSpolyacrylamide gel of native and recombinant a-bungarotoxins. The entire gel is shown. Native Bgtx appears as a band of -8 kDa. Both recombinant toxins (rBgtx and the D30A mutant) migrate more slowly upon SDS- polyacrylamide gel electrophoresis consistent with the presence of an additional 10 amino acids at the NH, terminus of the rBgtx. Each lane was loaded with 2 pg of total protein.

mass spectroscopic analysis, elutes as two major peaks of protein. The two forms isolated by HPLC may represent two con- formers of Bgtx that are resolvable by this chromatographic method or the material eluting later in the gradient may represent partially unfolded Bgtx (Corran, 1989). In any case, the rBgtx elutes similarly upon RP-HPLC. To compensate for possible variations in sample injections, authentic and rBgtx were mixed in a 1:l ratio and the mixture analyzed by RP-HPLC (Fig. 2). The similarity of the eluted protein peak profiles for the mixture as compared to both authentic and rBgtx suggests that the rBgtx is very similar to authentic Bgtx in its retention and elution characteristics upon RP-HPLC. In a further comparison of the relative ratio of the material in the two eluted peaks, a larger proportion of the rBgtx was found in the second peak eluting at the slightly greater acetonitrile concentration. Recovery of the eluted samples followed by reconstitution into aqueous buffer allowed us to determine that both the early and late-eluting peaks of protein from both authentic and rBgtx were able to bind to the nAChR with comparable high affinity as determined by a competition binding assay using Torpedo electric organ-derived membranes (data not shown).

A definitive comparison of specific binding activities of the recombinant toxins isolated after RP-HPLC purification was not possible as additional SDS-polyacrylamide gel electrophoresis analysis revealed the presence of a significant level of low molecular weight protein contaminants. We therefore decided to use cation-exchange (S-Sepharose) chromatography as a final purification step. The recombinant Bgtx eluted at the same position as authentic Bgtx (132 mM NaCl) on S-Sepha- rose, whereas the D30A mutant eluted at a slightly higher salt concentration (143 m~ NaCl), consistent with the removal of a negative charge. SDS-polyacrylamide gel electrophoresis of the recombinant proteins purified by S-Sepharose chromatography indicated that the final fractions were highly purified and essentially devoid of contaminating proteins as shown in Fig. 3.

Disulfide Refolding and Final Yields-A significant amount of rBgtx refolding, as evidenced by final recovery of binding activity, could be attained simply by dialyzing the fusion protein against MeSH either before or after the anion-exchange chromatography step. When the FPLC-Mono Q purified fusion protein was simply dialyzed against 10 m~ "is, pH 7, 100 mM NaCl, and 1 m~ MeSH, the final yields of the rBgtx following Factor Xa cleavage and HPLC isolation were 83 pg/liter of bacterial culture. Refolding of the fusion protein before FPLC under similar conditions increased the final yield slightly to 93 pgAiter of culture. The use of L-cystine to form mixed disulfides as the initial step of refolding has been reported as an impor-

Gena 9 Sequence sgtx codlng rsglon

E L A R G S S R V D I E G R E l I V C H ... gag ctc gcc cgg gga tcc tct aga gtc gac tc gag ggt cgt A l T G l T TGC CAC ...

1 1 observed cleavage Factor Xa Reeognltlon

Slte

FIG. 4. The fusion protein sequence surrounding the engineered Factor Xa recognition site. NH,-terminal amino acid sequencing of the purified rBgtx indicated that the observed enzyme cleavage occurred after the indicated arginine residue leaving 10 amino acids attached to the NH, terminus Ole) of Bgtx.

tant determinant of final yield (Kohno et al., 1990). "he cystine- based refolding scheme that we finally settled on is based on a protocol developed for the refolding of recombinant secretory leukocyte protease inhibitor, which contains 8 disulfide bonds (Seely and Young, 1991). Using this method and RP-HPLC for final purification, final yields were increased 3-fold to about 300 pg/liter of bacterial culture. With S-Sepharose chromatography replacing the RP-HPLC and with a batch anion-exchange column in place of FPLC, we were able to obtain 452 pg of toxiditer of starting culture.

Characterization of Recombinant Toxins-As indicated in Fig. 3, the relative migration of authentic and recombinant toxins upon SDS-polyacrylamide gel electrophoresis suggests that the recombinant toxins appear to have a greater molecular weight than authentic Bgtx. When analyzed by electrospray mass spectrometry, authentic Bgtx has an apparent mass of 7,974.5 daltons. The rBgtx contained no material with the mass of authentic Bgtx and instead contained a major species with an apparent mass of 9,032.9 daltons (data not shown). NH,- terminal sequencing of the purified rBgtx revealed the presence of 10 additional amino acids at the NH, terminus of the Bgtx sequence. These results suggest that the engineered IEGR protease recognition site may not be accessible to enzymatic cleavage. Fig. 4 illustrates the expected arginine cleavage site, following the recognition sequence IEGR, as well as the experimentally observed major cleavage site, 10 amino acids upstream within the Gene 9 polylinker coding region. We suspect that the NH,-torminol Ilc of Bgtx is not sufficiently exposed to allow access by Factor Xa protease. "his is consistent with the observation that the purified rBgtx can be exten- sively retreated with Factor Xa with no further change in apparent molecular weight. We did not attempt to use purified trypsin to cleave at the IEGR site of the fusion protein as authentic Bgtx appears to be sensitive to trypsin digestion as seen by SDS-polyacrylamide gel electrophoresis (data not shown). Tryptic release of the COOH-terminal tetrapeptide from Bgtx has previously been shown to reduce toxicity about 13-fold and further digestion caused peptide cleavage after Ar$6 (Wu et al., 1983).

The binding activities of the recombinant toxins were as- sayed by a competition binding assay using Torpedo nAChR- enriched membranes and '251-labeled native toxin. Unlabeled authentic Bgtx was used as a standard and all binding determinations were performed at least in triplicate. Fig. 5 summa- rizes a direct comparison of the competition data obtained for authentic Bgtx, rBgtx, and the D30A mutant. From a comparison of the mean values for the IC, values, it appears that the rBgtx is slightly less active than authentic Bgtx but the difference in IC,, values was clearly less than 2-fold. From four trials, the IC,, of authentic Bgtx was 15.1 2 4.4 n~ (&.E.) while the IC,, of the purified rBgtx was 25.2 2 6.5 n~ suggesting a statistically significant difference in the binding activities of these two preparations at a confidence level of 95% (Student's t test). In contrast, the IC,, of the D30A mutant was 10.5 k 2.8


100 X

B *

2 P

(r

M G z z e 50 - .- z $3 x

s OL 10"O 10-8 10" 10-6

Bgtx (M)

Bgtx, rBgtx, and the D30A mutant. Wells of microtiter plates were FIG. 5. Solid-phase competition binding assay of authentic

coated with 0.1 mg of Zbrpedo membranes and incubated with 3 n~

rBgtx (m), or D30Amutant (A). After a 2-h incubation at 25 "C, the wells '2SI-Bgt~ and various concentrations of unlabeled authentic Bgtx (O),

were washed 4 times and the bound radioactivity was determined. Each point is the average of quadruplicate determinations.

IIM ( n = 4) which is not significantly different from the IC,, of authentic Bgtx (15.1 2 4.4 m). Consistent with the observed 1.7-fold difference in the IC,, values between rBgtx and authentic Bgtx, the apparent affinity of the D30A mutant is significantly improved over its parental rBgtx construct (confidence level >99%). The use of IC,, determinations here indicates more accurately the raw data. IC,, values can be converted to Kd values if the dissociation constant for lZ5I-labeled Bgtx binding to solid-phase attached nAChR-enriched membranes is independently known. As the physical adsorp- tion of the membranes to the plastic wells of the microtiter plates may affect the affinity of lZ5I-labeled Bgtx binding to receptor-enriched membranes, we chose to use the observed IC, determinations as the basis for comparison. The ratio of the observed IC,, determinations clearly provides an index of the relative affinities for the nAChR. The range of IC, values that we observed in this study were comparable to those seen with other a-neurotoxins in which apparent Kd values in the subnanomolar range have been reported (Pillet et al., 1993).

DISCUSSION

The results presented indicate that we have developed a system for producing relatively high yields of a rBgtx which has a binding affinity comparable to authentic, snake venom-derived Bgtx. The final yields of active rBgtx (450 &liter of culture) are comparable to those obtained for recombinant charybdotoxin, a 37-residue scorpion venom polypeptide with three disulfides (Park et al., 1991), and to those reported for anthopleurin B, a 49-residue sea anemone venom polypeptide also with three disulfides (Gallagher and Blumenthal, 1992). In contrast, recombinant K-bungarotoxin, which is related in sequence to the long-type a-neurotoxin family and which, like Bgtx, contains five disulfides, has been produced in E. coli at a final yield of about 4 pg of active toxiditer based on 50-liter fermentations (Fiordalisi et al., 1991, 1992). Recombinant K-bungarotoxin was prepared, however, using a different fusion protein strategy than the one described here. Erabutoxin a, a 62-residue polypeptide with four disulfides and the first curaremimetic a-neurotoxin to be prepared as a recombinant protein (Ducancel et al., 19891, was produced as a CNBr-cleavable fusion to protein A. The final yield of HPLC-purified, active

recombinant erabutoxin a was about 33 pglliter of shaker culture (Ducancel et al., 1989; Boyot et al., 1990).

If we normalize the yield of active rBgtx (452 pg) to the amount of rBgtx-fusion protein at the ammonium sulfate precipitation step where we have about 75 mg of protein (per liter of culture), then our final yield is about 3% of that expected. This value represents only a minimal estimate of the overall refolding efficiency, however, as other factors such as fusion protein aggregation, possible endogenous protease contamina- tion, and inappropriate cleavages by the added Factor Xa protease may contribute to the overall losses. Nevertheless, the particular refolding strategy utilized remains an important consideration as overall yields of active rBgtx varied by about &fold depending on the refolding procedure. 'Ib test whether yields could be improved if another restriction protease were used in place of Factor Xa, we constructed a Bgtx-fusion protein in which the substrate site (DDDDK) for enterokinase-I11 (Biozyme) replaced the apparently Factor Xa-inaccessible IEGR sequence. Analysis of the enterokinase-111-cleaved rBgtx- fusion protein by mass spectroscopy revealed a minor peak with the appropriate molecular weight expected for authentic Bgtx (data not shown). The presence, however, of large quan- tities of other protein fragments in the digest made further purification from this mixture impractical. We are currently investigating other nonenzymatic strategies for site-specific peptide bond cleavage in Gene 9-fusion protein constructs. Chemical cleavage could offer a significant cost advantage for the production of the relatively large amounts of rBgtx, metabolically labeled with 15N, that would be required for hetero- nuclear NMR investigations of toxin structure and dynamics.

Although we did not intentionally set out to create a rBgtx with 10 additional amino acids at the NH, terminus, the properties of the BGTXlM3 construct, which does indeed contain the sequence GSSRVDIEGR appended to the NH, terminus of the Bgtx sequence, provide valuable new information on Bgtx structure and function. The 84-residue rBgtx showed chromatographic behavior, on reverse-phase HPLC, similar to that of authentic Bgtx (Fig. 2). The additional residues therefore did not affect the HPLC "footprint" of Bgtx but they did appear to influence somewhat the equilibrium between the two Bgtx con- formers separable by HPLC. Furthermore, the NH,-terminal addition of the 10 amino acids had a minimal but nevertheless detectable effect on receptor binding, increasing the measured IC,, by about 1.7-fold. This finding is consistent with chemical modification studies showing that trinitrophenylation of the amino terminus of the related a-neurotoxin, cobrotoxin, had little effect on biological toxic activity (Chang et al., 1971). I t also is consistent with the work of MBnez and co-workers (MBnez et al., 1984) who have suggested that the NH,-terminal part of the short a-neurotoxin, erabutoxin, is not directly in- volved in the "neurotoxic" site. In addition, the inability of Factor Xa to cleave the IEGR sequence, even in the purified rBgtx, suggests that protease access may be hindered by surrounding residues and thus the NH, terminus is unlikely to participate directly in receptor recognition. The present study of rBgtx argues that not only is the free a-amino group not important for receptor binding but also that additional NH,- terminal residues do not produce a major steric interference to receptor binding. Our findings suggest that the additional NH,- terminal residues in the rBgtx may provide a useful site for the preparation of fluorescent or other chemical conjugates of rBgtx with minimal effects on receptor binding activity.

Asp3' is a highly conserved residue among both short and long-type a-neurotoxins (Endo and Tamiya, 1987). Based on the x-ray crystal structure of a sea snake a-neurotoxin related to Bgtx, it has been suggested that Asp3, may interact with Arc to form a hydrogen-bonded ion pair that structurally mimics

11184 Site-directed Mutagenesis of a-Bungarotoxin

the stereochemical features of the agonist, acetylcholine, and which may thus be important for increased binding and specificity (Tsernoglou and Petsko, 1976; Tsernoglou et al., 1978; Low, 1979). In an attempt to determine the relative importance of the negative charge at position 30, we chose to mutate this residue to an alanine, adopting the charged to alanine scanning mutagenesis strategy utilized so effectively by Wells, Zoller, and co-workers (Bass et al., 1991; Gibbs and Zoller, 1991; Wells, 1991) to dissect other ligand-receptor and protein-protein interactions. Alanine, besides being found in all types of secondary structure and in both buried and exposed regions of proteins, has the further advantage that it does not introduce new hydrogen bonding possibilities nor is it sterically bulky, espe- cially in comparison to aspartate (Wells, 1991). Using alanine to substitute for Asp3’ provides important information on the specific role of the carboxyl group beyond the P-carbon common to both Ala and Asp. As clearly demonstrated in Fig. 5, the D30A mutant showed no significant change in binding affinity for the nAChR in comparison to the native toxin. This finding strongly suggests that the carboxyl group of Asp3’ in native Bgtx does not play a major functional role as a mimic of the ester and carbonyl oxygens of acetylcholine as had previously been proposed (Tsernoglou et al., 1978). In addition, as the final yield of the D30A mutant was comparable to that of the parental rBgtx (data not shown), the D30A substitution does not appear to affect adversely expression levels, protein folding, or disulfide formation.

Our results with the D30A mutant are in general agreement with experiments performed with a 13-mer peptide fragment of loop 2 from Bgtx (MetZ7-Val3’) in which the residue corresponding to Asp3’ was mutated to Ala with no change in the peptide’s binding affinity (Lentz, 1991). Similarly, when six carboxyl groups in cobrotoxin were simultaneously modified by the addition of a glycine methyl ester to the carboxyl groups only a 25% decrease in lethality was seen (Chang et al., 1971). These results, as well as our studies of the D30A mutant, appear somewhat contrary to a report that a Asp31-His mutation in recombinant erabutoxin a, leads to a 46-fold decrease in binding affinity (Pillet et al., 1993). To explain these findings, we propose that the chemical and steric properties of the functional group beyond the p-carbon at residue 30 in Bgtx does indeed influence binding, although a negative charge per se does not appear to be required for high affinity binding activity. The NMR solution structure of the stoichiometric complex formed between Bgtx and a dodecapeptide corresponding to nAChR a-subunit residues 185-196 (a185-196) has led to the identification of a number of Bgtx residues that are within -4 A of receptor peptide residues as evidenced by observable intermolecular NOES (Basus et al., 1993). A NOE cross-peak is observed between the amide NH of Asp3’ and the CGHs of the peptide residue corresponding to the conserved ’I’yrlw found in all functional a-subunits. Furthermore, the amide N H , the CaH, and one of the CPHs of Asp3’ show a large chemical shift (20.12 ppm) in resonance positions upon receptor-peptide binding indicating a large change in chemical environment (Basus et al., 1993). Together, these findings suggest that the side chain ofAsp3’ is indeed in close proximity to the bound receptor peptide, at least part of which appears to lie in the cleft formed between the NH,-terminal loop and the long middle loop of Bgtx (Fig. 6). The larger size of the His side chain in the A s p H i s mutant, possibly together with the hydrogen bonding characteristics of the imidazole ring, may contribute to the considerable effect of this particular substitution on receptor recognition in contrast to the D30A mutant results reported here. Simple removal of the carboxyl group from position 30 in Bgtx (Fig. 6), leaving the “truncated” p-carbon found in both Asp and Ala, does not inhibit receptor binding, whereas the

FIG. 6. Schematic diagram of the first 67 residues in Bgtx. Resi- dues 6 S 7 4 are out of the plane of the figure and are not illustrated here to s i m p w the representation. The location of the highly conserved Asprn which has been mutated in this study to an Ala is shown near the bottom tip of the middle “toxic loop.” A synthetic peptide derived from the nicotinic receptor’s a-subunit sequence and capable of micromolar binding to Bgtx has been shown to reside within -4 A of a number of Bgtx residues contributed by both the NH,-terminal loop and the middle toxic loop (Basus et al., 1993).

possible protrusion of the large and planar His side chain into the toxin’s “binding cleft” may explain the 46-fold reduction in binding affinity seen with the recombinant erabutoxin a (Pillet et al., 1993). Our findings with the D30A mutant clearly indicate that the negative charge at this position is not critical for binding.

The D30A mutant has a slightly but significantly higher apparent affinity (2.4-fold) for the nAChR than the parental rBgtx, both of which contain the additional 10 amino acids at the NH, terminus. As the parental rBgtx exhibits a 1.7-fold lower affinity in comparison to authentic Bgtx, the D30A mutation may alleviate some steric hindrance introduced by the NH,-terminal addition. It is possible that residues in the NH,- terminal addition may interact electrostatically with Asp3’ in the parental rBgtx leading to a decrease in binding affinity. Removal of the negative charge at position 30 may therefore prevent this mode of interference. Alternatively, the D30A mutation may itself lead to an increase in binding affinity by removing a charged group that normally interferes somewhat with binding. In fact, several alanine substitution mutations have been observed in other systems to produce 2-4-fold in- creases in apparent binding affinities (Wells, 1991).

The availability of an active rBgtx addresses to some extent concerns about the future availability of authentic snake venom-derived Bgtx due to the endangered status of Bungarus multicinctus. The final yields of active rBgtx that we have obtained compare favorably with other expressed neurotoxins (Howell and Blumenthal, 1989; Boyot et al., 1990; Park et al., 1991). Furthermore, the construct reported here should provide an important biochemical tool to probe the structure-function relationship of Bgtx residues through site-directed mutagenesis. A recombinant form of Bgtx also opens up the opportunity to label the protein metabolically with 15N and 13C. Hetero- nuclear NMR studies of toxin structure and dynamics should help to reveal more molecular details of the protein-protein recognition events underlying a-neurotoxin recognition and


specificity towards the nAChR. Finally, the recombinant neurotoxins may provide an important molecular scaffold (protein- folding motif) from which we can probe and engineer new receptor recognition sites.

the ear ly stages of this project. Acknowledgment-We thank J e r e m y J. Erdley for participation in

REFERENCES

Agard, D. A,, and Stroud, R. M. (1987)Acta Cryst. A38, 186194 Bass. S. H.. Mulkemn. M. G.. and Wells. J. A. (1991) Pmc. Natl. Acad. Sci. U. S. A.

~ ~~, aS, 4 4 9 k 5 0 2

Biochemistrv 27.22763-22771

, ,

Basus, V. J., Billeter, M., Love, R. A,, Stroud, R. M., and Kuntz, I. D. (1988)

Basus, V. J., Song, G., and Hawrot, E. (1993) Biochemistry 32, 12290-12298 Boyot, P., Pillet, L., Ducancel, F.. Boulain, J.-C., m m e a u , O., and MBnez, A. (1990)

Chang, C. C., Yang, C. C., Nakai, IC, and Hayashi, K. (1971) Eiochim. Eiophys.

Changeux, J.-P., Devillers-ThiCry, A,, Galzi, J.-L., and Bertrand, D. (1992) 2knd.s

Chiappinelli, V. A. (1993) in Natural and Synthetic Neurotoxins (Harvey, A., ed) pp.

Corran, P. H. (1989) in HPLCofMacromolecules (Oliver, R. W. A., ed) pp. 127-156,

Czajkowski, C., DiPaolo, M., Bodkin, M., Salazar-Jimenez, G., Holtzman, E., and

Ducancel, F., Boulain, J.-C., TrBmeau, O., and MBnez, A. (1989) Protein Eng. 3,

Endo, T., and Tamiya, N. (1987) P h a r m o l . Ther 34,403-451 Fiordalisi, J. J., Fetter, C. H., TenHarmsel, A,, Gigowski, R., Chiappinelli, V A,, and

Fiordalisi, J. J., Gigowski, R., and Grant, G. (1992) Protein Exp. Purif: 3, 28Z289 Gallagher, M. J., and Blumenthal, K. M. (1992) J. Biol. Chem. 267, 13958-13963

FEES Lett. 266,87-90

Acta 251, 334-344

Pharmocol. Sci. 13,299301

65-128, Academic Press, New York

IRL Press, Oxford

Karlin, A. (1989) Arch. Biochem. Biophys. 272,412420

13S143

Grant, G. (1991) Biochemistry 30, 10337-10343

Gibbs, C . S., and Zoller, M. J. (1991) Eiochemtstry 30, 5329-5334 Habeeb, A. F. S. A. (1972) Methods Enzymol. 26, 457-464 HervB, M., Pillet, L., Humbert, P., TrBmeau, 0.. Ducancel, F., Hirth, C., and MBnez,

Howell, M. L., and Blumenthal, K. M. (1989) J. Biol. Chem. 264, 15268-15273 Kohno, T., Carmichael, D. F., Sommer, A., and Tbampson, R. C. (1990) Method?

Kosen, P. A., Finer-Moore, J . , McCarthy, M. P., and Basus, V. J. (1988)Eiochemtstry

Lentz, T. L. (1991) Biochemistry 30, 10949-10957 Low, B. W. (1979) in Snake Venoms (Lee, C . Y., ed) pp. 21S257, Springer Verlag,

Love, R. A., and Stroud, R. M. (1986) Protein Eng. 1, 3 7 4 6 MBnez, A,, Boulain, J. C., Bouet, F., Couderc, J., Faure, G., Rouselet, A., TrBmeau,

Mdnez, A,, Bontems, F., Roumestand, C., Gilquin, B., and Toma, F. (1992) Proc. R. O., Gatineau, E., and Fmmageot, P. (1984) J. Physiol. 79, 196206

Oswald, R. E., Sutcliffe, M. J., Bamberger, M., Loring, R. H., Braswell, E., and Soc. Edinb. Sect. B Biol. 99, 83-103

Park, C. S., Hausdo#, S. F., andMiller, C. (1991)Proc. Natl. Acad. Sci. U. S. A. 8 8 , Dobson, C. M. (1991) Biochemistry 30,49014909

Pillet, L., Tremeau, O., Ducancel, F., Drevet, P., ZinnJustin, S., Pinkasfeld, S.. 2046-2050

Seely, R. J., and Young, M. D. (1991) in Protein Refolding (GeorDou, G., and De Boulain, J., and MCnes, A. (1993) J. Biol. Chem. 268,909-916

Bernardez-Clark, E., eds) pp. 206-216, American Chemical Society, Washing- ton, U. C .

Stroud, R. M., McCarthy, M. P., and Shuster, M. (1990) Biochemistry 29, 11009- 11023

Studier, F. W., and Moffatt, B. A. (1986) J. Mol. Biol. 189, 113-130 Tsernoglou, D., and Petsko, G. A. (1976) FEES Lett. 68, 1 4 Tsernoglou, D., Petsko, G. A., and Hudson, R. A. (1978) Mol. Pharmacol. 14,

A. (1992) E m J. Biochem. 208, 125-131

Enzymol. 185,187-195

27,2775-2781

New York

71&716 Wells, J. A. (1991) Methods Enzymol. 202, 39M11 Wilson, P. T., Lentz, T. L., and Hawrot, E. (1985) Pmc. Natl. Acad. Sci. U. S. A. 82,

Wu, S.-H, Chen, C. J., Tseng, M.-J., and Wang, K.-T. (1983)Arch. Biochem. Bio- 879M794

phys. 227, 111-117

functional expression and site-directed mutagenesis of a synthetic

Documents