Proc. Nati. Acad. Sci. USAVol. 88, pp. 2046-2050, March 1991Biochemistry

Design, synthesis, and functional expression of a gene forcharybdotoxin, a peptide blocker of K+ channelsCHUL-SEUNG PARK, SHARON F. HAUSDORFF, AND CHRISTOPHER MILLERHoward Hughes Medical Institute, Graduate Department of Biochemistry, Brandeis University, Waltham, MA 02254

Communicated by Clay M. Armstrong, December 6, 1990

ABSTRACT A gene encoding charybdotoxin (CTX), a K+channel blocker from scorpion venom, was designed, synthe-sized, and expressed as a cleavable fusion protein in Escherichiacoli. A sequence-specific protease, factor Xa, was used to cleavethe fusion protein and thus release the toxin peptide. Therecombinant toxin was purified, oxidized to form disulfidebonds, and treated to form N-terminal pyroglutamate. Recom-binant CTX is identical to the native venom CTX with respectto high-performance liquid chromatography mobility, aminoacid composition, and N-terminal modification. With singleCa2+-activated K+ channels as an assay system, recombinantCTX shows blocking and dissociation kinetics identical to thenative venom toxin. The synthetic gene and high-level expres-sion of functionally active CTX make it possible to study thefundamental mechanism of the toxin-ion channel interaction.

Charybdotoxin (CTX) is a basic peptide originally isolatedfrom the venom of the scorpion Leiurus quinquestriatus as ahigh-affinity blocker of the high-conductance Ca2l-activatedK+ channel from skeletal muscle (1, 2). This peptide alsoinhibits Ca2+-activated K+ channels found in neurons (3-5),kidney (6), and erythrocytes (7, 8); in addition, CTX blocksvoltage-activated K+ channels from a variety of sources (5,7, 9-12). In all cases, CTX inhibits in the nanomolar con-centration range by binding to the external part of the K+channel. Inhibition of the channel is known to be due to thephysical occlusion of the channel's conduction pathway bythe toxin (13-15). Because the channel-toxin interaction ismechanistically understood, CTX has proven to be an infor-mative probe of the external "mouth" of K+ channels(16-18).The amino acid sequence of CTX, determined by peptide

sequencing (19), is shown in Fig. 1. The toxin is a 37-residuepolypeptide of 4.3 kDa, with eight positively charged resi-dues, two negative charges, three disulfide bonds, and ablocked N terminus formed by a pyroglutamate residue. Thethree-dimensional structure of the toxin is known (20). Builton a foundation of a three-strand antiparallel f3-sheet, themolecule is roughly ellipsoidal, with major and minor axes of2.5 nm and 1.5 nm, respectively. The functional groups showa remarkable spatial segregation on the surface of the toxin;one face of the molecule is strongly polar, whereas most ofthe hydrophobic groups project off the opposite face.

Electrophysiological studies have yielded informationabout the mechanism of toxin block, but it has been difficultto probe the toxin biochemically. Purification ofCTX from itsnative source is costly and yields only a small amount ofprotein. In addition, because of lack of specificity ofchemicalmodifications, information on the functions of specific resi-dues is nonexistent. To overcome this problem, we havedeveloped a system to produce large quantities of geneticallymanipulable CTX.

We have designed a synthetic gene for CTX and haveachieved its high-level expression in Escherichia coli by usinga cleavable fusion protein strategy. The following three invitro posttranslational processing steps were required to yieldfully functional CTX: (i) proteolytic cleavage of the CTXcoding sequence, (ii) oxidation of the six cysteine residues toform three disulfide bonds, and (iii) formation of N-terminalpyroglutamate. This system will permit a close mechanisticstudy of the toxin-K+ channel interaction by a combinationof reconstitution of single K+ channels and site-directedmutagenesis of CTX.

MATERIALS AND METHODSBacterial Strains and Plasmids. E. coli DH1 was used for

plasmid propagation and BL21(DE3) was used for the expres-sion of the gene 9-X,-CTX fusion protein. E. coli BL21(DE3)is a A lysogen of BL21 wherein the prophage contains theRNA polymerase gene ofT7 bacteriophage under the controlof the lacUV5 promoter (21). Plasmid pSR9, a vector forproducing fusion proteins with the gene 9 protein of T7bacteriophage, was obtained from K. M. Blumenthal (22).

Design, Synthesis, and Construction ofa CTX Gene. A DNAsequence encoding the CTX peptide was designed to containthe maximum number ofunique restriction enzyme sites. Thegene was constructed from four overlapping oligonucleotidepairs, each of approximately 45 base pairs. Each oligonucle-otide was chemically synthesized on an Applied Biosystemsmodel 380A DNA synthesizer, gel-purified, and phosphoryl-ated. The eight oligonucleotides were annealed and ligatedusing T4 DNA ligase. The synthetic CTX gene, which isflanked by Sal I and HindIII sites, was inserted into pSR9 forthe gene 9-Xa-CTX construction pCSP105. After cloning,the sequence of the synthetic CTX gene was confirmed byDNA sequencing of both strands. General rules and con-straints for producing synthetic genes were followed (23).

Expression and Purification of CTX Fusion Protein. E. coliBL21(DE3) cells harboring pCSP105 were grown at 37°C inLB medium in the presence of ampicillin (100 ,ug/ml). Fusionprotein synthesis was induced at late logarithmic phase(OD650, 0.8 to 1.0) by the addition of 0.5 mM isopropylB-D-thiogalactoside (IPTG). Cells were harvested 2-3 hrlater, washed, and resuspended in 50 ml of 10 mM Tris HCI,pH 8.0/50mM NaCl/2mM Na2EDTA. Lysozyme was addedto 2 mg/ml and, after 20 min on ice, 2-mercaptoethanol,phenylmethylsulfonyl fluoride, pepstatin, and leupeptin wereadded to final concentrations of 10 mM, 1 mM, 1 ,M, and 1AM, respectively. After a brief sonication, the extract wasclarified by a centrifugation at 27,000 x g for 30 min. All ofthe following steps were performed at 4°C. Nucleic acidswere precipitated by the slow addition of 0.1 vol of 30%(wt/vol) streptomycin sulfate and subsequent centrifugation.The fusion protein was precipitated by the slow addition ofsolid ammonium sulfate to 50% saturation. After centrifuga-

Abbreviations: CTX, charybdotoxin; IPTG, isopropyl -D-thiogalactoside.

2046

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 88 (1991) 2047

Mael Avrll

Ampr

Ori

Sall 1

GTC GAC GGA TCC ATC GAA GGT CGT CAA TT1 ACC AAT GTT TCT TGT ACC ACTVal Asp Gly Ser Ik -C& jay AM Gin Phe Thr Asn Val Ser Cys Thr Thr

10 20

TCT AAG GAA TGT TGG TCC GTT TGT CAA CGT CTG CAT AAC ACC AGC CGC GGTSer Lys Glu Cys Trp Ser Val Cys Gin Arg Leu His Ann Ir Ser Aig Gly

30 37 HindIlIAAA TGC ATG AAC AAA AAA TGT CGT TGT TAC TCC TAG GAA TTC CAA GCI T

Lys Cys Met Asn Lys Lys Cys Ag Cys Tyr Ser End

FIG. 1. Design of synthetic gene for CTX and its expression in E.coli. (Upper) The expression vector of gene 9-Xa-CTX fusionprotein is shown. Vector pCSP105 was constructed by insertion of asynthetic CTX gene and flanking sequences into Sal I and HindIIIsites of pSR9. CTX coding region is hatched, Xa recognition se-quence is solid, and the last 12 bases of T7 gene 9 are stippled in theplasmid map. Ampr, ampicillin resistance; Ori, origin. (Lower) DNAsequence of synthetic gene and corresponding amino acid sequenceare shown. CTX coding sequence is in bold (from Gln-1 to Ser-37)and factor Xa recognition sequence is underlined (J-_hj-QIY-Arg).

tion, the precipitate was dissolved in 10 ml of 50 mMTris-HCI, pH 7.0/50mM NaCI/5 mM 2-mercaptoethanol anddialyzed against 2 liters of the same buffer. The dialysate wasfractionated on a DEAE-cellulose (2.5 x 12 cm, DE52,Whatman) column with the same buffer by using a 0.05 M-0.5M NaCI linear gradient (total volume, 300 ml). Fusion proteinfractions were pooled, concentrated, stored at -20°C.

Fusion Protein Cleavage by Factor X. The DEAE-cellulose-purified fusion protein pool was dialyzed against 2 liters of 50mM Tris HCI, pH 8.3/150 mM NaCl/0.5 mM 2-mercapto-ethanol. After dialysis, 3 mM CaC12 was added. CTX wascleaved from the fusion protein by factor Xa (restrictionprotease factor Xa, Boehringer Mannheim, 10 ,ug/mg offusion protein) at room temperature for up to 36 hr. Thedigestion mixture was loaded on a Mono-S FPLC columnequilibrated with 30mM sodium phosphate (pH 7.5). Materialwas eluted with a total of 20 ml of a 0-0.6 M NaCl lineargradient.N-Terminal Cyclization and Purification of Recombinant

CTX. To cyclize the N-terminal glutamine, 5% (vol/vol)acetic acid was added to the uncyclized CTX fraction fromthe Mono-S column. After incubation at 65°C for the appro-

priate time (usually 2 hr), the material was rechromato-graphed on Mono-S by FPLC as above. CTX with N-terminalpyroglutamate, which elutes earlier than uncyclized CTX,was further purified by reverse-phase HPLC on a C18 columnVydac, 4.6 x 250 mm, S,u). A strong-cation-exchange HPLCcolumn (The Nest Group, Southport, MA, polysulfoethylaspartamide, 4.6 x 200 mm, 5,u) was used to confirm thepurity of the final recombinant CTX.

Planar Bilayer Assay of Native and Recombinant CTX. Thechannel blocking activity of both native and recombinantCTX was measured on single high-conductance Ca2+-activated K+ channels from rat skeletal muscle inserted into

planar bilayers. Lipids, membrane preparation, nativevenom CTX preparation, and planar bilayer methods were asdescribed (2). Measurements of CTX blocking and dissocia-tion kinetics of single K+ channels were carried out byobserving at least 100 discrete blocking events caused byaddition of CTX to the external side of the bilayer. In eachexperiment, a single channel was incorporated into the planarbilayer by addition of plasma membrane vesicles underosmotic gradients; further channel insertion was suppressedby adjusting the salt concentration to equimolar. The finalsolutions on the two sides of the membrane were as follows:the internal solution contained 10 mM Hepes (pH 7.4), 150mM KCl, and 30 ,uM CaC12 and the external solution con-tained 10 mM Hepes (pH 7.4), 150 mM KCI, 0.1 mM EGTA,bovine serum albumin (30 pug/ml), and the desired concen-tration of CTX.

Gel Electrophoresis, Immunoblotting, and Analytical Meth-ods. Fusion protein and factor Xa digestion products wereanalyzed by SDS/polyacrylamide gel electrophoresis on 12%gels and Coomassie brilliant blue staining. Immunoblot anal-ysis was done by electroblotting proteins from the SDS gelonto nitrocellulose filters and treating them with total rabbitantiserum against native venom CTX. Amino acid analysiswas done using a Waters Pico-Tag system. The molar ex-tinction coefficients of both the native and the recombinantCTX were calculated based on two amino acid analysisresults. The peptide sequencing of the N-terminal uncyclizedand cyclized recombinant CTX was performed using anApplied Biosystems model 475A protein sequencer. Beforepeptide sequencing, cysteine residues of native and theN-terminal cyclized recombinant CTX were reduced withdithiothreitol and carboxamidomethylated with iodoaceta-mide, and the peptide was deblocked by pyroglutamateaminopeptidase (sequencing grade, Boehringer Mannheim)(19).

RESULTSDesign and Construction of CTX Gene Expression System.

The overall strategy for expressing recombinant CTX in E.coli is indicated in Fig. 1. The peptide was produced as afusion protein in which the CTX coding sequence was fusedonto the C-terminal portion of the T7 gene 9 protein, asdescribed by Howell and Blumenthal (22). The tetrapeptidesequence Ile-Glu-Gly-Arg, which is recognized by the "re-striction protease" blood coagulation factor Xa, was posi-tioned immediately upstream from the CTX sequence. TheN-terminal residue of native CTX is pyroglutamate, whichmay be formed by nonenzymatic cyclization of N-terminalglutamine. A translation termination codon was inserted atthe end of the CTX coding sequence.The nucleotide sequence encoding the 37 residues of CTX

was designed to maximize the number of unique restrictionsites. Nine restriction sites in the CTX coding region and foursites in the flanking region were inserted into the gene tofacilitate future cassette-mutagenesis studies. Eight chemi-cally synthesized oligonucleotides were individually gel-purified and phosphorylated, and the 5' end groups wereconfirmed by 5' end analysis. The oligonucleotide duplexeswere annealed and ligated in a single reaction. The full-lengthgene was inserted into the Sal I and HindIII sites of the fusionvector pSR9, downstream from the T7 promoter. The finalconstruct pCSP105 was propagated in E. coli DH1.

Expression and Purification of Recombinant CTX. The gene9-CTX fusion protein was expressed in E. coli BL21(DE3), thechromosome of which carries a lactose-inducible T7 RNApolymerase gene. Bacteria freshly transformed with pCSP105were induced in late logarithmic phase by addition ofIPTG. Thesoluble fusion protein was partially purified by ammoniumsulfate precipitation and DEAE-cellulose column chromatog-

Biochemistry: Park et al.

Proc. Natl. Acad. Sci. USA 88 (1991)

A B1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9

1

of.-i --

_u _ * _

- 0Q) \ \.-- o'!1c () \

u. _gI3 | | |_......... .. .

FIG. 2. Expression and cleavage of gene 9-Xa-CTX fusion protein. Fusion protein gene 9-Xa-CTX was expressed in E. coli BL21(DE3)by IPTG induction. The fusion protein was purified with ammonium sulfate precipitation and DEAE-cellulose column chromatography. CTXwas liberated from the fusion protein by factor Xa. Coomassie-stained gel (A) and immunoblot (B) of Laemmli 12% polyacrylamide gel ofuninduced cell-free extract of E. coli carrying pCSP105 (lane 1); total cell-free extract induced with IPTG for 2 hr (lane 2); soluble fraction (lane3); DEAE-cellulose pool of fusion protein at 0-hr factor Xa digestion (lane 4); 3-, 8-, 15-, and 24-hr factor Xa digestion (lanes 5-8, respectively);reverse-phase HPLC purified fully processed recombinant CTX (lane 9 of immunoblot). Molecular mass markers are indicated at the left at 84,47, 33, 24, 16, and 6.2 kDa.

raphy. After chromatography, the fusion protein was greaterthan 60% pure as judged by Coomassie staining of SDS/polyacrylamide gels (Fig. 2A). The yield of fusion protein wasapproximately 30 mg/liter of culture.The partially purified fusion protein was digested with a

sequence-specific protease, blood coagulation factor Xa.Over 24 hr, a major fusion protein band at 42 kDa wasgradually split into two products, the gene 9 product portionof the fusion protein at 37 kDa and a small protein at 5 kDa,as seen on Coomassie-stained gels and immunoblots (Fig. 2).Minor bands in the immunoblot are likely to be nonspecificbacterial proteins cross-reacting with anti-CTX antiserum.The recombinant CTX was further purified by cation-exchange chromatography, where a major peak was eluted atapproximately 360 mM NaCl. This is slightly later than theelution of native venom CTX (330 mM NaCl), the N terminusof which is blocked by pyroglutamate. Peptide sequencing ofthe recombinant material confirmed that the first five resi-dues were, as expected, Gln-Phe-Thr-Asn-Val.With its six cysteines in reduced form, CTX is inactive as

a channel-blocker (24), and so it is necessary to form the threedisulfide bonds properly after cleavage of the reduced CTXpeptide. We have found that inclusion of a low concentration(0.3-0.5 mM) of 2-mercaptoethanol in the factor Xa cleavagereaction mixture led to complete disulfide formation duringthis reaction (data not shown). As we demonstrate below, thecorrect disulfide bonds, yielding fully active material, areformed under these conditions.To cyclize the N terminus of recombinant CTX to pyro-

glutamate, we treated the purified toxin with 5% acetic acidfor 2 hr at 650C. The N-terminal glutamine was half-converted

A B- 1.0 M NaCI

0 O 10

to pyroglutamate at 1 hr and was fully cyclized in 4 hr. Whenthe acid-treated recombinant CTX was chromatographed ona Mono-S column, a new peak was eluted at the position ofnative CTX, as shown in Fig. 3A. This shift in elution time isa result of the single charge difference between the cyclizedand uncyclized forms of CTX.

Since the cyclized and oxidized CTX was fully active, aswe show below, we infer that the proper disulfide bonds wereformed during this treatment. The oxidized and cyclizedrecombinant CTX was purified to complete homogeneity onC18 reverse-phase HPLC (Fig. 3B). The final yield of recom-binant CTX was approximately 1 mg/liter of culture.Chemical Characterization of Recombinant CTX. The

amino acid compositions of native and recombinant CTX,shown in Table 1, were identical within experimental errorand were as expected from the sequence ofthe synthetic CTXgene. The 280-nm extinction coefficients ofthe native and therecombinant CTX were essentially identical (9000 and 9200M-1 cm-l, respectively). To characterize the recombinantCTX further, several cycles of peptide sequencing were donefrom its N-terminal end. Without deblocking by pyrogluta-mate aminopeptidase, no significant sequencing reactionoccurred. However, after denaturation, carboxamidometh-ylation, and cleavage of the N-terminal pyroglutamate, theCTX could be sequenced from the second amino acid resi-due, phenylalanine.

Functional Activity of Recombinant CTX. The activity ofCTX was examined at high resolution by the block of singleCa2+-activated K+ channels (2, 23). In this assay, shown inFig. 4A, single Ca2+-activated K+ channels were reconsti-tuted into planar lipid bilayer membranes, and discrete block-

| - 100 % B FIG. 3. Chromatographic profilesof recombinant CTX. (A) Mono-SFPLC profile of N-terminal uncy-clized (upper curve) and cyclized

r 50 CTX (lower curve). The N-terminalglutamine of recombinant CTX wascyclized in 5% acetic acid for 2 hr at

-L 65°C. (B) Reverse-phase HPLC pro-files of recombinant CTX. The N-ter-minal cyclized CTX was injected on aC18 column equilibrated with 0.1%trifluoroacetic acid and eluted with alinear gradient of 0-30% (vol/vol) ac-

20 30 40 etonitrile with 0.1% trifluoroaceticTime (min) acid over 40 min at 1 ml/min.

10 20 30 40Time (min)

2048 Biochemistry: Park et A

Proc. Nati. Acad. Sci. USA 88 (1991) 2049

Table 1. Amino acid composition of native andrecombinant CTX

Recombinant Expected fromAmino acid Native CTX CTX sequenceAsp + Asn 2.8 2.9 3Glu + Gln 2.9 3.0 3Ser 5.0 5.1 5Gly 1.5 1.0 1His 0.5 0.5 1Arg 3.0 3.3 3Thr 3.9 4.4 4Ala 0.3 0.1 0Pro 0.2 <0.1 0Tyr 0.9 0.9 1Val 1.9 2.0 2Met 0.8 0.7 1Cys ND ND 6lie 0.2 <0.1 0Leu 1.1 1.1 1Phe 1.0 1.0 1Lys 3.4 3.6 4Trp ND ND 1

Total 37ND, not determined.

ing events, due to binding of single molecules of CTX to thechannel, were observed electrically. Statistical distributionsof channel dwell-times in the blocked and unblocked statesmay be calculated from many such blocking events, and toxinassociation and dissociation rates may be readily measured(1, 2, 23). The raw data traces of Fig. 4A illustrate theseCTX-induced blocking events with preparations of toxinpurified from scorpion venom, fully processed recombinanttoxin, and uncyclized recombinant toxin at 25 nM, 25 nM,and 75 nM, respectively, in the external medium. It is clearthat fully processed recombinant toxin behaved similarly tonative venom-purified CTX. The uncyclized peptide was amuch poorer inhibitor, for two reasons. (i) The blocksinduced by uncyclized recombinant CTX were less frequentthan with the other two preparations of CTX. (ii) Theblocking events were approximately 5-fold shorter-lived withthe uncyclized material.These qualitative conclusions are supported by a quanti-

tative analysis of the dwell-time distributions for unblockedand blocked states (Fig. 4 B and C). All these distributionswere monoexponential, as demanded for block by a homo-geneous preparation of toxin. Both on-rates and off-rates ofnative and recombinant CTX were identical (Table 2), with adissociation constant of 12-15 nM. The uncyclized CTXpreparation displayed much weaker channel-blocking activ-ity, with an apparent dissociation constant of 110 nM. Thesimilarity in detailed function ofrecombinant and native CTXstrongly implies the identity in structure of these peptides.

DISCUSSIONWe designed and expressed a synthetic gene for CTX basedon its primary amino acid sequence. The toxin protein wasoverexpressed as a cleavable fusion protein with the gene 9product of phage T7, liberated by a sequence-specific pro-tease, and further processed to the functionally active formof the protein. The recombinant CTX was similar to nativeCTX chemically and functionally. To verify its chemicalidentity, we performed amino acid analysis, partial aminoacid sequencing, and enzymatic digestion of N-terminalpyroglutamate on both native and recombinant toxins. Thechannel blocking activities of both toxins were measured onsingle channels in planar bilayers. All ofthe results, includingHPLC profiles, show that recombinant and native CTX are

ANafiv Venm C1X, 2SnM

r~~~~I- F'! _- _-

Reobinant CIX (Cydiae), 2amM

I1 rr: I Ir _

L I

££: 1 ~-

Eu - tF--- B --- - ff

cbn-tCIX (U_1ic ), 7snm

TMC - l - o - P

B1.0001

D 0.100.

0

0.010

0.0050 5 10 15 20

Time(sec)

C0 v. CTX* r. CTX(cycl.)a r. CTX(uncycl.)

0.005 10 10 20 30 40

Time(sec)

FIG. 4. Channel blocking characteristics of native and recombi-nant CTX. Discrete blocking events on single Ca2+-activated K+channels by native and recombinant CTX (N-terminal cyclized anduncyclized) were observed in planar lipid bilayers (A). The cumu-lative probability distribution of unblocked events (B) and blockedevents (C) are shown. Exponential lifetime of blocking events is asfollows: 15.5 sec for native CTX, 16.4 sec for cyclized recombinantCTX, 3.1 sec for uncyclized recombinant CTX. Exponential lifetimeofunblocking events is as follows: 6.1 sec for native CTX, 6.2 sec forcyclized recombinant CTX, 9.5 sec for uncyclized recombinantCTX. For each experiment 35 nM CTX was added to the externalside of channel and 120 blocks were observed. Holding voltage was+35 mV. v, Native; r, recombinant.

indistinguishable. Due to difficulties in expressing the CTXgene directly in bacteria, we adopted a fusion protein strat-egy. In contrast to the high-level expression of the CTXfusion protein reported here, the original construct of CTXcontaining an ompA secretion sequence on its N-terminal endshowed no detectable protein expression despite adequatelevels ofmRNA (data not shown). We do not know whetherthe poor expression of this construct was caused by toxicityof CTX to host cells, by possible nonsense mutations as

Table 2. Kinetic parameters for block by native andrecombinant CTX

Native RecombinantCTX CTX

Dissociation constant, nM 14.0 ± 0.2 14.2 ± 0.1On rate (k0d), M-1sec-1 (x106) 4.6 ± 0.3 4.5 ± 0.2Off rate (kRff), sec-1 0.067 ± 0.002 0.064 ± 0.003

In each experiment, holding voltage was +35 mV and channelopen probability was adjusted between 0.3 and 0.5. Four experimentswere performed for both toxins and data are mean ± SEM.

Biochemistry: Park et al.

Proc. Natl. Acad. Sci. USA 88 (1991)

observed in other systems (22), or by other unknown factors.Aside from stable high-level expression, the gene 9 fusionprotein has two other advantages: the high solubility of thefusion protein and the absence of cysteine residues in thefusion vector coding sequence.

After release of CTX from the fusion protein, two addi-tional steps were required for functional activity: formationofdisulfide bonds and cyclization ofthe N-terminal glutamineto pyroglutamate. Under our preparation conditions (with alow concentration of mercaptoethanol present), oxidation ofsulfhydryl groups occurred during the factor Xa digestion. Itmay seem paradoxical that addition ofa thiol should promotedisulfide formation; however, dissolved 02 will rapidly oxi-dize the low amount of mercaptoethanol, which can thenefficiently react with protein to form disulfide bonds (25).After at least 24 hr of incubation in this condition, no freesulfhydryl group was detected. Without N-terminal cycliza-tion, the oxidized CTX showed poor affinity for the Ca2+-activated K+ channel (apparent Kd = 110 nM). Since theNMR-determined structure of the peptide shows the N-ter-minal pyroglutamate to be lying upon the hydrophobic face ofthe molecule, positive charge at a free N terminus mightplausibly disrupt a hydrophobic interaction between the toxinmolecule and the mouth of the channel (20). More detailedinformation on the solution structure of recombinant CTXshould be obtained from an NMR study. Preliminary one-dimensional NMR spectra show virtually identical patternsfor recombinant CTX and native toxin (data not shown).The high-level expression of functionally active CTX, a

K+-channel blocker, gives us not only a biochemical tool tostudy the interaction between CTX and K+-channels but alsoan example of the utility of the cleavable fusion proteinstrategy for expressing small proteins of less than 50 %iminoacid residues. In addition, because of the experimenter'sability to control many variables, including oxidation andN-terminal cyclization, this could be a useful model systemfor studying the folding process in small proteins. In the past,we have not been able to investigate in molecular detail theinteraction between CTX and K+ channels because of thelimited amounts of peptide available from scorpion venomand difficulties in obtaining clean reactions of protein-modifying reagents with this peptide. By using our expressionsystem in combination with site-directed alterations in thepeptide sequence, the molecular basis ofCTX recognition byand inhibition of the Ca2+-activated K+ channel may beattacked.

We are grateful to Dr. Dan Oprian for help and advice throughoutthis work and to Dr. Kenneth Blumenthal for a kind gift of the fusionprotein vector pSR9 and advice on its use. This work was supportedby National Institutes of Health Grant GM-31768.

1. Miller, C., Moczydlowski, E., Latorre, R. & Phillips, M. (1985)Nature (London) 313, 316-318.

2. Anderson, C., MacKinnon, R., Smith, C. & Miller, C. (1988) J.Gen. Physiol. 91, 317-333.

3. Hermann, A. & Erxleben, C. (1987) J. Gen. Physiol. 90, 27-47.4. Reinhart, P. H., Chung, S. & Levitan, I. B. (1989) Neuron 2,

1031-1041.5. Schneider, M. J., Rogowski, R. S., Krueger, B. K. &

Blaustein, M. P. (1989) FEBS Lett. 250, 433-436.6. Guggino, S. E., Guggino, W. B., Green, N. 4 Saktor, B. (1987)

Am. J. Physiol. 252, C128-C137.7. Wolff, D., Cecchi, X., Spalvins, A. & Canessa, M. (1988) J.

Membr. Biol. 106, 243-252.8. Castle, N. A. & Strong, P. N. (1986) FEBS Lett. 209, 117-121.9. MacKinnon, R., Reinhart, P. H. & White, M. M. (1988) Neu-

ron 1, 997-1001.10. Sands, S. B., Lewis, R. S. & Cahalan, M. D. (1989) J. Gen.

Physiol. 93, 1061-1074.11. Schweitz, H., Stansfeld, C. E., Bidard, J.-N., Fagni, L., Maes,

P. & Lazdunskj, M. (1989) FEBS Lett. 250, 519-522.12. Stuhmer, W., Ruppersberg, J. P., Schroter, K. H., Sakmann,

B., Stocker, M., Giese, K. P., Perschke, A., Baumann, A. &Pongs, 0. (1989) EMBO J. 8, 3235-3244.

13. MacKinnon, R. & Miller, C. (1988) J. Gen. Physiol. 91,335-349.

14. Miller, C. (1988) Neuron 1, 1003-1006.15. MacKinnon, R. & Miller, C. (1989) Biochemistry 28, 8087-

8092.16. MacKinnon, R. & Miller, C. (1989) Science 245, 1382-1385.17. MacKinnon, R., Heginbotham, L. & Abramson, T. (1990)

Neuron 5, 767-771.18. MacKinnon, R. & Yellen, G. (1990) Science 250, 276-279.19. Gimenez-Gallego, G., Navia, M. A., Reuben, J. P., Katz,

G. M., Kaczorowski, G. J. & Garcia, M. L. (1988) Proc. Natl.Acad. Sci. USA 85, 3329-3333.

20. Massefski, W., Redfield, A., Hare, D. R. & Miller, C. (1990)Science 249, 521-524.

21. Studier, F. W. & Moffatt, B. A. (1986) J. Mol. Biol. 189,113-130.

22. Howell, M. L. & Blumenthal, K. M. (1989) J. Biol. Chem. 264,15268-15273.

23. Oprian, D. D., Molday, R. S., Kaufman, R. J. & Khorana,H. G. (1987) Proc. Nat!. Acad. Sci. USA 84, 8874-8878.

24. Smith, C. & Miller, C. (1986) J. Biol. Chem. 261, 14607-14613.25. Scopes, R. K. (1987) Protein Purification-Principles and

Practice (Springer, New York), 2nd Ed., pp. 248-249.

2050 Biochemistry: Park et al.