Proc. Nati. Acad. Sci. USAVol. 86, pp. 481-485, January 1989Biochemistry

Recombinant system for overexpression of cholera toxin B subunitin Vibrio cholerae as a basis for vaccine development

(cholera vaccine/hybrid protein/gene fusion vector/immunogenic carrier/cholera toxin epitopes)

J. SANCHEZ AND J. HOLMGRENDepartment of Medical Microbiology, University of Goteborg, Guldhedsgatan 1OA, Goteborg S413 46 Sweden

Communicated by Sune Bergstrom, October 17, 1988

ABSTRACT We have constructed an overexpression sys-tem in which the gene encoding the B subunit of cholera toxin(CTB) was placed under the control of the strong tacP promoterin a wide host range plasmid. Recombinant nontoxigenicclassical and El Tor Vibrio cholerae strains of different sero-types harboring this plasmid excreted 10- to 100-fold higheramounts of CTB than any other wild-type or recombinantstrain tested and may therefore be useful killed oral vaccinestrains. The manipulations to place the CTB gene under tacPalso included, by design, the introduction of single enzymerestriction sites for gene fusions to the CTB amino terminus.Cloning into these sites allows construction of CTB-derivedhybrid proteins carrying various putative vaccine peptideantigens.

Cholera and diarrhea caused by enterotoxigenic Escherichiacoli are important causes of morbidity and mortality in manydeveloping countries. Vibrio cholerae of serogroup 01 andenterotoxigenic E. coli may induce diarrhea when multiplyingin the gut of infected individuals by releasing cholera toxin(CT) or heat-labile enterotoxin (LT), respectively. Thesetoxins are very similar structurally and functionally and arecomposed of two types of subunits, a single copy of A (CTAor LTA), which is responsible for activation of the adenylatecyclase in the intestinal cell of the host, and a pentamer of B(CTB or LTB), which binds the respective holotoxin to itsintestinal receptor (1, 2).CTB is an effective oral immunizing agent, which in a large

field trial has been shown to afford protection against bothcholera and enterotoxigenic E. coli-caused diarrhea. This hasmade CTB as such an important component, together withkilled whole vibrios, of an oral cholera vaccine (3, 4).Moreover, CTB has attracted much interest recently as animmunogenic carrier for various other peptide or carbohy-drate antigens (5, 6) and has also been used as a receptor-blocking and receptor-modulating agent for short term pro-phylaxis of cholera and enterotoxigenic E. coli diarrhea (7, 8).These findings have now emphasized the need to increase theyields of CTB for large-scale production, ideally avoiding atthe same time the drawback in currently used preparationmethods of having to purify the CTB protein from activetoxin. To facilitate vaccine development based on use ofCTB, we have constructed an overexpression system inwhich the CTB gene is under the control of the strong tacPpromoter in a wide host range plasmid. Recombinant V.cholerae of different biotypes and serotypes carrying thisplasmid produced increased amounts of CTB, which wassecreted to the cell culture medium. Overexpression of CTBmay now aid its purification by simplified means and assist inpreparing improved killed oral cholera vaccine strains. Inaddition, the overexpression system developed allows for

gene fusions at the CTB amino terminus, facilitating theconstruction of peptide-bearing immunogenic hybrid pro-teins.

MATERIALS AND METHODSGene Fusions and DNA Sequencing. The unphosphorylated

oligodeoxynucleotides used to join the Sac 1 3' end of theLTB leader to the 5' Nde I of CTB were purchased as singlestrands from the Department of Immunology, BiomedicalCentre, University of Uppsala. After pairing, the oligonucle-otides had Sac I and Nde I compatible single-strandedextensions and could be joined directly to Sac 1/HindIIIrestricted pMMB68 (9, 10). Ligation was performed over-night at 40C with T4 ligase and with a 10-fold molar excess ofoligonucleotide to plasmid. To the ligation mixture a purifiedNde I/H ndIII fragment from plasmid pCVD30 containingthe CTB gene (11) was added in equimolar amounts withrespect to vector plasmid and the ligase reaction continued at4°C for 18 hr. The ligated DNA was subsequently trans-formed into competent E. coli HB101 cells with selection forampicillin resistance (100 ,ug/ml). To verify that the predictedsequences had been generated after cloning, the hybrid genewas subcloned into M13 and sequenced by the dideoxynu-cleotide method (12).

Transfer of Plasmids to V. cholerae and Induction byIsopropyl f8-D-Thiogalactopyranoside (IPTG). Plasmid pJS162containing the CTB gene under the control of tacP (Fig. 1)was transferred by conjugation from a helper E. coli strain(13) to the El Tor V. cholerae 01 JBK70 (14), using polymyxinB (50 units/ml) and ampicillin (100 ,ug/ml) for counterselec-tion of the donor. When the plasmid was transferred toclassical strains, rifampicin derivatives of them were firstisolated. Selection for the recipient vibrio strains in thisinstance was done with rifampicin (50 ,g/ml) and ampicillin(100 ,ug/ml). For induction with IPTG, cultures were grownat 37°C to the desired optical density (A6w) in LB brothsupplemented with antibiotics and then IPTG was added tothe final concentrations needed. Growth was continued for 4hr and the cells and culture supernatants were separated bycentrifugation. Cell pellets were gently washed and resus-pended in 1 original vol of cold phosphate buffered saline (pH7.2) to be broken by two 30-sec sonic bursts with a miniprobe(Branson Sonifier).Immunoassay of CTB. Detection of recombinant CTB was

carried out by the monosialoganglioside (GM1) ELISA (15)using anti-CTB monoclonal antibodies, which do not cross-react with LTB (16). Samples tested were either culturesupernatants or sonicated cell pellets. Quantitation of recom-binant CTB was made against a standard curve establishedfor concurrently tested purified CTB from strain 569B (kindlyprovided by Institut Merieux, Lyon, France).

Abbreviations: aa, amino acid(s); CT, cholera toxin; CTA, CT Asubunit; CTB, CT B subunit; GM1, monosialoganglioside; IPTG,isopropyl ,-D-thiogalactopyranoside; LTB, B subunit ofEscherichiacoli heat-labile enterotoxin.

481

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.

482 Biochemistry: Sanchez and Holmgren

Purification of CTB and Amino Terminus Determination. Therecombinant CTB was purified from a culture supernatant ofJBK70 carrying plasmid pJS162 by affinity chromatography ona lyso-GM1 ganglioside Spherosil column (17). Purified CTBwas subjected to determination ofthe amino-terminal residue asdescribed by Von Bahr-Lindstrom et al. (18).NaDodSO4/PAGE. Purified recombinant CTB was elec-

trophoresed in NaDodSO4/13.5% polyacrylamide slab gelsand the proteins were stained with Coomassie blue. To testfor association of CTB into oligomers, samples were treatedjust prior to electrophoresis with 0.1% NaDodSO4 samplebuffer for 5 min at room temperature or subjected to boiling.

Association with CTA in Vitro. Purified CTA (prepared from569B CT; List Biological Laboratories, Campbell, CA) wasmixed with purified recombinant or 569B CTB to give a totalprotein concentration of 200 pug/ml (19) and in the molarCTB/CTA ratio normally found in intact CT. After mixing,samples were acidified with 0.2 M glycine buffer (pH 2.7) andthen neutralized by dialysis overnight against several changesof 0.05 M Tris buffer (pH 8.0). The neutralized samples werethen tested with subunit-specific monoclonal antibodies byGM1 ELISA. The amounts of 569B CTA that associated withrecombinant or 569B CTB to give holotoxin were calculatedusing untreated homologous CT as a reference.

Immunization of Rabbits and Serological Methods. Adultrabbits were immunized subcutaneously with 30 pug of therecombinant CTB or 569B CTB biweekly, using completeFreund's adjuvant for the first inoculation and then incom-plete adjuvant. After the third immunization, the animalswere bled and serum was collected for analysis.Double diffusion in gel immunoprecipitation analysis ad

modum Ouchterlony was performed using the microchambersystem described by Wadsworth (20).

Titration of anti-toxin antibodies was performed by GM1ELISA (15) using purified recombinant and 569B CTB asantigens attached to GM1-coated plates. Neutralizing anti-bodies were assayed by the skin test method of Craig (21)using purified 569B CT as test toxin. Heat-inactivated serumwas mixed in serial dilutions with equal volumes of purified569B CT (40 ng/ml in physiological phosphate buffer sup-plemented with 10 mg of bovine serum albumin per ml), and0.1 ml of the mixture was then tested for residual toxicity byintradermal injection.

RESULTSPlacement of the CTB Gene Under tacP. The DNA encoding

the LTB leader has a single EcoRI restriction site at its 5' endlocated just upstream of the ribosome binding site that wasused to insert the LTB gene after the tacP promoter (10). Toprofit from this strategically located EcoRI site (which ismissing in the CTB gene) for expression of CTB from thesame promoter, we fused genetically the mature CTB proteinto the leader peptide ofLTB in pMMB68. The CTB gene frompCVD30 (originating from V. cholerae strain 0395, classicalOgawa) has an Nde I site at the position for amino acid (aa)18 of the leader peptide, while the LTB gene has a Sac Irecognition sequence at the beginning of the mature protein.Fusion of the CTB gene by its 5' Nde I end to the 3' Sac I endin the LTB gene, via an ad hoc synthetic linker (Fig. 1A),leads to substitution of the CTB leader peptide by that inLTB. The resulting pJS162 plasmid (Fig. 1B) contained thehybrid CTB gene as an EcoRI/HindIII downstream of thetacP promoter.

Analysis of the CTB Gene Sequence. Sequencing of thehybrid CTB gene in plasmid pJS162 confirmed the sequencereported for the LTB leader portion (22) and showed a highdegree ofoverall homology with previously reported El Tor andclassical CTB mature sequences (23-25). A comparison be-tween our recombinant CTB (from a V. cholerae 0395 classicalstrain) and those other sequences is presented in Fig. 2.

A

? -HI -1 +

OGCA WAoC &A o6 tOC Ai-4 -~ i2 2

T.iTG.CA CAY .31A AI

B

FIG. 1. (A) Fusion joint between the LTB and CTB genes. Thenucleotides underlined with open squares indicate the DNA from theLTB gene, those underlined with solid squares are the syntheticoligodeoxynucleotide part of the linker, and those underlined withtriangles denote the CTB gene DNA. The numbering over the aminoacids refers to their former positions with respect to the first aa (+ 1)in their respective mature proteins. The stars indicate amino acidsnot originally encoded by any of the two fused genes. The syntheticlinker restored the Sac I site and introduced a Sma I recognitionsequence. After the fusion, a plasmid encoding CTB from tacP wasobtained (see B below). The sequence shown has been confirmed bydideoxynucleotide sequencing (12). (B) Map of plasmid pJS162 withthe CTB gene under the control of the tacP promoter. The plasmidhas an RSF1010 origin of replication (9) and is -10.2 kilobases (kb).The large arrow denotes the position of the tacP promoter. Thestarred box represents the gene portion encoding mature CTB. Thesection encoding the leader peptide (originating from LTB) is shownby the solid box. Approximate positions of the ampicillin resistanceand the 1acIq genes are indicated.

Expression of CTB from tacP in V. cholerae. Conjugaltransfer of pJS162 to the toxin-deleted JBK70 V. choleraestrains followed by induction with IPTG led to production of40-50 ,tg of CTB per ml, of which >95% was secreted intothe culture medium (Fig. 3). Expression of CTB to levels upto 50-100 Ag/ml was also achieved in several toxin- orCTA-deleted classical and El Tor strains (Table 1). Thisrepresented a marked overexpression of CTB in comparisonwith the levels produced by many wild-type or recombinantV. cholerae strains examined, including the hypertoxigenic569B strain that is currently used for purification of CTB forvaccine production. The recombinant CTB could then readilybe purified in high yield from the culture supernatants ofeither of these strains using receptor-specific affinity chro-matography on lyso-GM1 ganglioside (17).

Characterization of Recombinant CTB Protein. (i) Aminoterminus determinations. Cleavage of the precursor peptideof the recombinant CTB would have been naturally predictedto take place at either one or both of the original leaderpeptidase recognition sites in LTB or CTB. Identification ofthe first amino acid in the purified protein gave both tyrosineand alanine residues. This and the fact that the recombinantCTB was slightly larger than native CTB as determined byNaDodSO4/PAGE (Fig. 4) suggested that proteolytic pro-cessing of the leader peptide had taken place between theglycine encoded by the linker (position -5) and tyrosine(position -4) as well as between this latter amino acid andalanine at -3 (Fig. 2). When the remainder of the treated CTBwas subjected again to amino terminus determination alanineand histidine residues were now identified, confirming the

Proc. Natl. Acad. Sci. USA 86 (1989)

Proc. Natl. Acad. Sci. USA 86 (1989) 483

-20

SD Met Asn Lys Val Lys Phe Tyr Val Leu Phe Thr Ala Leu

AATTCGGGATGAATT ATG AAT AAA GTA AAA TTT TAT GTT TTA TTT ACG GCG TTA

Leu Ser Ser Leu

CTA TCC TCT CTA

-1 0

Cys Ala HisTGT GCA CAC

Gln Asn Ile Thr Asp Leu CysCAA AAT ATT ACT GAT TTG TGT

20 (Asn)

(+1) * * j -1 +1Gly Ala Pro Gly Tyr Ala His Gly Thr Pro

GGA GCT CCC GGG TAT GCA CAT GGA ACA CCT

10 (Tyr)Ala Glu Tyr His Asn Thr Gln Ile His Thr

GCA GAA TAC CAC AAC ACA CAA ATA CAT ACG

30

Leu Asn Asp Lys Ile Phe Ser Tyr Thr Glu Ser Leu Ala Gly Lys Arg Glu

CTA AAT GAT AAG ATA TTT TCG TAT ACA GAA TCT CTA GCT GGA AAA AGA GAG

40 (Ile) 50

Met Ala Ile Ile Thr Phe Lys Asn Gly Ala Thr Phe Gln Val Glu Val Pro

ATG GCT ATC ATT ACT TTT AAG AAT GGT GCA ACT TTT CAA GTA GAA GTA CCA

(Ser) 60 (Asn)

Gly Ser Gln His Ile Asp Ser Gln Lys Lys Ala Ile Glu Arg Met Lys AspGGT AGT CAA CAT ATA GAT TCA CAA AAA AAA GCG ATT GAA AGG ATG AAG GAT

80

Thr Leu Arg Ile Ala Tyr Leu Thr Glu Ala Lys Val Glu Lys Leu Cys Val

ACC CTG AGG ATT GCA TAT CTT ACT GAA GCT AAA GTC GAA AAG TTA TGT GTA

90 100

Trp Asn Asn Lys Thr Pro His Ala Ile Ala Ala Ile Ser Met Ala Asn End

TGG AAT AAT AAA ACG CCT CAT GCG ATT GCC GCA ATT AGT ATG GCA AAT TAA

GATATAAAAAAGCCCACCTCAGTGGGCTTTTTT

FIG. 2. Nucleotide sequence of the CTB gene in plasmid pJS162. Only the anti-sense strand is shown, and amino acids encoded are presentedabove their respective codons. The threonine residue numbered +1 is the first amino acid normally found in mature CTB, while the alanineresidue at position -7 (or + 1 in brackets) is the first amino acid in mature LTB. Amino acids not initially present in any of the two proteinsare indicated by stars. A potential ribosome binding site (Shine-Dalgarno sequence, SD) is underlined. The vertical arrows indicate the peptidebonds cleaved to release the mature recombinant CTB. Amino acids in the 569B CTB protein sequence (26, 27), which disagree with thosepredicted by our CTB DNA sequence, are in parentheses, residues in the CTB from El Tor strains (23-25), which differ from ours and thosein classical 569B CTB, are in parentheses and in boldface type.

postulated cleavage positions and providing evidence that therecombinant CTB carried short peptide extensions at itsamino terminus consisting of either 3 or 4 amino acids.

(ii) Receptor and CTA subunit recognition. The presenceof the few extra amino acids in the CTB did not affect its

50 *- SupernatantA-ACells x 10

40 -

t 30-_E_

Li 20 - )

U 2010

0 20 40 60 80 100IPTG (pg/ml)

FIG. 3. Induction of CTB expression by IPTG in V. choleraeJBK70 (pJS162). Levels ofCTB (ordinate), expressed as equivalents ofthe purified 569B CTB protein, were determined by GM1 ELISA usinga specific monoclonal antibody. The actual concentrations in the cellpellets were 10 times lower than shown in the graph (cells x 10).

recognition of the GM1 receptor. The binding affinity forplastic-coated GM1 ganglioside was compared for the recom-binant CTB and 569B CTB using the GM1 ELISA and nodifference was revealed (data not shown). Retention of highbinding affinity for GM1 ganglioside was in fact also takenadvantage of in the purification of CTB as described above.The recombinant CTB was also unaffected in its ability to

both oligomerize and associate with CTA, as shown by thehigh percentages of holotoxin recovered after acidificationand neutralization of the mixed subunits (Table 2).

(iii) Immunological properties of recombinant CTB. Todetermine whether the extra amino acids in CTB affected itsimmunoreactivity or even resulted in the appearance of newepitopes, we prepared rabbit antisera against recombinantand 569B CTB. The anti-CTB titers in GM1 ELISA weresimilar (range, 4 x 105 to 1 x 106) for anti-recombinant andanti-569B CTB as tested with either of the two CTB prepa-rations as solid-phase antigen. Likewise, the neutralizingantitoxin titers did not differ, the sera against either recom-binant or 569B CTB being able to completely neutralize 2 ngof CT at a concentration of 1 x 10-4. Ouchterlony immuno-diffusion tests performed with the purified recombinant CTBor 569B CTB and their corresponding rabbit immune sera asreactants showed immunoprecipitation bands of coalescencewithout any "spurs," indicating complete "identity" be-tween the recombinant and native CTB (Fig. 5).

Biochemistry: Sanchez and Holmgren

484 Biochemistry: Sanchez and Holmgren

Table 1. Overexpression of CTB from the tacP promoterin V. cholerae

Strain Phenotype* CTB, ,ug/mlExpressing CTBfrom tacP promoterJBK70 (pJS162) El Tor Inaba 75

CTA- CTB-JS1569 (pJS162) Classical Inaba 75

CTA- CTB+JS1395 (pJS162) Classical Ogawa 54

CTA- CTB+Expressing CT or CTBfrom own promoter569B Classical Inaba 1.28

CTA+ CTB+CVD103 Classical Inaba 0.88

(569B derivative) CTA- CTB+All strains were grown for the same periods at 30'C. Strains

carrying pJS162 were grown to A600 = 1.0 before addition of IPTGto 100 ug/ml, after which incubation was continued for 4 hr. StrainsJS1569 and JS1395 are rifampicin-resistance derivatives from strainCVD103 and CVD101 (11) (0395 CTA- derivative), respectively. Thelevels ofCTB produced by these latter strains in the absence of IPTGwere <0.5 ,tg/ml and therefore did not contribute significantly to thevalues in the presence of the inducer.*Nonfunctional (CTA-) or functional (CTA+) CTA gene and/ornonfunctional (CTB-) or functional (CTB+) CTB gene. Strainswere prepared by gene deletion (11).

DISCUSSIONThe demonstration that oral immunization with CTB pro-vides protection against cholera and enterotoxigenic E. colidiarrhea in humans (3, 4) and the evidence in experimentalsystems that CTB is also an effective immunogenic carrier forvarious other peptide or carbohydrate vaccine candidateantigens (5, 6) have emphasized the need to develop meansto produce high levels of this protein. We describe here asystem in which using recombinant DNA technology we have

MW97

66

43

31

21

14 Bm; 2 3 4

FIG. 4. PAGE of boiled and unboiled samples of CTB. Equalamounts of recombinant CTB protein (lanes 1 and 2) or reference569B CTB (lanes 3 and 4) were electrophoresed in a NaDodSO4/13.5% polyacrylamide gel after treatment in sample buffer for 5 minat 100°C (lanes 1 and 3) or at room temperature (lanes 2 and 4). Amolecular mass marker (Bio-Rad) with the approximate sizes ofprotein standards (kDa) is shown (lane MW). The slower migrationof the recombinant CTB as compared to the 569B CTB is only slightlynoticeable when examined as the monomers (Bin) but is more obviousin the oligomeric (pentameric) forms (Bp).

Table 2. Affinity of recombinant CTB for CTA

Amount of subunit in holotoxin

Association of as determined by GM1 ELISACTA with

569B CTBRecombinant CTB

CTA,,ug/ml CTB,,g/ml

43.3 (65%) 114.0 (86%)42.0 (63%) 109.6 (82%)

The affinity of recombinant or 569B CTB for CTA was tested bytheir ability to associate in vitro to form cholera holotoxin. Mixturesof CTA and CTB in a 1:5 molar ratio were acidified to pH 2.7 todissociate the CTB pentamers (19). Association between CTA andthe added CTB was subsequently favored by neutralization of thesolution by dialysis against Tris (pH 8.0). Neutralized samples wereassayed directly by GM1 ELISA with CTA- and CTB-specificmonoclonal antibodies. Values in parentheses refer to the fraction ofthe subunit found as CT calculated with relation to the maximumamount theoretically able to form holotoxin.

achieved overexpression of CTB by placing the CTB struc-tural gene under the control of the strong tacP promoter.Various V. cholerae strains harboring the wide host rangeplasmid (pJS162) containing the tacP-directed CTB overex-pression system expressed 10- to 100-fold higher levels ofCTB than any other wild-type or recombinant strains tested.The reason for the efficient expression of CTB attained by

us in contrast with previous attempts using the same pro-moter (25) has not been established. The removal of the CTBDNA upstream of the Nde I site used for cloning may haveeliminated sequences interfering with high expression byaffecting transcription (28) or translation (29). A positiveeffect on translation by the LTB ribosome binding site(Shine-Dalgarno sequence) seems unlikely since its comple-mentarity to the 16S rRNA 3' end is less than that of theoriginal CTB Shine-Dalgarno sequence (23). Although highexpression of CTB in V. cholerae has been consideredpossible only when directed from the original ctx promoter andin the presence of ancillary regulatory elements (25), ourdemonstration of CTB overexpression by tacP suggests thatcontrol by other strong foreign promoters might also be feasible.Indeed, we have preliminary data showing high levels of CTBexpression in V. cholerae determined by the T7 RNA poly-merase-dependent promoter (30) (unpublished data).

Sequencing of the CTB gene in plasmid pJS162 showed ahigh degree of homology with other published CTB se-quences but also revealed some interesting dissimilarities.DNA sequences of cloned CTB genes from several V.cholerae El Tor strains are available (23-25), whereas onlythe CTB protein sequence for the 569B classical strain has

FIG. 5. Ouchterlony double-diffusion in gel analysis of 569BCTB (wells B and F) and recombinant CTB (wells G and D) reactedwith rabbit antisera (wells A and E, anti-recombinant CTB; well C,anti-569B CTB).

Proc. Natl. Acad. Sci. USA 86 (1989)

Proc. Natl. Acad. Sci. USA 86 (1989) 485

been described (26, 27). Among the four (24) or five (25) aareported to differ between the 569B CTB protein sequenceand the various El Tor CTB DNA sequences, Asp-70 hasbeen proposed to determine an El Tor-specific antigenepitope (24). We found that in our CTB sequence aa 70 and22 were both aspartic acid as in the El Tor CTBs, whileHis-18, Thr-47, and Gly-54 (Tyr-18, Ile-47, and Ser-54 in theEl Tor CTB) were as in 569B CTB. This suggests that Asn-70(or Asp-70) plays no role in defining classical (or El Tor)specific CTB epitopes. His(Tyr)-18, Thr(Ile)-47, and Gly-(Ser)-54, on the other hand, might be relevant determinantsfor such epitopes, but before any conclusions can be drawnthere is a need to sequence at least a few additional classicalCTB genes to determine possible internal variability amongclassical strains.The strategy followed to replace genetically the CTB leader

peptide by that in LTB to achieve overexpression from tacPproduced a hybrid leader peptide in which the original leaderpeptidase cleavage sites in LTB and CTB were both present.Processing of the leader peptide occurred at two placesbetween these sites. As a result, the recombinant matureCTB carried an extension of 3 or 4 amino acids at its aminoterminus. These short peptide additions, however, did notcause any change in the immunological reactivity or otherproperties of recombinant CTB. Thus, as compared withnative 569B CTB our recombinant CTB exhibited a normalability to form pentamers and to associate with native CTAas well as unchanged GM1 ganglioside receptor bindingaffinity. Furthermore, the immunological analyses demon-strated that recombinant CTB was identical with the refer-ence 569B CTB in its immunological determinants and in itscapacity to give rise to anti-toxin antibodies.

Overexpression of CTB as reported here should nowfacilitate the production of this protein for use as a protectiveimmunogen in oral vaccination against cholera and LT-caused E. coli diarrhea. It is a particular advantage from thepoint of both purification and standardization of CTB as avaccine component that overexpression can be achieved byV. cholerae strains that stably, through genetic deletion, lackthe ability to produce cholera holotoxin and/or its A subunit.In addition, because overexpression of CTB was attained innontoxigenic El Tor and classical V. cholerae 01 of eitherInaba or Ogawa serotypes, it should be possible to preparevaccine analogous to the oral B subunit-whole cell choleravaccine recently field tested in Bangladesh (3) by simplygrowing and inactivating a combination of these CTB-hyperproducing strains.As mentioned, CTB has also attracted much interest

recently as an immunogenic carrier for other vaccine relevantantigen epitopes. Hybrid derivatives of CT and LT entero-toxins for immunoprotection have been made by chemicalcoupling (refs. 5, 31, and 32; C. Czerkinsky, M. W. Russell,N. Lycke, M. Lindblad, and J.H., unpublished data) or bygenetic fusion (33, 34). To create the possibility of obtainingCTB-derived hybrid proteins by gene fusion as an extrabonus of the tacP-directed overexpression system, we ar-ranged the constructions so as to have adjacent single Sac Iand Sma I recognition sequences at the amino terminus of theCTB gene. Use of these restriction sites allows linkage ofdiverse peptide antigens to CTB for immunoprotection stud-ies against, e.g., viral, bacterial, and parasitic diseases. The

accessibility of this approach has recently been demonstratedby the fusion and expression of a synthetic DNA sequenceencoding a nontoxic decapeptide antigen, derived from the E.coli heat-stable enterotoxin, to the pJS162 CTB gene (35).We thank Susanne Johansson for skillful technical assistance,

Mats Lake (Kabigen, Stockholm) for help in the DNA sequencing,and M. Lindblad and the Department of Clinical Chemistry, Uni-versity of Goteborg, for determination of the CTB amino terminus.We also thank B. E. Uhlin for useful discussions and J. B. Kaper forproviding various V. cholerae and E. coli strains. This work wasfinancially supported by the Swedish Medical Research Council andthe World Health Organization.1. Holmgren, J. (1981) Nature (London) 292, 413-417.2. Holmgren, J., Fredman, P., Lindblad, M., Svennerholm, A.-M. &

Svennerholm, L. (1982) Infect. Immun. 38, 424-433.3. Clemens, J. D., Sack, D. A., Harris, J. R., Chakraborty, J., Khan,

M. R., Stanton, B. F., Kay, B. A., Khan, M. U., Yunus, M. D.,Atkinson, W., Svennerholm, A.-M. & Holmgren, J. (1986) Lancet i, 124-127.

4. Clemens, J. D., Sack, D. A., Harris, J. R., Chakraborty, J., Neogy,P. K., Stanton, B., Huda, N., Khan, M. U., Kay, B. A., Khan, M. R.,Ansaruzzaman, Yunus, M., Rao, M. R., Svennerholm, A.-M. & Holm-gren, J. (1988) J. Infect. Dis. 158, 372-377.

5. McKenzie, S. J. & Halsey, J. F. (1984) J. Immunol. 133, 1818-1824.6. Aizpurua, H. J. & Rusell-Jones, G. J. (1988) J. Exp. Med. 167, 440-451.7. Glass, R. I., Holmgren, J., Khan, M. R., Hossain, K. M. B., Huq, M. I.

& Greenough, W. B. (1984) J. Infect. Dis. 149, 495-500.8. Donta, S. T., Damiano-Burbach, P. & Pointdexter, N. J. (1988) J. Infect.

Dis. 157, 557-564.9. Furste, J. P., Pansegrau, W., Frank, R., Blocker, H., Scholz, P.,

Bagdasarian, M. & Lanka, E. (1986) Gene 48, 119-131.10. Sandkvist, M., Hirst, T. R. & Bagdasarian, M. (1987) J. Bacteriol. 169,

4570-4576.11. Kaper, J. B., Lockman, H., Baldini, M. M. & Levine, M. M. (1984)

Biotechnology 2, 345-349.12. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. Acad. Sci.

USA 74, 5463-5467.13. Simon, R., Priefer, U. & P~ifler, A. (1983) Biotechnology 2, 784-791.14. Kaper, J. B., Lockman, H., Baldini, M. M. & Levine, M. M. (1984)

Nature (London) 308, 655-658.15. Svennerholm, A. & Holmgren, J. (1978) Curr. Microbiol. 1, 19-23.16. Lindholm, L., Holmgren, J., Wikstrom, M., Karlsson, U., Andersson,

K. & Lycke, L. (1983) Infect. Immun. 40, 570-576.17. Tayot, J. L., Holmgren, J., Svennerholm, L., Lindblad, M. & Tardy, M.

(1981) Eur. J. Biochem. 113, 249-258.18. Von Bahr-Lindstrom, H., Hempel, J. & Jornvall, H. (1982) J. Protein

Chem. 1, 257-262.19. Hardy, S. J. S., Holmgren, J., Johansson, S., Sanchez, J. & Hirst, T. R.

(1988) Proc. Natl. Acad. Sci. USA 85, 7109-7113.20. Wadsworth, C. (1957) Int. Arch. Allergy 17, 165-177.21. Craig, J. P. (1965) Nature (London) 207, 614-616.22. Leong, J., Vinal, A. C. & Dallas, W. S. (1985) Infect. Immun. 48, 73-77.23. Gennaro, M. L. & Greenaway, P. (1983) Nucleic Acids Res. 11, 3855-

3861.24. Lockman, H. & Kaper, J. B. (1983) J. Biol. Chem. 258, 13722-13726.25. Mekalanos, J. J., Swartz, D. J., Pearson, G. D. N., Harford, N.,

Groyne, F. & De Wilde, M. (1983) Nature (London) 306, 551-557.26. Kurosky, A., Markel, D. E. & Peterson, J. (1977) J. Biol. Chem. 252,

7257-7264.27. Lai, C. Y. (1977) J. Biol. Chem. 252, 7249-7256.28. McClure, W. R. (1985) Annu. Rev. Biochem. 54, 171-204.29. Yamamoto, T., Suyama, A., Mori, N., Yokota, T. & Wada, A. (1985)

FEBS Lett. 181, 377-380.30. Tabor, S. & Richardson, C. C. (1985) Proc. Natl. Acad. Sci. USA 82,

1074-1078.31. Klipstein, F. A., Engert, R. F. & Houghten, R. A. (1983) Infect. Immun.

40, 888-893.32. Svennerholm, A.-M., Wikstrom, M., Lindblad, M. & Holmgren, J. (1986)

J. Clin. Microbiol. 24, 585-590.33. Sanchez, J., Uhlin, B. E., Grundstrom, T., Holmgren, J. & Hirst, T. R.

(1986) FEBS Lett. 208, 194-198.34. Sanchez, J., Hirst, T. R. & Uhlin, B. E. (1988) Gene 64, 265-275.35. Sanchez, J., Svennerholm, A.-M. & Holmgren, J. (1988) FEBS Lett. 241,

110-114.

Biochemistry: Sanchez and Holmgren