Molecular Cloning of Vibrio Cholerae Enterotoxin Genes in Escherichia Coli K-12 (Cholera Toxin/Gene Duplication/Secretion/Proteolytic Processing) GREGORY D

Proc. Nati Acad. Sci. USA Vol. 79, pp. 2976-2980, May 1982 Genetics

Molecular cloning of Vibrio cholerae enterotoxin genes in Escherichia coli K-12 (cholera toxin/gene duplication/secretion/proteolytic processing) GREGORY D. N. PEARSON AND JOHN J. MEKALANOS Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, Massachusetts 02115 Communicated by A. M. Pappenheimer, Jr., January 27, 1982

ABSTRACT Hybridization probes derived from the A and B toxinogenic strains of E. coli (14). Cholera toxin is found pri- subunit genes ofthe heat-labile enterotoxin (LT) ofEscherichia coli marily as an extracellular protein and is presumably secreted were used to analyze DNA from Vibrio cholerae strain 569B for by V. cholerae. In contrast, the LT produced by E. coli remains cholera toxin gene sequences. Southern blot analysis indicated that tightly cell associated and may exist as an outer membrane or the cholera toxin A and B subunit genes were each duplicated in periplasmic protein (4, 15). A possibly related observation in- this strain. One of the two toxin subunit gene pairs was cloned as volves the relative extent ofproteolytic processing sustained by a 5. 1-kilobase DNA insert in plasmid pBR322. E. coli cells car- each of these toxins. Whereas the A subunit ofcholera toxin is rying the recombinant plasmid pJM17 were shown to produce generally found in a "nicked" form consisting ofdisulfide-linked cholera toxin, which was found to be largely cell associated. Pro- A1 and the A subunit of tein chemical analysis indicated that the toxin was in its unnicked (Mr 22,000) A2 (Mr 6,000) peptides, form and required additional proteolytic processing by trypsin to LT is usually composed of a single, unnicked A chain (Mr exhibit full toxicity in tissue culture. The alteration in E. coli ofthe 25,500-28,000) (3, 4, 16). secretion and proteolytic processing ofcholera toxin parallels that Do the reasons for these differences lie mainly at the level previously observed for LT. An in vitro generated insertion mu- ofthe specific toxin genes or might they be more related to the tation in the A subunit gene on pJM17 was shown to abolish pro- particular host organism in which the toxin genes reside? One duction of the A chain but still allow production of the B chain. possible means of answering this question as well as investi- These observations, together with restriction mapping data, have gating other aspects oftoxinogenesis is through the introduction demonstrated that the cholera toxin and LT genes are very similar ofthe cholera toxin gene into E. coli via recombinant DNA tech- in their genetic organization. nology. Moreover, the in vitro alteration of the cloned toxin gene would offer a powerful approach to the construction of Cholera toxin and the heat-labile enterotoxin (LT) of Esche- toxin mutations that might have practical value in cholera vac- richia coli share many structural and genetic properties. Both cine development. toxins are multimeric proteins composed of two types of sub- In this report we describe the cloning in E. coli ofa V. chol- units, A and B (1-4). The A subunits (Mr 25,000-28,000) are the erae DNA fragment that codes for production of cholera toxin. enzymically active moieties ofthe toxins and are known to pro- We have found that both the secretion and the proteolytic pro- mote the activation of adenylate cyclase in eukaryotic cells by cessing of the toxin subunits are altered in the E. coli clone in catalyzing the ADP-ribosylation of a GTPase regulatory com- a manner similar to that previously reported for LT. The genetic ponent ofthe cyclase complex (5, 6). The B subunits (Mr 11,600) organization of the cholera toxin A and B subunit genes is also are present in five copies per toxin molecule and display a high analogous to that of the LT cistrons. binding affinity for the toxins' probable cell surface receptor, ganglioside GM1 (2, 7). The two toxins have been found to pos- MATERIALS AND METHODS sess both common and unique antigenic determinants (8). The Bacterial Strains. V. cholerae 569B (prototrophic, Inaba se- DNA sequences of the cloned genes coding for the LT A and rotype) and E. coli MS371 (same as SK1592, from Sidney Kush- B subunits have been largely determined and have been pre- ner; F- gal thi endA sbcB hsdR4 hsdM+) were maintained at dicted to code for polypeptides that show a high degree of ho- -70'C in CYE broth (17) containing 15% (vol/vol) glycerol or mology with the known amino acid sequences of the cholera on LB plates (18). Plasmid content ofa strain is given in paren- toxin subunits (9, 10). Furthermore, Southern blot hybridiza- theses after its name. tion, utilizing probes derived from the two LT structural genes, Toxin Production and Assays. Bacteria were grown in CYE has demonstrated that toxinogenic strains of Vibrio cholerae broth at 300C for 18 hr with vigorous aeration. The cells were possess DNA sequences homologous to the LT genes (11, 12). collected by centrifugation and cell extracts were prepared as The conclusion that this homologous DNA contains the struc- follows. Cells were resuspended in an equal volume of25 mM tural genes for cholera toxin is supported by the observation that Tris-HCl buffer, pH 8.0/10 mM EDTA/8 ,Ag ofphenylmethyl- certain nontoxinogenic mutants of V. cholerae have lost all or sulfonyl fluoride per ml/3 mg of lysozyme per ml. After incu- part of these DNA sequences (13). bating 15 min at 220C the solution was frozen at -700C and Although the above discussion outlines some of the striking thawed at 37°C. This was repeated four times and then DNase similarities in the protein and nucleic acid chemistry ofLT and I and MgCl2 were added to 10 ,g/ml and 10 mM, respectively. cholera toxin, several differences can be noted. The level of Centrifugation at 20,000 X g for 15 min produced a clear su- toxin production by V. cholerae varies widely between strains pernatant extract, which was stored at -70°C until used in toxin but is generally much higher than LT production by entero- assays.

The publication costs ofthis article were defrayed in part by page charge Abbreviations: LT, heat-labile enterotoxin of E. coli; LT-A and LT-B, payment. This article must therefore be hereby marked "advertise- the A and B subunits of LT; kb, kilobase(s); ELISA, enzyme-linked im- ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact. munosorbent assay. 2976 Downloaded by guest on September 29, 2021 Genetics: Pearson and Mekalanos Proc. Natd Acad. Sci. USA 79 (1982) 2977

Toxin antigen was measured with an enzyme-linked immu- A probe B probe nosorbent assay (ELISA) performed essentially as described by kb 1 2 3 4 5 Std 1 2 3 4 5 Holmgren (19). Toxic activity was measured in the S49 mouse lymphosarcoma cell assay (20). Purified cholera toxin (17) was used as the standard in the above assays. DNA Preparation and Analysis. DNA was prepared from bacterial cells by the method of Brenner et aL (21). DNA re- 9.5 striction enzymes and phage T4 DNA ligase were purchased from New England BioLabs and T4 DNA polymerase from Be- 6.4 thesda Research Laboratories. These enzymes were used under conditions suggested by the suppliers. LT sequence probes 4.2 were prepared from plasmid EWD299 (22) by the methods of Moseley and Falkow (11). The LT-A probe was a 770-base pair Xba I/EcoRI fragment and the LT-B probe was a 590-base pair EcoRI/HindIII fragment ofEWD299 (Fig. 2). These fragments 2.2 and DNA molecular weight standards were labeled with [a- 1.8 32P]dCTP (New England Nuclear) by the method of Maniatis S et aL (23) to a specific activity of about 108 cpm/gg. Electro- phoresis in 0.7% agarose gels and Southern blot hybridization FIG. 1. Southern blot hybridization. DNA was digested with the indicated restriction enzymes, fractionated by electrophoresis in a on nitrocellulose sheets (Schleicher & Schuell) were performed 0.7% agarose gel, and transferred to a nitrocellulose sheet by the as described (24). method of Southern (23). Lanes 1-5 on the left were hybridized with Protein Analysis in Recombinant Strains. Strains were ra- lo0 cpm of 32P-labeled LT-A probe, and lanes 1-5 on the right were dioactively labeled by growing cells for 18 hr in 5 ml of M63 hybridized with 106 cpm of 32P-labeled LT-B probe. Lane 1, 1 ,ug of glucose minimal broth (18) containing 2 mM each of 19 non- 569B DNA digested with EcoRI; lane 2, 1 gg of 569B DNA digested radioactive amino acids (no methionine), and 100 pCi (1 Ci = withPst I; lane 3, 1 ,ug of 569B DNA digested with bothEcoRI andPst 3.7 x 1010 becquerels) of (New England Nu- I; lane 4, 0.1 jig of pJM17 digested with EcoRI and Pst I; lane 5, 0.1 [fS]methionine jtg of pJM17 digested withPst I. Std, molecular weight standard, 32p- clear). Cells were harvested and lysed as described above for labeled A phage DNA digested with HindM. toxin assays. Immunoadsorption with rabbit antitoxin and formalin-treated Staphylococcus aureus was performed essentially as described (25). The adsorbed proteins were solubilized from Cloning ofthe Cholera Toxin A and B Subunit Genes. Strain the immune complexes by boiling in NaDodSO4 sample buffer 569B DNA, doubly digested with EcoRI and Pst I, was frac- and were then electrophoresed in 15% polyacrylamide/ tionated by electrophoresis in a 0.7% agarose gel. The DNA in NaDodSO4 gels as described (26). The gels were treated with the area ofthe gel containing fragments in about the 4.5- to 5.5- EN3HANCE (New England Nuclear), dried, and autoradio- kb range was recovered from the gel by electroelution followed graphed at -70°C, using X-AR5 film (Eastman). by ethanol precipitation. This DNA fraction (20 ng) was mixed with pBR322 (27) DNA (100 ng) that had been previously di- RESULTS gested with both EcoRI and Pst I and treated with alkaline phos- Southern Blot Analysis of V. cholerae 569B. V. cholerae phatase. The mixture was treated with T4 ligase, introduced 569B is a strain that has been widely used in studies addressing into E. coli MS371 by transformation, and plated on LB plates the protein chemistry and genetics ofcholera toxin. On the basis containing tetracycline at 15 ,g/ml. Tetracycline-resistant col- ofresults ofSouthern blot hybridization utilizing LT-A and LT- onies that were ampicillin sensitive were identified and then B gene probes, Moseley and Falkow (11) have suggested that tested for the presence of the cholera toxin A subunit gene by this strain might possess multiple copies of the cholera toxin A using the LT-A probe and the colony hybridization method of and B subunit genes. Detailed hybridization analysis of 569B Gruenstein and Hogness (28). has recently demonstrated that this strain does indeed possess A colony displaying a positive hybridization signal was iden- a duplication ofthe toxin A and B subunit genes. This conclusion tified by this method and the plasmid it contained was purified. is based primarily on the observation that a large number of Southern blot analysis confirmed that this plasmid, pGP1, con- restriction endonucleases were found to produce two different tained a 5.1-kb Pst I/EcoRI fragment that hybridized to the LT- bands that hybridize to both LT-A and LT-B probes in Southern A probe (data not shown). However, additional restriction anal- blots. Fig. 1 illustrates this result for two enzymes, EcoRI and ysis indicated that pGP1 also contained two additional small Pst Pst I, in Southern blots of 569B DNA hybridized with LT-A I/EcoRB inserts ofabout 500 base pairs each (Fig. 2). These were probe. EcoRI produced two bands, 23 and 17 kilobases (kb) in deleted from pGP1 by digestion ofthe plasmid with EcoRI fol- size, and Pst I produced two bands, 9.5 and 5.4 kb in size. lowed by ligation to give plasmid pJM17, which contained only Digestion of 569B DNA with both enzymes simultaneously also a single 5.1-kb Pst I/EcoRI insert in pBR322. This insert was resulted in the generation of two bands, one of which comi- shown to contain both toxin A and B gene sequences in Southern grated with the 5.4-kb Pst I band and was presumably identical. blots and was found to comigrate with the 5.1-kb Pst I/EcoRI The 5.1-kb band in the double digest was smaller than either fragment detectable in 569B digests (Fig. 1). Digestion of one of the two Pst I or EcoRI bands and was therefore likely to pJM17 with Pst I alone produced the linearized plasmid, which be a Pst I/EcoRI subfragment derived from the 9.5-kb Pst I was 8.7 kb in length. fragment. Expression and Molecular Properties of Cholera Toxin En- An identical Southern blot was also hybridized with the LT- coded by pJM17. Cell-free culture fluid and cell extracts of B probe and the same bands were found to hybridize (Fig. 1). MS371(pJM17), MS371(pBR322), and V. cholerae 569B were Because the 5.1-kb Pst I/EcoRI fragment contained both toxin assayed for cholera toxin activity and antigen by the S49 tissue subunit gene sequences and was also ofa comparably small size, culture assay and the ELISA assay. Table, 1 shows that this DNA fragment presented an optimal target in our initial MS371(pJM17) produced cholera toxin detectable in both as- cloning experiments. says, whereas the control culture MS371(pBR322) was negative Downloaded by guest on September 29, 2021 2978 Genetics: Pearson and Mekalanos Proc. Nad Acad. Sci. USA 79 (1982) The cholera toxin produced by MS371(pJM17) was lower in specific toxicity than the toxin produced by 569B (Table 1). Pa .is However, treatment of the MS371(pJM17) cell extract with Eco RI trypsin resulted in a 260-fold increase in specific toxicity to a level equal to that ofpurified toxin (Table 1). Cholera antitoxin neutralized the toxic activity present in the MS371(pJM17) extract both before and after digestion with trypsin. Trypsin at the Hincil level used in the experiments was devoid ofS49 toxicity. These

0.65 that the toxin was EcoRI results suggested produced by MS371(pJM17) Pat 0.4 PvUl\ in a precursor form that required additional proteolytic pro- / pJMl7 cessing to express full toxicity. We examined the molecular nature ofthe toxin polypeptide / 2.7 1.06 chains produced by MS371(pJM17) in M63 medium containing 0.95 ~d [3S]methionine. Cell-free culture fluid and cell extracts were Xbal prepared as before for MS371(pJM17) and a portion of the MS371(pJM17) cell extract was also digested with trypsin under Pat I conditions known to enhance the toxicity of this extract. The I P4 Xba HincI EcoRI I PvNW~~~~~~I toxin-related polypeptides were prepared from the extracts by ) immunoadsorption using antitoxin and formalin-treated S. au- LT )o)I reus. Figure 3 shows that analysis ofthese preparations by electrophoresis in NaDodSOjpolyacrylamide gels and autoradi- ography revealed that MS371(pJMl7) produced two polypeptide FIG. 2. Restriction map of the cholera toxin and LT genes. Nu- chains that were specifically immunoadsorbed from the cell merical designations within pGP1 andpJM17 are the approximate size extracts. One polypeptide (Mr 28,000) comigrated with the un- of restriction fragments in kb. The thin line represents DNA derived nicked A chain of cholera toxin and the other comigrated with from the cloning vector pBR322. Arrows designate the approximate the B subunit of the toxin. Two peptides were also immunoad- locations of the toxins' A subunit (A) and B subunit (B) genes and the sorbed from the trypsin-treated MS371(pJM17) extract. One of direction of transcription and most probable location of the toxin promoters (P). TET, tetracycline resistance. The restriction map for the these again comigrated with the cholera toxin B subunit; how- LT genes is from Spicer et al. (10). ever, the other (Mr 22,000) comigrated with the Al peptide of cholera toxin. Peptides comigrating with the A2 polypeptide of cholera toxin are not resolved well from the B subunit in this in these assays. However, in contrast to 569B, which released gel system. These data indicate that MS371(pJM17) produces greater than 99% ofits toxin antigen into the culture fluid, the cholera toxin that is primarily in its unnicked form (16). As we toxin produced by MS371(pJM17) remained about 94% cell-as- previously reported, limited trypsin digestion converts this sociated. Other studies indicated that the cell-bound toxin was not substantially released by treatment of MS371(pJM17) cells with an osmotic shock (data not shown). These data suggest that 1 2 3 4 5 6 secretion of cholera toxin by 569B is probably a consequence of the physiology of V. cholerae and not an inherent property of the cholera toxin gene.

Table 1. Production of cholera toxin in liquid culture Toxin Specific Sample* antigent toxicityt * A1 Purified toxin 260,000 569B-S 12 270,000 569B-E 0.027 470,000 MS371(pJM17)-S 0.032 25,000 MS371(pJM17)-E 0.50 1,000 S- e B MS371(pBR322)-S <0.005 ND A2 MS371(pBR322)-E <0.005 ND MS371(pJM17)-E + Ab 0.50 <100 MS371(pJM17)-E + T 0.50 260,000 MS371(pJM17)-E + T + Ab 0.50 <100 MS371(pJM18)-S <0.005 ND MS371(pJM18)-E 0.10 ND FIG. 3. Autoradiograph of toxin polypeptides isolated by immu- MS371(pJM18)-E + T 0.10 ND noadsorption and NaDodSO4/polyacrylamide gel electrophoresis. Im- munoadsorption of toxin polypeptides from extracts of cells grown in * Strains were grown in CYE medium and supernatant (S) and cell the presence of ['S]methionine was performed as detailed in the text. extracts (E) were prepared. Extracts were treated with antibody (+ All samples were boiled in NaDodSO4 sample buffer containing mer- Ab)by addition of 1 A.l ofrabbitantitoxin per 50 1.l ofsample followed captoethanol prior to electrophoresis. Lanes 1 and 6, standard purified by incubation at 37TC for 30 min. Extracts were treated with trypsin cholera toxin (17) labeled with 125I as described (2); lane 2, MS371- (+ T) by addition of the enzyme to a final concentration of 200 Ixg/ (pJM17); lane 3, MS371(pJM17) treated with trypsin at 100 pg/ml for ml followed by incubation at 37TC for 15 min. 10 min at22C priorto immunoadsorption; lane4, MS371(pJM18); lane t Measured by ELISA (19) in units of jig of toxin equivalents per ml. 5, MS371(pBR322). Symbols on the side designate the location of the * Measured by the S49 lymphosarcoma assay (20) in units of S49 toxin unnicked A chain (Au) and the Al, B, and A2 polypeptides of cholera doses per tg of toxin antigen. ND, not detectable. toxin. Downloaded by guest on September 29, 2021 Genetics: Pearson and Mekalanos Proc. NatL Acad. Sci. USA 79 (1982) 2979 form of the toxin to the nicked form composed ofthe disulfide- A-G-A-3' by this treatment, yielding a four base pair insertion linked Al and A2 peptides (16). These data also suggest that the mutation in pJM17. Such a mutation, if located within the A increase in specific toxicity seen after trypsin digestion of subunit gene, should produce a one base pair shift in the trans- MS371(pJM17) extracts (Table 1) is also due to the cleavage of lational reading frame past the Xba I site and thus eliminate the unnicked A chain by trypsin. production of'the A subunit. Localization of the Cholera Toxin Genes on pJM17 by Hy- A typical plasmid obtained by this procedure was pJM18, bridization and in Vitro Mutagenesis. Restriction analysis of which was shown to have lost the Xba I site originally present pJM17 was performed with EcoRI, Pst I, Xba I, HincII, and Pvu on pJM17 but to have retained a Pst I/EcoRI insert 5.1 kb in I, and the DNA fragments in these digests containing sequences size. MS371(pJM18) produced about one-fifth as much toxin homologous to the LT-A probe were identified by Southern antigen as MS371(pJM17) did. However, the toxin antigen pro- hybridization (Fig. 4). A similar hybridization analysis was per- duced by MS371(pJMl8) was devoid-oftoxicity in the S49 assay formed with the LT-B probe (data not shown). A 1.8-kb Xba I/ both before and after trypsin treatment (Table 1). Analysis of Pvu [-fragment ofpJM17 was the smallest fragment containing the antigen produced by MS371(pJM18) by immunoadsorption all sequences homologous to both the LT-A and LT-B probes. and NaDodSO4polyacrylamide gel electrophoresis demon- Sequences homologous to the LT-A probe were found to reside strated that this antigen consisted only-ofthe B subunit ofchol- only on a 950-base pair Xba I/HincII fragment, whereas se- era toxin (Fig. 3). quences homologous to the LT-B probe were present on both We conclude that the Xba I site on pJM17 is localized within the latter fragment and a-1,050-base pair HincII/Pvu I frag- the A subunit gene because introduction ofan insertional frame- ment. These data, together with other restriction data pre- shift mutation at this site eliminated production of-the A chain. sented in Fig. 4, allowed us to deduce the restriction map of Furthermore, it is known that nonsense and frameshift muta- pJM17 and the location ofthe toxin subunit genes on this plas- tions in a proximal gene of an operon will frequently result in mid (Fig. 2). The relative order of the cholera toxin A and B reduced expression ofdistal genes due to p-mediated transcrip- subunit genes on pJM17 suggests that the Xba I site on this tional termination (29, 30). 'The reduction ofB subunit produc- plasmid may be analogous in location to the Xba I site used to tion to one-fifth observed with MS371(pJM18) therefore sug- make the LT-A probe. This Xba I site corresponds to amino gests that the order of transcription of the toxin subunit genes acids 10 and 11 of the LT-A chain and is located in a region is A followed by B. This in turn predicts that a promoter for the (amino acids 5-17) known to display 100% amino acid sequence toxin genes probably resides within the 2.7-kb Pst I/Xba I frag- homology with the cholera A subunit (10). ment of pJM17 as indicated in Fig. 2. We.have obtained additional evidence that the Xba I site on pJM17 is located within the A subunit gene by analyzing the DISCUSSION effect on toxin production ofan insertion mutation constructed Previous protein chemical analysis ofpurified cholera toxin and in vitro at this restriction enzyme site. Plasmid pJM17 was di- LT had established their remarkable biochemical similarity (3, gested withXba Ito completion and the single-stranded 5' ends 4). Our analysis of the cloned cholera toxin gene has indicated of the linearized plasmid were converted to duplex DNA by that the cholera toxin and LT genes are also highly similar in treatment with T4 DNA polymerase I. Blunt-end ligation ofthe their genetic organization (Fig. 2). Direct comparison of the modified ends of the.plasmid was performed with T4 DNA li- nucleotide sequences of the cholera toxin and LT genes is now gase and the resultant covalently closed plasmids were re- possible and may yield valuable insights into the evolution of covered by transformation of MS371. We expected the Xba I these two related toxins and possibly other ADP-ribosylating site 5'-T-G-T-A-G-A-3' to be converted to 5'-T-C-T-A-G-C-T- toxins such as diphtheria toxin and exotoxin A ofPseudomnas aeruginosa (1). Expression of the cholera toxin genes in E. coli has. allowed kb us to make several observations concerning the secretion and 23 A BCDE FGH JK proteolytic processing of'the toxin's polypeptides. The lack of significant release of cholera toxin by the E. coli cells suggests ABCDEFGH JK /9.5\ that V. cholerae possesses a protein secretory apparatus that <6.4 = - either is lacking in E. coli or is defective in its recognition ofthe 4.2 .. cholera toxin-polypeptide secretory signals. Although E. coli has shown the capacity to secrete a variety ofproteins into its peri- 2.2 plasm and outer membrane, examples oftrue extracellular pro- 1.8 - tein secretion by this species are rare (31-33). The fact that LT, 1.3 - the indigenous E.- coli enterotoxin, is also primarily cell associated further supports the existence ofan extracellular protein secretory defect in E. coli relative to other pathogenic Gram- negative organisms (e.g., members of Vibrio, Shigella, Pseu- domonas, and BordetellUa). 0.5 The observed defect in the proteolytic processing of the A subunits ofboth cholera toxin and LT (4) might also be related FIG. 4. Restriction and Southern blot analyses of NJM17. Purified to the failure ofE. coli to secrete these toxins. Because the un- pJM17 plasmid DNA was digested with the indicated restriction en- nicked forms of these toxins are deficient in both enzymic (16) zymes and then electrophoresed in a 0.7% agarose gel. On the left is and toxic activity (4), the defect in the secretion and processing shown the gel stained with ethidium bromide. Southern blot analysis ofthese enterotoxins might indeed explain the reduced severity of this gel with 32P-labeled LT-A probe is shown on the right. Lanes ofE. coli-induced enterotoxic disease when compared to chol- A and K, standards (32P-labeled A and qbX174 phage DNAs digested with HindMu and Hae El[, respectively); lanes B-J, pJM17 digested era (4). with the following enzymes: B, Pvu I and Xba I; C, Pvu I and HincII; It is ofinterest to note that the total amount ofcholera toxin D,Hindll andEcoRI; E HincHlandPstl; F,HincHandXbal; G,Hindll; produced by MS371(pJM18) was only 4% the amount made by H, Xba I and Pst I; I, Xba I and EcoRI; J, Xba I. V. cholerae 569B. Strain 569B possesses two copies ofthe toxin Downloaded by guest on September 29, 2021 2980 Genetics: Pearson and Mekalanos Proc. Natl. Acad. Sci. USA 79 (1982)

operon, and the relative expression of these two copies is not 10. Spicer, E. K., Kavanaugh, W. M., Dallas, W. S., Falkow, S., known. However, because the copy number of pJM17 is ap- Konigsberg, W. H. & Schafer, D. E. (1981) Proc. Nati Acad. Sci. proximately 10 plasmid molecules per cell, the amount oftoxin USA 78, 50-54. 11. Moseley, S. L. & Falkow, S. (1980)J. Bacteriol 144, 444-446. produced by MS371(pJM17) is still quite low by comparison. 12. Kaper, J. B., Moseley, S. L. & Falkow, S. (1981) Infect. ImmunoL Perhaps the secretory block in E. coli has also resulted in a net 32, 661-667. decrease in toxin accumulation due to the degradative turnover 13. Mekalanos, J. J., Moseley, S. L., Murphy, J. R. & Falkow, S. ofcell-bound or intracellular toxin. Alternatively, E. coli might (1982) Proc. Natl Acad. Sci. USA 79, 151-155; not recognize V. cholerae promoters efficiently or might lack 14. Rappaport, R. S., Sagin, J. F., Pierzchala, W. A., Bonde, G., some positive control element produced by V. cholerae that Rubin, B. A. & Tint, H. (1976)J. Infect. Dis. 133s, S41-S54. 15. Wensink, J., Gankema, H., Jansen, W. H., Guinee, P. A. M. promotes the high expression of the toxin genes (34). & Witholt, B. (1978) Biochim. Biophys. Acta 514, 128-136. The molecular cloning of the cholera toxin gene has allowed 16. Mekalanos, J. J., Collier, R. J. & Romig, W. R. (1979) J. BioL us to employ recombinant DNA technology in the construction Chem. 254, 5855-5861. of an in vitro generated mutation in the A subunit gene. This 17. Mekalanos, J. J., Collier, R. J. & Romig, W. R. (1978) Infect. Im- has demonstrated the feasibility of constructing plasmids that munot 20, 552-558. code for production ofonly the B subunit ofthe toxin. Because 18. Miller, J. H. (1972) Experiments in Molecular Genetics (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). this subunit is both nontoxic and highly immunogenic, plasmids 19. Homgren, J. (1973) Infect. Immunot 8, 851-859. such as pJM18 might be highly useful in the production ofchol- 20. Ruch, F. E., Murphy, J. R., Graf, L. H. & Field, M. (1978) J. era toxoids. Alternatively, the introduction ofthis or other anal- Infect. Dis. 137, 747-755. ogous in vitro altered toxin genes back into wild-type or mutant 21. Brenner, D. J., Fanning, G. R., Johnson, K. E., Citarella, R. V. strains (13) of V. cholerae could facilitate construction of a V. & Falkow, S. (1969) J. Bacteriol 98, 637-650. cholerae strain suitable for testing as a live oral cholera vaccine. 22. Dallas, W. S., Gill, D. M. & Falkow, S. (1979)J. Bacteriol 139, 850-858. We thank S. Lory for his helpful suggestions and R. Bacco for the 23. Maniatis, T., Jeffrey, A. & Kleid, D. G. (1975) Proc. Nati Acad. preparation of this manuscript. This investigation was supported by Sct. USA 72, 1184-1188. Grant AI-18045 from the National Institute of Allergy and Infectious 24. Southern, E. M. (1975) J. Mol Biol 98, 503-517. Disease and alsoby the Medical Foundation, Inc., Boston, Massachusetts. 25. Nichols, J. C., Tai, P. C. & Murphy, J. R. (1980)J. Bacteriol. 144, 518-523. 1. Collier, R. J. & Mekalanos, J. J. (1980) in Multifunctional Pro- 26. Mekalanos, J. J., Collier, R. J. & Romig, W. R. (1977) Infect. Im- teins, eds. Bisswanger, H. & Schmincke-Ott, E. (Wiley, New munol 16, 789-795. York), pp. 261-291. 27. Bolivar, F., Rodriguez, R. L., Greene, P. J., Betlach, M. C. & 2. Gill, D. M. (1976) Biochemistry 15, 1242-1248. Boyer, H. W. (1977) Gene 2, 95-113. 3. Dallas, W. S. & Falkow, S. (1979) Nature (London) 277, 28. Gruenstein, M. S. & Hogness, P. (1975) Proc. Natl Acad. Sci. 406-407. USA 72, 3961-3965. 4. Clements, J. D. & Finkelstein, R. A. (1979) Infect. Immunol 24, 29. Newton, W. A., Beckwith, J. R., Zipser, D. & Brenner, S. (1965) 760-769. J. Mol Biol 14, 290-296. 5. Cassel, D. & Pfeuffer, T. (1978) Proc. Natl Acad. Sci. USA 75, 30. Korn, L. J. & Yanofsky, C. (1976) J. Mol Biol 106, 231-241. 2669-2673. 31. Emr, S. D., Hall, M; N. & Shilhavy, T. J. (1980) J. Cell BioL 86, 6. Gill, D. M. & Richardson, S. H. (1980)J. Infect. Dis. 141, 64-70. 701-711. 7. Cuatrecases, P. (1973) Biochemistry 12, 3577-3581. 32. Wickner, W. T. (1980) Science 210, 861-868. 8. Clements, J. D. & Finkelstein, R. A. (1978) Infect. Immunol 22, 33. Davis, B. D. & Tai, P. C. (1980) Nature (London) 283, 433-438. 709-713. 34. Mekalanos, J. J. & Murphy, J. R. (1980) J. Bacteriol 141, 9. Dallas, W. S. & Falkow, S. (1980) Nature (London) 288, 499-501. 570-576. Downloaded by guest on September 29, 2021