Construction and Characterization of Bordetella pertussisToxin Mutants
WILLIAM J. BLACKt* AND STANLEY FALKOWDepartment of Medical Microbiology, Stanford University Medical School, Stanford, California 94305
Received 19 March 1987/Accepted 30 June 1987
Pertussis toxin is one of the major virulence determinants produced by Bordetella pertussis. The DNAencoding the structural genes for pertussis toxin was cloned in Escherichia coli, and pertussis toxin subunit S4was expressed under the control of the tac promoter. Mutations were introduced into the cloned toxin genes,
and a conjugative shuttle vector system was devised for delivering the mutations from E. coli back into B.pertussis. The mutations were introduced by allelic exchange into the chromosome of B. pertussis resulting ina series of B. pertussis strains which were isogenic except at the loci encoding the structural genes for pertussistoxin. These B. pertussis strains were utilized to study the biogenesis of pertussis toxin. Polar mutations in theS1 gene led to a lack of detectable S2 or S4 subunits in whole-cell lysates, suggesting a polycistronicarrangement for these genes. Mutations in the SS subunit gene resulted in a truncated S1 subunit, whilemutations in the S4 gene resulted in a lack of detectable S2 subunit, suggesting that physical relationshipsamong the toxin subunits are directly reflected in the stable biogenesis of the subunits.
Pertussis toxin, produced by Bordetella pertussis, is aprotein exotoxin which is central to the pathogenesis ofwhooping cough. Whooping cough is multiphasic. During aninitial catarrhal phase, B. pertussis colonizes the host naso-pharynx and damages local tissue. The subsequent paroxys-mal phase reflects a systemic intoxication of the host bypertussis toxin. Some aspects of the disease associated withthe paroxysmal phase, such as lymphocytosis, hypoglyce-mia, and reactogenicity, can be duplicated in animal modelsby administration of pertussis toxin alone (18, 22). The toxinalters host physiology by covalently attaching an ADP-ribose moiety to certain guanine nucleotide-binding proteinsinvolved in transduction of hormonal signals across eucary-otic membranes (2, 11).The structure of pertussis toxin conforms to the A-B
model for bacterial toxins (25). An "A" peptide monomer,designated S1, is the ADP-ribosyltransferase (10). A "B"component oligomer, composed of four different peptidesdesignated S2, S3, S4, and S5, mediates binding to hosttissue (26). The five subunits, S1 through S5, are present ina 1:1:1:2:1 ratio.The region of the B. pertussis chromosome which encodes
pertussis toxin was originally identified by a transposoninsertion that inhibited phenotypic expression of the toxin(28). Specifically, this insertion was shown to inactivateexpression of toxin subunit S3 (16). We have named thechromosomal region surrounding the transposon insertionptx (28). This chromosomal region has been cloned andsequenced, and the sequence data suggest a polycistronic,operonlike structure with open reading frames for all fivepertussis toxin subunits arranged contiguously (14, 19). Thefinding that the transposon insertion was in the open readingframe assigned to subunit S3 supported the sequence-basedassignments of the toxin genes (19).
In an earlier report, we described the genetic modificationof the B. pertussis chromosome by transposon mutagenesis(28). We have now extended this technique of genetic
* Corresponding author.t Present address: Department of Microbiology and Immunology,
Emory University School of Medicine, Atlanta, GA 30322.
modification and report here the allelic exchange of definedmutations into the ptx region of the B. pertussis chromo-some. These mutations enabled us to initiate an investigationof the pertussis toxin operon and its control.
MATERIALS & METHODS
Strains and media. The bacterial strains used in this studyare listed in Table 1. Escherichia coli was cultured on L agar(15), and B. pertussis was grown on Bordet Gengou agar (28)with 15% sheep blood at 37°C. B. pertussis was incubated injars loosely stoppered to maintain high humidity. Antibioticswere included, when appropriate, at the following concen-trations: ampicillin, 100 ,ug/ml for E. coli and 40 p.g/ml for B.pertussis; kanamycin, 40 ,ug/ml; rifampin, 50 ju.g/ml; strepto-mycin, 400 p,g/ml.
Nucleic acid manipulations. General cloning techniqueswere used in recombinant DNA molecule constructions (15).Large preparations of plasmid DNA for recombinant con-structions and small preparations for rapid screening ofplasmid constructions were prepared by alkaline lysis (15).For recombinant DNA constructions, plasmid DNA wasfurther purified by isopycnic CsC12 centrifugation (15). Bac-terial chromosomal DNA was prepared by the method ofMarmur (17), followed by CsCl gradient centrifugation.Restriction endonucleases and other DNA-modifying en-zymes were purchased from either Bethesda Research Lab-oratories, Inc. (Gaithersburg, Md.) or New England Bio-Labs, Inc. (Beverly, Mass.) and were used under conditionssuggested by the suppliers. E. coli HB101 (15) made compe-tent by CaC12 treatment was used as a recipient in transfor-mation experiments (15). DNA restriction fragments forsubcloning and for hybridization probes were recoveredfrom low-melting-point agarose gels (Sea Plaque; FMCCorp., Marine Colloids Div., Rockland, Maine) (15). Thenick translation kit from Bethesda Research Laboratorieswas utilized for radiolabeling DNA. Southern transfer hy-bridization (23) was at 68°C without formamide; the finalfilter wash was in 0.1x SSC (1x SSC is 0.15 M NaCl plus0.015 M sodium citrate)-0.1% sodium dodecyl sulfate at680C.
Bordetella pertussisBP370 ptx+ Rifr Strr 24TOX3302 A(ptxD-ptxC)3302 This studyTOX3305 AptxE3305 This studyTOX3311 A(ptxA-ptxC)3311 This studyTOX3316 A(ptxE-ptxC)3316 This studyTOX5105 ptx-5105 This studyTOX5119 ptxD5119 This studyTOX5148 ptx-5148 This studyTOX5167 ptx-5167 This studyTOX5171 ptxA5171 This study
PlasmidspEYDG1 Kanr rlx 29pRKTV5 Strr Spcr Tmpr Tra+ 28pRI133 Ampr Kanr tacP R. R. IsbergpHC79 Ampr Tetr cos 7p11.11 Ampr cos ptx A. A. WeisspKANrr Ampr Tetr Kanr This studypTOX1 Ampr Kanr ptx This studypTOX9 Ampr ptx This study
Plasmid constructions. Plasmid pTOX9 was constructed toserve as a shuttle vector for delivering ptx mutations into theB. pertussis chromosome (Fig. 1). pTOX9 was constructedfrom plasmids p11.11 (Fig. 1) (A. A. Weiss and S. Falkow,unpublished data) and pEYDG1 (29) and contains cloned B.pertussis DNA inclusive of the ptx locus and vector se-
quences which include a ColEl origin of replication, oriV; an
ampicillin resistance gene; and the P incompatability group(IncP) origin of transfer, rlx. ColEl oriV provided for repli-cation of the plasmid in E. coli but not in B. pertussis (27),and IncP rlx provided a means for conjugative mobilizationof pTOX9 from E. coli into B. pertussis (27). The cloned B.pertussis DNA, the ColEl oriV, and the ampicillin resistancegene were provided by the 10.0- and 3.6-kilobase-pair (kb)BglII fragments from p11.11. IncP rlx was provided by a0.7-kb fragment from pEYDG1. The three fragments werecombined such that the integrity of the ptx locus wasmaintained and the only remaining BglII site in pTOX9 wasin the pertussis toxin operon.
Plasmid pKANTr was constructed for production of apolylinker-flanked aminoglycoside-phosphotransferase-3'-I(APH) gene (a kanamycin resistance marker [20]) and was aninsertion of the polylinker-flanked APH gene into the EcoRIsite of pBR322 (15). A derivative of the ptacl2 expressionvector (15), pRI133 (provided by R. R. Isberg), was utilizedfor cloning random DNA fragments from cosmid p11.11.pRI133 contains the APH gene (20) inserted into the EcoRIsite of ptacl2 but is otherwise identical to ptacl2. Thefragments of p11.11 were generated by sonication by themethod of Deininger (5) and were cloned into the PvuII siteof pRI133 immediately adjacent to the tac promoter.The mutations generated in the cloned B. pertussis DNA
of pTOX9 were each marked with an inserted kanamycinresistance (APH) gene and were constructed such that the B.pertussis DNA flanking the mutation would facilitate recom-bination into the B. pertussis chromosome. The mutations
were either deletions, with the kanamycin resistance geneligated into the breach, or insertions of the kanamycinresistance gene at specific sites within the operon. Deletionmutations were created by digestion of pTOX9 with restric-tion endonuclease BglII, followed by digestion with Bal 31exonuclease (15). Bal 31 exonuclease-digested plasmidswere treated with Klenow polymerase (15) to form bluntends. The kanamycin resistance gene generated as a HincIIfragment from pKANnr was then ligated to the deletedpTOX9 to permit recircularization. For the single-pointinsertions, pTOX9 was partially digested with restrictionendonuclease Sau3A in the presence of ethidium bromide asdescribed by Parker et al. (21). This yielded a population ofcircularly permuted, linear pTOX9 molecules, each cut atonly one Sau3A site. The kanamycin resistance gene gener-ated as a BamHI-cut fragment from pKANmr was then ligatedto the linear pTOX9 molecules to permit recircularization.
Genetics. Mutations created in the pTOX9 shuttle vectorwere introduced into the B. pertussis BP370 chromosome byallelic exchange (24). The shuttle vector derivatives wereconjugatively mobilized into B. pertussis in triparental mat-ings utilizing the helper plasmid pRKTV5 (28). E. coli HB101(18 h old) and B. pertussis BP370 (60 h old) plate cultureswere swabbed together onto Bordet Gengou agar platescontaining 10.0 mM MgCl2. Mating mixtures were incubatedat 37°C for 3.5 h in loosely stoppered jars followed byreswabbing onto Bordet Gengou agar medium supplementedwith 40 ,ug of kanamycin per ml and 50 pug of rifampin per ml.The rifampin was used to counterselect donor E. coli. SincepTOX9 was unable to replicate autonomously in B. pertus-sis, plating on kanamycin-containing medium selected thoseexconjugants in which the plasmid had recombined into theB. pertussis chromosome. By subsequently scoring for lossof the ampicillin resistance encoded by the plasmid vectorsequence, while maintaining selection for kanamycin resis-tance, we identified exconjugants which had gained thekanamycin resistance-marked mutated ptx allele and con-comitantly lost the wild-type ptx allele and plasmid vectorsequence.Immunological assays. B. pertussis toxin mutants were
analyzed by Western immunoblotting (4). To detect allsubunits present, we assayed whole-cell lysates of the mu-tants. B. pertussis strains were grown on Bordet Gengouagar and were prepared for analysis by suspension in waterfollowed by heating in solubilization buffer (12) at 100°C for5 min. Preparations were electrophoresed in 15% polyacryl-amide gels by the method of Laemmli (12), transferred tonitrocellulose by the method of Burnette (4), and probedwith antibody and labeling reagents in BLOTTO buffer (9).Gel samples of whole-cell lysates were difficult to equili-brate, and thus autoradiographic signals were only usedqualitatively. Purified pertussis toxin for use as a standardwas purchased from List Biological Laboratories. Rabbitantiserum to pertussis toxin was provided by E. L. Hewlett.Monoclonal antibodies 10143x4 and 10126Dx3, specific forpertussis toxin subunits S1 and S4, respectively, were fromJ. G. Kenimer and J. L. Cowell, respectively. The specificityof these antibodies was confirmed by probing adjacent lanesof an immunoblot of purified pertussis toxin with the mono-clonal antibodies and the antiserum to holotoxin. The S2subunit-specific monoclonal antibody PllBlO of Frank andParker (6) was also utilized. Antiserum-probed transferswere labeled with 125I-protein A (Amersham Corp., Arling-ton Heights, Ill.). Monoclonal antibody-probed transferswere labeled with radioiodinated goat anti-mouse immuno-globulin (Amersham).
FIG. 1. Physical map of pertussis toxin genes and genetic constructions. The line adjacent to the BP370 chromosome denotes achromosomal restriction map of the ptx region. Restriction sites are: C, ClaI; E, EcoRI; S, Sall; X, XbaI; Bg, BgIII; B, BamHI. The locusorder and associated toxin subunit products are indicated below the chromosome restriction map. The transcriptional direction for the subunitgenes, deduced from the operon sequence (14, 19), is indicated by, and proceeds from, the 5' to the 3' designations. Plasmids pTOX1, pTOX9,and p11.11, described in the text, contain cloned ptx DNA. B. pertussis-derived DNA is denoted by a thin line; vector DNA is denoted bya thick line. The vector sequence of p11.11 has been truncated to facilitate inclusion in the diagram. The position and direction of the tacpromoter (tacP) in pTOX1 is indicated. The site of the transposon insertion which defined the ptx region, ptxCl::TnS (28) and the insertionmutations described herein are indicated by the vertical lines below the product designations. The deletion mutations and their extent areindicated by the open boxes. Arrows below the insertion mutations and adjacent to the deletion mutations denote the transcriptional directionof the inserted kanamycin resistance gene.
Pertussis toxin expressed from the tac promoter in E. coliwas labeled with [35S]methionine (Amersham) in vitro (3),using reagents prepared from E. coli HB101. Labeled pep-
tide products of the in vitro transcription-translation reac-
tions were isolated by radioimmunoprecipitation (8) utilizingmonoclonal antibody A2B5/G9 (provided by J. J. Munoz),specific for pertussis toxin subunit S4, and rabbit anti-mouseimmunoglobulin Immunobeads (Bio-Rad Laboratories,Richmond, Calif.). Labeled peptides were electrophoresedin 15% polyacrylamide gels (12) and autoradiographed (15).
RESULTS
The ptx region of the B. pertussis chromosome has beenshown by transposon mutagenesis (28) and DNA sequenceanalysis (14, 19) to encode the genes for the pertussis toxinpeptide subunits. The sequence data indicate that the genesare arranged contiguously and suggest a polycistronic,operonlike structure. Two putative promoters have beenidentified from the sequence data, one preceding the S1subunit gene (14, 19) and a second preceding the S4 subunitgene (13). To study the regulation of the toxin genes, we
cloned this chromosomal region in the cosmid p11.11 (Fig. 1)(A. A. Weiss and S. Falkow, unpublished data). The subunitgenes have previously been referred to by the peptidedesignations, S1 through S5 (14, 19). In this report, we
assigned the genetic loci associated with these genes as ptxA(S1), ptxB (S2), ptxC (S3), ptxD (S4), and ptxE (S5). Also,consistent with the transcriptional direction for the genes
predicted by the DNA sequence (14, 19), we call the ptxAend of the operon, 5', and the ptxC end, 3' (Fig. 1).
Expression of pertussis toxin subunit S4 in E. coli. We were
unable to detect expression of pertussis toxin in E. coli fromthe cloned sequence in cosmid p11.11. Therefore, we clonedrandomly generated fragments of the ptx region into an E.coli tac promoter expression vector, pRI133, to determinewhether translation of pertussis toxin was possible in E. coli.E. coli HB101 was transformed with the plasmid clones, andone, designated pTOX1, was found to be associated withproduction of pertussis toxin antigenic material as assayedwith an antisera by colony blot radioimmunoassay (data notshown). Restriction mapping of the B. pertussis DNA clonedin pTOX1 indicated that it included all the genes for subunitsS2, S4, S5, and S3, but only a portion of the gene for subunitS1 (Fig. 1). Furthermore, the toxin operon DNA was fusedto the tac promoter with the appropriate transcriptionalorientation predicted by the DNA sequence. Peptides pro-duced by pTOX1 and the vector pRI133 were labeled with[35S]methionine in vitro in transcription-translation reactionsand subsequently precipitated with a monoclonal antibodyspecific for pertussis toxin subunit S4. Polyacrylamide gelelectrophoresis indicated that a peptide encoded by pTOX1migrated with the apparent molecular weight for toxin sub-unit S4, ca. 12,000 (Fig. 2). Other [35S]methionine-labeledpeptides which were encoded by pTOX1 but not by pRI133(Fig. 2A, lane 3) were precipitable with antisera to holotoxin(data not shown) and may represent other toxin subunits.
Construction of B. pertussis mutants. Although we could
FIG. 2. Immunoprecipitation of peptides encoded by pTOX1. Invitro transcription-translation reactions were charged with no DNA(lanes 1), the vector pRI133 (lanes 2), or pTOX1 (lanes 3). Labeledproducts were then either directly electrophoresed in a sodiumdodecyl sulfate-polyacrylamide gel (A) or precipitated with mono-clonal antibody A2B5/G9, specific for pertussis toxin subunit S4, andthen electrophoresed (B). Apparent molecular size is indicated inkilodaltons (kD).
express a pertussis toxin subunit from the tac promoter in E.coli, our primary interest was to examine the regulation ofpertussis toxin expression from its own promoter. To do so,we utilized allelic exchange to introduce defined mutationsinto the B. pertussis chromosomal ptx loci. The ptx muta-tions were constructed in the shuttle vector pTOX9 (Fig. 1)by insertion of a kanamycin resistance gene into varioussites in the ptx DNA. The constructions were designed suchthat the B. pertussis DNA flanking each mutational insertionfacilitated recombination into the B. pertussis chromosome.The mutations were of two general forms: deletions markedwith an inserted kanamycin resistance gene, and pointinsertions of the kanamycin resistance gene (Fig. 1). Thekanamycin resistance gene cassette contained its own pro-moter and so was not dependent on the transcriptionalcontrol of the pertussis toxin operon. A ClaI site in thekanamycin resistance gene, located asymmetrically withrespect to the ends of the cassette, allowed us to determinethe transcriptional direction of the gene insertions (Fig. 1).We confirmed that the mutations were introduced at the
chromosomal ptx loci of the exconjugants by Southerntransfer analysis. Chromosomal DNA from B. pertussis ptxmutants was digested with ClaI and probed with the 1.2-kbkanamycin gene cassette and the 4.7-kb EcoRI fragmentfrom the ptx region. The bands to which the kanamycin genecassette hybridized were identical in size to the bandsdetected with the ptx probe (Fig. 3). This indicated that thekanamycin resistance gene, and the mutations, were intro-duced at the ptx loci.We were thus able to generate a series of B. pertussis
strains which were isogenic except at their ptx loci. Inaddition to genetically defining the ptx loci, these mutationsprovided a means by which the genetic regulation of theoperon could be studied. The insertions of the kanamycinresistance gene, whether at a Sau3A site or into a deletion,interrupted the transcriptional control of the operon at thepoint of insertion. It was thus possible to determine howinterruption of the operon with any given point insertion ordeletion affected biogenesis of the toxin.
Pertussis toxin biogenesis. The pertussis toxin operonsequence data suggested the presence of two promoterlikeregions, one preceding the Si gene and a second preceding
the S4 gene (13, 14, 19). If only the promoter preceding theSi gene was active, all five subunit genes would form a singletranscriptional unit, and any interruption of the operonproximal to this promoter would be polar for all genes in theoperon. Alternatively, if the second promoter, preceding theS4 gene, was also active, then S4 biogenesis could conceiv-ably occur independently of Si and S2 biogenesis. There-fore, we were interested in determining the effect of themutation ptxA571 on the biogenesis of subunit S4. ptxA5171is an insertion of the kanamycin resistance gene into the Sigene, proximal to the first putative promoter (Fig. 1). B.pertussis TOX5171, which contains the ptxA5171 allele, wasscored by immunoblot for the presence of toxin subunits.We could not detect Si, S2, or S4 subunits in lysates ofTOX5171 (Fig. 4, lane d), although they were clearly evidentin the parental strain BP370 containing the wild-type operon(Fig. 4, lane b). Furthermore, insertions of the kanamycinresistance gene outside of the toxin operon, such as instrains TOX5105 (ptx-5105) and TOX5167 (ptx-5167), had noeffect on S1, S2, or S4 biogenesis (Fig. 4, lanes c and g). Itappeared that biogenesis of toxin subunit S4 was dependenton the integrity of the operon proximal to the first promoter.This is consistent with the notion that only the firstpromoterlike sequence was active in controlling S4 expres-sion.
It should be noted that lysates of wild-type strains con-tained two peptides which bound the anti-Si monoclonalantibody. One peptide comigrated with the Si peptide ofpurified pertussis toxin at an apparent molecular weight ofca. 27,000 (Fig. 4, lane a). The other peptide, at a lowerapparent molecular weight of ca. 20,000, was not found inpurified pertussis toxin preparations. Neither of the Sipeptides was present in lysates of the operon deletion mutantTOX3311 [A(ptxA-ptxC)3311] or of the Si gene insertionmutant TOX5171 (Fig. 4). We believe that the peptide oflower apparent molecular weight represents a truncated Sisubunit, owing to either incomplete synthesis or degrada-
\N ¶~%D
,\A ->
~~Z~>O+ O+I,() Ill\9
_ do
karn PROBE ptx PROBE
FIG. 3. Southern analysis of B. pertussis ptx mutants. Chromo-somal DNA from B. pertussis ptx mutants was digested with ClaIand probed with either the 4.7-kb EcoRI fragment from the ptxregion (ptx probe panel) or the 1.2-kb kanamycin resistance genecassette (kan probe panel). Southern analysis of the wild-type strainBP370 plus four representative mutant strains, TOX3305, TOX3302,TOX5171, and TOX5148, is presented.
FIG. 4. Western immunoblot analysis of B. pertussis ptx mu-
tants. Control lanes (C) contained purified pertussis toxin and were
probed with antiserum to holotoxin. Other lanes were probed withmonoclonal antibodies and contained either purified pertussis toxinor lysates of the mutants. Purified pertussis toxin was in lanes a. Thelysates were: lanes b, BP370; lanes c, TOX5167; lanes d, TOX5171;lanes e, TOX5119; lanes f, TOX5148; lanes g, TOX5105; lanes h,TOX3305; lanes i, TOX3302; lanes j, TOX3316; and lanes k,TOX3311. The monoclonal antibodies (McAb) used were: S1 anti-body, 10143Cx4; S2 antibody, PllB10; and S4 antibody, 10126Dx3.
tion, since it is missing in all deletion or insertion mutationswhich interrupt the Si gene. Also, a mutation which intro-duced additional codons into the Si gene has been found toresult in larger peptides of both electrophoretic mobilities(W. J. Black et al., manuscript in preparation).Another issue of pertussis toxin biogenesis which we
investigated regarded the biosynthesis of a limited array ofthe toxin subunits by B. pertussis strains. That is, would a
strain which lacked the genes for certain of the subunits, say
S5 or S4, stably produce the other subunits?We first examined strains containing mutations which
either deleted or interrupted the S5 gene. These includedTOX3305 (AptxE3305) and TOX3316 [A(ptxE-ptxC)3316](Fig. 1). The mutations of these strains were well distal to theSi gene and also, of course, to the proposed promoter for theSi gene. Yet strains harboring these mutations containedonly the truncated form of subunit S1 (Fig. 4, lanes h and j).In contrast, the B. pertussis strain with the ptx-5148 allele,an interruption of the operon downstream (or 3') of the S5gene (Fig. 1), produced wild-type Si (Fig. 4, lane f). Thus,interruption of the S5 gene apparently adversely affectedsynthesis or stability of subunit Si.A second such linkage was found between toxin subunits
S4 and S2. Mutations in the strains TOX5119 (ptxD5119) andTOX3302 [A(ptxD-ptxC)3302] which interrupted the S4 genewere distal (or 3') to both the S2 gene and the promoterpreceding the Si gene (Fig. 1). Yet strains harboring thesemutations contained no discernable S2 subunit (Fig. 4, lanese and i). In contrast, the B. pertussis strain with theAptxE3305 allele, an interruption of the operon downstream(or 3') of the S4 gene (Fig. 1), produced wild-type S2 (Fig. 4,lane h). These data are consistent with the view that inter-
ruption of the S4 subunit gene adversely affected the syn-thesis or stability of subunit S2.
DISCUSSION
Pertussis toxin is one of the major virulence determinantsproduced by B. pertussis. The genetic locus encoding pertus-sis toxin has been identified (28), and the genes have beencloned and sequenced (14, 19). Previous reports (14, 19) andour own work indicated that pertussis toxin was not ex-pressed from toxin genes cloned in E. coli. We have nowbeen able to express pertussis toxin subunit S4 in E. coliunder control of an E. coli promoter. Expression of the toxinunder the control of an exogenous promoter is consistentwith a recent report of Barbieri et al. (1) who expressed Si asa fusion peptide under control of the lac promoter. Moreimportantly, we were able to examine toxin gene expressionby creating defined mutations in the chromosomal pertussistoxin genes. These mutations enabled us to establish severalpoints regarding pertussis toxin biogenesis.The first is that biogenesis of toxin subunits S2 and S4
requires an intact and uninterrupted Si gene. This mayindicate that the first three genes of the operon form a singletranscriptional unit, that is, they are polycistronic. Twoputative promoters have been identified in the pertussistoxin operon DNA sequence (13, 14, 19). One precedes theSi gene, the other precedes the S4 gene. Our data suggestthat only the promoter sequence preceding the Si gene isused. Alternatively, it is possible that S2 and S4 are inde-pendently transcribed, but require the Si message or prod-uct as a transcriptional activator, or are unstable in theabsence of Si. In any event, our data indicate that thebiogenesis of subunits S2 and S4 requires an intact Si gene.Another feature of pertussis toxin biogenesis suggested by
our work is that mutations in the S5 subunit gene result in atruncated Si subunit, while mutations in the S4 gene result ina lack of detectable S2 subunit. These biogenesis relation-ships between subunits are possibly notable in their reflec-tion of the physical associations of the subunits. Tamura etal. (25) have shown that in vitro, pertussis toxin subunitsassociate with each other very specifically. Subunits S2 andS4 and subunits S3 and S4 assemble into relatively stableheterodimers. The two heterodimers, S2/S4 and S3/S4, willcombine with S5 to form a heteropentamer. This S2/S4-S5-S3/S4 heteropentamer is a stable structure and is believed tomediate binding of the toxin to host tissue; for example, ithas been used in vitro to block activity of the holotoxin (26).The heteropentamer will stably associate with the Si sub-unit. In the absence of S5, the Si subunit will not associatewith any of the other subunits. Also, in the absence ofsubunit S4, the S2 subunit will not associate with any of theother subunits. We find, then, that the physical relationshipsamong the toxin subunits-Si with S5, and S2 with S4-maybe directly reflected in the stable biogenesis of the subunits.The truncation of the Si or A subunit of pertussis toxin in theabsence of the S5 subunit is reminiscent of a similar degra-dation of cholera toxin subunit A to the Al form when thecholera toxin A subunit gene is cloned into E. coli in theabsence of the cholera toxin B subunit gene (J. J. Mekal-anos, personal communication). These toxin subunit pep-tides have evolved to interact stably with each other and somay be conformationally unstable and subject to proteolysiswhen unpaired.
Utilizing allelic exchange, we constructed defined B.pertussis toxin mutants. These mutants facilitated our inves-tigation of pertussis toxin biogenesis. These and similar
mutants should be valuable tools in analyzing the role ofpertussis toxin in pathogenic and immunoprotective pro-cesses.
ACKNOWLEDGMENTS
This study was supported by Public Health Service grantAI-23945 from the National Institutes of Health and by a Bank ofAmerica-Giannini Foundation fellowship to W.J.B.We thank M. L. Palmer for providing the S30 in vitro transcrip-
tion-translation extracts.
LITERATURE CITED
1. Barbieri, J. T., R. Rappuoli, and R. J. Collier. 1987. Expressionof the S1 catalytic subunit of pertussis toxin in Escherichia coli.Infect. Immun. 55:1321-1323.
2. Bokoch, G. M., T. Katada, J. K. Northup, E. L. Hewlet, andA. G. Gilman. 1983. Identification of the predominant substratefor ADP-ribosylation by islet activating protein. J. Biol. Chem.258:2072-2075.
3. Brusilow, W. S. A., R. P. Gunsalus, and R. D. Simoni. 1983.Biogenesis of an oligomeric membrane protein complex: theproton translocating ATPase of Eschericia coli. Methods En-zymol. 97:188-195.
4. Burnette, W. N. 1981. "Western blotting": electrophoretictransfer of proteins from sodium dodecyl sulphate-polyacryl-amide gels to unmodified nitrocellulose and radiographic detec-tion with antibody and radioiodinated protein A. Anal. Bio-chem. 112:195-203.
5. Deininger, P. L. 1983. Random subcloning of sonicated DNA:application to shotgun DNA sequence analysis. Anal. Biochem.129:216-223.
6. Frank, D. W., and C. D. Parker. 1984. Interaction of monoclo-nal antibodies with pertussis toxin and its subunits. Infect.Immun. 46:195-201.
7. Hohn, B., and J. Collins. 1980. A small cosmid for efficientcloning of large DNA fragments. Gene 11:291-298.
8. Ito, K., P. J. Bassford, and J. Beckwith. 1981. Protein translo-cation in E. coli: is there a common step in the secretion ofperiplasmic and outer membrane proteins? Cell 24:707-717.
9. Johnson, D. A., J. W. Gautsch, J. R. Sportsman, and J. H.Elder. 1983. Improved technique utilizing nonfat dry milk foranalysis of proteins and nucleic acids transferred to nitrocellu-lose. Gene Anal. Tech. 1:3-8.
10. Katada, T., M. Tamura, and M. Ui. 1983. The A protomer ofislet-activating protein, pertussis toxin, as an active peptidecatalyzing ADP-ribosylation of a membrane protein. Arch.Biochem. Biophys. 224:290-298.
11. Katada, T., and M. Ui. 1982. Direct modification of the mem-brane adenylate cyclase system by islet-activating protein dueto ADP-ribosylation of a membrane protein. Proc. Natl. Acad.Sci. USA 79:3129-3133.
12. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)
227:680-685.13. Locht, C., P. A. Barstad, J. E. Coligan, L. Mayer, J. J. Munoz,
S. G. Smith, and J. M. Keith. 1986. Molecular cloning ofpertussis toxin genes. Nucleic Acids Res. 14:3251-3261.
14. Locht, C., and J. M. Keith. 1986. Pertussis toxin gene: nucleo-tide sequence and genetic organization. Science 232:1258-1264.
15. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning, a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.
16. Marchitto, K. S., J. J. Munoz, and J. M. Keith. 1987. Detectionof subunits of pertussis toxin in Tn5-induced Bordetella mutantsdeficient in toxin biological activity. Infect. Immun. 55:1309-1313.
17. Marmur, J. 1961. A procedure for the isolation of ribonucleicacid from microorganisms. J. Mol. Biol. 3:208-216.
18. Munoz, J. J., H. Arai, R. K. Bergman, and P. L. Sadowski. 1981.Biological activities of crystalline pertussigen from Bordetellapertussis. Infect. Immun. 33:820-826.
19. Nicosia, A., M. Perugini, C. Franzini, M. C. Casagli, M. G.Borri, G. Antoni, M. Almoni, P. Neri, G. Ratti, and R. Rappuoli.1986. Cloning and sequencing of the pertussis toxin genes:operon structure and gene duplication. Proc. Natl. Acad. Sci.USA 83:4631-4635.
20. Oka, A., H. Sugasaki, and M. Takanami. 1981. Nucleotidesequence of the kanamycin resistance transposon Tn9O3. J.Mol. Biol. 147:217-226.
21. Parker, R. C., R. M. Watson, and J. Vinograd. 1977. Mapping ofclosed circular DNAs by cleavage with restriction endo-nucleases and calibration by agarose gel electrophoresis. Proc.Natl. Acad. Sci. USA 74:851-855.
22. Pittman, M. 1984. The concept of pertussis as a toxin-mediateddisease. Pediatr. Infect. Dis. 3:467-486.
23. Southern, E. 1975. Detection of specific sequences among DNAfragments separated by gel electrophoresis. J. Mol. Biol. 98:503-511.
24. Stibitz, S., W. J. Black, and S. Falkow. 1986. The constructionof a cloning vector designed for gene replacement in Bordetellapertussis. Gene 50:1765-1774.
25. Tamura, M., K. Nogimori, S. Murai, M. Yajima, K. Ito, T.Katada, M. Ui, and S. Ishii. 1982. Subunit structure of islet-activating protein, pertussis toxin, in conformity with the A-Bmodel. Biochemistry 21:5516-5522.
26. Tamura, M., K. Nogimori, M. Yajima, K. Ase, and M. Ui. 1983.A role of the B-oligomer moiety of islet-activating protein,pertussis toxin, in development of the biological effects on intactcells. J. Biol. Chem. 258:6756-6761.
27. Weiss, A. A., and S. Falkow. 1982. Plasmid transfer to Bor-detella pertussis: conjugation and transformation. Infect. Im-mun. 152:549-552.
28. Weiss, A. A., E. L. Hewlett, G. A. Myers, and S. Falkow. 1983.Tn5-induced mutations affecting virulence factors of Bordetellapertussis. Infect. Immun. 42:33-41.
29. Yakobson, E. A., and D. G. Guiney. 1984. Conjugal transfer ofbacterial chromosomes mediated by the RK2 plasmid transferorigin cloned into transposon Tn5. J. Bacteriol. 160:451-453.
