(~) INSTITUT PASTEUR/ELSEVIER Res. MicrobioL Paris 1990 1990, 141, 787-796

A R O M A T I C - D E P E N D E N T SALMONELLA AS L i b IF VACCINE

P R E S E N T E R S OF FOREI G N E P I T O P E S AS INSERTS IN F L A G E L L I N

B.A.D. Stocker

Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, CA 94305-5402 (USA)

Summary.

Synthetic oligonucleotides specifying amino acid sequences identified as epitopes of various foreign antigens (cholera toxin subunit B, hepatitis B surface protein and others) have been inserted at an EcoRV-EcoRV deletion site iF a cloned Salmonella flagellin gene; the resulting plasmids, when placed in flage!l~a-negative Escherichia coli or Salmonella sp. strains, caused production of flagellin expressing the epitope. If the chimeric flagellin allowed formation of flagella, the epitope was exposed at the surface of the flagellar filaments. A AaroA flagellin-negative S. dublin live vaccine strain given plasmids carrying various chimeric flagellin genes was administered to mice, etc. Serum antibody specific for the foreign epitope was in all cases evoked by parenteral administration; oral route administration was effective in the case of two epitopes of hepatitis B surface protein but not effective for several other epitopes. Several i.p. inocula of the live vaccine strain with an insert corresponding to the 15 N-terminal amino acids of the M protein of Strep- tococcus pyogenes type 5 evoked M-specific antibody with opsonic activity, and the mice were (incompletely) protected against a lethal challenge of S. pyogenes type 5.

The non-virulence of Salmonella sp. strains with complete blocks in the aromatic biosynthesis pathway, even for animals with genetically determined or other defects in host defences, can be completely accounted for by their requirement for p-aminobenzoic acid, since non-leaky pabB mutations caused similar attenuation. Two transposon insertions at aroE caused little or no attenuation, presumably be- cause they did not result in complete block of the relevant s;ep in biosynthesis. The limited growth of AaroA strains in mouse tissues parallels that which precedes the bacteriostasis caused by addition of a sulphonamide to a growing broth culture of a sulphonamide-sensitive strain; the final cessation of growth in each case presuma- bly results from inability to initiate new protein chains with a formyl-me,hio,nine unit when the original folic acid content of the bacteria has been diluted out by residual growth.

KEY-WORDS: Live Vaccine, Flagellin, Salmonella, Epitope; Inserts.

788 B.A.D. STOCKER

Introduction.

Most work in my laboratory on the use of live vaccine Salmonella as deliverers of foreign antigens or epitopes has concerned vaccines given by injection rather than by the oral route. I first describe our system by which peptides constituting epitopes of unrelated organisms are made integral components of Salmonella flagellar pro- tein or flagellin and the results of their presentation by live vaccine strains of S. typhimurium or S. dublin, then mention some observations on the aromatic- dependent live-vaccine strains of Salmonella constructed in my laboratory and dis- cuss some theoretical considerations on the use of aromatic-dependent Salmonella as vaccines, either to protect against Salmonella infection or to obtain an immune response to other antigens.

Materials and methods.

For methods used to insert epitope-specifying oligonucleotides into the cloned d flagellin gene, see Newton et aL, 1989 and Wu et al., 1989, and for methods used to construct aroA live vaccine strains, including those with AaroA148, see Hoiseth and Stocker, 1981; Edwards and Stocker, 1988; Sigwart et al., 1989.

Results.

The system devised in my laboratory (Newton et al., 1989, 1990) for inducing an immune response to peptide epitopes by administration of a live vaccine strain of Salmonella sp. involves the use of a cloned chromosomal fragment from a strain of S. muenchen which includes afliC+ (formerly called H-l) allele specifying flagel- lin which when polymerized into flagella r filaments constitutes flagellar antigen d (Wei and Joys, 1985). The sequencing ot the ca. 1,500 base pairs of this flagellin gene and of three others specifying flagellins which, as filaments, constitute flagellar antigens a, c and i, showed identity or near identity between the alleles for several hundred base pairs at the C-terminal and likewise at the N-terminal end, but increas- ing divergence towards the middle, with a "hypervariable" segment of ca. 360 base pairs where the homology between alleles in respect o f predicted amino acid sequence is no greater than 30 °70, for any pairwise comparison (Wei and Joys, 1985; Joys, 1985).

The great variation in amino acid sequence in this region, segment IV of Joys (1985), and the fact that single amino acid substitution mutations within this seg- ment and likewise in-frame deletions in it, could be obtained by selection for ability to swim in the presence of anti-flagella serum (Joys and Martin, 1973; Newton and Stocker, unpublished) suggested that substantial alterations in sequence in this part of the molecule might be tolerated without loss of function. To test this, Dr. Salete Newton, then a postdoctoral fellow in my laboratory, transferred the EcoRI-EcoRI chromosomal fragment containing the d flagellin gene from its original plasmid to

AMP = adenosine monophosphate. 1 i.p. = intraperitoneal(ty). CFU = colony-forming unit. [ pAB = p-aminobenzoic acid. ELISA = enzyme-linked immunoassay, p.o. = per os. GMP = guanosine monophosphate. TMP = thymidine monophosphate.

S A L M O N E L L A A S L I V E V A C C I N E E P I T O P E - P R E S E N T E R S 789

plasmid pUCI9, which has no EcoRV sites. The presence of two EcoRV sites, 48 base pairs apart, in region IV of the d flagellin gene then allowed generation of a 48-base-pair deletion by EcoRV digestion followed by re-ligation. The d flagellin plasmid with this deletion, called pLS408, conferred motility on flageUin-negative E. coil and Selmonella sp. strains, though not to the same extent as plasmid pLS405, with the wild-type d flagellin gene.

Synthetic complementary nucleotides, either fuli-length or overlapping, specify- ing the amino acid sequence of the epitope of interest, with codon usage chosen to correspond to that in major protein genes of E. coil, were purchased, and after ap- propriate treatment the double-stranded molecules were added to plasmid pLS408 cut by EcoRV endonuclease. The ligation mixture was then used to transform com- petent cells of a flagellin-negative strain of E. coli, C L~ 7 , with selection for the ampicillin resistance trait of the plasmid. Transformants were tested for motility and representative plasmids were transferred, via an S. typhimurium intermediate which is restriction-negative but modification-proficient for all three systems of this spe- cies, to a flagellin-negative live vaccine strain of S. typhimurium or 5. dublin. The relevant part of the hybrid flageUin gene was then sequenced by a standard method to determine the orientation of the inserted fragment. (When overlapping nucleo- tides had been used, a few clones were found to have inserts lacking one or two C-terminal or N-terminal codons).

If a recombinant flagellin gene allowed formation of functional flagella, expres- sion of the foreign epitope (and therefore insertion of oligonucleotide in desired orien- tation) could be recognized by the bacterial immobilization and immunogold labelling of flagellar flaments by antipeptide serum. If the recombinant gene did not confer motility, then expression was tested by looking for reaction of both anti-flageUar- antigen serum and anti-peptide antibody with a protein band of the expected size in Western blots of bacterial lysates. Only about half of the inserts confirmed as cor- rectly inserted in desired orientation conferred motility but expression of the foreign epitope was detected by Western blot even when motility was not observed.

Table I summarizes results obtained (at Stanford or in a collaboration with the laboratory of the late Dr. E. Beachey) by immunization with an aroA(deletion) strain of S. dublin made flagellin-negative by transductional replacement of its single flagellin gene by an allele inactivated by insertion of transposon TnlO, and given plasmid pLS408 with inserts specifying various epitopes. Our first experiments (Newton et al., 1989) concerned an amino acid sequence corresponding to residues 50 to (,4 of subunit B of cholera toxin, identified as a major neutralizing epitope, CTP3, of cholera toxin (Jacob et al., 1983). A recombinant plasmid, pLS411, consisting of pLS408 with an oligonucleotide specifying the CTP3 sequence inserted at its EcoRV site, conferred motility on a flagellin-negative S. typhimurium strain and on the live vaccine strain of S. dublin, SL5928; anti-CTP3 monoclonal antibody arrested this motility and bound to the flagellar filaments, as shown by immunogold labelling. Each of five C57B 16 mice given three doses of the live vaccine strain by i.p. injection developed substantial serum titres by ELISA assay with either the synthetic peptide or whole cholera toxin as test antigen, as did mice similarly immunized but with killed vaccine given by the same route. In a single test, however, mice given three doses of the live vaccine strain by mouth did not show any such response.

In an investigation in the laboratory of Dr. W. Robinson (Division of Infectious Disease, Department of Medicine, Stanford University School of Medicine), four chimeric flagellin genes, each with one or more than one insert of an oligonucleotide specifying one or another of two epitopes of hepatitis B sarface protein, either S(122-137) or pre-S2 (121-145), one at least in correct orientation, caused produc- tion of flagellin expressing the epitope. All four versions of the epitope-expressing live vaccine strain caused production of antibody reactive with the relevant peptide, protein and virion, when injected i.p. and also when fed to mice or guinea pigs (Wu et al., 1989). These results first showed that some recombinant flagellin genes cause

790 B . A . D . S T O C K E R

TABLE I. - - Epitope inserts in the plasmid-horne flagellin gene: ability to c~nfer motility on a flagellin-negative aroA host and immunogeneic effect

of vaccine administration.

Vaccination trial (**)

Epitope Motil- No. of ELISA (residues) Plasmid ity(*) Species Dose doses Route (+/tested)

Cholera toxin subunit B CTP3 pLS411 + C57BL/6 5 x 106, live × 3 i.p. 5/5 (50-64) C57BL/6 5× 106 killed ×3 i.D. 5/5

C57BL/6 1'09', live x 3 p.o, 0/5

Hepatitis B surface protein S(122-137) S16 -

Pre-S2 pS21 (120-145)

S(122-137) $20

Pre-S2 pS8 (121-145)

rabbit BALB/cJ guinea pigs rabbit B10/BR guinea pigs rabbit BALB/cJ guinea pigs rabbit B 10/BR guinea pigs

109 live 5 x l0 s live

109 live 109 live

5 x l0 s live 109 live 109 live

5 x I0 g live 109 live 109 live

5 × 10 s live 109 live

x 5 i.m. 2/2 x 4 p.o. 10/10 x 4 p.o. 3/3 x 5 1.m. 2/2 × 4 p.o. i0 /10 x 4 p.o. 3/3 x 5 t.m. 2/2 x 4 p.o. 10/10 × 4 .p.o. 3/3 x 5 ~.m. 2/2 x 4 p.o. 10/10 x 4 p.o. 3/3

Streptococcus pyogenes type-5 M protein 42-57 pLS435 + rabbit 10 s, kilIed × 3 i.m. 2/2

BALB/c 5 x 106, live ×3 i.m. 5/5 BALB/c l09, live × 3 p.o. 0/5

HIV surface protein gp160 735-752 pLS439 + rabbit 108, killed ×3 i.m. 1/2

BALB/c 5 x 106, live x 3 i.p. 5/5 BALB/c 5 x 109, live ,'< 5 p.o. 0/5

Murine eytomegalovirus immediate-early protein pp89 168-176 pLS462 + BALB/c 5 x 10 °, live

ByJ x 3 i.p. n.t.

The flagellin-determining plasmids were pLS408, with synthetic oligonucleotides specifying the indi- cated epitope sequences inserted at the EcoRV site in the d flagellin gene with the deleted EcoRV-EcoRV fragment.

(')Motility of flagellin-negativc Salmonella sp. strains given the indicated plasmid. (")The aroA flagellin-negative live-vaccine carrier strain used was in all trials S. dublin SL5928. C57BL/6, BALB/cJ, BI0.BR and BALB/c are all strains of mice. n.t. = not tested.

production o f flagellin but not of flagella and that such flagellin, presumably ac- cumulated within the bacterial cell, may be an effective immunogen when presented by a live vaccine strain given either by injection or the oral route.

Another epitope tested as insert in fla~ellin comprised residues 735-752 of pro- tein gpl60 of HIV, from the C-terminal half of its gp41 fragment. This sequence was chosen because the corresponding synthetic peptide coupled to a protein carrier had been shown to cause production of HIV-neutralizing antibody when injected

S A L M O N E L L A A S L I V E V A C C I N E E P I T O P E - P R E S E N T E R S 791

into rabbits with adjuvant (Kennedy et ai., 1986). Full-length complementary syn- thetis- nucleotides, kindly given to us by Dr. T.M. Joys, were annealed; the double- straMed product was successfully ligated with cut pLS408. A clone with a correctly oriemed insert conferred motility, whereas one in reverse orientation did not. The serum of one of two rabbits given three injections of a killed culture of the S. dublin live vaccine strain, SL5928, made motile by the plasmid conferring motility, pLS439, reacted well by ELISA with the synthetic peptide and in tests in the laboratory of Dr. R.C. Kennedy (Southwest Foundation for Biomedical Research, San Antonio), had ELISA titre against gp 120 about the same as that obtained earlier by immuniza- tion with the peptide conjugate; the other rabbit made only a low titre response in the same test. All of five BALB/c mice given three i.p. doses, each 5 x 10 ° CFU, developed substantial ELISA titres against the synthetic peptide and several of these sera, tested in Dr. Kennedy's lab, had ELISA titres against gpl60. However, again no such response was seen in five mice fed the live vaccine, 5 x 109 CFU, five doses. Four other oligonucleotides specifying sequences from gpl20 have been inserted into the flagellin gene in plasmid pLS408 modified by insertion of a linker at the EcoRV site. (These oligonucleotides were given to us by Dr. Maurice Hofnung, who had designed them for insertion at linker sites in gene lamb (Charbit et al., 1986, 1988) and the modified version of the flagellin plasmid was provided by Dr. R. Brey, of Praxis Biologics, Inc.). Only one of the four recombinant piasmids with insert con- ferred motility on the flagellin-negative live vaccine strain. The immune response of mice to these four chimeric flagellins is at present being tested.

Earlier ~vork in the laboratory of the late Dr. Ed Beachey, at the University of Tennessee, Memphis, showed that oral administration to BALB/c mice of an aroA live vaccine' strain carrying a low-copy-number plasmid which included the gene for the M surface protein of Streptococcuspyogenes type 5 caused production of serum and salivary anti-M5 antibody, with opsonizing activity. Vaccinated mice survived a lethal challenge, i.p. or intranasal, with type 5 streptococci (Poirier et al., 1988). In a recent collaboration, Drs. M. Kotb and T. Poirier, with Ed Beachey, and Salete Newton in my laboratory, have expressed an epitope of the M5 protein as an insert in flagellin. The amino acid sequence used is that of the N-terminal 15 residues of the mature form of the M5 protein. Plasmid pLS408 with the corresponding oligonucleotide inserted in correct orientation introduced into the S. dublin flagellin. negative live vaccine strain rendered ;t motile. Expression of the epitope at the sui'- face of the flagellar filaments was shown by the immobilization test and immunogold microscopy. Table II records the ELISA titres, with the synthetic peptide or with whole live-vaccine bacteria (without insert) as test antigen, of sera collected after i.p. vaccination of BALB/c mice. A control group of mice were similarly vaccinated with strain SL5928 given the flagellin plasmid without any insert, to allow detection of any non-specific protection resulting from live vaccine administration. The opsoniz- ing activity of the sera was also tested (table II). After several doses, the ELI~iA titres and opsonizing activity were such as to predict protection. Two weeks after the fifth booster dose, groups of five vaccinated mice were challenged with type 5 streptococ- ci, i.p., or with type 24 streptococci, i.p., or with S. dublin, wild-type ancestor of the aroA live vaccine strain, i.p. All of the vaccinated mice tested survived the S. dublin challenge, which killed all of five non-vaccinated mice. All the vaccinated mice challenged with type 24 streptococci died, as did the five vaccinated with the strain lacking insert in the flagellin gene and challenged with the type 5 strain. By contrast, four of the five mice given the live vaccine strain with the M5 epitope insert survived the type 5 challenge. Thus, the immune response to the M5 epitope as a component of the flagellin was sufficient to cause substantial protective immunity.

Severa~othere Ro ~ fromavatlet ofant~ n • p" Fe., " y "ge s, have been expressed in the flagel- lin system, at Stanford or elsewhere, and are being !nvestigated as to immunogenic efficacy ~'hen given as a component of a live vaccine or as concentrated flagella. See Majarian et al. (1989) for a preliminary account of some of the results obtained by workers at Praxis Biologics, Inc.

792 B.A.D. STOCKER

TABLE II. - - Summary of immune responses of BALB/c mice given S. dublin live vaccine strain with M5 insert in flagellin, 5 × 106 CFU, i.p., weekly, or given

control live-vaccine strain, without insert in flagellin,

ELISA titres(*) Opsonization("°)

Week Peptide S. dublin %

0 <100 <100 0 2 200 3,200 6 4 800 12,800 52 6( '") 6,:400 25,600 96

Mice given live-vaccine strain SL5928 without insert in flagellin gene. 0 < 100 < 100 0 6("*) < 100 12,800 0

(')Titres of pooled sera from 5 mice in ELISA test with synthetic M5 peptide or whole bacteria of S. dublin live-vaccine strain with no insert in flagellin gene used as test antigen.

(")Percent of human aeutrophila with one or more associated type 5 s)reptococci. ('")Two weeks after the last vaccine dose, these mice were challenged by i.p. injection of 1.6x 106

type 5 streptococci. None of the 5 mice given the live vaccine strain without M5 insert flagellin survived; only ,one of the 5 given the live vaccine strain with insert died.

Discussion

It seems to me that our results to date, described above, call not for discussion, but for more experiments ! I therefore take this opportunity to discuss the theoreti- cal background for the use of aromatic-dependent Salmonella sp. as live vaccines, with mention of some recent relevant observations.

Strains with complete, irreversible blocks in the common aromatic biosynthesis pathway (fig. 1) cannot make chorismic acid, the precursor of all endogenous aro- matic metabolites in bacteria and plants. Strains of S. typhimurium with such com- plete blocks were originally tested for attenuation and possible utility as live vaccines with the idea that each of their two requirements for compounds not known to be metabolites in higher animals, that is, for p-aminobenzoic acid (pAB) and for 2,3-dihydroxybenzoic acid, would prevent their growth in host tissues and thus cause loss of virulence, at least in respect of systemic infections (Hoiseth and Stocker, 1981). However, it now seems that the non-virulence of such strains can be completely ac- counted for by their requirement for p-aminobenzoic acid (pAB). Two lines of evi- dence suggest this conclusion. One is the effect on virulence of non-leaky pab mutations, which block conversion of chorismic acid to pAB. The pab mutant of S. typhi found of reduced virulence by Bacon el al., in 1950, when later tested by i.p. inoculation of mice with hog gastric mucin as adjuvant had LDs0 ca. 10~-fold greater than that of isogenic pab + strains (Brown and Stocker, 1987); similarly a non-leaky pabB mutation transduced into a mouse-virulent S. typhimurium strain caused attenuation about as complete as that caused by areA deletion mutations (Stocker, unpublished data). Secondly, the inability of strains with blocks in aro- matic biosynthesis to make the siderophor enterobactin from chorismic acid via 2,3-dihydroxybenzoic acid appears not to cause loss of virulence. A non-leaky ent mutation preventing enterobactin synthesis introduced by transduction into a viru- lent strain of S. typhimurium had no effect on its virulence, as tested by i.p. inocula-

S A L M O N E L L A A S LIVE VACCINE EPITOPE-PRESENTERS 793

pep

i cooH cooM i

c . - o -~o~ ? .o o e,oo coo~ ,roe COO, CHz ~oFGH , ~ ~ - ~

+ w, Z"X)-C-H Z - 3 o OH I MO OH c l io I o OH H-C-0H "1 " o~ i oH OH

H-C-Or4 H-C-OH J I~SHIK IMIC ACID

Ct'~O~ POSH2 ~H~.O.POzH Z OHO Otis . TR¥1~roPt#AN

EP OAHP S~P aroL, taro: TYf~DS|NE

PHENYLALANIN|

COOH : toA C00H #toC COOH / / . . ~ FOLICACID

"~ ~ ' . ~ | N T E R O C H E L I N o H OH ~OOH OH COOH H20]P"O OH $ HzOsP'O 0"~ ' O'C

SAP [PSAP CHORISMIC ACID ~ " " t L UIIIDUINONE

MENADUINONE

FIG. 1. - - Aromatic biosynthesis pathway.

(PRAB) p-aminobenzoic acid; (DHB) 2,3-dihydroxybenzoic acid. eprinted, with permission of Butterworth & Co., from Stocker, 1988, Vaccine, 6, 141-145.)

tion of BALB/c mice, even when the ent mutation was combined with a deletion mutation affecting gene tonB (= chr), whose function is required for uptake of iron bound to enterobactin (S.K. Hoiseth, Ph.D. thesis, Stanford, 1982).

Complete blocks in the aromatic biosynthesis pathway caused by mutation of genes aroA, aroC or aroD have all been shown to cause about the same complete or almost complete loss of mouse virulence in S. typhimurium and other serotypes, as tested by i.p. inoculation of genetically susceptible (ity-s) mice. The case is different for gene aroE, though the protein specified by gene aroE+ catalyses a reaction in the same pathway; two different transposon insertions causing loss of function of this gene were recently tested (Stocker, unpublished; C. Hormaeche, pers. comm.) and found to cause no or only a sfight degree of attenuation of S. typhimurium. No proven deletions of this gene are available, so two explanations are possible. (i) Each of the transposon insertions may be in a part of the gene, perhaps promoter region or near the C-terminus of the coding sequence, such that a protein product is still produced, though in very small amount or with only very slight activity; (ii) alternatively, the reactzon normally catalysed by the product of wild-type gene aroE+ may proceed at a rate sufficient to provide the very small amount of pAB required for synthesis of folic acid, either without assistance of any enzyme or by a side activity of some enzyme specified by a gene other than those of the aro pathway.

A naive expectation would be that a Aaro mutant, because it required pAB for growth, would not multiply at all when introduced into an environment not contain- ing this compound, such as the tissues of a vertebrate host. In fact, however, the number of CFU recovered from liver and spleen of mice killed at intervals after in- oculation with an aroA live-vaccine strain may decline by a factor of about ten dur- ing the first 24 h, then slowly increases to a plateau level about 100-fold the number found 24 h after vaccine injection. This limited multiplication parallels the effect of addition of a sulphonamide to a growing culture of a sulphonamide-sensitive strain, a treatment which blocks synthesis of folic acid from pAB by competitive inhibition. Growth is arrested but only after a delay which is shortest in simple defined medium,

794 B.A.D. S T O C K E R

but prolonged if the products of folate-dependent l-carbon transfer reactions in biosynthetic pathways (table III) are provided, preformed, in the medium. These metabolites, with one exception, are known or expected to be available in mammalian tissues, though probably at concentrations so low as to impede bacterial multiplica- tion in the case of thymine and purines, since thy, pur, and gua mutants are all of reduced virulence. The one product of a folate-dependent reaction which is not avail- able from the host is fMet-tRNAmetf, required by nearly all bacteria as precursor of the formyl-methionine residue used to initiate new protein chains. Perhaps inability to initiate new protein chains when the intracellular concentration of folic acid has been diluted out by growth accounts for the final cessation of multiplication of art) mutants, even in hosts with greatly reduced defence mechanisms.

TABLE Ill. - - Products of folate.dependent biosynthetic pathways.

Probable availability in host Product tissues (*)

AMP and GMP Limiting TMP Limiting Methionine Non-limiting Glycine Non-limiting Pantothenate Non-limiting (?) fMet-tRNAMetf Not available

(')As indicated by attenuation or non-attenuation of relevant class of auxotrophic mutant.

Strains attenuated by requirement for a metabolite not available in host tissues might be expected to be non-virulent even in abnormally susceptible hosts. As predict- ed, aroA live-vaccine strains of Salmonella sp. have been found to be greatly attenu- ated both in mice with genetically determined defects, ity-s, xid, or Ips, and in those made susceptible by treatments which impair host defence mechanisms, including injection of cyclophosphamide, X-irradiation or pretreatment with microparticulate silica given i.v. (S.K. Hoiseth, Ph.D. thesis, Stanford, 1982; C. Hormaeche, pers. comm. ; A.D. O'Brien, pers. comm.). To this list, we may perhaps add the SCID mouse; in a current experiment, my colleague at Stanford, D r Hope Rugo, working at DNAX, has found that the AaroA live-vaccine S. typhimurium strain SL3261 given i.p. is no more virulent for SCID mice than for the nearly isogenic BALB/c line.

The ability of aroA and other attenuated S. typhimuriura strains given by feeding or gavage to protect mice against later challenge with wild-type strains, given either by the oral route or by injection, is striking, but we know little about some steps in the process by which this protection is produced. For instance, some workers have used sodium bicarbonate, given before or with the bacterial inoculum, to protect the bacteria against killing by gastric acid while others have found this .precaution un- necessary. Ducluzeau and his colleagues (1970) used inocula comprising a mixture of equal numbers of E. coil, Shigella, etc., and of spores of an obligate thermophilic Bacillus strain, reasonably assumed to pass through the gut without either multipli- cation or death, thus serving as a biological marker. Their figure 1 shows the num- ber of CFU of Shigella, and of Bacillus, per gram of freshly passed pellet, in Call mice allowed to drink 1 ml of bacterial suspension after overnight deprivation of drinking water. In normal mice, both species appeared abruptly at high concentra- tion about 4 h after ingestion; by 24 h, the concentration of each had fallen to about

S A L M O N E L L A A S L I V E V A C C I N E E P I T O P E - P R E S E N T E R S 795

1 % of its initial value, which I presume indicates that by that t ime 99 % of the in- gested dose had been excreted. In the 24 h sample, the rat io of viable Shigella to Bacillus was almost the same as in the 4 h specimen. This strongly suggests that in the condit ions o f their experiment there was at most a trivial extent o f killing or mul- t iplication o f the Shigella in the gut lumen. Similar results were obtained with E. coli and several other species which might have been expected to suffer killing by gastric acid. These workers also showed by the same procedure that adminis trat ion o f an inoculum of S. typhimurium which caused no adverse effects when ingested by the mouse resulted in systemic infection and death o f some mice if given by garage.

Acknowledgements.

The work of Dr. Stocker's laboratory has been supported by U.S. Public Health Service grants AI-18872 and A1-27722 from the National Institute of Allergy and Infectious Diseases, grant 000553 from the American Foundation for AIDS Research and gifts from Praxis Biologics, Inc.

References.

BACON, G.A., BURROWS, T.W. & YATES, M. (1950), The effects of biochemical mutation on the virulence of Bacterium typshosum: the virulence of mutants. Brit. J. Exp. Path., 31, 714-724.

BROWN, R.F. & STOCKER, B.A.D. (1987), Salmonella typhi 205aTy, a strain with two attenu- ating auxotrophic characters, for use in laboratory teaching. J. Bact., 55, 892-898.

CHARmT, A., BOULAIN, J.C., RYTER, A. & HOFNUN6, M. (1986), Probing the topology of a bac- terial membrane protein by genetic insertion of a foreign epitope; expression at the cell surface. EMBO J., 5, 3029-3037.

CHARBIT, A., VANDER WERF, S., MIMIC, V., BOULAIN, J.C., GIRARD, M. & HOFNUNO, M. (1988), Expression of a poliovirus neutralization epitope at the surface of recombinant bac- tena: first immunization results. Ann. Inst. Pasteur/Microbiol., 139, 45-58.

DUCLUZEAU, R., BELLIER, M. & RAIBAUD, P. (1970), Transit digestif de divers inoculums bac- t6riens introduits per os chez des souris ax~niques et ~ holox6niques >~ (convention- nelles): effet antagoniste de la microflore du tractus gastroinstestinal. Zeut. Bakt. AO, 213, 533-548.

EDWARDS, M.F. & STOCKER, B.A.D. (1988), Construction of AaroA his Apur strains of Salmonella typhi. J. Bact., 170, 3991-3995.

HOISETN, S.K. & STOCKER, B.A.D. (1981), Aromatic-dependent Salmonella typhimurium are non-virulent and effective as live vaccir.e. Nature (Lond.), 291, 238-239.

JACOB, C.O., SELA, ~i. & ARNON, R. (1983), Antibodies against synthetic peptides of the B subunit of cholera toxin: cross reaction and neutralization of the toxin. Proc. nat. Acad. Sci. (Wash.), 80, 7611-7615.

Joys, T.M. (1985), The covalent structure of the phase-I flagellar filament protein of Salmonella typhimurium and its comparison with other flageUins. J. biol. Chem., 260, 15758-15761.

JoYs, T.M. & MARTIN, J.F. (1973), Identification of amino acid changes in serological mu- tants of the i flagellar antigen of Salmonella typhimurium. Microbios., 7, 71-73.

KENNEDY, R.C., HENKEL, R.D., PAULETTI, D., ALLAN, J.S., LEE, T.H. & DREESMAN, R. (1986), Antiserum to synthetic peptide recognizes the HTLV-III envelope glycoprotein. Science, 231, 1556-1559.

MAJARIAN, W.R., gASPER, S.J. & BREY, R.N. (1989), Expression of heterologous epitopes as recombinant flagella on the surface of attenuated Salmonella, in "Vaccines 89: modern approaches to new vaccines including prevention of AIDS" (R.A. Lerner et al.), (pp. 277-281). Cold Spring Harbor Laboratory, New York.

NEWTON, S.M.C., JACOB, C.O. & STOCKER, B.A.D. (1989), Immune response to cholera toxin epitope inserted in Salmonella flagellin. Science, 244, 70-72.

NEWTON, S.M.C., MANNING, W.C., Hovi, M. & STOCKER, B.A.D. (1990), Aromatic-dependent Salmonella with foreign epitope insert in flagellin as live vaccine, in "Vaccines 90: modern Approaches to New Vaccines Including Prevention of AIDS" (F. Brown et aL) (pp. 439-445). Cold Spring Harbor Laboratory, New York.

796 B . A . D . S T O C K E R

PO,RIER, T.P., KEHOE, M.A. & BEACHEY, E.H. (1988), Protective immunity evoked by oral ad- ministration of attenuated aroA Salmonella typhimurium expressing cloned strep- tococcal M. protein. J. exp. Meal., 168, 25-32.

S~GWART, D.F., STOCKER, B.A.D. & CLEMaNTS, J.D. 0989), Effect of a purA mutation on the efficacy of Salmonella live-vaccine vectors. Infect. Imraun., 57, 1858-1861.

WE,, L.-N. & JoYs, T.M. 0985), Covalent structure of three phase-I flagellar filament pro- teins of Salmonella. J. tool. Biol., 186, 791-803.

Wu, J.Y., NEWTON, S., JUDD, A., STOCKER, B. & ROBINSON, W.S. 0989), Expression of im- munogenic epitopes of hepatitis B surface antigen with hybrid flagellin proteins by a vaccine strain of Salmonella. Proc. nat. Acad. Sci. (Wash.), 86, 4726-4730.