US 5,695,983 AGrant

Salmonella Vaccines

Issue Date:1997-12-09

•5 Claims

•20 Drawing Sheets

Abstract

A bacterial cell the virulence of which is attentuated by a first mutation in a PhoP regulon and a second mutation in an aromatic amino acid synthetic gene and bacterial cells the virulence of which is attenuated by a mutation in one or more PhoP-activated genes or one or more PhoP-repressed genes.

Metadata

Assignees

The General Hospital Corporation
President and Fellows of Harvard College

Inventors

Samuel I. Miller
John J. Mekalanos

Application Information

Application Number:US 2713543

Filing Date:1994-07-06

Priority Date:1990-12-18

Art Unit:185

Classifications

IPC:

C12N 120C12N 136

Field of Search:

435252.3;245;252.8

Patent Drawings (20 sheets)

Description

BACKGROUND OF THE INVENTION

The invention relates to vaccines.

Enteric fevers and diarrheal diseases, e.g., typhoid fever and cholera, are major causes of morbidity and mortality throughout the developing world, Hook et al., 1980, In Harrison's Principles of Internal Medicine, 9th Ed., 641-848, McGraw Hill, New York. Traditional approaches to the development of vaccines for bacterial diseases include the parenteral injection of purified components or killed organisms. These parenterally administered vaccines require technologically advanced preparation, are relatively expensive, and are often, because of dislike for needle-based injections, resisted by patients. Live oral vaccine strains have several advantages over parenteral vaccines: low cost, ease of administration, and simple preparation.

The development of live vaccines has often been limited by a lack of understanding of the pathogenesis of the disease of interest on a molecular level. Candidate live vaccine strains require nonrevertable genetic alterations that affect the virulence of the organism, but not its induction of an immune response. Work defining the mechanisms of toxigenesis of Vibrio cholerae has made it possible to create live vaccine strains based on deletion of the toxin genes, Mekalanos et al., 1983, Nature 306:551, Levine et al., 1988, Infect. Immun. 56:161.

Recent studies have begun to define the molecular basis of Salmonella typhimurium macrophage survival and virulence, Miller et al., 1989, Proc. Natl. Acad. Sci. USA 86:5054, hereby incorporated by reference. Salmonella typhimurium strains with mutations in the positive regulatory regulon phoP are markedly attenuated in virulence for BALB/c mice. The phoP regulon is composed of two genes present in an operon, termed phoP and phoQ. The phoP and phoQ gene products are highly similar to other members of bacterial two-component transcriptional regulators that respond to environmental stimuli and control the expression of a large number of other genes. A mutation at one of these phoP regulatory region regulated genes, pagC, confers a virulence defect. Strains with pagC, phoP, or phoQ mutations afford partial protection to subsequent challenge by wild-type S. typhimurium.

Salmonella species cause a spectrum of clinical disease that includes enteric fevers and acute gastroenteritis, Hook et al., 1980, supra. Infections with Salmonella species are more common in immunosuppressed persons, Celum et al., 1987, J. Infect. Dis. 156:998. S. typhi, the bacterium that causes typhoid fever, can only infect man, Hook et al., 1980, supra. The narrow host specificity of S. typhi has resulted in the extensive use of S. enteriditis typhimurium infection of mice as a laboratory model of typhoid fever, Carter et al., 1984 J. Exp. Med. 139:1189. S. typhimurium infects a wider range of hosts, causing acute gastroenteritis in man and a disease similar to typhoid fever in the mouse and cow.

Salmonella infections are acquired by oral ingestion. The organisms, after traversing the stomach, replicate in the small bowel, Hornik et al., 1970, N. Eng. J. Med. 283:686. Salmonella are capable of invasion of the intestinal mucosal cells, and S. typhi can pass through this mucosal barrier and spread via the Peyer's patches to the lamina propria and regional lymph nodes. Colonization of the reticuloendothelial cells of the host then occurs after bacteremia. The ability of S. typhi to survive and replicate within the cells of the human reticuloendothelial system is essential to its pathogenesis, Hook et al., 1980, supra, Hornick et al., 1970, supra, and Carter et al., 1984, supra.

Immunity to Salmonella typhi involves humoral and cell-mediated immunity, Murphy et al., 1987, J. Infect. Dis. 156:1005, and is obtainable by vaccination, Edelman et al., 1986, Rev. Inf. Dis. 8:324. Recently, human field trials demonstrated significant protective efficacy against S. typhi infection after intramuscular vaccination with partially purified Vi antigen, Lanata et al., 1983, Lancet 2:441. Antibody-dependent enhancement of S. typhi killing by T cells has been demonstrated in individuals who received a live S. typhi vaccine, indicating that these antibodies may be necessary for the host to generate a cell-mediated immune response, Levine et al., 1987, J. Clin. Invest. 79:888. The cell-mediated immune response is important in typhoid immunity since killed vaccines that do not induce this immune response are not protective in man, Collins et al., 1972, Infect. Immun. 41:742.

SUMMARY OF THE INVENTION

The invention provides a Salmonella vaccine which does not cause transient bacteremia. In general, the invention features a bacterial cell, preferably a Salmonella cell, e.g., a S. typhi, S. enteritidis typhimurium, or S. cholerae-suis cell, the virulence of which is attenuated by a first mutation in a PhoP regulon and a second mutation in an aromatic amino acid synthetic gene. As used herein, PhoP regulon is defined as a DNA which comprises a unit of Salmonella virulence gene expression characterized by two regulatory genes, phoP and phoQ, and structural genes, the expression of which is regulated by phoP and phoQ, e.g., phoP regulatory region repressed genes (prg) or phoP regulatory region activated genes (pag). Such a bacterial cell can be used as a vaccine to immunize a mammal against salmonellosis.

The Salmonella cell may be of any serotype, e.g., S. typhimurium, S. paratyphi A, S. paratyphi B, S. paratyphi C, S. pylorum, S. dublin, S. heidelberg, S. newport, S. minnesota, S. infantis, S. virchow, or S. panama.

The first mutation may be a non-revertable null mutation in the PhoP/PhoQ locus. Preferably, the mutation is a deletion of at least 100 nucleotides; more preferably, the mutation is a deletion of at least 500 nucleotides; even more preferably, the mutation is a deletion of at least 750 nucleotides; and, most preferably, the mutation is a deletion of nucleotides 376 to 1322 of the PhoP/PhoQ regulatory locus.

The second mutation may be a non-revertable null mutation in an aroA locus or a non-revertable null mutation in an aroC/aroD locus, or another locus involved in the biosynthesis of aromatic amino acids.

To further attenuate the virulence of the bacterial cell of the invention, the cell may contain yet another mutation, e.g., a deletion, in a non-aromatic amino acid synthetic gene, e.g., a mutation which renders the cell auxotrophic for a non-aromatic amino acid, e.g., histidine. In preferred embodiments, the bacterial cell of the invention is a S. typhi cell with the genotype AroA.sup.-, His.sup.-, PhoP/PhoQ.sup.-, e.g., TyLH445.

The invention may also include a Salmonella cell, the virulence of which is attenuated by the constitutive expression of a gene under the control of a two-component regulatory system. In preferred embodiments the constitutive expression is the result of a mutation at a component of the two-component regulatory system. In preferred embodiments the bacterial cell includes a second mutation which attenuates virulence.

In yet other preferred embodiments of the vaccine the two-component regulatory system is the phoP regulatory region, and the gene under the control of the two-component system is a phoP regulatory region regulated gene, e.g., a prg gene, e.g., prgA, prgB, prgC, prgE, or prgH, or pag gene, e.g., pagC. In preferred embodiments constitutive expression is the result of a change or mutation, e.g., a deletion, (preferably a non-revertible mutation) at the promoter of the regulated gene or of the phoP regulatory region, e.g., a mutation in the phoQ or the phoP gene, e.g., the PhoP.sup.c mutation.

In another aspect, the invention features a vaccine including a bacterial cell which is attenuated by decrease of expression of a virulence gene under control of a phoP regulatory region, e.g., a prg gene, e.g., prgA, prgB, prgC, prgE, or prgH.

In preferred embodiments of the vaccine the Salmonella cell includes a first mutation, e.g., a deletion, which attenuates virulence, e.g., a mutation in a phoP regulatory region gene, e.g., a mutation in the phoP or phoQ gene, e.g., PhoP.sup.c, or a mutation in a phoP regulatory region regulated gene, and a second mutation which attenuates virulence, e.g., a mutation in an aromatic amino acid synthetic gene, e.g., an aro gene, a mutation in a phoP regulatory region regulated gene, e.g., a mutation in a prg gene, e.g., prgA, prgB, prgC, prgE, or prgH, or pag locus, e.g., a pagC mutation.

In yet other preferred embodiments the bacterial cell includes a first mutation in a phoP regulatory region gene and a second mutation in an aromatic amino acid synthetic gene, e.g, an aro gene.

In another aspect, the invention features a vaccine, preferably a live vaccine, including a bacterial cell, the virulence of which is attenuated by a mutation, e.g., a deletion, in a gene under the control of a two-component regulatory system. In preferred embodiments the bacterial cell includes a virulence attenuating mutation in a second gene, e.g., in an aromatic amino acid synthetic gene, e.g., an aro gene.

In yet other preferred embodiments of the vaccine the bacterial cell is a Salmonella cell, the two-component regulatory system is the phoP regulatory region, and the gene under its control is a prg gene, e.g. prgA, prgB, prgC, prgE, or prgH, or a pag gene, e.g., the pagC gene.

In another aspect the invention features a vaccine, preferably a live vaccine, including a Salmonella cell e.g., a S. typhi, S. enteritidis typhimurium, or S. cholerae-suis cell, including a first virulence attenuating mutation in an aromatic amino acid biosynthetic gene, e.g., an aro gene, and a second virulence attenuating mutation in a phoP regulatory region gene, e.g., a phoP.sup.- mutation.

In another aspect the invention features a bacterial cell, or a substantially purified preparation thereof, preferably a Salmonella cell, e.g., a S. typhi, S. enteritidis typhimurium, or S. cholerae-suis cell, which constitutively expresses a gene under the control of a two-component regulatory system and which includes a virulence attenuating mutation, e.g., a deletion, which does not result in constitutive expression of a gene under the control of the two-component regulatory system. In preferred embodiments the bacterial cell includes a mutation in a component of the two-component regulatory system.

In preferred embodiments the bacterial cell is a Salmonella cell which expresses a phoP regulatory region regulated gene constitutively (the constitutive expression preferably caused by a mutation, preferably a non-revertible mutation, e.g., a deletion in the phoP regulatory region, e.g., a mutation in the phoQ or phoP gene, e.g., phoP.sup.c), and which includes a virulence attenuating mutation, preferably a non-revertible mutation, e.g., a deletion, preferably in an aromatic amino acid synthetic gene, e.g., an aro gene, or in a phoP regulatory region regulated gene, e.g., a prg gene, e.g., prgA, prgB, prgC, prgE, or prgH or pag gene, e.g., pagC which does not result in the constitutive expression of a gene under the control of the phoP regulatory region.

In another aspect, the invention features a bacterial cell, or a substantially purified preparation thereof, e.g., a Salmonella cell, e.g., a S. typhi cell, an S. enteritidis typhimurium or a S. cholerae-suis cell, including a virulence attenuating mutation in a gene regulated by a two-component regulatory system. In preferred embodiments the virulence attenuating mutation is in a phoP regulatory region regulated gene, e.g., a prg gene, e.g., prgA, prgB, prgC, prgE, or prgH or pag gene, e.g., pagC.

In preferred embodiments the bacterial cell includes a second mutation, e.g., in an aromatic amino acid synthetic gene, e.g., an aro gene, in a phoP regulatory region gene, e.g., the phoP or phoQ genes, or in a phoP regulating region regulated gene, e.g., a prg gene, e.g., prgA, prgB, prgC, prgE, or prgH or a pag gene, e.g., pagC, which attenuates virulence but which does not result in constitutive expression of a phoP regulatory region regulated gene.

The invention also features a live Salmonella cell, or a substantially purified preparation thereof, e.g., a S. typhi, S. enteriditis typhimurium, or S. cholerae-suis cell, in which there is inserted into a virulence gene, e.g., a gene in the phoP regulating region, or a phoP regulating region regulated gene, e.g., a prg gene, e.g., prgA, prgB, prgC, prgE, or prgH or a pag locus, e.g., pagC, a gene encoding a heterologous protein, or a regulatory element thereof.

In preferred embodiments the live Salmonella cell carries a second mutation, e.g., an aro mutation, e.g., an aroA mutation, e.g., aroA.sup.- or aroADEL407, that attenuates virulence.

In preferred embodiments the DNA encoding a heterologous protein is under the control of an environmentally regulated promoter. In other preferred embodiments the live Salmonella cell further includes a DNA sequence encoding T7 polymerase under the control of an environmentally regulated promoter and a T7 transcriptionally sensitive promoter, the T7 transcriptionally sensitive promoter controlling the expression of the heterologous antigen.

The invention also features a vector capable of integrating into the chromosome of Salmonella including: a first DNA sequence encoding a heterologous protein; a second (optional) DNA sequence encoding a marker e.g., a selective marker, e.g., a gene that confers resistance for a heavy metal resistance or a gene that complements an auxotrophic mutation carried by the strain to be transformed; and a third DNA sequence, e.g., a phoP regulon encoded gene, e.g., a prg gene, e.g., prgA, prgB, prgC, prgE, or prgH or a pag locus, e.g., pagC, encoding a phoP regulatory region regulated gene product necessary for virulence, the third DNA sequence being mutationally inactivated.

In other preferred embodiments: the first DNA sequence is disposed on the vector so as to mutationally inactivate the third DNA sequence; the vector cannot replicate in a wild-type Salmonella strain; the heterologous protein is under the control of an environmentally regulated promoter; and the vector further includes a DNA sequence encoding T7 polymerase under the control of an environmentally regulated promoter and a T7 transcriptionally sensitive promoter, the T7 transcriptionally sensitive promoter controlling the expression of the heterologous antigen.

In another aspect the invention includes a method of vaccinating an animal, e.g., a mammal, e.g., a human, against a disease caused by a bacterium, e.g., Salmonella, including administering a vaccine of the invention.

The invention also includes a vector including DNA which encodes the pagC gene product; a cell transformed with the vector; a method of producing the pagC gene product including culturing the transformed cell and purifying the pagC gene product from the cell or culture medium; and a purified preparation of the pagC gene product.

In another aspect the invention includes a method of detecting the presence of Salmonella in a sample including contacting the sample with pagC encoding DNA and detecting the hybridization of the pagC encoding DNA to nucleic acid in the sample.

The invention also includes a vector including DNA which encodes the prgH gene product; a cell transformed with the vector; a method of producing the prgH gene product including culturing the transformed cell and purifying the prgH gene product from the cell or culture medium; and a purified preparation of the prgH gene product.

In another aspect the invention includes a method of detecting the presence of Salmonella in a sample including contacting the sample with prgH encoding DNA and detecting the hybridization of the prgH encoding DNA to nucleic acid in the sample.

In another aspect the invention features a method of attenuating the virulence of a bacterium, the bacterium including a two-component regulatory system, including causing a gene under the control of the two-component system to be expressed constitutively. In preferred embodiments the bacterium is Salmonella, e.g., S. typhi, S. enteritidis typhimurium, or S. cholerae-suis, and the two-component system is the phoP regulatory region.

In yet another aspect, the invention features a substantially pure DNA which includes the sequence given in SEQ ID NO: 5 or a fragment thereof.

The invention also includes a substantially pure DNA which includes a sequence encoding pagD, e.g., nucleotides 91 to 354 of SEQ ID NO: 5 (pagD open reading frame (ORF)) and degenerate variants thereof that encode a product with essentially the amino acid sequence given in SEQ ID NO: 6, as well as the pagD ORF and its 5' non-coding region, nucleotides 4 to 814 of SEQ ID NO: 15) which contains the pagD promoter. DNA in the region between the pagC ORF and the pagD ORF (nucleotides 4 to 814 of SEQ ID NO: 15), DNA which includes the pagC promoter (nucleotides 562 to 814 of SEQ ID NO: 15), and DNA which includes the pagD promoter alone (nucleotides 4 to 776 of SEQ ID NO: 15) are also within the claimed invention.

The invention also includes a substantially pure DNA which includes a sequence encoding envE, e.g., nucleotides 1114 to 1650 of SEQ ID NO: 5 (envE ORF) and degenerate variants thereof that encode a product with essentially the amino acid sequence given in SEQ ID NO: 7.

Another aspect of the invention features a substantially pure DNA which includes a sequence encoding msgA, e.g., nucleotides 1825 to 2064 of SEQ ID NO: 5 (msgA ORF) and degenerate variants thereof which encode a product with essentially the amino acid sequence given in SEQ ID NO: 8, as well as the msgA ORF with its 5' non-coding region, nucleotides 1510 to 1824 of SEQ ID NO: 5 which contains the msgA promoter. Also within the invention is a substantially pure DNA comprising the msgA promoter alone (nucleotides 1510 to 1760 of SEQ ID NO: 5).

In yet another aspect, the invention features a substantially pure DNA which includes a sequence encoding envF, e.g., nucleotides 2554 to 3294 of SEQ ID NO: 5 (envF ORF) and degenerate variants thereof which encode a product with essentially the amino acid sequence given in SEQ ID NO: 9, as well as the envF ORF with its 5' non-coding region, nucleotides 2304 to 2553 of SEQ ID NO: 5 containing the envF promoter.

Also within the invention is a substantially pure DNA which includes the sequence given in SEQ ID NO: 10 or a fragment thereof.

The invention also includes a substantially pure DNA which includes a sequence encoding prgH, e.g., nucleotides 688 to 1866 of SEQ ID NO: 10 (prgH ORF) and degenerate variants thereof which encode a product with essentially the amino acid sequence given in SEQ ID NO: 11, as well as the prgH ORF with its promoter region (nucleotides 1 to 689 of SEQ ID NO: 10).

The invention also includes a substantially pure DNA which includes a sequence encoding prgI, e.g., nucleotides 1891 to 2133 of SEQ ID NO: 10 (prgI ORF) and degenerate variants thereof which encode a product with essentially the amino acid sequence given in SEQ ID NO: 12, as well as the prgI ORF with its promoter region (nucleotides 1 to 689 of SEQ ID NO: 10.

In another aspect, the invention features a substantially pure DNA which includes a sequence encoding prgJ, e.g., nucleotides 2152 to 2457 of SEQ ID NO: 10 (prgJ ORF) and degenerate variants thereof which encode a product with essentially the amino acid sequence given in SEQ ID NO: 13, as well as the prgJ ORF and its promoter region (nucleotides 1 to 689 of SEQ ID NO: 10.

In yet another aspect, the invention features a substantially pure DNA which includes a sequence encoding prgK, e.g., nucleotides 2456 to 3212 of SEQ ID NO: 10 (prgK ORF) and degenerate variants thereof which encode a product with essentially the amino acid sequence given in SEQ ID NO: 14, as well as the prgK ORF with its promoter region (nucleotides 1 to 689 of SEQ ID NO: 10.

The invention also encompasses a bacterial cell the virulence of which is attenuated by a mutation, e.g., a deletion, in one or more genes selected from the group consisting of pagD, pagE, pagF, pagG, pagH, pagI, pagJ, pagK, pagL, pagM, pagN, pagP, envE, and envF. Also included is a bacterial cell which is attenuated by a mutation, e.g., a deletion, in one or more genes selected from the group consisting of pagC, pagD, pagJ, pagK, pagM, and msgA. A bacterial cell, the virulence of which is attenuated by a mutation, e.g., a deletion, in one or more genes selected from the group consisting of prgH, prgI, prgJ, and prgK is also within the claimed invention.

Two-component regulatory system, as used herein, refers to a bacterial regulatory system that controls the expression of multiple proteins in response to environmental signals. The two-components referred to in the term are a sensor, which may, e.g., sense an environmental parameter and in response thereto promote the activation, e.g. by promoting the phosphorylation, of the second component, the activator. The activator affects the expression of genes under the control of the two-component system. A two-component system can include, e.g., a histidine protein kinase and a phosphorylated response regulator, as is seen in both gram positive and gram negative bacteria. In E. coli, e.g., 10 kinases and 11 response regulators have been identified. They control chemotaxis, nitrogen regulation, phosphate regulation, osmoregulation, sporulation, and many other cellular functions, Stock et al., 1989 Microbiol. Rev. 53:450-490, hereby incorporated by reference. A two-component system also controls the virulence of Agrobacterium tumefasciens plant tumor formation, Leroux et al. EMBO J 6:849-856, hereby incorporated by reference). Similar virulence regulators are involved in the virulence of Bordetella pertussis Arico et al., 1989, Proc. Natl. Acad. Sci. USA 86:6671-6675, hereby incorporated by reference, and Shigella flexneri, Bernardini et al., 1990, J. Bact. 172:6274-6281, hereby incorporated by reference.

Environmentally regulated, as used herein refers to a pattern of expression wherein the expression of a gene in a cell depends on the levels of some characteristic or component of the environment in which the cell resides. Examples include promoters in biosynthetic pathways which are turned on or off by the level of a specific component or components, e.g., iron, temperature responsive promoters, or promoters which are expressed more actively in specific cellular compartments, e.g., in macrophages or vacuoles.

A vaccine, as used herein, is a preparation including materials that evoke a desired biological response, e.g., an immune response, in combination with a suitable carrier. The vaccine may include live organism, in which case it is usually administered orally, or killed organisms or components thereof, in which case it is usually administered parenterally. The cells used for the vaccine of the invention are preferably alive and thus capable of colonizing the intestines of the inoculated animal.

A mutation, as used herein, is any change (in comparison with the appropriate parental strain) in the DNA sequence of an organism. These changes can arise e.g., spontaneously, by chemical, energy e.g., X-ray, or other forms of mutagenesis, by genetic engineering, or as a result of mating or other forms of exchange of genetic information. Mutations include e.g., base changes, deletions, insertions, inversions, translocations or duplications.

A mutation attenuates virulence if, as a result of the mutation, the level of virulence of the mutant cell is decreased in comparison with the level in a cell of the parental strain, as measured by (a) a significant (e.g., at least 50%) decrease in virulence in the mutant strain compared to the parental strain, or (b) a significant (e.g., at least 50%) decrease in the amount of the polypeptide identified as the virulence factor in the mutant strain compared to the parental strain.

A non-revertible mutation, as used herein, is a mutation which cannot revert by a single base pair change, e.g., deletion or insertion mutations and mutations that include more than one lesion, e.g., a mutation composed of two separate point mutations.

The phoP regulatory region, as used herein, is a two-component regulatory system that controls the expression of pag and prg genes. It includes the phoP locus and the phoQ locus.

phoP regulatory region regulated genes, as used herein, refer to genes such as pag and prg genes.

pag, as used herein, refers to a gene which is positively regulated by the phoP regulatory region.

prg, as used herein, refers to a gene which is negatively regulated by the phoP regulatory region.

An aromatic amino acid synthetic gene, as used herein, is a gene which encodes an enzyme which catalyzes a step in the synthesis of an aromatic amino acid. aroA, aroC, and aroD are examples of such genes in Salmonella. Mutations in these genes can attenuate virulence without the total loss of immunogenicity.

Abnormal expressions, as used herein, means expression which is higher or lower than that seen in wild type.

Heterologous protein, as used herein, is a protein that in wild type, is not expressed or is expressed from a different chromosomal site, e.g., a heterologous protein is one encoded by a gene that has been inserted into a second gene.

Virulence gene, as used herein, is a gene the inactivation of which results in a Salmonella cell with less virulence than that of a similar Salmonella cell in which the gene is not inactivated. Examples include the phoP, pagC, prgH genes.

A marker, as used herein, is gene product the presence of which is easily determined, e.g., a gene product that confers resistance to a heavy metal or a gene product which allows or inhibits growth under a given set of conditions.

Purified preparation, as used herein, is a preparation, e.g., of a protein, which is purified from the proteins, lipids, and other material with which it is associated. The preparation is preferably at least 2-10 fold purified.

Constitutive expression, as used herein, refers to gene expression which is modulated or regulated to a lesser extent than the expression of the same gene in an appropriate control strain, e.g., a parental or in wild-type strain. For example, if a gene is normally repressed under a first set of conditions and derepressed under a second set of conditions constitutive expression would be expression at the same level, e.g., the repressed level, the derepressed level, or an intermediate level, regardless of conditions. Partial constitutive expression is included within the definition of constitutive expression and occurs when the difference between two levels of expression is reduced in comparison in what is seen in an appropriate control strain, e.g., a wild-type or parental strain.

A substantially purified preparation of a bacterial cell is a preparation of cells wherein contaminating cells without the desired mutant genotype constitute less than 10%, preferably less than 1%, and more preferably less than 0.1% of the total number of cells in the preparation.

The invention allows for the attenuation of virulence of bacteria and of vaccines that include bacteria, especially vaccines that include live bacteria, by mutations in two-component regulatory systems and/or in genes regulated by these systems. The vaccines of the invention are highly attenuated for virulence but retain immunogenicity, thus they are both safe and effective.

The vectors of the invention allow the rapid construction of strains containing DNA encoding heterologous proteins, e.g., antigens. The heterologous protein encoding DNA is chromosomally integrated, and thus stable, unlike plasmid systems which are dependent on antibiotic resistance or other selection pressure for stability. Live Salmonella cells of the invention in which the expression of heterologous protein is under the control of an environmentally responsive promoter do not express the heterologous protein at times when such expression would be undesirable e.g., during culture, vaccine preparation, or storage, contributing to the viability of the cells, but when administered to humans or animals, express large amounts of the protein. This is desirable because high expression of many heterologous proteins in Salmonella can be associated with toxicity to the bacterium. The use of only a single integrated copy of the DNA encoding the heterologous protein also contributes to minimal expression of the heterologous protein at times when expression is not desired. In embodiments where a virulence gene, e.g., the pagC gene or the prgH gene, contains the site of integration for the DNA encoding the heterologous protein the virulence of the organism is attenuated.

A substantially pure DNA, as used herein, refers to a nucleic acid sequence, segment, or fragment, which has been purified from the sequences which flank it in a naturally occurring state, e.g., a DNA which has been removed from the sequences which are normally adjacent to the fragment, e.g., the sequences adjacent to the fragment in the genome in which it naturally occurs. The term also applies to DNA which has been substantially purified from other components which naturally accompany the DNA, e.g., DNA which has been purified from proteins which naturally accompany it in a cell.

Other features and advantages of the invention will be apparent from the following description of the preferred embodiments and from the claims.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

The drawings will first be described.

Drawings

FIG. 1 is a graph of the survival of Salmonella strains within macrophages.

FIG. 2 is a map of the restriction endonuclease sites of the pagC locus.

FIGS. 3A-3B are a map of the DNA sequence of the pagC region (SEQ ID NO: 1).

FIG. 4 is a map of the location of prgH within the hil locus. The arrows indicate the direction of orientation of the neomycin promoter of Tn5B50 insertions within the hil locus and the direction of transcription of the prgH1::TnphoA fusion protein. Restriction endonuclease sites are represented by B, BamH1; H, HindIII; X, XhoI; S, SacI; V, EcoRV.

FIGS. 5A-5B are a DNA sequence from the prgH gene (plasmid pIB01) (SEQ ID NO: 3).

FIG. 6 is a bar graph showing a comparison of the sensitivity of wild type (ATCC 14028), PhoP-null mutant (CS015), and pag::TnphoA mutant strains to NP-1 defensin. The y-axis represents the Defensin Killing Index (DKI) which is a measure of bacteria killed on exposure to NP-1. The DKI is defined as the logarithmic function of the ratio of control bacteria to surviving bacteria incubated with NP-1 DKI=log (CFU without NP-1/CFU with NP-1)!. The individual bars represent the mean and standard error of five separate experiments. The x-axis indicates the allele mutated. The mean DKI for each of the pag::TnphoA strains tested was determined not be different from that of wild type Salmonella. (P<0.05). In contrast, the phoP mutant was significantly different (P<0.0001).

FIG. 7 is a diagram showing a partial physical map of the restriction endonuclease sites of the pagC chromosomal region. The mouse 50% lethal doses (LD.sub.50) for strains with transposon insertions in pagD, envE, msgA, and pagC are shown above each gene. Horizontal arrows demonstrate the direction of transcription. Vertical arrows denote TnphoA insertions and the hollow triangle denotes a MudJ insertion. Below the chromosomal map is a representation of the DNA insert in plasmid pCAA9, which was mutagenized with TnphoA and MudJ. Letter designations: A, AccI; C, ClaI; E, EcoRI; H, HpaI; P, PstI; and V, EcoRV.

FIGS. 8A-8C are a DNA sequence of the region upstream of pagC and the translation of each ORF. The HpaI and ClaI sites at the beginning and end of the region are indicated. Shine-Delgarno regions are underlined and stem loop structures (potential Rho-independent terminators) are indicated with a line below and above the sequence. Arrow heads denote the location of the representative transposon insertion within each gene. Horizontal arrows in the pagD and msgA promoter regions mark the transcriptional start sites, and asterisks mark the -10 and -35 sequences. The consensus lipid attachment site in EnvF is enclosed in brackets. The pagD ORF begins at nucleotide 91 and ends at nucleotide 354 of SEQ ID NO: 5; the envE ORF begins at nucleotide 1114 and ends at nucleotide 1650 of SEQ ID NO: 5; the msgA ORF begins at nucleotid 1825 and ends at nucleotide 2064 of SEQ ID NO: 5; and the envF ORF begins at nucleotide 2554 and ends at nucleotide 3294 of SEQ ID NO: 5.

FIGS. 9A-9E are a DNA sequence containing the prgH, prgI, prgJ, and prgK genes. The start codon (ATG) of each gene is underlined, and the stop codon is indicated with an asterisk. The prgH ORF begins at nucleotide 688 and ends at 1866 of SEQ ID NO: 10; the prgI ORF begins at nucleotide 1891 and ends at nucleotide 2133 of SEQ ID NO: 10; the prgJ ORF begins at nucleotide 2152 and ends at nucleotide 2457 of SEQ ID NO: 10; and the prgK ORF begins at nucleotide 2454 and ends at nucleotide 3212 of SEQ ID NO: 10.

FIG. 10 is a line graph showing the growth rates of the parent Salmonella strain (AroA-) and the vaccine strain (AroA-, PhoP-).

FIG. 11 is a bar graph showing defensin sensitivity of mouse vaccine strains (S. typhimurium).

FIG. 12 is a bar graph showing phoP activation as measured by LacZ activity using the PagB:LacZ recorder fusion construct.

FIG. 13 is a bar graph showing defensin sensitivity of S. typhi vaccine strain TyLH445 compared to the AroA.sup.- parent strain.

FIG. 14A is a graph showing the relative expression of constitutive expression (610 and 617) and phoP regulated (PagC and pagD) expression of AP fusion proteins.

FIG. 14B is a graph showing the immune response to lipopolysaccharide (LPS).

FIG. 14C is a graph showing the immune response to the model heterologous antigen, AP.

FIG. 15 is a DNA sequence containing the pagC-pagD intergenic region. pagC translational start site (ATG on the opposite DNA strand) is underlined (nucleotides 1-3 of SEQ ID NO: 15). The pagC transcriptional start (nucleotide 562) is indicated with an arrow pointing left. The pagD translational start (ATG) is underlined (nucleotides 815-817 of SEQ ID NO: 15). The pagD transcriptional start is indicated with an arrow pointing right (nucleotide 776).

STRAIN DEPOSIT

Under the terms of the Budapest Treaty on the International Recognition of the Deposit of Microorganisms for the Purpose of Patent Procedure, deposit of the following materials has been made with the American Type Culture Collection (ATCC) of Rockville, Md., USA.

Applicant's assignee, Massachusetts General Hospital, represents that the ATCC is a depository affording permanence of the deposit and ready accessibility thereto by the public if a patent is granted. All restrictions on the availability to the public of the material so deposited will be irrevocably removed upon the granting of a patent. The material will be available during the pendency of the patent application to one determined by the Commissioner to be entitled thereto under 37 CFR 1.14 and 35 U.S.C. .sctn.122. The deposited material will be maintained with all the care necessary to keep it viable and uncontaminated for a period of at least five years after the most recent request for the furnishing of a sample of the deposited plasmid, and in any case, for a period of at least thirty (30) years after the date of deposit or for the enforceable life of the patent, whichever period is longer. Applicants' assignee acknowledges its duty to replace the deposit should the depository be unable to furnish a sample when requested due to the condition of the deposit.

PhoP.sup.c strain CS022 (described below) has been deposited with the American Type Culture Collection (Rockville, Md.) and has received ATCC designation 55130.

The plasmid, pIB01, containing the prgH gene has been deposited on Jul. 9, 1993 with the American Type Culture Collection (Rockville, Md.) and has received ATCC designation ATCC 75496.

Constitutive Expression of the PhoP Regulon Attenuates Salmonella Virulence and Survival within Macrophages

The phoP constitutive allele (PhoP.sup.c), pho-24, results in derepression of pag loci. Using diethyl sulfate mutagenesis of S. typhimurium LT-2, Ames and co-workers isolated strain TA2367 pho-24 (all strains, materials, and methods referred to in this section are described below), which contained a phoP locus mutation that resulted in constitutive production of acid phosphatase in rich media, Kier et al., 1979, J. Bacteriol. 138:155, hereby incorporated by reference. This phoP-regulated acid phosphatase is encoded by the phoN gene, a pag locus, Kier et al., 1979, supra, Miller et al., 1989, supra. To analyze whether the pho-24 allele increased the expression of other pag loci the effect of the pho-24 allele on the expression of other pag loci recently identified as transcriptional (e.g., pagA and pagB) and translational (e.g., pagC) fusion proteins that required phoP and phoQ for expression, Miller et al., 1989, supra, was determined. pag gene fusion strains, isogenic except for the pho-24 allele, were constructed and assayed for fusion protein activity. PhoP.sup.c derivatives of the pagA::Mu dJ and pagB::Mu dJ strains produced 480 and 980 U, respectively, of .beta.-galactosidase in rich medium, an increase of 9- to 10-fold over values for the fusion strains with a wild-type phoP locus, see Table 1.

The pagC::TnphoA gene fusion produced 350 U of AP, an increase of three- to fourfold over that produced in strain CS119, which is isogenic except for the pho-24 mutation, Miller et al., 1989, supra. These results compare with a ninefold increase in the acid phosphatase activity in strain CS022 on introduction of the pho-24 allele. Therefore, these available assays for pag gene expression document that the pho-24 mutation causes constitutive expression of pag loci other than phoN.

Identifications of protein species that are repressed as well as activated in the PhoP.sup.c mutant strain

Whole-cell proteins of strain CS022 were analyzed to estimate the number of protein species that could be potentially regulated by the PhoP regulon. Remarkably, analysis by one-dimensional polyacrylamide gel electrophoresis of the proteins produced by strains with the PhoP.sup.c phenotype indicated that some protein species were decreased in expression when many presumptive pag gene products were fully induced by the pho-24 mutation. The proteins decreased in the PhoP.sup.c strain might represent products of genes that are repressed by the PhoP regulator. Genes encoding proteins decreased by the pho-24 allele are designated prg loci, for phoP-repressed genes. Comparison of wild-type, PhoP.sup.-, and PhoP.sup.c mutant strain proteins shows that growth in LB medium at 37.degree. C. represents repressing conditions for pag gene products and derepressing conditions for prg gene products.

To estimate the total number of potentially PhoP-regulated gene products, the total cell proteins of wild-type and PhoP.sup.c mutant strains grown in LB were analyzed by two-dimensional gel electrophoresis. At least 40 species underwent major fluctuation in expression in response to the pho-24 mutation.

Virulence defects of the PhoP.sup.c strain

Remarkably, strains with the single pho-24 mutation were markedly attenuated for virulence in mice (Table 2). The number of PhoP.sup.c organisms (2.times.10.sup.5) that killed 50% of BALB/c mice challenged (LD.sub.50) by the intraperitoneal (i.p.) route was near that (6.times.10.sup.5) of PhoP.sup.- bacteria, Miller et al., 1989, supra. The PhoP.sup.c strains had growth comparable to wild-type organisms in rich and minimal media. The PhoP.sup.c mutants were also tested for alterations in lipopolysaccharide, which could explain the virulence defect observed. Strain CS022 had normal sensitivity to phage P22, normal group B reactivity to antibody to O antigen, and a lipopolysaccharide profile identical to that of the parent strain, as determined by polyacrylamide gel electrophoresis and staining.

Since the TA2367 pho-24 strain was constructed by chemical mutagenesis and could have another linked mutation responsible for its virulence defect revertants of the PhoP.sup.c were isolated to determine whether the pho-24 allele was responsible for the attenuation of virulence observed. Phenotype PhoP.sup.c revertants, identified by the normal levels of acid phosphatase in rich medium, were isolated among the bacteria recovered from the livers of mice infected with strain CS022. Six separate phenotypic revertants, designated CS122 to CS128, were found to be fully virulent (LD.sub.50 of less than 20 organisms for BALB/c mice). The locus responsible for the reversion phenotype was mapped in all six revertants tested for virulence by bacteriophage P22 cotransduction and had linkage characteristics consistent with the phoP locus (greater than 90% linkage to purB). These data indicate that these reversion mutations are not extragenic suppressors but are intragenic suppressors or true revertants of the pho-24 mutation. Thus, the virulence defect of PhoP.sup.c mutants is probably the result of a single revertible mutation in the phoP locus and not the result of a second unrelated mutation acquired during mutagenesis.

Reversion frequency of the PhoP.sup.c phenotype

The reversion frequency of the PhoP.sup.c mutation in vivo in mice was investigated to assess whether reversion could reduce the LD.sub.50 of this strain. The presence of the revertants of strain CS022 was tested for by administering 10.sup.6, 10.sup.4, and 10.sup.2 challenge organisms to each of eight animals by i.p. injection. On day 7, three animals died that received 10.sup.6 PhoP.sup.c organisms. On that day, the livers and spleens of all animals were harvested and homogenized in saline. After appropriate dilution, 10% of the tissue was plated on LB plates containing the chromogenic phosphatase substrate XP. Revertants were identified by their lighter blue colonies compared with PhoP.sup.c bacteria and were confirmed by quantitative acid phosphatase assays. An estimated 10.sup.7, 10.sup.5, and 10.sup.3 organisms per organ were recovered from animals at each of the three respective challenge doses. Revertants were identified only at the highest dose and comprised 0.5 to 1%, or 10.sup.5 organisms per organ, at the time of death. It is likely that revertants are able to compete more effectively for growth in these macrophage-containing organs, since strain CS022 is deficient in survival within macrophages (see below). However, revertants were not identified if fewer than 10.sup.5 organisms were administered in the challenge dose, suggesting that the reversion frequency must be approximately 10.sup.-5. The reversion rate of the PhoP.sup.c phenotype for CS022 bacteria grown in LB is in fact 6.times.10.sup.-4 when scored by the same colony phenotypes. The percentage of revertants recovered from animals near death suggests that pressure is applied in vivo that selects for revertants of the PhoP.sup.c phenotype and implies that the virulence defect observed could be much greater quantitatively for a strain with a nonrevertible PhoP.sup.c mutation.

The PhoP.sup.c strain is deficient in survival within macrophages

Because of the importance of survival within macrophages to Salmonella virulence Fields et al., 1986, Proc. Natl. Acad. Sci. USA 83:5189, hereby incorporated by reference, PhoP.sup.c bacteria were tested for this property. Strain CS022 was defective in the ability to grow and persist in macrophages as compared with wild-type organisms (FIG. 1). In FIG. 1 the survival of strain CS022 (PhoP.sup.c) (triangles) in cultured macrophages is compared with that of wild-type S. typhimurium ATCC 10428 (circles). The experiment shown is a representative one. The difference between the two strains at 4 and 24 hours is significant (P<0.05). PhoP.sup.- bacteria seemed to have a macrophage survival defect qualitatively similar to that of PhoP.sup.c bacteria but survived consistently better by two- to threefold in side-by-side experiments. The increased recovery of organisms that reverted to PhoP.sup.c phenotype in mouse organs rich in macrophage content is consistent with the reduced macrophage survival of PhoP.sup.c mutants in vitro.

Use of the PhoP.sup.c strain as a live vaccine

It has been previously reported that PhoP.sup.- strains are useful as live vaccines in protecting against mouse typhoid, Miller et al., 1989, supra. The immunogenicity of PhoP.sup.c when used as live attenuated vaccines in mice was compared with the of PhoP.sup.-. This was done by simultaneous determination of survival, after graded challenge doses with the wild-type strain ATCC 10428, in mice previously immunized with graded doses of the two live vaccine strains. CS015 phoP::Tn10d-Cam and CS022 pho-24, as well as a saline control. The results obtained (Table 2) suggest the following conclusions: (i) small i.p. doses of the PhoP.sup.c strain (e.g., 15 organisms) effectively protect mice from challenge doses as large as 5.times.10.sup.5 bacteria (a challenge dose that represents greater than 10.sup.4 i.p. LD.sub.50 s), (ii) large doses of PhoP.sup.c organisms given orally completely protect mice from an oral challenge consisting of 5.times.10.sup.7 wild-type bacteria (over 200 oral wild-type LD.sub.50 s) and (iii) by comparison, a large dose of PhoP.sup.- organisms (5.times.10.sup.5) does not provide similar protection. The reversion of the PhoP.sup.c mutation in vivo somewhat complicates the analysis of the use of these strains as vaccines, since revertants of the CS022 strain (i.e., wild-type cells) could increase immunogenicity). However, we were unable to identify revertants by examining 10% of the available spleen and liver tissue from those mice that received 10.sup.4 or fewer organisms.

Strains, Materials and Methods

The strains, materials, and methods used in the PhoP regulon work described above are as follows.

American Type Culture Collection (ATCC) strain 14028, a smooth virulent strain of S. typhimurium, was the parent strain for all virulence studies. Strain TT13208 was a gift from Nang Zhu and John Roth. Strain TA2367 was a generous gift of Gigi Stortz and Bruce Ames, Kier et al., 1979, supra. Bacteriophage P22HT int was used in transductional crosses to construct strains isogenic except for phoP locus mutations, Davis et al., 1980, Advanced Bacterial Genetics, p. 78, 87. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y., hereby incorporated by reference. Luria broth was used as rich medium, and minimal medium was M9, Davis et al., 1980, supra. The chromogenic phosphatase substrate 5-bromo-4-chloro-3indolyl phosphate (XP) was used to qualitatively access acid and AP production in solid media.

Derivatives of S. typhimurium ATCC 10428 with the pho-24 mutation were constructed by use of strain TA2367 as a donor of the purB gene in a P22 transductional cross with strain CS003 .DELTA.phoP .DELTA.purB, Miller et al., 1989, supra. Colonies were then selected for the ability to grow on minimal medium. A transductant designated CS022 (phenotype PhoP.sup.c) that synthesized 1,750 U of acid phosphatase in rich medium (a ninefold increase over the wild-type level in rich medium) was used in further studies.

Derivatives of strains CS022 and CS023 pho-24 phoN2 zxx::6251Tn10d-Cam, and acid phosphatase-negative derivative of CS022, containing pag gene fusions were constructed by bacteriophage P22 transductional crosses, using selection of TnphoA- or Mu dJ-encoded kanamycin resistance. Strains were checked for the intact pag gene fusion by demonstration of appropriate loss of fusion protein activity on introduction of a phoP105::TN10d or phoP102::TN10d-Cam allele.

Assays of acid phosphatase, AP, and .beta.-galactosidase were performed as previously described, Miller et al., 1989, supra and are reported in units as defined in Miller, 1972, Experiments in molecular genetics, p. 352-355, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y., hereby incorporated by reference.

In the mouse virulence and vaccination studies bacteria grown overnight in Luria broth were washed and diluted in normal saline. The wild-type parent strain of CS022 (ATCC 10428) was used for all live vaccine challenge studies. This strain has a 50% lethal dose (LD.sub.50) for naive adult BALB/c mice of less than 20 organisms when administered by intraperitoneal (i.p.) injection and 5.times.10.sup.4 when administered orally in NaHCO.sub.3. Mice were purchased from Charles River Breeding Laboratories, Inc. (Wilmington, Mass.) and were 5 to 6 weeks of age at initial challenge. All i.p. inoculations were performed as previously described, Miller et al., 1989, supra. Oral challenge experiments were performed with bacteria grown in LB broth and concentrated by centrifugation. The bacteria were resuspended in 0.1M NaHCO.sub.3 to neutralize stomach acid, and administered as a 0.5-ml bolus to animals under ether anesthesia. Colony counts were performed to accurately access the number of organisms administered. All challenge experiments were performed 1 month after i.p. inoculation and 6 weeks after oral challenge. Challenge inocula were administered by the same route as vaccinations. The care of all animals was under institutional guidelines as set by the animal are committees at the Massachusetts General Hospital and Harvard Medical School.

Protein electrophoresis was performed as follows. One-dimensional protein gel electrophoresis was performed by the method of Laemmli, 1970, Nature 227:680, hereby incorporated by reference, on whole-cell protein extracts of stationary-phase cells grown overnight in Luria broth. The gels were fixed and stained with Coomassie brilliant blue R250 in 10% acetic acid-10% methanol. Two-dimensional protein gel electrophoresis was performed by method of O'Farrell, 1975, J. Biol. Chem. 250:4007, hereby incorporated by reference, on the same whole-cell extracts. Isoelectric focusing using 1.5% pH 3.5 to 10 ampholines (LKB Instruments, Baltimore, Md.) was carried out for 9,600 V h (700 V for 13 h 45 min). The final tube gel pH gradient extended from pH 4.1 to pH 8.1 as measured by a surface pH electrode (BioRad Laboratories, Richmond, Calif.) and colored acetylated cytochrome pI markers (Calbiochem-Behring, La Jolla, Calif.) run in an adjacent tube. The slab gels were silver stained, Merril et al., 1984, Methods Enzymol. 104:441, hereby incorporated by reference.

In the macrophage survival assays experiments were performed as previously described, Miller et al., 1989, supra, by the method of Buchmeier et al., 1989, Infect. Immun. 57:1, hereby incorporated by reference, as modified from the method of Lissner et al, 1983, J. Immunol. 131:3006, hereby incorporated by reference. Stationary-phase cells were opsonized for 30 min in normal mouse serum before exposure to the cultured bone marrow-derived macrophages harvested from BALB/c mice. One hour after infection, gentamicin sulfate (8 .mu.g/ml) was added to kill extracellular bacteria. All time points were done in triplicate and repeated on three separate occasions.

PhoP.sup.c Mutant Strains Are More Effective as Live Vaccines

PhoP.sup.c mutant S. typhimurium are very effective when used as a live vaccine against mouse typhoid fever and are superior to PhoP.sup.- bacteria. As few a 15 PhoP.sup.c bacteria protect mice against 10.sup.5 LD.sub.50 (lethal doses 50%) of wild type organisms by the intraperitoneal route (Table 3). This suggests that pag gene products are important antigens for protective immunity against mouse typhoid. Preliminary results have documented that antigens recognized by serum of chronic typhoid carriers recognizes some phoP-regulated gene products of S. typhi. If protective antigens are only expressed within the host, then dead vaccines only grown in rich media may not induce an immune response against these proteins.

The use of different S. typhimurium dead vaccine preparations containing different mutations in the phoP regulon was evaluated. As can be seen in Table 3 no dead cell preparations (even those containing mixtures of PhoP.sup.- and PhoP.sup.c bacteria) are as effective vaccines as are live bacteria. This suggests that there are other properties of live vaccines that increase immunogenicity or that important non-PhoP-regulated antigens are not in these preparations. The only protection observed in any animals studied was at the lowest challenge dose for those immunized with PhoP.sup.c bacteria. This further suggests that phoP activated genes are important protective antigens.

aroA PhoP Regulon Double Mutant Strains

Recent efforts by Stocker, Levine, and colleagues have focused on the use of strains with auxotrophic mutations in aromatic amino acid and purine pathways as live vaccines, Hoseith et al., 1981, Nature 291:238, hereby incorporated by reference, Stocker, 1988, Vaccine 6:141, hereby incorporated by reference, and Levine et al., 1987, J. Clin. Invest. 79:888, hereby incorporated by reference. Purine mutations were found to be too attenuating for immunogenicity, likely because purines are not available to the organism within the mammalian host, Sigwart et al., 1989, Infect. Immun. 57:1858, hereby incorporated by reference. Because auxotrophic mutations may be complemented by homologous recombination events with wild type copies donated from environmental organisms or by acquiring the needed metabolite within the host, it would seem prudent for live vaccines to contain a second attenuating mutation in a different virulence mechanism, (i.e., not just a second mutation in the same metabolic pathway). Additionally, in mice the aroA mutants have some residual virulence. Various strains with aroA mutations combined with phoP regulon mutations were investigated for virulence attenuation and immunogenicity. Table 4 demonstrates that a PhoP.sup.- or PhoP.sup.c mutation further attenuates aroA mutant S. typhimurium by at least 100-fold and that, at least at high levels of vaccinating organisms, immunogenicity is retained. Strains with both a pagC.sup.- and phoP.sup.c phenotype are also further attenuated than either mutation alone. Therefore, phoP regulon mutations may increase the safety of aroA live vaccine preparations.

Salmonella typhi phoP Regulon Mutations

The phoP regulon is at least partially conserved in S. typhi DNA hybridization studies as well as P22 bacteriophage transductional crosses have documented that the phoP, phoQ, and pagC genes appear highly conserved between S. typhi and S. typhimurium mutations in these genes in S. typhi have been made.

Salmonella Live Vaccines as Delivery Systems for Heterologous Antigens

The vector used in the vaccine delivery system is a derivative of pJM703.1 described in Miller et al., 1988, J. Bact. 170:2575, hereby incorporated by reference. This vector is an R6K derivative with a deletion in the pir gene. R6K derivatives require the protein product of the pir gene to replicate. E. coli that contain the pir gene present as a lambda bacteriophage prophage can support the replication of this vector. Cells that do not contain the pir gene will not support the replication of the vector as a plasmid. This vector also contains the mob region of RP4 which will allow mobilization into other gram negative bacteria by mating from E. coli strains such as SM10lambda pir, which can provide the mobilization function in trans.

The pagC region is shown in FIGS. 2 and 3. FIG. 2 shows the restriction endonuclease sites of the pagC locus. The heavy bar indicates pagC coding sequence. The TnphoA insertion is indicated by a inverted triangle. The direction of transcription is indicated by the arrow and is left to right. The numbers indicate the location of endonuclease sites, in number of base pairs, relative to the start codon of predicted pagC translation with positive numbers indicating location downstream of the start codon and negative numbers indicating location upstream of the start codon. A is AccI, B is BglI, C is ClaI, D is DraI, E is EcoRI, H is HpaI, N is NruI, P is PstI, S is SspI, T is StuI, U is PvuII, V is EcoRV, and II is Bg1II. FIG. 3 shows the DNA sequence (Sequence I.D. No. 1) and translation of pagC::TnphoA. The heavy underlined sequence indicates a potential ribosomal binding site. The single and double light underlines indicate sequences in which primers were constructed complementary to these nucleotides for primer extension of RNA analysis. The asterisk indicates the approximate start of transcription. The arrow indicates the direction of transcription. The boxed sequences indicate a region that may function in polymerase binding and recognition. The inverted triangle is the site of the sequenced TnphoA insertion junction. The arrow indicates a potential site for single sequence cleavage.

3 kilobases of DNA containing the pagC gene (from the PstI restriction endonuclease site 1500 nucleotides 5' to the start of pagC translation to the EcoRI restriction endonuclease site 1585 nucleotides downstream of pagC translation termination) were inserted into the pJM703.1 derivative discussed above. The pagC sequence from the ClaI restriction endonuclease site was deleted (490 nucleotides) and replaced with a synthetic oligonucleotide polylinker that creates unique restriction endonuclease sites. DNA encoding one or more heterologous proteins, e.g., an antigen, can be inserted into this site. This creates a vector which allows the insertion of multiple foreign genes into the DNA surrounding pagC.

The vector can be mobilized into Salmonella by mating or any other delivery system, e.g., heat shock, bacteriophage transduction or electroporation. Since it can not replicate, the vector can only insert into Salmonella by site specific recombination with the homologous DNA on both sides of the pagC gene. This will disrupt and inactivate the native pagC locus and replace it with the disrupted pagC DNA carried on the vector.

Such recombination events can be identified by marker exchange and selective media if the foreign DNA inserted into the pagC locus confers a growth advantage. The insertion of antibiotic resistance genes for selection is less desirable as this could allow an increase in antibiotic resistance in the natural population of bacteria. Genes which confer resistance to substances other than antibiotics e.g., to heavy metals or arsenic (for mercury resistance, see Nucifora et al., 1989, J. Bact., 171:4241-4247, hereby incorporated by reference), can be used to identify transformants. Alternatively, selection can be performed using a Salmonella recipient strain that carries an auxotrophic mutation in a metabolic pathway and a vector that carries DNA that complements the auxotrophic mutation. Many Salmonella live vaccine prototypes contain mutations in histidine or purine pathways thus complementation of these metabolic auxotrophies can be used to select for integrants. (Purine mutations specifically have been shown to be too attenuated for use in man.) Further proof of marker exchange can be documented by loss of the ampicillin resistance (carried on the plasmid backbone) or by blot hybridization analysis.

A gene useful for selection can be cloned by complementation of a vaccine strain with a metabolic auxotrophy. Specific examples include the cloning of the DNA encoding both purB and phoP by complementation of a strain deleted for function of both these genes. Salmonella gene libraries have been constructed in a pLAFR cosmid vector (Frindberg et al., 1984, Anal. Biochem. 137:266-267, hereby incorporated by reference) by methods known to those skilled in the art. pLAFR cosmids are broad host range plasmids which can be mobilized into Salmonella from E. coli. An entire bank of such strains can be mobilized into Salmonella vaccine strains and selected for complementation of an auxotrophic defect (e.g., in the case of purB growth on media without adenine). The DNA able to complement this defect is then identified and can be cloned into the antigen delivery vector.

As discussed above heterologous genes can be inserted into the polylinker that is inserted into the pagC sequence of the vector. The heterologous genes can be under the control of any of numerous environmentally regulated promotor systems which can be expressed in the host and shut off in the laboratory. Because the expression of foreign proteins, especially membrane proteins (as are most important antigens), is frequently toxic to the bacterium, the use of environmentally regulated promoters that would be expressed in mammalian tissues at high levels but which could be grown in the laboratory without expression of heterologous antigens would be very desirable. Additionally, high expression of antigens in host tissues may result in increased attenuation of the organism by diverting the metabolic fuel of the organism to the synthesis of heterologous proteins. If foreign antigens are specifically expressed in host phagocytic cells this may increase the immune response to these proteins as these are the cells responsible for processing antigens.

The promoter systems likely to be useful include those nutritionally regulated promoter systems for which it has been demonstrated that a specific nutrient is not available to bacteria in mammalian hosts. Purines, Sigwart et al., 1989, Infect. Immun., 57:1858 and iron, Finklestein et al., 1983, Rev. Infect. Dis. 5:S759, e.g., are not available within the host. Promoters that are iron regulated, such as the aerobactin gene promoter, as well as promoters for biosynthetic genes in purine pathways, are thus excellent candidates for testing as promoters that can be shut down by growth in high concentrations of these nutrients. Other useful environmentally regulated Salmonella promoters include promoters for genes which encode proteins which are specifically expressed within macrophages, e.g., the DnaK and GroEL proteins, which are increased by growth at high temperature, as well as some phoP activated gene products, Buchmeier et al., 1990, Science 248:730, hereby incorporated by reference. Therefore, promoters such as the pagC 5' controlling sequences and the better characterized promoters for heat shock genes, e.g., GroEL and DnaK, will be expected to be activated specifically within the macrophage. The macrophage is the site of antigen processing and the expression of heat shock genes in macrophages and the wide conservation of heat shock genes in nature may explain the immunodominance of these proteins. A consensus heat shock promoter sequence is known and can be used in the vectors (Cowling et al., 1985, Proc. Natl. Acad. Sci. USA 82:2679, hereby incorporated by reference).

The vectors can include an environmentally regulated T7 polymerase amplification system to express heterologous proteins. For example, the T7 polymerase gene (cloned by Stan Tabor and Charles Richardson, See Current Protocols in Molecular Biology ed. Ausubel et al., 1989, (page 3.5.1.2) John Wiley and Sons, hereby incorporated by reference) under control of an iron regulated promoter, can be included on the vectors described above. We have inserted the aerobactin gene promoter of E. coli with the sequence CATTTCTCATTGATAATGAGAATCATTATTGACATAATTGTTATTATTTTACG (SEQ ID NO: 2), Delorenzo et al. J. Bact. 169:2624, hereby incorporated by reference, in front of the T7 polymerase gene and demonstrated iron regulation of the gene product. This version of the vector will also include one or more heterologous antigens under the control of T7 polymerase promoters. It is well known that RNA can be synthesized from synthetic oligonucleotide T7 promoters and purified T7 in vitro. When the organism encounters low iron T7 polymerase will be synthesized and high expression of genes with T7 promoters will be facilitated.

pagC-fusion proteins in S. typhimurium

Expression of heterologous antigens within macrophages under the control of phoP regulated promotors can be used as an effective method of both attenuating Salmonellae and enhancing immunogenicity of foreign antigens. As discussed above, the expression of PagC is induced in antigen processing cell, i.e., a macrophage. Thus, expression of a heterologous antigen under the control of the pagC promoter is also likely to be inducible in macrophages.

To evaluate the immune response to a heterologous antigen expressed under the control of inducible pag promoters, mice were inoculated with bacteria which expressed the antigen, AP, under the control of the pagC or pagD regulatory sequences. Pag-AP fusion proteins were produced in these bacteria from a single chromosomal copy of the gene encoding AP. The bacteria were generated utilizing two methods: TnphoA mutagenesis, and genetic engineering techniques using a suicide vector, both of which are described above.

As a control, mice were innoculated with bacteria which expressed an AP fusion protein under the control of constitutive promoters. The constitutive promoter was completely independent of regulation by genes in the PhoP regulon. Two such strains of bacteria, Strain 610 and Strain 617, were constructed using methods described above. AP expression in Strain 610 was moderate, whereas AP expression in Strain 617 was high (see FIG. 14C).

These strains were injected intraperitoneally into BABL/C mice. Serum samples were taken three weeks after inoculation. Normal mouse serum (MNS) was used as a control. Standard ELISA assays were used to test the sera for the presence of AP-specific antibodies. Sera was also tested for LPS-specific antibodies using S. typhimurium LPS. Antibodies directed to LPS were detected in all the murine sera tested, but only those strains in which AP was expressed as a Pag fusion protein from a single chromosomal gene copy engendered an immune response against the model heterologous antigen, AP (see FIGS. 14A and FIG. 14B).

Despite approximately 10-fold higher constitutive expression of the AP fusion in strain 617, only a minimal immune response to this antigen was noted after immunization with strain 617. In contrast, a strong response was observed in mice inoculated with strains which expressed the Pag-AP fusion protein. These data indicate that phoP-regulation which results in in vivo induction of protein expression within macrophages increases the immunogenicity of heterologous antigens expressed under the control of the pag promoters. Any promoter which directs cell-specific, inducible expression of a protein in macrophages or other antigen presenting cells, e.g., pag described herein, can be used to increase the immunogencity of an antigen expressed in Salmonella.

The pagC gene and pagC Gene Product Strains, materials, and methods

The following strains, materials, and methods were used in the cloning of pagC and in the analysis of the gene and its gene product.

Rich media was Luria broth (LB) and minimal media was M9, Davis et al., 1980, supra. The construction of S. typhimurium strain CS119 pagC1::TnphoA phoN2 zxx::6251 TN10d-Cam was previously described, Miller et al., 1989, supra. American Type Culture Collection (ATCC) S. typhimurium strain 10428 included CS018 which is isogenic to CS119 except for phoP105::TN10d, Miller et al., 1989, supra, CS022 pho-24, Miller et al., 1990, J. Bacteriol. 172:2485-2490, hereby incorporated by reference, and CS015 phoP102::Tn10d-cam, Miller et al., 1989, supra. Other wild type strains used for preparation of chromosomal DNA included S. typhimurium LT2 (ATCC 15277), S. typhimurium Q1 and S. drypool (Dr. J. Peterson U. Texas Medical Branch, Galveston), and Salmonella typhi Ty2 (Dr. Caroline Hardegree, Food and Drug Administration). pLAFR cosmids were mobilized from E. coli to S. typhimurium using the E. coli strain MM294 containing pRK2013, Friedman et al., 1982, Gene 18:289-296, hereby incorporated by reference. AP activity was screened on solid media using the chromogenic phosphatase substrate 5-bromo-4-chloro-3-indolyl phosphate (XP). AP assays were performed as previously described, Brickman et al., 1975, J. Mol. Biol. 96:307-316, hereby incorporated by reference, and are reported in units as defined by Miller, Miller, 1972, supra, pp. 352-355.

One dimensional protein gel electrophoresis was performed by the method of Laemmli, 1970, Nature, 227:680-685, hereby incorporated by reference, and blot hybridization using antibody to AP was performed as previously described, Peterson et al., 1988, Infect. Immun. 56:2822-2829, hereby incorporated by reference. Whole cell protein extracts were prepared, from saturated cultures grown in LB at 37.degree. C. with aeration, by boiling the cells in SDS-pagE sample buffer, Laemmli, 1970, supra. Two dimensional gel electrophoresis was performed by the method of O'Farrell, 1975, J. Biol. Chem. 250:4007, hereby incorporated by reference. Proteins in the 10% polyacrylamide slab gels were visualized by silver staining, Merril et al., 1984, Methods in Enzymology, 104:441, hereby incorporated by reference.

Chromosomal DNA was prepared by the method of Mekalanos, 1983, Cell, 35:253-263, hereby incorporated by reference. DNA, size fractionated in agarose gels, was transferred to nitrocellulose (for blot hybridization) by the method of Southern, 1975, J. Mol. Biol. 98:503-517, hereby incorporated by reference. DNA probes for Southern hybridization analysis were radiolabeled by the random primer method, Frinberg et al., 1984, supra. Plasmid DNA was transformed into E. coli and Salmonella by calcium chloride and heart shock, Mekalanos, 1983, supra, or by electroporation using a Genepulser apparatus (BioRad, Richmond, Calif.) as recommended by the manufacturer, Dower et al., 1988, Nucl. Acids Res. 16:6127-6145, hereby incorporated by reference. DNA sequencing was performed by the dideoxy chain termination method of Sanger et al., 1977, Proc. Natl. Acad. Sci. USA, 74:5463-5467, hereby incorporated by reference, as modified for use with SEQUENASE.RTM. (U.S. Biochemical, Cleveland, Ohio). Oligonucleotides were synthesized on an Applied Biosystems Machine and used as primers for sequencing reactions and primer extension of RNA. Specific primers unique to the two ends of TnphoA, one of which corresponds to the AP coding sequence and the other to the right IS50 sequence, were used to sequence the junctions of the transposon insertion.

Construction of a S. typhimurium cosmid gene bank in pLAFR3 and screening for clones containing the wild type pagC DNA was performed as follows. DNA from S. typhimurium strain ATCC 10428 was partially digested using the restriction endonuclease Sau3A and then size selected on 10-40% sucrose density gradient. T4 DNA ligase was used to ligate chromosomal DNA of size 20-30 kilobases into the cosmid vector pLAFR3, a derivative of pLAFR1, Friedman et al., 1982, Gene 18:289-296, hereby incorporated by reference, that was digested with the restriction endonuclease BamHI. Cosmid DNA was packaged and transfected into E. coli strain DH5-.alpha. using extracts purchased from Stratagene, La Jolla, Calif. Colonies were screened by blot hybridization analysis.

The analysis of proteins produced from cloned DNA by in vitro transcription/translation assays was analyzed as follows. These assays were performed with cell free extracts, (Amersham, Arlington Heights, Ill.), and were performed using conditions as described by the manufacturer. The resultant radiolabeled proteins were analyzed by SDS-pagE.

RNA was purified from early log and stationary phase Salmonella cultures by the hot phenol method, Case et al., 1988, Gene 72:219-236, hereby incorporated by reference, and run in agarose-formaldehyde gels for blot hybridization analysis, Thomas, 1980, Proc. Natl. Acad. Sci. USA 77:5201, hereby incorporated by reference. Primer extension analysis of RNA was performed as previously described, Miller et al., 1986, Nuc. Acids. Res. 14:7341-7360, hereby incorporated by reference, using AMV reverse transcriptase (Promega, Madison, Wis.) and synthesized oligonucleotide primers complementary to nucleotides 335-350 and 550-565 of the pagC locus.

Identification of an 18 kDa protein missing in a pagC mutant of S. typhimurium

pagC mutant strain CS119 was analyzed by two dimensional protein electrophoresis to detect protein species that might be absent as a result of the TnphoA insertion. Only a single missing protein species, of approximately 18 kD and pI-8.0, was observed when strains, isogenic except for their transposon insertions, were subjected to this analysis. This 18 kDa species was also missing in similar analysis of Salmonella strains with mutations phoP and phoQ. Though two-dimensional protein gel analysis might not detect subtle changes of protein expression in strain CS119, this suggested that a single major protein species was absent as a result of the pagC::TnphoA insertion.

Additional examination of the 2-dimensional gel analysis revealed a new protein species of about 45 kDa that is likely the pagC-Ap fusion protein. The pagC-AP fusion protein was also analyzed by Western blot analysis using antisera to AP and found to be similar in size to native AP (45 kDa) and not expressed in PhoP-S. typhimurium.

Cloning of the pagC::TnphoA insertion

Chromosomal DNA was prepared from S. typhimurium strain CS119 and a rough physical map of the restriction endonuclease sites in the region of the pagC::TnphoA fusion was determined by using a DNA fragment of TnphoA as a probe in blot hybridization analysis. This work indicated that digestion with the restriction endonuclease ecoRV yielded a single DNA fragment that included the pagC::TnphoA insertion in addition to several kilobases of flanking DNA. Chromosomal DNA from strain CS119 was digested with EcoRV (blunt end) and ligated into the bacterial plasmid vector pUC19 (New England Biolabs) that had been digested with the restriction endonuclease SmaI (blunt end). This DNA was electroporated into the E. coli strain Dh5-.alpha. (BRL) and colonies were plated onto LB agar containing the antibiotics kanamycin (TnphoA encoded and ampicillin (pUC19 encoded). A single ampicillin and kanamycin resistant clone containing a plasmid designated pSM100 was selected for further study.

A radiolabeled DNA probe from pSM100 was constructed and used in Southern hybridization analysis of strain CS119 and its wild type parent ATCC 10428 to prove that the pagC::TnphoA fusion had been cloned. The probe contained sequences immediately adjacent to the transposon at the opposite end of the AP gene HpaI endonuclease generated DNA fragment that included 186 bases of the right IS50 of the transposon and 1278 bases of Salmonella DNA (FIG. 2). As expected, the pSM100 derived probe hybridized to an 11-12 kb AccI endonuclease digested DNA fragment from the strain containing the transposon insertion, CS119. This was approximately 7.7 kb (size of TnphoA) larger than the 3.9 kB AccI fragment present in the wild type strain that hybridizes to the probe. In addition, a derivative of plasmid pSM100, pSM101 (which did not allow expression of the pagC-PhoA gene fusion off the lac promoter), was transformed into phoP- (strain Cs015) and phoN- (strain CS019) Salmonella strains and the cloned AP activity was found to be dependent on phoP for expression. Therefore we concluded that the cloned DNA contained the pagC::TnphoA fusion.

The presence of the pagC gene was also demonstrated in other strains of S. typhimurium, as well as in S. typhi, and S. drypool. All Salmonella strains examined demonstrated similar strong hybridization to an 8.0 kb EcoRV and a 3.9 kb AccI restriction endonuclease fragment suggesting that pagC is a virulence gene common to Salmonella species.

The pagC gene probe from nucleotides -46 (with 1 as the first base of the methionine to 802 (PstI site to the BglII site) failed to cross hybridize to DNA from Citrobacter freundii, Shigella flexneri, Shigella sonnei, Shigella dysenterial, Escherichia coli, Vibrio cholerae, Vibrio vulnificus, Yersenia entero colitica, and Klebsiella pneumonia.

Cloning of the wild type pagC locus DNA and its complementation of the virulence defect of a S. typhimurium pagC mutant

The same restriction endonuclease fragment described above was used to screen a cosmid gene bank of wild type strain ATCC 10428. A single clone, designated pWP061, contained 18 kilobases of S. typhimurium DNA and hybridized strongly to the pagC DNA probe. pWP061 was found to contain Salmonella DNA identical to that of pSM100 when analyzed by restriction endonuclease analysis and DNA blot hybridization studies. Probes derived from pWP061 were also used in blot hybridization analysis with DNA from wild type and CS119 S. typhimurium. Identical hybridization patterns were observed to those seen with pSM100. pWP061 was also mobilized into strain CS119, a pagC mutant strain. The resulting strain had wild type virulence for BALB/c mice (a LD.sub.50 less than 20 organisms when administered by IP injection). Therefore the cloned DNA complements the virulence defect of a pagC mutant strain.

Since, a wild type cosmid containing pagC locus DNA was found to complement the virulence defect of a pagC mutant S. typhimurium strain, it was concluded that the pagC protein is an 188 amino acid (18 kDa) membrane (see below) protein essential for survival within macrophages and virulence of S. typhimurium.

Physical mapping of restriction endonuclease sites, DNA sequencing, and determination of the pagC gene product

Restriction endonuclease analysis of plasmid pSM100 and pWP061 was performed to obtain a physical map of the pagC locus, and, in the case of PSM100, to determine the direction of transcription (FIG. 2). DNA subclones were generated and the TnphoA fusion junctions were sequenced, as well as the Salmonella DNA extending from the HpaI site, 828 nucleotides 5' to the phoA fusion junction, to the EcoRI site 1032 nucleotides 3' to the TnphoA insertion (FIG. 2 and 3). The correct reading frame of the DNA sequence was deduced from that required to synthesize an active AP gene fusion. The deduced amino acid sequence of this open reading frame was predicted to encode a 188 amino acid protein with a predicted pI+8.2. This data were consistent with the 2-D polyacrylamide gel analysis of strain CS119 in which an 18 kDa protein of approximate pI+8.0 was absent. No other open reading frames, predicted to encode peptides larger than 30 amino acids, were found.

The deduced amino acid sequence of the 188 amino acid open reading frame contains a methionine start codon 33 amino acids from the fusion of pagC and AP (FIG. 3). This 33 amino acid pagC contribution to the fusion protein was consistent with the size observed in Western blot analysis and contains a hydrophobic N-terminal region, identified by the method of Kyle et al., 1982, J. Mol. Biol. 157:105-132, hereby incorporated by reference, that is a typical bacterial signal sequence, Yon Heinje, 1985, J. Mol. Biol. 184:99-105, hereby incorporated by reference. Specifically, amino acid 2 is a positively charged lysine, followed by a hydrophobic domain and amino acid 24 is a negatively charged aspartate residue. A consensus cleavage site for this leader peptide is predicted to be at an alanine residue at amino acid 23, Von Heinje, 1984, J. Mol. Biol. 173:243-251, hereby incorporated by reference. The DNA sequence also revealed a typical ribosomal binding site, Shine et al., 1974, Proc. Natl. Acad. Sci. USA 71:1342-1346, hereby incorporated by reference, at 6-2 nucleotides 5' to the predicted start of translation (FIG. 3) nucleotides 717-723). This suggested that the open reading frame was, in fact, translated and further supported the assumption that this was the deduced amino acid sequence of the pagC protein interrupted by the TnphoA insertion (FIG. 3).

In vitro synthesis of proteins by the cloned pagC locus

To detect if other proteins were encoded by pagC and to determine the approximate size of the pagC gene product, an in vitro coupled transcription/translation analysis was performed. A 5.3 kilobase EcoRI fragment of pWP061 was inserted into pUC19 so that the pagC gene would not be expressed off the lac promotor. This plasmid was used in an in vitro coupled transcription-translation assay. A single protein of approximately 22 kilodaltons was synthesized by the cell free system. The size was compatible with this being the precursor of the pagC protein containing its leader peptide. These data further support the conclusion the single and the single pagC gene product had been identified.

Identification of the pagC encoded RNA

An approximately 1100 nucleotide RNA is encoded by pagC. The pagC gene is highly expressed by cells with a phoP constitutive phenotype of pag activation, as compared to wild type and phoP constitutive phenotype of pag activation, as compared to wild type and phoP.sup.- bacteria. In these blot hybridization experiments pagC is only detected in wild type cells grown in rich media during stationary growth. This result, coupled with previous work, Miller et al., 1989, supra, Miller et al., 1990, supra, demonstrates that pagC is transcriptionally regulated by the phoP gene products and is only expressed during early logarithmic phase growth in rich media by cells with a phoP constitutive phenotype.

The size of the pagC transcript is approximately 500 nucleotides greater than that necessary to encode the 188 amino acid protein. Primer extension analysis of Salmonella RNA using oligonucleotide primers specific for pagC sequence was performed to determine the approximate start site of transcription and to determine whether these nucleotides might be transcribed 5' or 3' to the 188 amino acid pagC gene product. Primer extension analysis with an oligonucleotide predicted to be complementary to nucleotides 550-565 of pagC, 150 nucleotides 5' to the predicted start codon, resulted in an approximately 300 nucleotide primer extension product. Therefore a primer further upstream was constructed complementary to nucleotides 335-350 of pagC and used in a similar analysis. A primer extension product of 180 nucleotides was observed to be primer specific. This is consistent with transcription starting at nucleotide 170 (FIG. 3). Upstream of the predicted transcriptional start, at nucleotides 153-160, a classic RNA polymerase binding site was observed with the sequence TATAAT at -12 nucleotides as well as the sequence TAATAT at -10 nucleotides. No complete matches were observed for the consensus RNA polymerase recognition site (TTGACA) 15-21 nucleotides upstream from the -10 region. AT -39 (126-131) nucleotides (TTGGAA), -38 (127-132) nucleotides (TTGTGG), and -25 (135-140) nucleotides (TTGATT) are sequences that have matches with the most frequently conserved nucleotides of this sequence.

Based on the above results transcription was predicted to terminate near the translational stop codon of the 188 amino acid protein (nucleotide 1295, FIG. 3). Indeed, a stem loop configuration was found at nucleotides 1309-1330 that may function as a transcription terminator. This was consistent with the lack of evidence of open reading frames downstream of the 188 amino acid protein and the lack of synthesis of other transcription/translation using the cloned pagC DNA. This further suggests that the pagC::TnphoA insertion inactivated the synthesis of only a single protein.

Similarity of pagC to Ail and Lom

A computer analysis of protein similarity using the National Biomedical Research Foundation/Protein Identification Resource, George et al., 1986, Nucleic Acids Res. 14:11-15, hereby incorporated by reference, protein sequence base was conducted to identify other proteins that had similarity to pagC in an attempt to find clues to the molecular function of this protein. Remarkably, pagC was found to be similar to a bacteriophage lambda protein, Lom, that has been localized to the outer membrane in minicell analysis, Court et al., 1983, Lambda II, Hendrix, R. W. et al. ed. Cold Spring Harbor Laboratory (Cold Spring Harbor N.Y.), pp. 251-277, hereby incorporated by reference, and demonstrated to be expressed by lambda lysogens of E. coli, Barondess, et al., 1990, Nature 346:871-874, hereby incorporated by reference. Recently, the deduced amino acid sequence of the cloned ail gene product of Y. enterocolitica was determined and found to also be similar to Lom, Miller et al., 1990b, J. Bacteriol. 172:1062-1069. Therefore, a protein family sequence alignment was performed using a computer algorithm that establishes protein sequence families and consensus sequences, Smith et al., 1990, Proc. Natl. Acad. Sci. 87:118-122, hereby incorporated by reference. The formation of this family is indicated by the internal data base values of similarity between these proteins: pagC and Lom (107.8), pagC and Ail (104.7), and Ail and Lom (89.8). These same proteins were searched against 314 control sequences in the data base and mean values and ranges were 39.3 (7.3-52.9) pagC, 37.4 (7.3-52.9) Ail, and 42.1 (7.0-61.9) Lom. The similarity values for this protein family are all greater than 3.5 standard deviations above the highest score obtained for similarity to the 314 random sequences. No other similarities or other family members were found in the database. Regions of similarity are located not only in the leader peptide transmembrane domains but throughout the protein.

pag Mutant Strains Are Attenuated For Virulence

Salmonella typhimurium strains of the invention with a pagC mutation were attenuated for virulence by least 1,000-fold.

In addition pagC, other pag genes described herein may be useful in the development of live Salmonella vaccines. Mutations in phoP-activated genes could be used to construct attenuated, live Salmonella vaccines. In constructing multivalent Salmonella vectored vaccines, PhoP-activated promoters could increase immunogenecity by targeting foreign protein expression to antigen presenting macrophages.

Identification of novel phoP-activated genes

To further analyze the role of phoP-activated genes in bacterial virulence, a bank of strains with active phoA gene fusions was generated by TnphoA mutagenesis. Strain CS019 was the parent strain for TnphoA mutagenesis because it has wild-type bacterial virulence and carries the phoN2 allele, which results in minimal background phosphatase activity. Strains with active phoA gene fusions were identified by blue colony phenotype after growth in agar containing XP. Such strains were then screened for decreased fusion protein activity on acquisition of the phoP12 allelle that results in a PhoP-null phenotype.

Two thousand and sixty-four AP expressing strains were isolated and colony purified from two hundred and forty independent matings. Strains with AP activity were isolated at a frequency of 0.8% from the total pool of kanamycin resistant (TnphoA containing) bacteria. A total of fifty-four candidate pag::TnphoA insertions were isolated from the AP expressing strain bank, and forty-nine of these were determined to have greater than six-fold reduction in AP activity in the absence of functional phoP/phoQ. Therefore, approximately 2% of the colonies expressing AP were identified as pag-phoA gene fusions.

Identification of thirteen unique pag loci

Three methods were used to determine whether the forty-nine TnphoA insertions defined unique pag loci. First, physical maps of the EcoRI and HindIII restriction endonuclease sites 5' to the TnphoA insertions were defined. Second, linkage analysis to transposon insertions highly linked to known pag loci was performed. Third, strains determined to be unique by the above methods were screened for linkage to a bank of strains with transposon insertions of known chromosomal location.

Blot hybridization analysis demonstrated that thirteen of the forty-nine strains had unique restriction endonuclease sites 5' to the TnphoA insertion. The numbers of strains with similar physical maps 5' to the TnphoA insertion ranged from 1-7. One of the thirteen physical maps was similar to that expected for an insertion in pagC and was noted in seven of the strains isolated as containing candidate pag::TnphoA insertions. Analysis of these seven strains indicated that only three of these were pagC::TnphoA insertions, since blot hybridization analysis with a fragment of pagC as a probe and linkage analysis to transposon insertions highly linked to pagC indicated that four of these insertions were not in pagC. Another of the pag::phoA fusions, denoted pagP, had the physical 5' restriction-endonuclease map that would be expected for phoN. However, this insertion was determined not to be within phoN by linkage analysis and blot hybridization. A

transductional cross was performed between wild type bacteria and strain CS1247 containing pagP::TnphoA and zxx::6215TN10d-cam. These transductants were selected on kanamycin, insuring the inheritance of the pagP::TnphoA which encodes kanamycin resistance. These colonies were then screened for choramphenicol resistance which would indicate linkage of zxx:6215TN10d-cam to pagP. No linkage was found indicating that pagP was not linked to phoN. Blot hybridization using a portion of phoN as a probe was also performed on CS1247 and indicated that this strain contained a wild type phoN locus. Thirteen pag loci were defined and designated pagD-P.

To further define the PhoP regulation of the 13 pag::TnphoA fusion proteins, AP activity was assayed in strains isogenic except for the phoP locus. AP activity was assayed during bacterial growth in rich medium in logarithmic and stationary growth phase (Table 13). The dependence of an intact phoP locus for full expression remained constant for the different stages of growth; however, the relative amount of AP expression increased as growth was limited. The difference in expression of pag gene fusions varied from six to forty-eight fold when isogenic strains with a wild type and null phoP locus were compared.

Of the five previously identified pag loci, only phoN, pagC, and pagA have known chromosomal locations. Linkage analysis of the 13 newly identified pag loci was performed using strains containing transposon insertions linked to pagC (AK3233 and AK3140), and to pagA (AK3255). Three pag::TnphoA insertions were found to be linked to AK3140 which is in a region near pagC at 24-25 minutes on the chromosome. These were designated pagD, pagE, and pagF. PagD::TnphoA was similarly linked to the transposon insertion of AK3233 (83%) and AK3140 (33%) as was previously reported for pagC. The TnphoA insertion of this strain has been further defined and is divergently transcribed from pagC. page and pagF exhibited different linkage to the insertions of AK3233 and AK3140 than pagC and pagD suggesting a significantly different chromosomal location. The pagE::TnphoA insertion is 39% linked to the transposon insertion of AK3233 and 99.1% linked to that of AK3140, while pagF::TnphoA is 31% linked to the insertion of AK3140 but not to that of AK3233. These different linkages in addition to the physical maps of the restriction endonuclease sites 5' to the TnphoA insertion indicated that these were new pag loci. Therefore, three new pag loci were found in the region of 25 minutes, one of which is highly linked to the previously defined pagC.

Linkage analysis was then performed using a group of defined random Tn10.DELTA.16 .DELTA.17 insertions on the ten strains with TnphoA insertions of no known location. Of these ten pag::TnphoA alleles only two demonstrated linkage to the bank of Tn10.DELTA.16.DELTA.17 insertions. The pagG::TnphoA insertion was demonstrated to have 97% linkage to the transposon insertion of AK3258 located at approximately 30 minutes. The pag::TnphoA insertion, designated pagH, exhibited 23% linkage to the insertion of AK3091. The linkage to the transposon insertion of AK3091 was similar to linkage previously demonstrated for prgE (26%). Therefore, this chromosomal region contains both PhoP-activated and repressed genes. This Tn.DELTA.16.DELTA.17 insertion was analyzed using pulse field gradient electrophoresis of chromosomal DNA from AK3091 digested with the restriction endonuclease XbaI and BlnI. These data indicate that the transposon insertion of AK3091 was located in the region of 20-25 minutes and that pagH and prgE are located in this region of the chromosome.

Strains with pag:TnphoA insertions have wild type sensitivity to the rabbit NP-1 defensin

S. Typhimurium strains with null mutations in the phoP operon have increased sensitivity to a variety of cationic antimicrobial peptides including defensins, magainins, and protamine. The defensins are a family of mammalian peptides present in the granules of neutrophils, lung macrophages, and intestinal Paneth cells. Resistance to these peptides may contribute to bacterial virulence and the ability to colonize mucosal surfaces. Strains with pag::TnphoA insertions were tested for sensitivity to the highly active rabbit defensin NP-1. None of the strains with single pag::TnphoA insertions demonstrated increased sensitivity to NP-1 defensin (see FIG. 6). Thus despite the demonstrated sensitivity of PhoP-null mutants to rabbit defensin NP-1, no defined mutations in pag loci were associated with sensitivity to defensins.

Four strains with pag::TnphoA insertions demonstrate marked attenuation for mouse virulence

To further define whether these new pag loci contributed to mouse virulence, the 13 strains with pag transposon insertions were screened in vivo. Mice were injected intraperitoneally with approximately 100 organisms. Four strains with transposon insertions in pagD, pagJ, pagK, and pagM demonstrated attenuated virulence. Mice injected with these strains all survived and showed no signs of systemic infections, such as hepatosplenomegaly and scruffiness (piloerection due to fever). These four strains were subjected to further virulence testing by intraperitoneal injection of multiple doses of organisms in a total of ten mice on two separate occasions. The mean LD.sub.50 was determined from these subsequent injections and is listed in Table 14. One of these strains, containing the pagD::TnphoA insertion, has a LD.sub.50 10,000 fold greater than wild-type S. typhimurium. The other three strains were also markedly attenuated for mouse virulence with LD.sub.50 values greater than 1000-10,000 times that of wild type organisms. These data indicated that the PhoP-regulated loci, pagD, pagJ, pagK, and pagM, when mutated, result in attenuation of bacterial virulence.

pag::TnphoA strains attenuated for mouse virulence have reduced survival within macrophages

Since PhoP mutant Salmonella are deficient in survival within macrophages, strains containing mutations in pag genes that had attenuated mouse virulence were tested for reduced viability within macrophages. As shown Table 14, all strains with pag mutations demonstrated significantly reduced survival within macrophages. Decreased intracellular survival of pag mutants was not observed until a time when pag are predicted to be maximally expressed.

Four strains with mutations in the pagC, pagD, pagJ, pagK and pagM loci were found to be attenuated for mouse virulence and survival within macrophages. Strains with mutations in these five pag all had varying degrees of virulence attenuation. Strains with a mutation in pagJ had a virulence defect comparable to that observed for pagC mutants (greater than 1000.times.the LD.sub.50 of wild type organisms). The pagD::TnphoA insertion resulted in the greatest attenuation of virulence, comparable to that of a PhoP null mutation (greater than 10,000.times.the LD.sub.50 of wild type organisms). pagK and pagM mutants had virulence attenuation that was intermediate between the pagJ and pagD mutants. The cumulative effect of deletion of pagC, pagD, pagJ, pagK, and pagM, if additive and similar to the

attenuation observed with TnphoA insertions, may be much greater than that observed by deletion of phoP alone. The observation that many of these genes are somewhat expressed in stationary phase in the absence of PhoP suggests that functional Pag proteins could be produced in vivo in the absence of PhoP. One virulence gene pagM is significantly expressed in the absence of PhoP, though it may still require PhoP/PhoQ for induction within macrophage phagosomes. This data suggests that deletion of pag gene products could lead to greater virulence attenuation than deletion of the regulatory proteins.

Salmonella envelope proteins as virulence factors: Defensin sensitivity

Based on the methods used to identify pag loci, i.e., translational gene fusions to phoA, and the observation that the pagC gene fusions produce AP, it has now been discovered that many pag encode bacterial envelope proteins. No strains have been found with single pag mutations that confer sensitivity to defensins or other cationic peptides. The data suggest that an alteration of the bacterial envelope as a result of the change in synthesis of the entire aggregate of envelope proteins mediated by PhoP/PhoQ may be important to S. typhimurium virulence.

Defensins are small amphipathic cationic peptides of approximately 30-35 amino acids in length whose anti-microbial action involves penetration and disruption of membranes, possibly by forming selective anionic channels. Though defensins are largely found in neutrophils and Paneth cells these or other related molecules likely contribute to non-oxidative killing of phagocytosed bacteria by macrophages. Though it remains possible that a single unidentified pag encodes a protein responsible for defensin resistance, it seems more likely that a cumulative effect of expression of several pag encoded envelope proteins could result in resistance to defensins. An aggregate change in a large number of bacterial envelope proteins could alter the membrane charge, electrical potential, or lipid content such that defensin interaction with bacterial membranes could be changed.

Identification of transcriptional units linked to paqC

To identify genes upstream of pagC, E. coli carrying plasmid pWPL17 containing 2.8 kb of DNA 5' to pagC (Table 15 and FIG. 7) was mutagenized with the transposons MudJ and TnphoA, and strains with AP or .beta.-galactosidase activity were identified on chromogenic substrates. In addition, as part of an effort to identify additional PhoP-activated genes, random mutagenesis of the Salmonella chromosome with TnphoA was performed, and strains with AP activity were screened for TnphoA insertions linked to the Tn10.DELTA.16.DELTA.17 of strain AK3233, which is 75% linked to pagC. Several strains that contained plasmids with active MudJ or TnphoA generated gene fusions were identified. In addition, two strains were identified that contained active chromosomal TnphoA insertions closely linked to pagC. Physical maps of the restriction endonuclease sites surrounding the transposon insertions in strains with active plasmid or chromosomal lacZ and phoA gene fusions were performed to determine the relationship of the transposon insertions to pagC. This analysis revealed that several regions of the DNA were transcribed oppositely to pagC (FIG. 7). Several TnphoA insertions that resulted in active phoA gene fusions were identified. These data indicated that pagC-linked genes encoded membrane or secreted proteins.

Genes linked to pagC encode four novel proteins

To further analyze the genes defined by transposon insertions, the DNA sequence of this region was determined (FIG. 8). DNA containing this region was cloned; 4 kb of DNA between the HpaI site 737 bp upstream of the start codon of pagC to a ClaI site far upstream was sequenced. The DNA sequence of the fusion junctions of all TnphoA and MudJ gene fusions was also determined. Based on these data, the correct reading frame of each gene was determined. The DNA sequence data revealed four ORFs predicted to be transcribed and translated based on the data derived from the TnphoA and MudJ insertions. All ORFs revealed typical ribosome binding sites 6 to 11 bases from the predicted start of translation. The translation of the ORF immediately upstream and oppositely transcribed to pagC, pagD, indicates that a short envelope protein of 87 amino acids (unprocessed) is encoded. It is followed by a second ORF (envE) which encodes an envelope protein of 178 amino acids (unprocessed). This ORF is followed by a structure that could function as a Rho-independent transcriptional terminator (see FIG. 8). The third ORF, msgA (macrophage survival gene), encodes a small protein similar in size to that of the first gene product (79 amino acids) and is also followed by a structure that could function as a Rho-independent transcriptional terminator (see FIG. 8). The DNA sequence predicts that this protein is composed of several charged residues with a large number of negatively charged amino acids residing at the carboxy terminus. The predicted protein product does not contain a structure resembling a signal sequence at its amino terminus nor any hydrophobic stretches; therefore, the third ORF is unlikely to encode an envelope protein. The final ORF (envF) encodes an envelope protein of 278 amino acids (unprocessed). A computer search of known protein motifs revealed that EnvF contains a consensus prokaryotic membrane lipid attachment site and, therefore, is likely to be a lipoprotein (see FIG. 8 for consensus site location).

The predicted proteins produced by pagD, envE, and envF contain a typical bacterial signal sequence structure. In addition, hydrophobic profiles confirmed the hydrophobic nature of the amino-termini of these proteins. The EnvE and EnvF proteins also contain hydrophobic stretches that could function as membrane spanning domains. The G+C content of the genes in this region are: pagC, 43.4%; pagD, 42.1%; envE, 45.9%; msgA, 46.8%; and envF, 40.5%, which is considerably lower than the average G+C content of S. typhimurium (52%). A complete search of the database with the predicted protein sequences of these four ORFs showed no significant similarities. Strains containing three distinct TnphoA insertions and one MudJ insertion, each located in one of the four genes, were chosen for further characterization.

A gene pagD, oppositely transcribed to pagC, is positively regulated by PhoP/PhoQ

Representative strains with transposon insertions were examined to evaluate whether genes transcribed oppositely to pagC were increased in synthesis in the presence of PhoP. To accurately determine if these genes were PhoP regulated, it was necessary to recombine plasmid insertions onto the Salmonella chromosome. Upon replacement of the wildtype gene with the gene containing the transposon insertion, P22HTint lysates made on these strains were transduced into a PhoP deleted (PhoP.sup.-) strain and AP or .beta.-galactosidase levels were monitored. One of these transposon generated gene fusions demonstrated a significant increase in activity between PhoP.sup.- and WT backgrounds, while the other insertions showed no PhoP regulation (Table 16).

The pagD gene is adjacent to and divergently transcribed from pagC.

The representative transposon insertion in envF was unable to be recombined onto the chromosome, likely due to an insufficient amount of homologous DNA downstream of the transposon. In order to examine the possibility of PhoP regulation of the envF gene, a region upstream of this gene through and including the phoA gene of the TnphoA transposon was cloned as a 3-kb PvuI (blunt-ended)-XhoI fragment into the EcoRV-SalI sites of the suicide vector pGP704. This clone was mated into Salmonella strain CS019, and ampicillin-resistant recombinants were selected (creating a strain designated envF::pGPP2). A phoP105::Tn10d-Tet mutation was transduced into this strain to create an isogenic pair differing only in the ability to produce a functional PhoP protein. As shown in Table 16, the introduction of the phoP105::TN10d-Tet had no effect on the AP levels of these two strains, demonstrating that envF is not a PhoP-activated gene.

Transposon insertions in pagC-linked genes attenuate virulence and cause reduced survival within macrophages

Since transposon insertions in pagC significantly increase the LD.sub.50 of S. typhimurium in BALB/c mice, strains containing transposon insertions linked to pagC were evaluated for attenuation of mouse virulence. As shown in FIG. 7, while the transposon insertion in envE had no affect on strain virulence, a TnphoA insertion in pagD and the MudJ insertion 1.8 kb downstream in msgA attenuate S. typhimurium virulence by greater than 300 fold as compared to wild-type organisms (LD.sub.50 <20 organisms). These data suggested that these two loci are essential to virulence.

To examine the survival capabilities of those strains having a virulence defect, S. typhimurium containing insertions in either pagD or msgA were used to infect bone marrow-derived macrophages. The results, shown in Table 15, demonstrate a macrophage survival defect for these two strains. The survival defect is greater for the pagD insertion (MSI=0.002) compared with the msgA insertion (MSI=0.01), and both defects are equal to or greater than that of the PhoP.sup.- strain (MSI=0.01).

Transposon insertions in this gene could not be recombined onto the chromosome. Thus, it was necessary to demonstrate that the virulence and macrophage survival defects of msgA was not due to a polar effect of the MudJ insertion on envF transcription. Therefore, pGPP2 was recombined into the msgA::MudJ strain and AP activity of this strain was compared to that of CSO19 containing the recombinant pGPP2. This data (shown in Table 16) demonstrates that the transcription of the envF gene is unaffected by the msgA::MudJ insertion and is transcribed from its own promoter. However, it is possible that under different environmental conditions, other promoters may be activated that could place msgA and envF on the same transcript.

Determination of the msgA and pagD transcriptional start sites

The 5' regions of these genes were examined to define the transcriptional start sites of msgA and pagD. Oligonucleotides complimentary to the 5' end of each ORF or upstream region were used in a primer extension analysis. The results of this analysis revealed that the pagD transcript begins 39 bases upstream of the translational start. The predicted -10 (TTCCAT) and -35 (TTGAAT) regions were found to be similar to the known consensus sequences for E. coli promoters. The pagD transcript was detected only in PhoP.sup.c Salmonella RNA and not in RNA from PhoP.sup.- Salmonella. The msgA transcriptional start was found to begin 58 bases upstream of the translational start and contain predicted -10 (CAAAAC) and -35 (TTACGT) sequences. These regions do not conform well to consensus -10 and -35 sequences; however, the cDNA from this transcript was easily detected using various primers in primer extensions of both PhoP.sup.c and PhoP.sup.- RNA and appears to produce an abundant RNA.

Distribution of pagD and msgA genes in the Enterobacteriaceae and in two G+C content organisms

The G+C content of the pagC chromosomal region is much lower than the average G+C content of Salmonella. The gene encoding the PhoP-regulated acid phosphatase of S. typhimurium (phoN) also has a low G+C content (39%), and DNA homologous to phoN was found only in two low G+C organisms of several genera tested. The DNAs of several members of the Enterobacteriaceae and two low G+C organisms were examined for similarity to pagD and msgA by blot hybridization. PCR fragments highly specific to each ORF were labeled and used as probes. This analysis demonstrated hybridization at high stringency to all Salmonella species examined as well as Shigella sonnei, Shigella flexneri, Klebsiella pneumoniae and Citrobacter freundii. No hybridization was seen to the low G+C organisms Morganella morganii or Providencia stuartii. Identical hybridization patterns were seen with probes specific for both genes indicating that these genes are also linked in organisms other than Salmonella.

A virulence gene cluster required for Salmonells typhimurium survival within macrophage macrophages

Four genes upstream and oppositely transcribed to the pagC gene of Salmonella typhimurium have now been identified. Three genes (pagD, envE and envF) are predicted to be envelope proteins based on the isolation of active TnphoA insertions in these loci and the presence of a typical signal sequence at the amino-terminus of each protein. None of the four proteins possess significant homology to any protein in the database.

Only the gene immediately upstream of pagC and oppositely transcribed (pagD) was determined to be PhoP regulated. Transposon insertions in this gene greatly attenuate virulence and the ability of the organism to survive within murine macrophages. The transcription of several pag (including pagC) has been shown to be induced when Salmonella are within macrophage phagosome. In addition, analysis of proteins produced by Salmonella after infection of macrophage-derived cell lines indicate that pag products are induced and that pagC may be among the most abundant gene products induced upon macrophage infection. Since pagD is required for macrophage survival, it is likely that the transcription of this gene also will be induced within macrophage phagosomes. The pagD protein is small (87 amino acids, unprocessed) and has no strong hydrophobic domains; therefore, it is likely that it is a periplasmic or secreted protein.

Transposon insertions in the msgA gene were found to have an effect on mouse virulence and macrophage survival. It is likely that this gene may also be induced within acidified macrophage phagosomes as are other genes necessary for macrophage survival. If this gene is induced by the macrophage environment, its expression (as well as other genes necessary for macrophage survival) may be controlled by a regulatory system separate from the PhoP/PhoQ system.

These pagC-linked genes do not appear to form an operon. Because none of the genes downstream of pagD are PhoP regulated, they appear not be transcribed from the pagD promoter. The presence of a potential transcriptional terminator at the end of the envE gene makes it unlikely that msgA is co-transcribed with envE. The data suggest that the msgA::MudJ insertion is not polar on envF, which suggests that envF has its own promoter. Additionally, a potential transcriptional terminator following msgA as well as a 493 bp intergenic region makes it unlikely that these genes are co-transcribed. Primer extension analysis of these genes confirms that all four genes are transcribed from their own promoter.

The other two genes identified in this region, envE and envF, appear to produce membrane proteins that contain characteristic membrane spanning regions. The envF gene product is likely to be a lipoprotein based on the presence of a consensus lipid attachment site, and is likely to play a role in Salmonella virulence.

The low G+C content of the genes in the pagC region suggests that they may have been acquired by horizontal transmission. Southern blot analysis of low G+C organisms probed with the msgA or pagD genes showed no homology, but this does not eliminate the possibility that they were acquired from another low G+C content organism. The possibility also exists that these genes reside on a mobile genetic element acquired from another source. The msgA and pagD probes hybridized in identical patterns to some members of the Enterobacteriaceae other than Salmonella. However, the pagC gene has been shown to be unique to Salmonella species. This may indicate that the products of the genes upstream of pagC do not form a complex with PagC or that their functions do not require PagC interaction. Alternatively, because proteins that have homology to PagC exist in other Enterobacteriaceae (in the absence of any DNA homology), a PagC homolog may be linked to msgA and pagD in other species which was not detected by the DNA hybridization experiments.

pagC/pagD promoter region: expression of heterologous proteins

pagC and pagD are divergently transcribed and are both PhoP activated. Other divergently transcribed, regulated genes are known in the art (Beck et al., 1988, Microbiol. Rev. 52:318-326), e.g., the Klebsiella pneumoniae pulA-malX region (Chapon et al., 1985, J. Bacteriol. 164:639-645). Transcription of most of such genes require accessory proteins, such as CAP, in addition to the regulator to activate transcription. These two genes are divergently transcribed, and their promoters are arranged back-to-back. A region of 134 bp exists between transcriptional start sites of these genes, which is similar to the intergenic region between pagC and pagD. The pulA-malK promoter region is predicted to contain two MalT (the regulatory protein of this system) binding sites, one for each gene. Other MalT-activated genes require the CAP protein for expression, but the pulA and malX genes do not, possibly because of the high local concentration of the MalT regulator. Since the region between the transcriptional start sites of pagC and pagD (the predicted -35 sequences) is only 137 bp (nucleotides 562 to 776 of SEQ ID NO: 15), it is likely that only PhoP binding sites exist in the intergenic region, and that binding of one or more phosphorylated PhoP molecules positively regulates both genes. This pagC/pagD intergenic region which contains the divergent promoters can be used to construct vectors to express two heterologous proteins, one in each direction.

prg genes

As discussed above, phoP/phoQ constitutive mutations (phenotype PhoP.sup.c) increase the expression of pag and repress the synthesis of approximately 20 proteins encoded by phoP-repressed genes (prg). PhoP.sup.c bacteria are attenutated for mouse virulence suggesting that prg are virulence genes.

By use of the transposon, TnphoA, five unlinked prg loci were identified. In general, media conditions (starvation) that activate pag expression repress prg expression. One prg locus, prgH, was demonstrated to contribute to mouse virulence by both the oral and the intraperitoneal route. Both PrgH as well as PhoP.sup.c mutant S. typhimurium were found to be defective in induction of endocytosis by epithelial cells. Identification and mutation of such virulence genes will be useful in vaccine development.

Nucleotide sequence of the prg H, prgI, prgJ, and prgK genes

SEQ ID NO: 10 represents the nucleotide sequence of a 5100-bp HindIII fragment that contains the hyperinvasive hil locus. Four ORFS encoding four prg genes are located within this DNA (see FIG. 9). The ATG start codon is underlined; the asteriks indicate the positions of the prgH, prgI, prgJ, and prgK stop codons. These prg loci are required for bacterial invasion of epithelial cells, full mouse virulence, and transepithelial neutrophil migration. A bacteria attenuated by a mutation in one or more of these loci can be used to vaccinate individuals against infection by the wild type pathogen.

Strains, materials and methods

All bacterial strains used in the characterization of prg genes are listed in Table 5.

Bacteria were grown as follows: Luria-Bertani (LB) broth was used as rich medium. Antibiotics were used in the following concentrations in growth media or agar: ampicillin 100 .mu.g/ml (Ap), chloramphenicol 25 .mu.g/ml (Cm), gentamicin 30 .mu.g/ml (Gm), kanamycin 45 .mu.g/ml (Km), and tetracycline 25 .mu.g/ml (Tc). The chromogenic substrate 5-bromo-4-chloro-3-indolyl-phosphate (p-toluidine salt) (XP) was used to detect phosphatase activity on agar at a final concentration of 40 .mu.g/ml. p-nitrophenyl phosphate (p-NPP) was used as a substrate for quantitative measurement of AP activity. Media was buffered to various pH ranges with 1M sodium citrate. E media (Vogel-Bonner minimal) was prepared as described by Davis et al., 1980, Advanced Bacterial Genetics: A Manual for Genetic Engineering. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. Nitrogen-, carbon-, and phosphate free medium (N.sup.- C.sup.- P.sup.-) was prepared as described by Kier et. al., 1977, J. Bacteriol. 130:399, herein incorporated by reference.

This starvation medium was supplemented with 0.04% (wt/vol) glucose as the carbon source, 10 mM NH.sub.4 Cl as the nitrogen source, and 1 mM NaH.sub.2 PO.sub.4.H.sub.2 O as the phosphate source. The carbon concentration is one log less than described by Kier et al., supra.

AP activity of strains isogenic except for mutations in the phoP locus was measured in cultures grown from a single colony inoculum under various oxygen tensions with or without shaking at 37.degree. C. Anaerobic cultures were grown in an anaerobic chamber (Coy Laboratories Products, Inc.) with a gas mixture of 80% N.sub.2, 10% O.sub.2, and 10% CO.sub.2 at 37.degree. C. For acid regulation, aliquots of mid-logarithmic cultures were removed to measure initial pH and AP activity. 1M sodium citrate (pH>6.0) or 1M citric acid (pH 4.7) were added to equivalent amounts of culture to a final concentration of 50 mM citrate. Cultures were grown aerobically for two hours at 37.degree. C. and then pH and AP measurements were taken. AP activity was measured as described previously (Michaelis et al., 1983, J. Bacteriol. 154:366-374, herein incorporated by reference). AP units were calculated by the following formula: units ={OD.sub.420 /time (minutes).times.volume.times.OD.sub.600 !}.times.1000 as defined by Miller for .beta.-galactosidase (Miller et al., 1972, Experiments in molecular genetics, p. 352-355. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.).

Standard bacterial genetic techniques were used to study prg loci. Bacteriophage P22HTint-mediated transduction was performed as according to methods known in the art. TnphoA mutagenesis was performed using a broad host range plasmid (pRT291) to deliver TnphoA (Taylor et al., 1989, J. Bacteriol. 171:1870, herein incorporated by reference). Transpositions of TnphoA into Salmonella DNA were identified by use of the incompatibility plasmid pPH1JI (Taylor et al., supra). Screening for phoP-repressed genes was performed using CS031, the donor strain of the pho-24 allele. CS031 was constructed by a P22 bacteriophage transductional cross between strains AD154 and CS022 which contains the purB::Tn10 allele and the pho-24 allele, respectively. The linkage of pho-24 and purB::Tn10 was 70%, similar to the linkage of purB to other phoP alleles. Therefore, when P22 bacteriophage transductional crosses were performed between CS031 and the strains containing active gene fusions to phoA, strains could be screened for loss of fusion protein activity on acquisition of tetracycline resistance. Initial screening involved detection of loss of AP activity in approximately 70% of colonies that acquired tetracycline resistance, as they were presumed to contain the pho-24 allele. In addition, controls were performed using strain AD154 that contains the same purB::Tn10 allele linked to a phoP null allele, phoP12. Plasmid DNA was transformed into S. typhimurium strain LB5010 by the calcium chloride and heat shock procedure (Maclachlan et al., 1985, J. Bacteriol. 161:442).

Isolation of strains with TnphoA insertions in phoP-repressed genes

Constitutive mutations in the phoP locus (phenotype PhoP.sup.c) that result in increased expression of pag in an unregulated fashion also markedly attenuate S. typhimurium virulence and survival within macrophages. The virulence defect of PhoP.sup.c strains can be explained by their decreased expression of approximately 20 polypeptides encoded by phoP-repressed genes (prg).

A PhoP.sup.- PhoN.sup.- strain (IB001) was constructed by a P22 transductional cross between CS019 and CS003. IB001 was then mutagenized with TnphoA (so that background acid phosphatase, encoded by phoN, would not interfere with the measurement of fusion protein activity on alteration of the phoP locus) and 1800 individual blue colonies with PhoA fusion protein activity were isolated on LB agar plates containing XP. These colonies were the result of 18 separate matings with approximately 20 pools in each. These strains were tested for reduction of fusion protein activity on acquisition of the pho-24 allele (CS031), which resulted in a PhoP.sup.c phenotype. AP assays were then performed on strains isogenic except for the phoP locus.

The PhoP.sup.c phenotype was confirmed in these strains by preparation of whole cell protein extracts and SDS-PAGE analysis. All strains with a PhoP.sup.c phenotype demonstrated the expected distinctive pattern of protein expression in PhoP.sup.c strains, i.e. repressed protein species of specific sizes.

Eight strains were identified with gene fusions to phoP-repressed genes. As shown in Table 6, the synthesis of most prg::TnphoA fusion proteins was fully repressed by the pho-24 allele. While two loci had complete repression of fusion protein activity, others demonstrated only partial repression. The expression of pag in PhoP.sup.c strains is 5-10 fold less than that observed after bacteria are phagocytosed by macrophages suggesting that the degree of repression of some prg loci may be greater when pag are maximally activated within acidified macrophage phagosomes.

Lower values for prgB -phoA fusions in strains with a wildtype phoP locus (Table 7B) compared to PhoP.sup.- strains (Table 7) may represent some degree of repression in the presence of PhoP.

In Table 6, a comparison of the effect of phoP locus mutations on Prg-PhoA fusion protein activity is made. PhoP.sup.- indicates that the strain assayed contains the phoP12 allele (CS030) and PhoP.sup.c indicates the strain assayed contains the pho-24 allele (CS031). Values were calculated from stationary phase cultures. The numbers denote representative values of experiments performed on three separate occasions and represent activity in units of AP as defined above.

Table 7 demonstrates the effects of environmental conditions on the in vitro regulation of prg loci.

Table 7A shows the effect of starvation on prg and pag expression. Starvation medium (N.sup.- C.sup.- P.sup.-) (17) contained 0.04% glucose, 10 mM NH.sub.4 Cl, and 1 mM NaH.sub.2 PO.sub.4.H.sub.2 O. The fusion protein activity for starvation media was measured after 48 hours of growth (OD.sub.600 =0.5) while that in rich media (LB) was measured in late-logarithmic growth (OD.sub.600 =1.0).* All cultures were grown aerobically.

Table 7B shows the effect of oxygen tension on expression of phoP-activated and phoP-repressed genes. Expression in rich medium is compared under aerobic conditions at stationary phase (OD.sub.600 >1.4), microaerophilic (OD.sub.600 =0.8), and strict anaerobic conditions with 80% N.sub.2, 10% O.sub.2, and 10% CO.sub.2 (OD.sub.600 =0.6) after 24 hours of growth.* Table 7C shows the effect of pH on the expression of fusion protein activity of prg and pag loci. Expression was measured from cultures grown to logarithmic growth (OD.sub.600 =0.5) in LB media buffered to various pHs with sodium citrate. All the numbers represent activity in units of AP as defined above.

Chromosomal location of prg::TnphoA loci

prg::TnphoA linkage analysis was performed to a bank of strains with randomly spaced Tn10.DELTA.16.DELTA.17 insertions to determine chromosomal locations and whether prg::TnphoA alleles were unlinked loci. The prg::TnphoA insertions were in five distinct linkage groups. Three alleles, prgE1-3::TnphoA were identically linked to the Tn10.DELTA.16.DELTA.17 insertion of AK3091 (26%) and two other alleles, prgB1-2::TnphoA were similarly linked to the Tn10.DELTA.16.DELTA.17 insertion of AK3190 (94%), AK3249 (89%), and AK3186 (50%). Another allele, prgH1::TnphoA, was found to be 37% linked to the Tn10.DELTA.16.DELTA.17 insertion of strain AK3304. The other two prg alleles did not demonstrate linkage to the bank of strains tested. The chromosomal DNA of these two strains was analyzed by Southern hybridization analysis using a portion of TnphoA as a probe, and a rough physical map of the sites located adjacent to the TnphoA insertion was determined. These alleles, prgA and prgC, had different restriction endonuclease sites surrounding the TnphoA insertions. In addition, the repression of prgA and prgC fusion protein activity in strains with the pho-24 mutation was different; prgC was completely repressed, while prgA was only partially repressed indicating that these loci are different. Thus, five unlinked loci encoding envelope proteins repressed in the PhoP.sup.c phenotype were identified.

Though three prg loci were identified that were linked to transposon insertions, none of the Tn10.DELTA.16.DELTA.17 insertions had a known map location. The physical map location of two of these transposon insertions, AK3249 and AK3304, was analyzed using XbaI restriction endonuclease digestion and pulse field gel electrophoresis (PFGE). Since Tn10.DELTA.16.DELTA.17 contains a single XbaI site, these Tn10.DELTA.16.DELTA.17 insertions can be assigned to a specific XbaI fragment of known map location (Liu et al., 1992, J. Bacteriol. 174:16622). AK3249 was assigned to 28-32 min, while AK3304 was assigned to either end of the 58-70 minute fragment. Further P22 transduction to known markers in those regions was performed. The Tn10.DELTA.16.DELTA.17 insertion of strain AK3249 and pgrB1::TnphoA were found not to be linked to the Tn10 insertion of strain TN3061 (6% linked to dcp), which has a transposon insertion at 28 min, or to the ompD::Tn5 insertion of strain SH7782 at 32 min. prgH1::TnphoA was found to be very weakly linked to the srl202::Tn10 insertion of strain TT520 (<0.1%) at 59 minutes. These data indicate that prg are unlinked on the Salmonella chromosome, consistent with the function of PhoP/PhoQ as global regulators.

The chromosomal location of TnphoA insertions in phoP-repressed genes (prg::TnphoA) was determined by linkage analysis to a bank of strains with Tn10.DELTA.16.DELTA.17 insertions (Kukral et al., 1987, J. Bacteriol. 169:1787, herein incorporated by reference). Cells with TnphoA insertions were spread on LB agar plates containing 10 .mu.g/ml tetracycline and 40 .mu.g/ml XP. Then P22 lysates grown on strains with Tn10.DELTA.16.DELTA.17 insertions were spotted onto plates with a multiprong inoculator. After overnight inoculation, plates were reviewed for linkage by looking for mixed blue and white colonies. Linkage was confirmed and quantitated by carrying out individual transductional crosses between the Tn10.DELTA.16.DELTA.17 containing strains and the strain with the TnphoA insertion. After selection for the Tn10.DELTA.16.DELTA.17 encoded tetracycline resistance, strains were scored for loss of blue color and TnphoA encoded kanamycin resistance. Some TnphoA strains were found to be linked to Tn10.DELTA.16.DELTA.17 strains with no known map location. Two of these Tn10.DELTA.16.DELTA.17 insertions were physically mapped using PFGE following XbaI restriction endonuclease digestion. Based on physical mapping, linkage analysis to other transposon insertions by P22 bacteriophage transduction was determined as necessary.

Chromosomal DNA was prepared as described by Mekalanos, 1983, Cell 35:253, herein incorporated by reference, using Proteinase K instead of Pronase. Purification of plasmid DNA was performed by standard methods. Restriction endonuclease digestion was performed according to the recommendations of the manufacturer (New England Biolabs). DNA, size fractionated in agarose gels, was transferred to Genescreen Plus membranes (New England Nuclear/Dupont, Boston, Mass.) for blot hybridization by the method of Southern well known in the art. DNA probes were purified from agarose gels by the freeze-squeeze method (Tautz et al., 1983, Anal. Biochem. 132:14) and radiolabelled with .sup.32 P!dCTP by the random primer method (Feinberg et al., 1983, Anal. Biochem. 132:6).

Cloning genes from Tnpho A fusions

The gene encoding prgH has been cloned using methods described below. The plasmid, pIB01, containing the prgH gene has been deposited with the American Type Culture Collection on Jul. 9, 1993 (Rockville, Md.) and has received ATCC designation ATCC 75496. FIG. 5 shows the partial DNA sequence of prgH (SEQ ID NO: 3). FIG. 9 shows the location and sequence of the entire prgH gene.

The genes described herein which have been identified by ThphoA insertion can be cloned using methods known in the art (Beattie et al., 1990, J. Bacteriol. 172:6997). Chromosomal For example, DNA of each strain containing a prg::TnphoA gene fusion is digested with a restriction enzyme such as BamH1 which cuts at a single site in TnphoA maintaining the fusion junction, phoA sequences and the neo gene. Similarly, a plasmid such as pUC19 is digested with the same enzyme. Digested chromosomal and plasmid DNA are ligated overnight at 15.degree. C. and transformed into competent E. coli. Transformations are plated on LB agar containing ampicillin and kanamycin to select for the bla gene of pUC19 and the neo gene of TnphoA. The chromosomal DNA containing the prg::TnphoA gene fusion can then be sequenced using standard methodology described above, such as the SEQUENASE.RTM. (United States Biochemical) kit. Universal primer (United States Biochemical) corresponding to DNA sequences in the plasmid or TnphoA primer (5'-AATATCGCCCTGAGCA-3') (SEQ ID NO: 4) corresponding to bases 71 to 86 of TnphoA can be used as primers.

To clone the wild type gene, a fragment of chromosomal DNA flanking TnphoA sequences can be used to screen a cosmid gene bank of wild type Salmonella strain ATCC 10428 using methods described above for cloning wild type pagC.

Environmental regulation of prg loci

Since PhoP/PhoQ are environmentally responsive regulators, the effects of different growth conditions on prg::TnphoA expression were tested. The growth rate of strains with prg::TnphoA insertions was comparable to wild-type organisms under all conditions. The expression of all prg loci was maximal in late logarithmic growth phase when bacteria were grown in rich (LB) media. An example of this is the comparison of values of prgH::TnphoA expression in Table 7A (rich media and stationary growth) and Table 7C (pH 7.0, log phase). Since the expression of pag loci was maximal in starvation (which only reaches a maximal OD.sub.600 =0.5) and stationary growth phase, this was consistent with a reciprocal relationship between the expression of pag and prg. Further analysis of prg loci expression under starvation conditions confirmed this reciprocal relationship (Table 7A). prgH expression was repressed (Table 7A) and other prg were minimally affected under starvation conditions, in contrast to the induction of pag expression when bacteria were starved (Table 7A).

Because of its role in bacterial-mediated endocytosis (BME), the effect of oxygen tension in rich medium on pag and prg expression was also tested (Table 7B). Different but not reciprocal regulation of pag and prg loci was found on growth at different oxygen tensions. Though pagA and pagB loci were minimally affected by growth at different oxygen tensions, the pagC virulence locus was approximately 5 fold repressed when bacteria were grown anaerobically as compared to aerobic growth (Table 7B). Variability was also noted in the expression of prg loci in response to growth conditions in the absence of oxygen. One loci, prgH, was repressed three-fold in anaerobic growth, while another locus, prgB, was induced almost 50-fold when grown anaerobically (Table 7B). Other prg loci had minimal change in fusion protein expression as a result of different oxygen tensions in the growth media.

Low pH conditions also had a variable effect on prg expression (Table 7C). The expression of pagC fusion protein activity was induced under acid conditions as previously known. When bacteria were grown to mid-logarithmic growth, no significant induction of the relative repression of prgH expression was noted in media of low pH, while prgB expression was induced on exposure of bacteria to low pH (Table 7C). Hence, loci maximally expressed under diverse environmental conditions can all be repressed by the PhoP.sup.c phenotype.

Acid sensitivity was tested by the method of Foster et. al., 1990, J. Bacteriol. 172:771, herein incorporated by reference. Strains were grown aerobically in E media and 0.4% glucose at 37.degree. C. to an OD.sub.600 of 0.5. The pH of the bacterial culture was decreased to near 3.3 by the addition of 1M hydrochloric acid. An aliquot was taken immediately (t.sub.o), the remainder of the culture was incubated further at 37.degree. C. with subsequent aliquots removed at 40 min (t.sub.40) and 80 min (t.sub.80) time points. The pH of the cultures remained near 3.3. The aliquots were diluted 1:10 in cold PBS, washed and resuspended in normal saline prior to plating serial dilutions for colony forming units.

prgH is a virulence locus for S. typhimurium

Since the PhoP.sup.c phenotype resulted in virulence attenuation and repressed the synthesis of approximately 20 proteins, the virulence of strains with single mutations in prg loci was tested (Table 8). Strains with prg::TnphoA insertions were screened for virulence defects by intraperitoneal injection of approximately 150 organisms into BALB/c mice. Controls were also performed with wild-type bacteria. A significantly longer time course of clinical disease progression was observed with a prg mutant strain compared to wild type bacteria. Mice injected intraperitoneally with strains containing the prgH1::TnphoA insertion developed clinical signs of typhoid fever, such as a "scruffy" phenotype (fever and piloerection) and hepatosplenomegaly in approximately 10-14 days, compared to approximately 24 hours for the wild type bacteria. Despite the extended time course of disease development, all the mice eventually died. Disease progression of mice injected with other strains containing prg::TnphoA insertions showed a similar pattern of illness to that of wild type bacteria.

Table 8 shows the effect of the prgH1::TnphoA mutation on Salmonella mouse virulence. Strains were isogenic and administered by intraperitoneal injection and oral inoculation in 35 day old BALB/c mice. The number of animals used at bacterial dilutions near the LD.sub.50 for each allele is listed in parentheses. The LD.sub.50 determinations were repeated on three separate occasions.

Further testing of the LD.sub.50 of strains containing prgH mutations was performed. prgH mutants were determined to have an LD.sub.50 of approximately 60 organisms compared to a value of <10 for wild type bacteria. Due to the difficulty in accurately delivering organisms in small doses to mice, a strain with a mutation in both prgH and phoP was constructed. The PrgH.sup.- PhoP.sup.- strain had greater than a 10 fold increase in LD.sub.50 compared to CS015, an isogenic PhoP.sup.- strain (Table 8). The combined effect of the two mutations further documented that the prgH1::TnphoA mutation attenuated S. typhimurium virulence and indicated that mutations which affected two phases of PhoP/PhoQ regulated gene expression were additive in their effect on virulence. Strains with prgH1::TnphoA insertions were also tested for virulence when administered by the oral route. A 10 fold decrease in virulence (increase in LD.sub.50) was observed (Table 8).

Further analysis of the efficiency of strains with prgH1::TnphoA insertions in crossing the mucosal barrier was tested by competition experiments with wild-type bacteria. During the first 72 hours after oral inoculation with mutant bacteria, no prgH1::TnphoA mutants were recovered from the bloodstream of mice compared to control experiments in which organisms were routinely isolated from the blood of mice inoculated with wild type bacteria. Other strains with prg mutations were also tested for virulence defects by the oral route, but no significant change in virulence was observed.

Mouse virulence studies were carried out as follows. Bacteria were grown aerobically at 37.degree. C. to stationary phase, washed with LB, and diluted in normal saline. 35 days old (16-18 g) female BALB/c mice were purchased from the Charles River Breeding Laboratories, Inc. (Wilmington, Mass.). Diluted bacterial samples in saline were injected intraperitoneally with an inoculum of 0.1-0.15 ml. Bacteria were administered orally as a 0.5 ml bolus to mice fasted for 2 hours, via a 2 inch straight, 18 gauge stainless steel animal oral feeding needle (Harvard Apparatus, Inc., South Natick, Mass.) under mild 2-bromo-2-chloro-1,1,1-trifluoroethane (Halothane) anesthesia. The number of organisms administered was quantitated by plating for cfu/ml on LB agar. Mouse 50% lethal dose (LD.sub.50) values were determined by standard methods (Reed and Muench, 1938, Amer. J. Hygiene 27:493). The LD.sub.50 determinations were repeated on three separate occasions. Competition assays were performed after bacteria were administered orally to mice as above. Bacteremia was assessed on days 1-4 from tail bleeds or intracardiac punctures with 50 .mu.l of blood plated immediately and after growth in LB broth at 37.degree. C. overnight. Spleen and intestinal harvests were performed on days 1-6 with organs homogenized in 3 mls of 0.9% sodium chloride. Samples and cultures were plated in serial dilutions. S. typhimurium was confirmed by characteristic growth (black colonies) on Hektoen-enteric agar (Difco Laboratories) and by the macroscopic slide agglutination test with Salmonella rabbit serum Group B (Antigens 4, 5, 12) (Fisher Scientific).

Mutations in oxygen-induced genes do not affect mouse virulence

Both prgH and pagC loci were shown to be repressed by anaerobic growth and required for full virulence, thus suggesting that a shift from anaerobic to aerobic conditions might serve as a general signal for induction of virulence genes. Strains with mutations in oxygen-inducible loci (Aliabadi et al., 1986, J. Bacteriol. 165:780) were constructed. ATCC14028s derivatives with oxiA, oxiC, and oxiE mutations were made (termed CS032, CS033, CS034, respectively). These strains were as virulent as wild type bacteria. Though these gene fusions could still mark operons containing virulence genes, this data suggests that these loci are not essential to full virulence and that oxygen induction is not always correlated with virulence function.

prgH mutants have normal survival within macrophages

Since the PhoP.sup.c phenotype resulted in a defect in bacterial survival within macrophages, the effect of this mutation on the synthesis of a prgH-encoded protein was tested. A strain with the prgH1::TnphoA insertion was tested for intracellular survival within bone marrow-derived macrophages from BALB/c mice and J774.2 cells, a macrophage derived cell line. No defect in intracellular survival was observed. A strain with a pgrB1::TnphoA insertion was also tested and found not to have a defect in survival within macrophages.

Assays to determine bacterial survival within macrophages were performed as described by Buchmeier al., 1989, Infect. Immun. 57:1, herein incorporated by reference. Bacteria grown to stationary-phase were opsonized for 30 minutes in normal mouse serum before exposure to cultured bone marrow-derived macrophages harvested from BALB/c mice. One hour after infection, gentamicin 10 .mu.g/ml was added to kill extracellular bacteria. All time points (1, 4, and 24 hr) were done in triplicate and repeated on three separate occasions.

Cultured bone marrow macrophages were harvested from BALB/c mice purchased from the Charles River Breeding Laboratories. J774.2 macrophages were cultured in Dulbecco's minimal essential medium with 10% fetal bovine serum (DMEM/10%FBS).

prg::TnphoA insertions do not suppress the phenotypes of PhoP mutants

Several phenotypes of phoP mutants, including defensin and acid sensitivity as well as mouse virulence attenuation, were tested for suppression on addition of a prg::TnphoA mutation. To test the ability of a phoP mutation to suppress the synthesis of prg products, PhoP mutant strains isogenic except for prg::TnphoA mutations were constructed and tested for mouse virulence, where suppression would involve an increase in virulence, or decreased acid and defensin sensitivity. prg::TnphoA insertions had no effect on the virulence phenotypes of PhoP.sup.- bacteria. These results indicate that the prg::TnphoA mutations tested did not suppress the PhoP null phenotype as single mutations.

PrgH and PhoP.sup.c mutants are defective in bacterial-mediated endocytosis by cultured epithelial cells

The BME of prg::TnphoA and PhoP.sup.c S. typhimurium strains was tested. The following observations (described herein) suggested that prg genes may be involved in bacterial-mediated uptake by eucaryotic cells: prgH1::TnphoA was shown to be located at 59' on the bacterial chromosome, a location where other genes essential to invasion are clustered; prgH mutants were shown to be defective in competition with wild type organisms on reaching the bloodstream of mice in the first 72 hours after oral ingestion; and the expression of one prg locus, prgB, was dramatically induced under anaerobic growth conditions. Strains with prgH and pho-24 mutations had a significant reduction (p-value<0.01) in their ability to induce uptake by Madin-Darby canine kidney (MDCK) polarized epithelial cells compared to wild-type bacteria. Other prg strains with TnphoA insertions did not demonstrate a statistically significant defect in BME by epithelial cells (Table 9). The adherence of strains defective in BME was unaffected by the prgH::TnphoA insertion when determined by cell-associated cfu/ml before the administration of gentamicin (Table 9) and by microscopy.

To assay bacterial adherence and uptake of bacteria by epithelial cells, bacterial strains were grown at 37.degree. C. without shaking (microaerophilic) to a final density of approximately 2.times.10.sup.8 colony forming units (cfu)/ml. Assays were performed by seeding 10.sup.5 MDCK cells/well in 24-multiwell tissue culture plates. Cells were incubated overnight at 37.degree. C. in 5% CO.sub.2 /95% air atmosphere in DMEM/10%FBS without antibiotics until >80% confluent. The adherence and invasion assays were carried out according to the protocol of Lee and Falkow, 1990, Proc. Natl. Acad. Sci. USA 87:4304, herein incorporated by reference.

In Table 9, the effect of prg::TnphoA insertions on Salmonella-mediated endocytosis by MDCK epithelial cells is shown. Microaerophilically grown bacterial strains were assessed for changes in adherence and invasion. Adherence was determined as the percentage of bacteria adhered to the cells after centrifugation and 30 minute 4.degree. C. incubation/total number of bacteria added to each well. Invasion was determined as the percentage of bacteria that had invaded after a two hour incubation with gentamicin/total number of bacteria added to each well. There was no difference between S. typhimurium wildtype and S. enteritidis CDC5 wildtype strains with respect to adherence and invasion frequency. The asterisk (*) represents statistical significance by variance analysis of the invasion data done in triplicate compared to wild-type (p-value<0.01).

The confluent MDCK monolayers were washed three times with PBS, then 0.9 ml of cold DMEM/10%FBS was added to each well. Bacteria were washed in LB and resuspended in an equivalent volume of DMEM/10%FBS. Approximately 5.times.10.sup.7 bacteria were added/well. The plates were spun at 500 rpm at 4.degree. C. for 10 minutes, then incubated at 4.degree. C. for 30 minutes. Adherent bacteria were recovered by washing the plates three times with phosphate-buffered saline (PBS), lysing the epithelial cells in 0.5 ml of 1% Triton-X-100/PBS, and plating for cfu/ml on LB agar. A morphologic assessment of adherence was also performed by staining bacterially infected epithelial cell monolayers grown overnight on coverslips for 7 minutes in 1 .mu.g/ml 4' 6-diamidino-2-phenylindole (DAPI). These DAPI stained coverslips were examined by both fluorescent and phase contrast microscopy using a Leitz Laborlux 12 microscope.

Invasion or bacterial-mediated endocytosis (BME) was assessed by allowing bacteria to adhere as described above. Plates containing bacteria and epithelial cells were incubated for two hours at 37.degree. C. in a 5% CO.sub.2 /95% air atmosphere. Each well was washed three times with PBS to remove bacteria not associated with cells. DMEM/10%FBS supplemented with 10 .mu.g/ml gentamicin was then added to kill extracellular bacteria. After 90 minutes of incubation, the cell monolayers were washed three times with PBS and the viable intracellular bacteria were released by vigorously pipetting with 0.5 ml of 1% Triton X-100/PBS. An invasion deficient Salmonella enteritidis mutant and an invasive clinical wild-type isolate of S. enteritidis were used as controls for BME. Viable bacteria were quantitated by plating for cfu/ml on LB agar medium. All assays were done in triplicate and repeated at least three times.

MDCK epithelial cells were used between passage 40-58 to maximize bacterial adherence and invasion. Epithelial cell lines were cultured in DMEM/10% FBS and 1% penicillin/streptomycin solution at 37.degree. C. in a 5% CO.sub.2 atmosphere.

To assay bacterial defensin sensitivity, NP-1 defensin was purified from rabbit peritoneal neutrophils according to methods known in the art (Selsted et al., 1985, J. Biol. Chem. 260:4579; Selsted et al., 1984, Infect. Immun. 45:655). Typically, 10.sup.5 bacteria in 0.5% tryptone in 100 .mu.l volume were exposed to 50-100 .mu.g of defensin/ml at 37.degree. C. for 2 hours. The reactions were stopped by diluting the reaction in 0.9% NaCl. Appropriate dilutions were plated to determine the cfu/ml of surviving bacteria. Assays were performed in duplicate at least twice for each strain. Appropriate assays with sensitive (PhoP.sup.-) and resistant (wild-type) strains were performed as controls.

Mapping of prgH

The location of prgH relative to other invasion loci at 59 minutes was determined using linkage analysis. P22 transduction linkage analysis indicated that the Tn10.DELTA.16.DELTA.17 of strain AK3304 had similar linkage to invA (40%) and prgH (37%); however, invA was not linked to sorbital. The prgH1::TnphoA insertion was found to be linked (99.6%) to the transposon insertion of EE517, a strain with a 8.5 kilobase deletion adjacent to the Tn5B50-378 insertion of hil.

A physical map of the restriction endonuclease sites surrounding the TnphoA insertion of strain IB037 was made (FIG. 4) revealing no similarities to the known restriction endonuclease map of the invA-E region. Plasmids containing the Cloned inv and hil DNA were then used as probes in Southern hybridization analysis of chromosomal DNA from wild type ATCC10428s and IB040 bacteria containing the prgH1::TnphoA insertion. When a plasmid which contains other invasion loci highly linked to invA-E (invH, invF, and part of invG) was used as a probe, no differences in hybridization pattern was found between wild type bacteria and strain IB040 indicating that prgH was not located within the inv region. However, when a plasmid containing a 5 kb region immediately downstream of the Tn5B50-380 insertion of hil was used as a probe, the prgH1::TnphoA insertion was demonstrated to be located within this region. By use of the known restriction map of the hil locus (Lee et al., 1992, Proc. Natl. Acad. Sci. USA 89:1847) and the known restriction endonuclease sites of TnphoA, the physical map of this area and the relationship of prgH1::TnphoA within it were further defined (FIG. 4). The prgH1::TnphoA insertion was oriented so that the direction of transcription of the phoA fusion protein was opposite to that of the Tn5B50 insertions that confer the hil phenotype and contain a constitutive neomycin promoter that is transcribed out of the transposon (FIG. 4). Although prgH was found to be located within the hil locus, this gene is unique in that it is oppositely transcribed and unlike any other genes identified within the hil locus, prgH is regulated by the phoP regulon.

Since it was possible that a protein whose expression was altered by the Tn5B50-380 insertion might alter the expression of prgH, strains containing both insertions were constructed and the prgH-phoA fusion protein activity compared under different environmental conditions. When bacteria were starved or grown anaerobically, derepression of fusion protein activity was observed. Table 11 shows the effect of the Tn5B50-380 insertion on expression of prgH fusion protein activity.

This data demonstrates that the Tn5B50-380 insertion increased prgH expression, even though prgH transcription was opposite to that of the Tn5B50-380 encoded neomycin promoter. Starvation (repressing conditions for prg) indicates that bacteria were grown aerobically for 48 hours in starvation medium (N.sup.- C.sup.- P.sup.-) containing 0.04% glucose, 10 mM NH.sub.4 Cl, and 1 mM NaH.sub.2 PO.sub.4.H.sub.2 O. LB (aerobic) indicates that bacteria were grown in Luria-Bertani broth (rich media) to late logarithmic growth (nonrepressing conditions) (OD.sub.600 =1.0). LB (anaerobic) indicates that bacteria were grown under strict anaerobic conditions for 24 hours (OD.sub.600 =0.6). All the numbers represent activity in units of AP as described above.

To rule out the possibility that the BME defect of the prgH mutant was an artifact of the PhoA fusion protein produced, complementation analysis was performed with a plasmid (pWKSH5) containing a 5.1 kb HindIII fragment which included the hil and prgH loci. The plasmid was crossed into PrgH (IB040) and PhoP.sup.c (CS022) mutant bacteria to create strains IB043 and IB044, respectively. The BME phenotype of the PrgH mutant was similar to wild-type with the same plasmid insertion. The BME phenotype of the PhoP.sup.c mutant was not complemented by this plasmid. These results indicate that a gene product altered in synthesis as a result of the prgH::TnphoA insertion was necessary for BME.

Using a strain with a phoP/phoQ locus mutation that constitutively simulates the environmental activation of pag (phenotype PhoP.sup.c), five unique phoP-repressed loci encoding envelope proteins were defined. phoP-repressed genes (prg) were found to be widely spaced on the chromosome and the expression of prg loci was repressed under starvation conditions, when pag loci were induced (Table 10).

PrgH was shown to lie between two Tn5B50 insertions that confer the hil phenotype. Since deletion mutants in this region have been demonstrated to also have defects of BME, and the BME defect of prgH mutants can be complemented with a plasmid containing this locus, it is possible that a protein not synthesized as a result of the prgH1::TnphoA insertion promotes BME (FIG. 4).

Contrary to the expectation that genes essential to the hil phenotype would be induced under microaerophilic conditions similar to what was found for prgB, prgH expression was maximal during aerobic growth and the Tn5B50-380 insertion, which results in a hil phenotype, derepressed expression of prgH. In addition, the direction of transcription predicted by the prgH1::TnphoA insertion is opposite to that of the Tn5B50-380 encoded neomycin promoter associated with the hil phenotype suggesting that a regulatory protein interrupted by or transcribed from the Tn5B50-380 insertion affects the expression of prgH.

In view of the observation that pWKSH5, a plasmid containing prgH (hil), did not complement PhoP.sup.c bacteria for BME, it is possible that other invasion genes may also be regulated by PhoP/PhoQ. If prgH was expressed from pWKSH5, despite the presence of the pho-24 mutation, this suggest that other genes repressed as part of the PhoP.sup.c phenotype are necessary for BME.

The identification and characterization of prgH has shown that PhoP/PhoQ oppositely regulate factors necessary for bacteria to enter or to survive within mammalian cells, further documenting the importance of gene regulation to bacterial virulence. The identification of prg loci can be used to study the regulation of bacterial genes after infection of mammalian cells. Understanding the regulation of virulence genes, such as prgH can also be used to attenuated pathogenic bacteria for the development of new live vaccines for typhoid fever.

Role of prg genes in virulence

The prg locus, prgH, was found to contribute to mouse virulence when S. typhimurium was administered by both the oral and intraperitoneal routes. PrgH as well as PhoP.sup.c mutants were further found to be defective in bacterial-mediated uptake by epithelial cells suggesting that an inability to cross epithelial barriers might contribute to the attenuation of virulence observed. Competition studies in mice after oral ingestion of bacteria further supported that prgH mutants were defective in transcytosis across the intestinal epithelial barrier. Therefore, at least two phases of PhoP/PhoQ regulated protein expression essential to bacterial virulence have been defined. In one phase, prg expression promotes bacterial mediated endocytosis by epithelial cells (Table 10), while in another phase, pag expression promotes survival within macrophages.

Systemic pathogens, such as Salmonella, may encounter more complex and varied environments than may be encountered by mucosal pathogens. The achievement of intermediate states of pag and prg expression could be essential to virulence at some stage of the infectious cycle. Consistent with this concept was the lack of uniformity observed in the expression of pag and prg on growth at different oxygen tensions and pH conditions. These data may also indicate that not all regulation of pag and prg is mediated directly through PhoP and PhoQ. Given the function of PhoP as a transcriptional regulator, it is likely that prg loci repression occurs at the level of transcription.

The approach of defining genes repressed by the pho-24 mutation has led to the discovery of at least one virulence locus, prgH, which can be mutated to attentuate the bacteria for vaccine purposes.

Attenuation of Bacterial Virulence by Constitutive Expression of Two-component Regulatory Systems

The virulence of a bacterium can be attenuated by inducing a mutation which results in the constitutive expression of genes under the control of a two-component regulatory system or by inducing a mutation that inactivates a gene under the control of the two-component systems. A balance between the expression of the genes under the control of the two-component system, e.g., between pag and prg gene expression, and possibly between two-component system regulated genes and other genes, is necessary for full virulence. Mutations that disrupt this balance, e.g., mutations that cause the constitutive expression of a gene under the control of the two-component system, or a mutation that inactivates a gene under the control of the two-component system, e.g., the pag gene, reduce virulence.

Constitutive mutations in two-component regulators can be identified by the use of a strain containing a recorder gene fusion to a gene regulated by the two-component system. Such gene fusions would most typically include DNA encoding the lacZ gene or AP fused to a gene under the control of the two-component system. Strains containing fusions that are (as compared to wild type or parental strains) highly expressed in an unregulated fashion, i.e., constitutive, can be detected by increased color on chromogenic substrates for the enzymes. To detect constitutive mutations a cloned virulence regulator could be mutagenized e.g., by passage through an E. coli strain defective in DNA repair or by chemical mutagenesis. The mutated DNA for the regulator would then be transferred to the strain containing the gene fusion and constitutive mutations identified by the high gene fusion expression (blue color in the case of a lacZ fusion grown on media containing X-gal). Constitutive mutations in a component of a two-component regulatory system could also be made by in vitro mutagenesis after other constitutive mutations have been sequenced and a specific amino acid change responsible for the constitutive phenotype identified. Putting several amino acid changes that all result in a PhoP constitutive phenotype would result in a decreased frequency of reversion by spontaneous base changes. A constitutive mutation could also be constructed by deletion of the portion of the amino terminus of the phospho-accepting regulator which contains the phosphoacceptor domain e.g., deletion of sequences encoding amino acids amino terminal to amino acid 119 in the phoP gene or deletion of analogous phospho accepting sequences in genes of other two-component regulatory systems. This could result in a conformational change similar to that induced by phosphorylation and result in increased DNA binding and transcriptional activation.

Attenuation of virulence: deletion in the phoP/phoQ regulon

As discussed above, the PhoP regulon is essential to full virulence of Salmonella. This regulon is composed of two genes, PhoP and PhoQ located in an operon, and over 40 genes they positively and negatively regulate (pag and prg, respectively).

PhoP null S. typhimurium mutants have been demonstrated to be markedly attenuated and also effective vaccine strains when studied in the BALB/c mouse model of typhoid fever. This phenotype is likely the result of multiple, phoP-activated virulence genes, as transposon insertions in multiple different phoP-activated genes have been independently demonstrated to decrease S. typhimurium virulence. S. typhimurium mutants deleted for genes essential to aromatic amino acids (aroA null or aroC/aroD null mutants) are also markedly attenuated in the mouse model. However, testing of aroC/aroD mutants in humans has shown that although these strains are immunogenic, bacteremias and side effects such as fever have been noted at doses as low as 10.sup.5 to 10.sup.7 organisms administered as a single oral dose (Hone et al., J. Clin. Invest. 90:412-420).

It has now been found that a large deletion in a global regulator of Salmonella virulence, i.e., the PhoP/PhoQ operon, significantly decreases the virulence of the bacteria. This mutation, the result of a 1 kB deletion of DNA within the phoP/phoQ locus, was initially made in S. typhimurium and subsequently transferred via homologous recombination to S. typhi. In order to confer an even greater margin of safety in construction of this vaccine, it was created in a strain background deleted for genes essential to aromatic amino acids and carrying the histidine G46 mutation, a mutation rendering the organism auxotrophic for histidine. The resulting strain, S. typhi TyLH445, offers several advantages over existing vaccine candidates, most notably, immunogenicity without transient bactermia.

Use

The Salmonella cells of the invention are useful as sources of immunological protection against diseases, e.g., typhoid fever and related diseases, in an animal, e.g., a mammal, e.g., a human, in particular as the basis of a live-cell vaccine capable of colonizing the inoculated animal's intestine and provoking a strong immune reaction. Appropriate dosages and conditions of administration of such a live, attenuated vaccine are known in the art, e.g., as described in Holem et al., Acute Enteric Infections in Children, New Prospects for Treatment and Prevention, (1981) Elsevier/North-Holland biomedical Press, Ch. 26, pp. 443 et seq. (Levine et al.), hereby incorporated by reference, and are described in the examples below.

Advantages

One advantage of the invention is that the bacterial cells are attenuated as a result of a mutation(s), i.e., the phoP/phoQ operon, that directly affect a virulence pathway. Another advantage is that the bacterial cells have mutations in two completely different attenuating genes, i.e., the aromatic amino acid synthesis pathway (Aro), and in an operon important to Salmonella virulence (PhoP/Q). As a result, the bacteria appear to be extremely attenuated; doses as high as 1.times.10.sup.9 cfu appear to be very safe. Other vaccines under development, such as CVD 908, have caused some systemic symptoms, e.g., fever or bacteremia, at doses as low as 1.times.10.sup.7 cfu.

In addition to the phoP/phoQ deletion and the AroA-mutation, the bacterial cells of the invention may also contain a histidine mutation to further alternate virulence, although absence of the histidine mutation may improve immunogenicity. The bacterial cells of the invention are the most promising vaccine candidates to date because they are strongly immunogenic and safe, i.e., extremely attenuated.

EXAMPLE 1

Construction of vaccine strain

The bacterial cells of the invention were made by deleting approximately 1 kb of DNA in the phoP/phoQ regulon.

PhoP/phoQ deleted suicide vectors were constructed using methods known in the art. A DNA fragment containing the phoP/phoQ locus was obtained by PCR using wild type S. typhimurium chromosomal DNA as a template. PCR primers flanking the phoP/phoQ locus were engineered to contain terminal restriction enzyme recognition sites, e.g., recognition site for EcoRI, to facilitate subsequent cloning. Following amplification, the PCR product was digested with EcoRI and cloned into the EcoRI site in the polylinker of a high copy vector. The plasmid containing the phoP/phoQ DNA fragment was named pLH356.

Sequence analysis and restriction mapping of the phoP/phoQ locus revealed four HpaI sites within the locus; no HpaI sites were found in the vector. To create an internal deletion within the phoP/phoQ locus, pLH356 DNA was cut to completion with HpaI, and religated, to yield with an internal deletion from nucleotides 376-1322 (pLH418). This deletion was confirmed by restriction digestion of the plasmid.

A DNA fragment containing the internally deleted phoP/phoQ locus was excised from pLH418 using the SacI/SphI restriction sites within the polylinker region of the vector. This fragment was cloned into compatible sites in the plasmid CVD442, which carries the sacB gene to allow positive selection for allelic exchange. The resulting suicide vector was called pLH423.

pLH423 was transformed into E. coli lambda pir SY327, and subsequently into E. coli lambda pir SM10 (strain LH425). E. coli strain LH425 was mated with S. typhimurium strain CS019. Single recombinants carrying plasmid sequences integrated onto the S. typhimurium chromosome were selected by plating on agar containing ampicillin and chloramphenicol (Strain LH428). These strains were confirmed to be ampicillin resistant and sucrose sensitive, i.e., death on 20% sucrose plates containing no NaCl when incubated at 30.degree. C. These data confirm the integration of plasmid sequences into the Salmonella chromosome.

A P22 bacteriophage lysate was made from strain LH428; phage particles were concentrated 20.times. by high speed centrifugation and transduced into S. typhi strain 522Ty2 (a strain with a deletion in the aroA gene, and the G646 mutation which renders the organism auxotrophic for histidine). Single recombinant S. typhi organisms were selected by plating on LB plates supplemented with aromatic amino acids, cystine, histidine, and ampicillin (strain LH435).

Strain LH453 was grown with aromatic amino acids, cystine, and histidine (but without ampicillin) to mid logarithmic growth phase. Serial dilutions were plated on LB 20% sucrose plates lacking NaCl, and on LB plates lacking NaCl. The number of bacteria that grew on plates without sucrose was greater than the number that grew on sucrose-supplemented plates by a factor of three logs. These data suggest that many colonies lost plasmid sequences containing the sacB gene.

Multiple colonies from the sucrose selection were picked and confirmed to be ampicillin sensitive and sucrose resistant. Chromosomal DNA from approximately 10 colonies was purified and subjected to Southern blot analysis, utilizing the 2.3 kb fragment of wild type phoP/phoQ as a probe.

Southern blotting revealed the loss of two HpaI sites and an XmnI site known to be within the 1 kb deleted fragment of phoP/phoQ in several strains. One of these strains was designated TyLH445.

EXAMPLE 2

in vitro evaluation of TyLH445

TyLH445 was extensively characterized in vitro using standard clinical microbiological tests. The nutritional requirements of TyLH445 were evaluated. TyLH445 did not grow on M-9 plates unless supplemented with aromatic amino acid mix, cystine (S. typhi grows better with cystine), and histidine. These data confirmed that TyLH445 was AroA-, His-.

TyLH445 was found to agglutinate with polyclonal serum against Salmonella and polyclonal serum against S. typhi Vi antigen. Group D agglutination was found to be variable, perhaps due to excess Vi antigen. TyLH445 was also found to be indole negative (as are all Salmonellae), and to produce very little hydrogen sulfide (as do many S. typhi). Biochemical testing utilizing both the VITEK system as well as the BBL Crystal Enteric organism identification system was also carried out. These data indicated that the TyLH445 strain was S. typhi.

Growth characteristics of TyLH445 were also evaluated. TyLH445 was found to grow just as quickly as its parent, 522Ty2, (phoP/phoQ locus intact). Growth in vitro was measured in aromatic amino acid/histidine/cystine-supplemented Luria broth at 37.degree. C. Growth curves of the parent and vaccine strain were found to be essentially identical (see FIG. 10).

Standardized clinical testing methods were use to determine antibiotic sensitivity. TyLH445 and the parent strain, 522Ty2, were found to be sensitive to ampicillin, trimethoprim-sulfamethoxazole, ciprofloxacin, aminoglycosides, and third generation cephalosporins. No difference in zone sizes was detected between the parent and vaccine strains, suggesting that no other antibiotic resistance mechanisms, e.g., modification of antibiotic transport systems, or modification of the cell wall of the bacterium, were affected by introduction of the mutated phoP/phoQ locus into S. typhi.

The phoP/phoQ HpaI deletion mutants were tested for defensin sensitivity, a phenotype of PhoP null mutants. Defensin sensitivity assays were performed as follows.

Liquid cultures of strains to be tested were grown overnight. Cultures were then diluted 1:200, and grown to an optical density (OD.sub.600) of approximately 0.2, after which the cells were diluted to concentration of approximately 1.times.10.sup.5 organisms per 0.05 ml.

Two reactions were carried out for each strain: (1) vehicle alone (0.01% acetic acid in sterile water) and (2) defensin NP-1 solution (70 ug/ml in 0.01% acetic acid). An equal volume of bacterial suspension in tryptone was added and the test tubes were incubated on a roller at 37.degree. C. for 2 hours. The final volume in each reaction tube was 0.1 ml, making the final concentration of defensin 35 ug/ml.

Defensin is inactivated by the high salt and high protein concentration present in bacterial growth media, e.g. LB broth. Thus, defensin activity was stopped by adding 900 ul of Luria broth to each tube. Serial dilutions of each tube were plated and cfu/ml was determined for both the control tube and treatment tube for each strain. Results were expressed as log of bacteria killed for each strain. Typically, 1.0-1.5 log of wild type bacteria were killed. PhoP null mutants generally exhibit 2-4 logs of killing. Since strains with slower growth rates appear less susceptible to defensin killing, the growth rate of each strain tested in the defensin sensitivity assay was measured. Strains with similar growth rates were compared in the defensin sensitivity assay.

The HpaI deletion was evaluated both in an S. typhimurium background and in the S. typhi background. In both backgrounds, the deletion mutation conferred sensitivity to rabbit defensin NP-1 at a concentration of 35 ug/ml. See FIG. 11 and FIG. 13. The difference between PhoP+ and HpaI deleted PhoP null mutants was less pronounced in the S. typhi strain, an effect that may reflect the slower growth rate of the less hardy S. typhi strain compared to the S. typhimurium strain which lacks the additional auxotrophies.

The state of phoP activation in bacteria with the HpaI phoP/phoQ deletion was tested utilizing a LacZ recorder gene fused to phoP-activated gene B (pagB). Since the efficiency of transduction utilizing P22 in S. typhi is low, these studies were performed in S. typhimurium rather than S. typhi. PhoP activation was found to be 40-60 Miller units (Miller et al., 1972, Experiments in Molecular Genetics, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y., pp. 352-355) in the presence of an intact phoP/phoQ locus, and just barely detectable in strains with the HpaI deletion (3cfusee FIG. 12).

EXAMPLE 3

In vivo evaluation of S. typhimurium HpaI deleted strain

As S. typhi strains are not pathogenic for mice, the HpaI phoP/phoQ deletion mutation was evaluated in both wild type and aroA- S. typhimurium. Female BALB/c mice were injected intraperitoneally with various dilutions of S. typhimurium LH430, a wild type S. typhimurium carrying the HpaI deletion. The LD.sub.50 of this strain was determined to be between 8.2.times.10.sup.5 and 8.2.times.10.sup.6. (All mice receiving 8.2.times.10.sup.5 cfu survived, and all receiving 8.2.times.10.sup.6 died.) These data are consistent with the LD.sub.50 data obtained with strains harboring transposon insertions at the phoP/phoQ locus.

Immunogenicity of the HpaI phoP/phoQ deletion was evaluated in S. typhimurium aroA::tet (LH481), a strain comparable to the human vaccine strain. Mice were inoculated intraperitoneally with 2.3.times.10.sup.5 and 2.3.times.10.sup.6 cfu of LH481 (4 mice per vaccine dose), and challenged 30 days later with 30.times. the LD.sub.50 of wild type organisms. All mice but one mouse survived. The mouse that died was in the group that received the lower vaccine dose. No animal receiving the higher vaccine dose became ill.

EXAMPLE 4

Phase I study human studies

The vaccine strain was administered to human volunteers at doses of 1.times.10.sup.5 to 1.times.10.sup.10 cfu/single oral dose. Two volunteers received each dose; 3 volunteers were given a dose of 1.times.10.sup.8 cfu/ml. Volunteers were evaluated at various time points following administration of the vaccine.

Safety

To detect the presence of the vaccine strain in patient blood, Bactec blood cultures were performed in duplicate on days 4, 6, 8, 10, 12 after taking vaccine. Bacteremia was not detected in any of the volunteers.

Thirteen adult human volunteers have received escalating single oral doses of this new attenuated typhoid fever vaccine. No individuals have had side effects of any sort. Specifically, there have been no fevers, no gastrointestinal symptoms, and no constitutional symptoms. Volunteers have been subjected to serial blood cultures on a preset schedule after receiving the oral vaccine 2 sets of BACTEC blood cultures performed on each of days 4, 6, 8, 10 and 12 after receiving the vaccine, and no positive blood cultures have been noted. Volunteers have been followed up at 2 months after receiving the vaccine, and no late symptoms have been reported.

Colonization

Stool samples were tested for the presence of the vaccine strain TyLH455 using methods known in the art. Primary stool was evaluated for the presence of the vaccine strain on culture plates. In some cases, it was necessary to enrich stool samples for the vaccine strain by incubating the stool overnight in BBL Selenite F broth supplemented with Aro/His/Cystine in order to detect the bacteria. This medium is somewhat inhibitory for E. coli and but promotes Salmonella growth.

Volunteers have been colonized for various time periods from 1-6 days after receiving the vaccine. With the highest doses (10.sup.9 or 10.sup.10) volunteers have had positive primary culture plates in the initial 1-3 days post vaccination, whereas at lower doses, only selenite enrichment broth cultures (selective medium for Salmonella which inhibits other enterics) have been positive for the vaccine organism. No volunteer studied thus far has had prolonged carriage of the vaccine organism at 2 months of followup.

Immunogenicity

Induction of an immune response to the vaccine strain was measured by standard ELISA assays. Sera was collected from volunteers 0, 7, 14, 21, and 28 days after receiving a single oral dose of the vaccine. ELISA assays were carried out using whole bacteria TyLH445 and S. typhi LPS (SIGMA, St. Louis, Mo.) as antigens. Day 0 serum from each volunteer was used as an internal negative control. Convalescent sera from patients previously infect with wild type S. typhi (most from Mexico) were used as positive controls.

Several volunteers had documented seroconversion at 21 days after receiving the vaccine, as measured by ELISA in which IgG antibodies directed against whole vaccine organisms or against S. typhi LPS were detected. Sera taken from patients prior to administration of the vaccine (preimmune sera) were tested and the data used to establish a baseline. Patient sera taken at various time points after vaccination were considered positive if the test results were 0.2 ELISA OD units greater than that of the preimmune serum.

Other Embodiments

Other embodiments, e.g., strains of Salmonella which contain only a deletion in the phoP/phoQ regulatory locus to attenuate virulence, and strains which, in addition to a phoP related mutation or genetic alteration, also contain an attenuating mutation in another gene, e.g., cya gene (adenylate cyclase) or crp gene (adenylate cyclase receptor), are also within the claims.

Claims

What is claimed is:

1. A Salmonella cell, the virulence of which is attenuated by a mutation in one or more genes selected from the group consisting of pagJ, pagK, pagM, and msgA.

2. The Salmonella cell of claim 1, wherein said mutation is in pagJ.

3. The Salmonella cell of claim 1, wherein said mutation is in pagK.

4. The Salmonella cell of claim 1, wherein said mutation is in pagM.

5. The Salmonella cell of claim 1, wherein said mutation is in msgA.

Patent Citations (9)

Patent	Date	Inventor
US4550081	1985-10-01	Stocker
US4735801	1988-04-01	Stocker
US4764370	1988-08-01	Fields et al.
US4837151	1989-06-01	Stocker
US5077044	1991-12-01	Stocker
US5210035	1993-05-01	Stocker
US5424065	1995-06-01	Curtiss, III et al.
WOXWO90/11687	1990-10-01
WOXWO92/17785	1992-10-01

Non-Patent Literature (25)

Miller et al. Proc. Natl. Acad. Sci. USA 86 5054-5058 (1989) A Two-Component Regulatory System (phoP phoQ) Controls . . . .
Crosa et al., "Molecular Relationships Among the Salmonelleae," J. of Bacteriology, 115(1):307-315.
Groisman et al., "Salmonella typhimurium phoP virulence gene is a transcriptional regulator," Proc. Natl. Acad. Sci. USA, 86:7077-7081 (1989).
Miller et al., "Salamonella Vaccines With Mutations in The PhoP Virulence Regulon," Res. Microbiol. 141:817-821 (1990).
Pulkkinen et al., "A Salmonella typhimurium Virulence Protein Is Similar to a Yersinia enterocolitica Invasion Protein and a Bacteriophage Lambda Outer Membrane Protein," J. of Bacteriology 173(1):86-93, 1991.
Supplementary European Search Report, Application No. EP 92 90 4089, date 15 Jun. 1995.
PCT Written Opinion, Application No. PCT/US94/07658, mailed 17 Jul. 1995.
Alpuche-Aranda, C., et al., "Salmonella Stimulate Macrophage Macropinocytosis and Persist Within Spacious Phagosomes", 1994, J. Exp. Med., 179:601-08.
Behlau, I., et al., "A PhoP-Repressed Gene Promotes Salmonella typhimurium Invasion of Epithelial Cells", 1993, J. Bacteriology, 175:4475-84.
David, S., et al., "Leuconostoc Lactis .beta.-Galactosidase Is Encoded by Two Overlapping Genes", 1992, J. Bacteriology, 174:4475-81.
Edwards, M., et al., "Construction of .DELTA.aroA his .DELTA.pur Strains of Salmonella typhi", J. Bacteriology, 170:3991-95.
Groisman E., et al., "Horizontal Transfer of a Phosphatase Gene as Evidence for Mosaic Structure of the Salmonella Genome", 1992, EMBO J., 11:1309-16.
Groisman E., et al., "Cognate Gene Clusters Govern Invasion of Host Epithelial Cells by Salmonella typhimurium and Shigella flexneri", 1993, EMBO J., 12:3779-87.
Gulig, P., et al., "Identification, Genetic Analysis and DNA Sequence of a 7.8-kb Virulence Region of the Salmonella typhimurium Virulence Plasmid", 1992, Molecular. Microbiol., 6:1395-411.
Heffernan, E., et al., "The Salmonella typhimurium Virulence Plasmid Complement Resistance Gene rck is Homologous to a Family of Virulence-Related Outer Membrane Protein Genes, Including pagC and ail", 1992, J. Bacteriol., 174:84-91.
Hone, D., et al., "Evaluation in Volunteers of a Candidate Live Oral Attenuated Salmonella typhi Vector Vaccine", 1992, J. Clin. Invest., 90:412-20.
Levine, M., et al., "Safety, Infectivity, Immunogenicity, and In Vivo Stability of Two Attenuated Auxotrophic Mutant Strains of Salmonella typhi, 541Ty and 543Ty, as Live Oral Vaccines in Humans", 1987, J. Clin. Invest., 79:888-902.
Mekalanos, J., "Environmental Signals Controlling Expression of Virulence Determinants in Bacteria", 1992, J. Bacteriology, 174:1-7.
Miller, S., et al., "The PhoP Virulence Regulon and Live Oral Salmonella Vaccines", 1993, Vaccine, 11:122-25.
Miller, S., et al., "Constitutive Expression of the PhoP Regulon Attenuates Salmonella Virulence and Survival Within Macrophages", 1990, J. Bacteriology, 172:2485-90.
Miller, V. et al., "An Unusual pagC::TnphoA Mutation Leads to an Invasion- and Virulence-Defective Phenotype in Salmonellae", 1992, Infection and Immunity, 60:3763-70.
Miller, S., "PhoP/PhoQ:macrophage-specific Modulators of Salmonella Virulence?", 1991, Molecular Microbiology 5:2073-78.
Sugino, H., et al., "A Monoamine-Regulated Klebsiella Aerogenes Operon Containing the Monoamine Oxidase Structural Gene (maoA) and the maoC Gene", 1992, J. Bacteriology, 174:2485-92.
Tacket, C., et al., "Comparison of the Safety and Immunogenicity of .DELTA.aroC .DELTA.aroD and .DELTA.cya .DELTA.crp Salmonella typhi Strains in Adult Volunteers", 1992, Infection and Immunity, 60:536-41.
Tacket, C., et al., "Clinical Acceptability and Immunogenicity of CVD 908 Salmonella typhi Vaccine Strain", 1992, Vaccine, 10:443-46.