INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial live-vector vaccines represent a vaccine development strategy that offers exceptional flexibility. In this approach, genes that encode protective antigens of unrelated bacterial, viral, or parasitic pathogens are expressed in a live vector that carries the foreign antigens to the immune system, thereby eliciting an appropriate immune response. The attenuated Salmonella typhi vaccine strain CVD 908-htrA is a particularly attractive live vector in that it is well tolerated and elicits a broad immune response to S. typhi antigens, which includes intestinal soluble immunoglobulin A antibodies, serum immunoglobulin G antibodies, and cellular immune responses (47, 48). In addition, genetic methods have been developed to express foreign antigens within attenuated S. typhi vaccine strains, and a murine intranasal model has been developed as a practical animal model for examining the immunogenicity and protective efficacy of a wide variety of heterologous antigens within S. typhi-based live-vector vaccines, prior to initiating clinical trials (3, 16).

The efficacy of any bacterial live-vector vaccine rests with its ability to present sufficient foreign antigen to the human immune system to initiate the desired protective immune response. Controlled expression of heterologous antigens from multicopy expression plasmids represents one obvious solution for synthesis of high levels of antigen within live vectors. However, these plasmids may become unstable in vivo, resulting in the loss of foreign genes and a decrease in the intended immune response.

One method of enhancing the inheritance of expression plasmids by live vectors involves construction of a "balanced-lethal" system for plasmids expressing heterologous antigens (34). In a plasmid-based balanced-lethal system, plasmids replicating in the cytoplasm of the bacterium express a critical protein required by the bacterium to grow and replicate; loss of such plasmids removes the ability of the bacterium to express the critical protein and results in cell death. This phenomenon of plasmid loss during bacterial replication, which results in the death of any plasmidless bacterium, is also referred to as "postsegregational killing." Such a system has been successfully employed in Salmonella typhimurium and is based on expression of the asd gene encoding aspartate -semialdehyde dehydrogenase (Asd) (15, 34). Asd is a critical enzyme involved in the synthesis of structural components essential for the formation of the cell wall in gram-negative bacteria. Therefore, loss of plasmids encoding such a critical enzyme would be lethal for any bacterium incapable of synthesizing Asd from the chromosome. Although the asd balanced-lethal system has been successfully employed in attenuated S. typhimurium-based live-vector strains for immunization of mice with a variety of prokaryotic and eukaryotic antigens (12, 28, 45), use of this method for stabilizing plasmids within attenuated S. typhi vaccine strains has, to date, been unsuccessful (47).

Here, we present the design and initial testing of a novel set of isogenic multicopy expression plasmids into which we have incorporated a noncatalytic postsegregational killing function, coupled with both active (14, 27) and passive plasmid partition functions (1, 31, 54), to provide a plasmid maintenance system designed to optimize expression of heterologous antigens within CVD 908-htrA for delivery to the human immune system. Since this method of improving plasmid maintenance involves no chromosomal mutagenesis of the live vector strain, in principle, such stabilized plasmids can be introduced into any live vector strain to improve the immunogenicity of heterologous antigens expressed, without additional genetic manipulations.

The approach is based on the use of the naturally occurring hok-sok postsegregational killing system residing on the R factor pR1 (19, 20). The hok-sok system is a two-component toxin-antitoxin system in which hok encodes a lethal pore-forming Hok protein. Synthesis of Hok is blocked by hybridization of a small antisense sok mRNA to hok mRNA, preventing translation and synthesis of Hok. However, sok mRNA is highly susceptible to degradation by nucleases, and its protective intracellular concentration must be maintained by constitutive transcription from resident plasmids carrying hok-sok. Therefore, bacteria that spontaneously lose such plasmids are postsegregationally killed because existing levels of the protective sok mRNA rapidly drop and levels of the more stable toxin-encoding hok mRNA quickly lead to Hok synthesis and cell death.

Inheritance of these expression plasmids has been enhanced through insertion of at least one partition function. A passive par locus from pSC101 (4, 31, 32, 52) was tested, in which no de novo partitioning proteins were encoded. In addition, the parA centromere-like active partitioning system from pR1 (14, 27) was investigated, since the combination of hok-sok and parA naturally occurs within pR1 and was not expected to present any compatibility complications. Plasmid maintenance was monitored by using flow cytometry to measure synthesis of a variant green fluorescent protein (GFPuv) test antigen within plasmid-bearing CVD 908-htrA. We show that although individual maintenance functions contribute to various degrees to the observed stability of expression plasmids, the highest levels of sustained synthesis of the test heterologous antigen GFPuv were detected from expression plasmids carrying the full complement of maintenance functions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and culture conditions. All plasmid constructions were recovered in Escherichia coli DH5 (Gibco BRL). Construction of the hok-sok gene cassette used pR1 template DNA isolated from E. coli J53(pR1), a generous gift from James B. Kaper. The live vector S. typhi CVD 908-htrA is an auxotrophic derivative of the wild-type strain Ty2 with deletions in aroC, aroD, and htrA (48). All strains used in this work were grown in media supplemented with 2,3-dihydroxybenzoic acid (DHB) as previously described (16, 26). When grown on solid medium, plasmid-bearing strains of CVD 908-htrA were streaked from frozen (70°C) master stocks onto 2× Luria-Bertani (LB) agar containing 20 g of Bacto Tryptone, 10 g of Bacto yeast extract, and 50 mM NaCl (2× LB agar) plus carbenicillin at a concentration of 50 µg/ml. Plates were incubated at 30°C for 24 to 36 h to obtain isolated colonies ~2 mm in diameter; strains were incubated at 30°C to minimize the toxicity of GFPuv expression in CVD 908-htrA.

Molecular genetic techniques. Standard techniques were used for the construction of the plasmids represented here (42). Unless otherwise noted, native Taq DNA polymerase (Gibco BRL) was used in PCRs. S. typhi strains were prepared for electroporation of recombinant plasmids after being harvested from Miller's LB broth (Gibco BRL) supplemented with DHB; after pelleting of bacteria, the cells were washed three times with one culture volume of sterile distilled water and resuspended in sterile distilled water to a final volume of 1/100 of the original culture volume. Electroporation of strains was performed in a Gene Pulser apparatus (Bio-Rad) set at 2.5 kV, 200 , and 25 µF. Following electroporation, bacteria were repaired with SOC medium (Biofluids) and incubation at 37°C and 250 rpm for 45 min; bacteria were then plated on 1× LB medium containing DHB plus 50 µg of carbenicillin per ml, and incubated at 30°C for 24 h. Isolated colonies were then swabbed onto supplemented 2× LB medium and incubated at 30°C for 16 h. Frozen master stocks were prepared by harvesting bacteria into SOC medium without further supplementation and freezing at 70°C.

Construction of expression plasmids. (i) Construction of pJN1 and pJN2. All primers used in this work are listed in Table 1; essential plasmids created from these primers are listed in Table 2 to illustrate the flow of logic in designing the final isogenic expression plasmids. The expression plasmids constructed for these studies are composed of three basic cassettes encoding (i) expression of a heterologous antigen, (ii) a plasmid origin of replication, and (iii) selection and maintenance functions. To accomplish this, a basic replicon was constructed in which these cassettes were separated by unique restriction sites. pTETnir15 (Table 2) (38) was reengineered such that the oriE1 origin of replication and bla gene were separated by a unique SpeI site. Toward this end, an oriE1 cassette was synthesized by PCR with Vent polymerase with primers 1 and 2 and pCVD315 (17) as the template. The resulting 735-bp fragment carries engineered SpeI and BglII sites 5' proximal to the promoter controlling transcription of RNA II, and an engineered AvrII site 675 bases from these sites. A separate PCR was carried out by using primers 3 and 4 to create a 1,234-bp bla cassette containing an engineered XbaI site 5' proximal to the original EcoRI site. The products from these two PCRs were gel purified and used in an overlapping PCR with primers 1 and 4 to yield a final 1,916-bp oriE1-bla fragment which was self-ligated to create pJN1. The P_nir15-toxC fragment from pTETnir15 was excised as an EcoRI (partial digestion)-AvaI fragment, in which the AvaI terminus was polished, and inserted into the multiple cloning region from pSL1180 (9) cleaved with EcoRI and StuI; this cassette was then reexcised as an EcoRI (partial digestion)-AvrII fragment and inserted into pJN1 cleaved with EcoRI-AvrII, creating pJN2 (Table 2).

(ii) Construction of pGFPompC. To facilitate screening of a functional osmotically regulated P_ompC allele from E. coli, an aphA-2 cassette was constructed, encoding resistance to the aminoglycosides neomycin and kanamycin (44). A PCR was carried out with primers 5 and 6 with the template pIB279 (5) to generate a 1,044-bp product, from which a promoterless 903-bp aphA-2 BglII-NheI fragment was cleaved for replacement of a BglII-NheI toxC cassette encoding fragment C of tetanus toxin in pTETnir15. The anaerobically regulated P_nir15 promoter was replaced with a 459-bp EcoRI-BglII P_ompC allele constructed with primers 7 and 8 with chromosomal template DNA from E. coli DH5 to create pKompC. After confirming osmotic induction of P_ompC by examining the increase in resistance to kanamycin with increasing osmolarity, the aphA-2 cassette was then replaced with a gfpuv gene encoding a prokaryotic codon-optimized GFPuv allele (Clontech) (13). The gfpuv gene was recovered by PCR with primers 9 and 10 with the template pGFPuv to generate a 751-bp BglII-NheI fragment, which was inserted into pKompC, to generate pGFPompC. Colonies were screened for functional GFPuv, and the brightest colonies were then examined for induction of fluorescence with increasing concentrations of NaCl. A P_ompC1-gfpuv cassette (see Results below) was cleaved from pGFPompC1 as an EcoRI-NheI fragment and inserted into a derivative of pJN2 cleaved with EcoRI-NheI to create pJJ4.

(iii) Construction of pNRB1, pGEN2, and pGEN3. Since it was intended that copy number not be influenced by transcription originating from promoters outside the origin of replication, it was necessary to ensure that all replication cassettes were flanked at both ends by transcription terminators. Because the origin and antigen cassettes of pJN2 are separated by the trpA terminator, it was only necessary to insert one additional terminator between the origin and bla cassettes.

(iv) Construction of pJN5, pGEN51, and pGEN91. The principle set of isogenic expression plasmids, to which individual elements of a plasmid maintenance system were sequentially added, was composed of pGEN51 (containing oriE1) and pGEN91 (containing ori15A). The basic replicon from which these two plasmids were constructed was pJN5, which was assembled by cleaving the P_ompC-gfpuv cartridge as an EcoRI-NheI fragment from pGFPompC to replace the P_nir15-toxC cassette of pJN2. Construction of pGEN51 was then accomplished by removal of the replication cassette from pGEN2 as a BamHI fragment and replacement of the origin of replication within pJN5 digested with BglII and BamHI, thereby regenerating the gfpuv gene. Construction of pGEN91 was accomplished in an identical manner by excision of the origin cassette from pGEN3 as a BamHI fragment and insertion into pJN5 (Fig. 1 and Table 2).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Genetic maps of a representative set of isogenic expression plasmids, carrying an ori15A origin of replication, which differ only in the introduction of increasingly complex combinations of maintenance functions. The plasmids pGEN111, pGEN121, pGEN193, and pGEN222 contain maintenance functions encoded by the hok-sok, hok-sok plus par, hok-sok plus parA, and hok-sok plus par plus parA loci, respectively. Other expression plasmids were constructed by replacing each ori15A origin of replication cassette (copy number, ~15) with an oriE1 origin (copy number, ~60); a list of the plasmids used in this work is given in Table 2. The key restriction sites shown represent unique sites in the expression plasmids. PompC1, modified osmotically controlled ompC promoter from E. coli; gfpuv, gene encoding the prokaryotic codon-optimized GFPuv; T1, transcriptional terminator from the rrnB rRNA operon of E. coli; par, passive partitioning system from pSC101; bla, -lactamase gene conferring resistance to carbenicillin; hok-sok, postsegregational killing locus from the multiple antibiotic resistance R plasmid pR1; parA, active partitioning system from pR1.

(v) Construction of pJN6, pGEN71, and pGEN111. The hok-sok locus was then excised as an XbaI-SalI fragment from pGEN2 and inserted into pJN5 cleaved with XbaI and SalI, again regenerating the gfpuv gene to create pJN6. Construction of pGEN71 and pGEN111 was then carried out exactly as described for pGEN51 and pGEN91 by insertion into pJN6 of origin cassettes as BamHI fragments from pGEN2 and pGEN3, respectively (Fig. 1 and Table 2).

(vi) Construction of pJN7, pGEN84, and pGEN121. Construction of oriE1 and ori15A expression plasmids containing a plasmid maintenance system, composed of both a postsegregational killing system and at least one partition function, was first attempted by using the par function from pSC101. A 377-bp BamHI-BglII fragment was synthesized with primers 18 and 19 with pSC101 template DNA; this fragment was inserted into pJN6 cleaved with BglII to create pJN7. As in the constructions above, origin cassettes from pGEN2 and pGEN3 were then excised as BamHI fragments and inserted into pJN7 digested with BglII and BamHI to create pGEN84 and pGEN121.

(vii) Construction of pJN8, pGEN183, pGEN193, pGEN211, and pGEN222. The final expression plasmids were constructed by introduction of the parA active partitioning locus from pR1. As with hok-sok, initial attempts at recovering the parA locus after PCR were unsuccessful. It was necessary to use overlapping PCR to generate an aphA-2-parA cassette, in which aphA-2 and parA were divergently transcribed and separated by XbaI and XhoI sites, to enable subcloning of the parA locus. A 1,737-bp parA fragment was synthesized by using primers 20 and 21 with pR1 template; a 1,076-bp aphA-2 fragment was recovered in a separate PCR by using primers 22 and 23 with pIB279 as the template. The products from these two PCRs were then used in an overlapping PCR with primers 21 and 23 to yield the final 2,743-bp aphA-2-parA fragment. This fragment was inserted as a 2,703-bp EcoRI-SpeI fragment into pJN6. The parA cassette was then reexcised as an XhoI fragment and inserted again into pJN6 cleaved with XhoI, regenerating the gfpuv gene and creating pJN8.

Flow cytometry. Quantitation of GFPuv and plasmid maintenance were analyzed by measuring the fluorescence of plasmid-bearing live vectors by using an Epics Elite ESP flow cytometer/cell sorter system (Coulter) with the argon laser exciting bacteria at 488 nm and emissions detected at 525 nm. Twenty-five-milliliter 1× LB cultures grown as described above were pelleted, and bacteria were resuspended in 1 ml of PBS. Cells were then diluted 1:1,000 in PBS prior to determination of viable counts and flow analysis. Forward versus side light scatter, measured with logarithmic amplifiers, was used to gate on bacteria. A minimum of 30,000 events were acquired from each sample at a collection rate of approximately 3,500 events per second. Mean fluorescence intensity for a given bacterial population was determined by using the Epics Elite Software Analysis Package. The fluorescence for plasmidless S. typhi CVD 908-htrA and E. coli DH5 strains was quantitated to establish that autofluorescence from either host strain was negligible and that fluorescence measured for plasmid-bearing strains was directly related to synthesis of GFPuv.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Rationale for construction of the expression plasmids. Although balanced-lethal plasmid stabilization systems based on expression of Asd have been created to maintain plasmids within Salmonella, a potential limitation of the asd system is its reliance on synthesis of an enzyme with catalytic activity. Since complementation with only a single copy of asd is sufficient to remove auxotrophy (15), it is not clear why all copies of a multicopy expression plasmid should remain stable and maintain maximum gene dosage, especially if they encode an especially problematic antigen which inhibits growth of the bacterium. Another potential limitation of the asd system is that it does not enhance the inheritance of resident plasmids, which continue to segregate randomly with or without the presence of the asd system. Therefore, if resident expression plasmids carrying asd genes are inherently unstable for some undetermined reason, they will be lost regardless of the requirement of the bacterium for Asd. Here, we present the design and initial testing of a set of isogenic multicopy expression plasmids into which we have incorporated a noncatalytic postsegregational killing function, coupled with both active (14, 27) and passive plasmid partition functions (1, 31, 54), to provide a plasmid maintenance system designed to optimize expression of heterologous antigens within CVD 908-htrA for delivery to the human immune system.

Osmotic control of P_ompC. It was intended that any promoter controlling transcription of a heterologous gene be responsive to an environmental signal of biological relevance. For the expression plasmids described here, an ompC promoter cassette (P_ompC) from E. coli was used, which is induced by increases in osmolarity. Construction of this cassette was based on the sequence of P_ompC published by Norioka et al. (37) and was engineered to control expression of a test antigen cassette containing the gfpuv allele encoding GFPuv; this P_ompC-gfpuv cassette was inserted into a derivative of pBR322 to create pGFPompC. During the visual screening of E. coli DH5(pGFPompC) colonies subilluminated with UV light, one very brightly fluorescing colony and another representative fluorescent colony were chosen for further study, designated clones 1 and 3, respectively. Upon purification of the plasmids involved, it was determined that clone 1 contained a plasmid that no longer carried a BglII site separating P_ompC and gfpuv, while clone 3 carried the expected BglII site. We examined the induction of GFPuv expression when clones 1 and 3 were grown on nutrient agar in the presence or absence of NaCl and determined by visual inspection that clone 3 displayed very little fluorescence when grown in the absence of NaCl but fluoresced brightly when plated on medium containing 300 mM NaCl (data not shown). Clone 1, however, had a higher background level of fluorescence when uninduced, but fluoresced intensely when induced with 300 mM NaCl. To rule out mutations within the gfpuv gene which might affect fluorescence, we replaced P_ompC from clone 1 with P_ompC from clone 3 and confirmed the expected decrease in fluorescence as judged by subillumination (data not shown). We therefore concluded that differences in observed fluorescence were controlled by two genetically distinct versions of the P_ompC promoter, which we designate as P_ompC1 (higher transcription levels with less osmotic control) and P_ompC3 (moderate transcription levels but more responsive to osmolarity); we have designated the plasmids containing these expression cassettes as pGFPompC1 and pGFPompC3, respectively.

View this table: [in this window] [in a new window]   TABLE 3.   Comparison of induction of PompC1 and PompC3, controlling expression of GFPuv, within the host strains E. coli DH5 and CVD 908-htrA

Stability of expression plasmids in the absence of selection. Since the broad objective of the research presented here is to develop a noncatalytic plasmid maintenance system to enhance the stability of multicopy expression plasmids encoding foreign antigens within CVD 908-htrA, we initiated experiments to monitor plasmid stability by quantitating expression of GFPuv by flow cytometry when strains were passaged in the absence of antibiotic selection. These experiments were designed to address three fundamental questions. (i) What is the effect of copy number on the stability of plasmids expressing GFPuv? (ii) What is the effect of the induction level of P_ompC1 on the stability of plasmids encoding a heterologous antigen such as GFPuv? (iii) How do the hok-sok, par, and parA maintenance functions affect plasmid retention, both as individual components and synergistically?

View larger version (34K): [in this window] [in a new window]   FIG. 2.   Flow cytometry histograms of GFPuv fluorescence for populations of attenuated S. typhi CVD 908-htrA containing expression plasmids with the ori15A origin of replication. Histograms are arranged in rows indicating analysis of bacterial populations grown under inducing conditions with either 50, 150, or 300 mM NaCl. The histogram columns indicate CVD 908-htrA carrying the expression plasmid listed at the bottom of the column, which contains the maintenance function(s) listed at the top of the column. Results are plotted as the log of GFPuv relative fluorescence intensity versus the number of fluorescing bacteria. The mean fluorescence intensity for a given subpopulation is indicated under each curve; these data are fully quantitated in Table 4.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The broad objective of the research presented here is to develop a plasmid maintenance system for the stabilization of multicopy expression plasmids encoding foreign antigens in an S. typhi live-vector vaccine strain, without additional modification of the chromosome. Attempts were made to enhance the maintenance of expression plasmids at two independent levels. First, dependence upon balanced-lethal maintenance systems that involve catalytic enzymes expressed from multicopy plasmids was removed; this was accomplished through incorporation into expression plasmids of a postsegregational killing system based on the noncatalytic hok-sok plasmid addiction system from the antibiotic resistance factor pR1. At least one naturally occurring plasmid partition function was also introduced into these expression plasmids, to potentially eliminate random segregation of such plasmids, thereby enhancing their inheritance and stability.

Although these expression plasmids are ultimately intended to express immunogenic and protective antigens for delivery to the human immune system, GFPuv was selected as a test reporter antigen, because quantitation of mean fluorescence in a population of growing live vectors could be used as a measure of the stability of resident plasmids within the live vector. All expression plasmids carried an identical antigen expression cassette, with a P_ompC1 allele controlling transcription and with translation optimized by incorporation of a consensus ribosome binding site. Because no catalytic activity is associated with the fluorescence of GFPuv, the level of fluorescence intensity measured by flow cytometry within individual bacteria could be correlated directly with gene dosage and copy number. In addition, use of an osmotically regulated ompC promoter allowed an assessment of plasmid stability and live-vector viability as increasing osmolarity induced higher levels of GFPuv synthesis and presumably higher levels of metabolic stress on the live vector. It was surprising and encouraging that although the P_ompC1 allele was engineered from the chromosomal locus of E. coli, it appeared to function more efficiently in S. typhi (Table 3).

The contributions of several plasmid maintenance functions to the stability of plasmids within CVD 908-htrA, growing in the absence of antibiotic selection, were examined. No combination of maintenance functions could stabilize plasmids containing oriE1 origins of replication; in fact, constructs containing maintenance functions were difficult to propagate even in the presence of antibiotic. These observations cast doubt upon the rationale for using higher-copy-number plasmids to optimize expression of heterologous antigens within the cytoplasm of S. typhi-based live vectors, a strategy that heretofore has been followed by other groups investigating salmonellae as live vectors (12).

Incorporation of plasmid maintenance functions into plasmids carrying an ori15A origin of replication was more encouraging. When live vectors carrying such plasmids were passaged without selection for 24 h at 37°C, the effects of various combinations of maintenance functions became apparent. In the absence of maintenance functions, the ori15A replicon pGEN91 was lost from greater than 85% of the population, regardless of the level of induction of P_ompC1 (Table 4 and Fig. 2). With incorporation of the hok-sok postsegregational killing locus in pGEN111, the percentage of bacteria expressing GFPuv tripled under all induction conditions, confirming the observations of others that the hok-sok locus enhances the stability of ori15A replicons (18, 20, 21). However, it was still noted that regardless of induction conditions, greater than 50% of the bacterial population no longer fluoresced. Since it was confirmed that at least a portion of this nonfluorescing population was still viable and lacked drug resistance, these data confirm previous reports (22, 39, 53) that the presence of a hok-sok postsegregational killing system is insufficient by itself to ensure that plasmidless viable bacteria will not arise in a growing population.

One possible mechanism that allows for escape from the influence of hok-sok involves spontaneous point mutations arising within the lethal Hok ORF, which could conformationally inactivate Hok and thereby allow plasmid loss to occur without lethality. This point emphasizes the requirement of multiple mechanisms for enhancing the stability of resident plasmids within growing bacteria; should one maintenance function become inactivated, the probability of other independent functions simultaneously becoming inactivated becomes vanishingly small. Indeed, such redundancy in maintenance functions is widespread within naturally occurring low-copy-number plasmids (36). For example, the E. coli sex factor F contains one active partitioning function (sop) and two killing systems (ccd and flm) (23, 29, 35, 51). Similarly, the drug resistance plasmid pR1 contains the active partitioning function parA, as well as the postsegregational killing system hok-sok; in addition, it carries yet another recently defined kis-kid killing system (7, 8, 41). We have demonstrated in the work reported here that insertion into multicopy ori15A replicons of a more complete maintenance system, composed of both a postsegregational system and two partition functions, dramatically improves the stability of these expression plasmids in the absence of selection, regardless of induction conditions for heterologous antigen expression. However, after passage without selection for 48 h, plasmids were eventually lost from the bacterial population, possibly due to escape from the lethality of Hok. This problem has recently been addressed by Pecota et al. (39), who reported that incorporation of dual killing systems significantly improved plasmid stability compared to the use of hok-sok alone; no partition functions were present in these plasmids. Perhaps inclusion of the kis-kid killing system, to more fully represent the complement of pR1 stability functions, may be required for optimal stability of higher-copy expression plasmids within S. typhi live vectors.

The efficiency of eliciting an immune response directed against a heterologous antigen will depend in part upon the ability of the live vector to present such antigens to the immune system. The ability of a live vector to present antigens will in turn depend upon the stability of multicopy expression plasmids that encode the heterologous antigens. However, we hypothesize that a significant metabolic burden is placed upon CVD 908-htrA carrying a multicopy expression plasmid; as copy number and/or level of gene expression increases, metabolic burden increases. Studies with E. coli have clearly established that plasmid-bearing bacteria grow more slowly than plasmidless bacteria (6, 30, 39, 46, 53). It has also been demonstrated that as copy number increases, the growth rate of such strains decreases; similarly, as induction of heterologous genes increases, growth rate decreases further (39, 53). Clearly, spontaneous plasmid loss would remove any metabolic burden and allow plasmidless bacteria to quickly outgrow the population of plasmid-bearing bacteria. Such a shift in antigen expression within a population of live-vector bacteria would be expected to reduce the efficiency of stimulating any immune response specific to the foreign antigen. Such reasoning suggests that the goal for an effective multivalent S. typhi-based live vector vaccine is to optimize viability by using stabilized lower-copy-number expression plasmids, capable of expressing high levels of heterologous antigen in response to an environmental signal likely to be encountered in vivo after the vaccine organisms have reached an appropriate ecological niche. We are currently testing this strategy by using the murine intranasal model to examine the immunogenicity of protein fusions involving fragment C of tetanus toxin fused at the carboxyl terminus to antigens from the malaria agent Plasmodium falciparum, expressed within CVD 908-htrA by using expression plasmids derived from pGEN222. It is hoped that such experiments will point the way toward development of single-dose, oral S. typhi-based live-vector vaccines capable of inducing protective immune responses against multiple unrelated human pathogens.