Salmonella Vaccines: Conduits for Protective Antigens

Infectious diseases have had deleterious effects on humankind, other animals, and plants from time immemorial and have shaped the history of life on earth. The necessity of defeating these diseases led to the development and evolution of host immune responses. For most of the history of life, individuals had to rely on the strengths of their immune systems to combat infections, but infectious diseases continue to adversely affect the human population. Worldwide, infectious diseases caused by microbes accounted for ∼7.6 million deaths in 2012 (1). In high-income countries, the main causes of death are chronic diseases, but in low-income countries, people predominantly die of infectious diseases [lower respiratory tract infections, diarrheal diseases, tuberculosis, HIV/AIDS, and malaria account for almost one third of all deaths in these countries (1)].

In the twentieth century, effective means to deal with infections were discovered and developed, in the form of antibiotics (2–8). Antibiotics were truly “wonder drugs,” enabling life-saving treatments of individuals suffering from acute infectious diseases caused by bacterial pathogens that had previously been fatal. However, with the improved health afforded by the availability of these wonder drugs came overuse and misuse of antibiotics. During the past 10–15 y, infectious diseases caused by pathogenic bacteria resistant to antibiotics usually used to treat them have been increasing at an alarming rate (9, 10). Writers in the popular press and scientific journals have noted that we are rapidly moving toward a time that will be like the preantibiotic era, in which survival will depend upon the robustness of one’s immune system (9–13). The increases in incidence of infectious diseases caused by antibiotic-resistant strains of bacterial pathogens are the consequences of multiple factors, including the misuse of antibiotics (using antibiotics to treat viral infections), lack of compliance by patients (failure to take antibiotics for the prescribed duration of treatment), and increased amounts of antibiotics in the environment (a result of the use of antibiotics in feed to promote rapid growth of agriculturally important animals and to treat bacterial infections of farm animals and plants) (9–13). Resistance to antiviral and antiparasite drugs is also increasing (9, 10). The high costs of discovery and development of new effective antimicrobial and antiviral drugs that are nontoxic to animals and humans, coupled with the propensity of pathogens to develop drug resistance, have become a hindrance to treatment of infectious diseases.

An alternative and much more cost-effective approach to combatting infections is to stimulate the immune system via vaccines against specific microbial pathogens. Since the times of Edward Jenner and Louis Pasteur (14), inoculation of individuals with weakened or closely related attenuated microbes or inactivated microbial toxins or viruses has been amply demonstrated to protect vaccinated individuals from infection by stimulating protective immune responses to Ags shared by the vaccine and the pathogen (14, 15). Vaccines often elicit long-lasting protection that precludes infection by microbial and viral pathogens, thereby reducing the need to use drugs to treat infections. In addition, vaccines provide protection against drug-sensitive and drug-resistant pathogens. Based on the belief that immunization to protect individuals from infection is superior to the continued development of new drugs to treat bacterial, viral, and parasite infections that inevitably acquire resistance to currently available or newly designed antibiotics and drugs, improved recombinant attenuated Salmonella vaccine (RASV) vectors have now been developed to deliver Ags and DNA vaccines encoding Ags from numerous bacterial, viral, and parasitic pathogens to elicit protective immune responses. Live attenuated vaccines are generally more efficacious than killed and subunit vaccines in inducing long-lasting protective immunity (16–20). However, most classical means of attenuation of live vaccines to ensure safety and eliminate induction of adverse reactions often resulted in diminished immunogenicity due to impaired abilities to colonize and persist in lymphoid tissues (21). Realization of this problem has recently led, as described below, to the development of vaccines designed to circumvent and abrogate this problem.

An important critical attribute for a bacterial vaccine vector is to be derived from a highly virulent invasive bacterial species that is able to colonize, after oral administration, internal effector lymphoid tissues within the vaccinated individual (16, 22, 23). As previously reviewed (23, 24), Salmonella is ideal as a bacterial vector because it invades internal effector lymphoid tissues after oral delivery, can stimulate mucosal and systemic Ab– and cell-mediated immune responses in immunized individuals, and can be readily manipulated, because much is known about its genetics, physiology, pathogenesis, and genomics. These features enable rational design and construction of RASVs that are completely attenuated but can target specific organs, cells, and cellular compartments to maximize immune responses, using recently developed strategies. Furthermore, attenuated Salmonella vaccines and RASVs derived from them can be preserved by lyophilization in a thermostable form to be reconstituted at the time and place of oral needle-free delivery, contributing to their cost effectiveness (23).

Salmonella has been the most widely studied among bacterial genera for delivering recombinant protective Ags and DNA vaccine vectors because of its capacity to be delivered orally, thereby precluding the use of needles for immunization (16, 17, 21–37). An important consideration in the selection of a strain to use in the development of a live bacterial vector is its ability to induce protection, not just against itself but against most or all other strains within the same species (38, 39). Live attenuated bacterial vaccines should consist of strains that possess two or more stable attenuating mutations (ideally, dispersed in the bacterial chromosome) and plasmids (without drug-resistant traits) into which genes encoding heterologous Ags from microbial pathogens can be inserted. The attenuating mutations should be gene deletions whenever possible and should be in genes that encode components of bacterial cell structures or biosynthetic pathways for essential nutrients that are not readily available in any environment that the RASV may inhabit outside of the laboratory (21, 23, 40). RASVs should be designed and constructed to be as invasive as wild-type strains to colonize internal lymphoid tissues and persist there for an adequate time to synthesize and deliver recombinant protective Ags. RASVs need to be replication proficient to provide longer periods (41) in which Ags are continuously produced and presented to the immune system, with the expectation of triggering stronger immune responses (16, 17, 23, 27, 28, 32, 33). To eliminate the need for antibiotic selection for maintenance of plasmids in RASVs, balanced lethal vector–host systems were developed to ensure maintenance of the plasmid vectors in vivo and in vitro, without reliance on antibiotic-resistance markers (42, 43). Balanced lethal vector–host systems consist of RASV strains that carry a deletion mutation in an essential gene into which a plasmid carrying the wild-type gene is introduced. One of the most frequently used mutations is Δasd, whose product is essential for the synthesis of diaminopimelic acid (DAP), an integral unique component of the bacterial cell wall peptidoglycan (42, 43). Unless the RASVs are supplied with DAP in growth media, or include plasmids with the wild-type asd gene, the strains will lyse as a result of the inability to make peptidoglycan. Because DAP is not present in in vivo environments, the RASVs will not survive unless a plasmid with the wild-type gene is present within the bacteria, constituting a powerful selection for maintenance of the plasmids within the RASV (42, 43). DNA sequences encoding Ags from heterologous pathogens are included in these plasmids. An added benefit of the balanced lethal vector–host systems is that the RASV remains sensitive to antibiotics used to treat Salmonella infections.

To ensure safety when using live bacterial vaccines, it has long been considered necessary that the vaccine be completely attenuated, induce no disease symptoms, but maintain high immunogenicity. Attenuation can lead to decreased survival of orally delivered live bacterial vaccines during passage through the stomach (44–63), increased sensitivity to bile (64–68), and/or increased sensitivity to defensins in the gastrointestinal tract (69–72). Traditional means of attenuation often decreased the ability of orally delivered attenuated bacteria to compete with the microbiota in the intestinal tract, invade through mucin to reach intestinal epithelial cells, or attach to and invade M cells of the GALT and, thus, diminished the ability of the vaccine to access and persist in lymphoid tissues (23, 35, 51). Recently, several strategies have been developed to achieve the goals of producing attenuated RASVs with high immunogenicity. The first strategy is to enhance the ability of RASVs to colonize internal lymphoid tissues. This is achieved by having the RASV express a highly invasive phenotype to enhance colonization of lymphoid tissues while reducing the dose of vaccine needed to induce protective immunity (73), by expression of arginine decarboxylase and/or glutamate decarboxylase at the time of vaccine administration to neutralize stomach acidity to enable a higher percentage of the RASV to reach the ileum (74, 75), by regulated delayed in vivo expression of attenuated phenotypes (51, 76–81), and by regulated delayed in vivo synthesis of rAgs (80–84). Immunogenicity can be further enhanced by regulated delayed in vivo lysis of the RASV (73, 83) and by decreasing the ability to suppress or modulate induction of immunity (77–79). These alterations also decrease inflammation of the intestinal tract and decrease the ability of the RASVs to form biofilms, thereby lessening in vivo persistence (85–88).

Several strategies have been developed to ensure that, at the time of oral administration, the vaccine strains would be as nearly like wild-type Salmonella as possible and so withstand the host defenses and stresses encountered when traversing the gastrointestinal tract to invade the intestinal epithelium and colonize appropriate effector lymphoid tissues. The first strategy was to construct strains with regulated delayed in vivo attenuation, using one of three ways. 1) Use Salmonella vaccine strains with mutations or deletion–insertion mutations that allow the strains to synthesize LPS O-antigen and/or outer core when supplied with an appropriate sugar during growth in vitro, but be unable to continue synthesis when the sugar was unavailable in vivo. Consequently, after six to eight generations in vivo, the RASV strain would become susceptible to complement-mediated cytotoxicity (89, 90) and be more readily phagocytized and killed by macrophages (91). 2) Delete promoters of genes that encode products essential for virulence and pathogenicity, replacing them with sugar-regulated promoters so that expression of the genes would be dependent on the presence of arabinose or rhamnose during in vitro growth of the vaccine strains, but expression would immediately cease in vivo due to the absence of the sugars, and the gene products would be diluted by at least half at each cell division (23, 35, 80). 3) Construct Salmonella vaccine strains that undergo regulated delayed lysis in vivo, which provides total biological containment, with no vaccine persistence in vivo and no survival if the vaccine strain is shed in feces (73, 83). This strategy uses arabinose-regulated synthesis of muramic acid and DAP, two unique essential constituents of the peptidoglycan layer of the cell wall. This system is designed for release of a bolus of protective Ag(s) or a DNA vaccine vector (23, 73, 83). Vaccine strains exhibiting the regulated delayed-lysis in vivo phenotype have been shown to induce better immune responses and higher levels of protective immunity to challenge than RASVs that do not undergo regulated delayed lysis (73, 92–95). In studies comparing mice immunized with nonlysis RASVs with those immunized with lysis RASVs, where both kinds of RASVs produce the same Ag(s), we have consistently observed that the lysis RASVs elicit significantly higher titers of total IgG (92, 94–96) and IgA (95). Immunization with either lysis or nonlysis RASVs usually results in levels of IgG2a or IgG2b (depending upon the mouse strain used) that are two to three times greater than IgG1 levels (73, 92, 94–96). When several of the early lysis RASVs were compared with nonlysis RASVs in terms of protection against challenge with the pathogen to which the RASV was directed [e.g., influenza virus (95) or Mycobacterium tuberculosis (94)], the nonlysis RASVs conferred protection that was equivalent to or slightly better than that elicited by lysis RASVs producing the same Ag (94, 95). However, with improved-lysis RASVs, more recent studies have shown that mice immunized with lysis RASVs producing influenza virus strain A/WSN/33 nucleoprotein or hemagglutinin Ags confer 100% survival for ≥21 d after intranasal challenge with influenza virus (73, 92). In studies in which improved-lysis RASVs producing three M. tuberculosis Ags were used to orally immunize mice, followed by challenge with low-dose aerosols (50–100 bacteria per mouse) of virulent M. tuberculosis, the titers of M. tuberculosis bacteria in the lungs of immunized mice 6 wk after challenge were 1.2–1.7 logs lower than the titers in the lungs of unimmunized and challenged control mice (S. Sanapala, S. Wang, and J.E. Clark-Curtiss, unpublished observations). The reduction in titers of M. tuberculosis in the RASV-immunized mice was similar to or slightly greater than those in the lungs of mice immunized with Mycobacterium bovis bacillus Calmette-Guérin, which served as positive vaccine controls in these experiments. We speculate that induction of better immune responses and protection may be due, in part, to release by lysis of CpG, peptidoglycan constituents, lipid A, flagellin, and other effectors of TLRs and nucleotide-binding oligomerization domain–like receptors.

Synthesis of heterologous Ags by vaccine vector strains invariably decreases their rates of growth, which lessens efficient colonization of effector lymphoid tissues. In addition, synthesis of some Ags can be toxic to the vaccine cells and can further impede their ability to efficiently access internal lymphoid tissues. These problems can be alleviated by selectively deleting portions of Ag genes present in the vaccine vector cells, such as those encoding protein sequences with multiple cysteine residues or hydrophobic domains (97–99), but this approach may diminish the immune responses to the Ag if the deleted portions encode key epitopes. Alternatively, one can use promoters that are activated in vivo [e.g., P_nirB (100) and P_pagC (101)] to regulate expression of Ag-encoding genes. This approach has the benefit of being universally usable. An additional universal means for regulated delayed synthesis in vivo of recombinant protective Ags is to have expression of plasmid-encoded Ag genes controlled by a regulatable strong promoter, such as P_trc. During in vitro growth of RASV strains, this promoter is repressed by LacI, which is produced from an araC P_BAD lacI cassette inserted into the chromosome of the RASV. When arabinose is present in the growth medium, LacI is synthesized to preclude the P_trc-controlled expression of the Ag-encoding sequence. In vivo, in the absence of arabinose, LacI is no longer produced and is gradually diluted out, thereby allowing synthesis of the Ag (99, 102, 103). Wang et al. (103) compared three RASV strains with chromosomal arabinose-regulated insertion–deletion mutations that produced different levels of LacI and that each carried the same plasmid that produced the Streptococcus pneumoniae PspA Ag (pYA4088) to a Salmonella strain that did not produce LacI but did have pYA4088. These studies demonstrated the effects of the presence or absence of arabinose in in vitro growth media on the synthesis of PspA: after 10 generations of growth in medium without arabinose, LacI had been diluted out of the in vitro growth media, and maximal levels of PspA were produced in the RASVs with arabinose-regulated expression of lacI, whereas the strain without the lacI gene produced PspA constitutively (103). When these Salmonella strains were used to orally immunize mice, all of the strains carrying pYA4088 induced strong IgG responses, but the mucosal IgA responses correlated with the amount of LacI that the strains produced (103). All mice immunized with the Salmonella strains producing PspA had higher levels of IgG2a than IgG1, suggestive of a Th1 response (103). When immunized mice were challenged with virulent S. pneumoniae, mice immunized with the Salmonella strains producing higher levels of LacI (and hence, with tighter regulation of expression of the pspA gene) were significantly better protected than mice immunized by Salmonella producing low levels or no LacI [40–50% survival 15 d after challenge compared with 18–20% survival (103)]. Regulated delayed Ag-synthesis systems were subsequently used in numerous RASV constructs to enhance the survival and colonization abilities of the RASVs prior to production of Ags (92–94, 99, 101, 103).

When using RASVs with the regulated delayed-attenuation capabilities to immunize mice, we observed that strains producing higher amounts of the Fur protein (due to araC P_araBAD regulation of fur) at the time of mucosal immunization were more invasive and colonized Peyer’s patches better than strains that produced less Fur (80). Fur is an iron uptake–regulatory protein that governs expression of genes whose products are necessary to maintain adequate, but not excessive, concentrations of iron in bacterial cells (104). Several investigators (50, 105) provided an explanation for enhanced invasiveness and colonization when they reported that Fur activates expression of hilD. HilD is a transcriptional regulator of hilA, which encodes the main transcriptional activator of genes in Salmonella pathogenicity island (SPI)-1, and SPI-1 gene products facilitate invasion (50, 105). Subsequently, Kong et al. (73) discovered that wild-type strains of Salmonella in which the SPI-1 hilA promoter was replaced with P_trc lacking the lacO sequence (ΔP_hilA::P_trcΔlacO hilA) were more invasive, colonized all lymphoid tissues to higher levels, and had a lower LD₅₀ than the wild-type parent. In RASVs with other attenuating mutations, introduction of this construction resulted in increased invasiveness and better tissue colonization, but without the enhanced virulence (73).

Orally administered RASVs must survive the inhospitable low pH environment of the stomach while traversing the gastrointestinal tract. Although Salmonella possesses a low acid response and the arginine decarboxylase acid resistance systems, the latter response is typically repressed under routine culture conditions (74). In the S. Typhi RASV-Pneumococcus vaccine for humans, Brenneman et al. (74) placed the arginine decarboxylase system under the control of a sugar-regulated promoter to enable expression of this system prior to oral administration of the vaccine. This feature rendered the S. Typhi grown in the presence of the sugar acid resistant on demand by coadministration of arginine, leading to enhanced survival and increased colonization (74). Although exposure of RASVs to low gastric pH can be circumvented by coadministering the vaccine with an antacid (e.g., sodium bicarbonate) or encapsulating the vaccine in a protective capsule, this is not totally advantageous for survival of the RASV and for enabling efficient invasion into epithelial cells (74). Exposure of Salmonella to the low pH of the stomach serves as an important signal to the bacteria that they have entered the host environment (74). Exposure to acid and other host signals induces expression of genes whose products confer resistance to short-chain fatty acids (106), antimicrobial peptides (70–72), and osmotic stresses (49) found in the intestines. Moreover, induction of the acid-tolerance response induces expression of Salmonella SPI-1 and SPI-2 genes and facilitates increased invasion of the RASV into intestinal epithelial cells (96, 107–109). Regulated expression of the arginine and/or glutamate decarboxylase systems in S. Typhi RASVs, coupled with using Ensure to modulate the stomach pH (110), should facilitate successful immunization of humans (which have a lower stomach pH after fasting than do mice).

In addition to the acid-tolerance response systems discussed above, Salmonella possesses many other genes whose products confer acid tolerance and resistance to acid stress (52–59, 107). Expression of these genes is controlled by the regulatory genes Fur (50, 59), RpoS (45–49), PhoPQ (60, 61), and OmpR (62, 63). Fur, in addition to its role in regulating iron concentration in Salmonella cells (discussed previously), also regulates production of a subset of acid shock proteins involved in Salmonella’s response to organic acid challenge (59); thus, the inability to synthesize Fur causes RASVs to be more susceptible to stomach acid (52, 59). Inclusion of cassettes that regulate production of Fur in RASVs enables the RASVs to produce adequate Fur to withstand stomach acid while traveling through the gastrointestinal tract; however, after colonization, the RASVs cease to synthesize Fur and begin to produce increased amounts of iron-acquisition proteins, contributing to the attenuation of the RASVs, presumably due to iron overload (23). RpoS is a key regulator of the acid-tolerance response (49); thus, rpoS mutants are more susceptible to stomach acid, rapidly dying when exposed to pH 3 (49, 52). As a result, S. Typhimurium rpoS mutants are avirulent in mice, because they are severely impaired in their capacity to invade M cells and colonize Peyer’s patches and spleens of infected mice (45, 46, 48, 49). Moreover, S. Typhimurium rpoS mutants have reduced immune responses and protective capacity compared with wild-type strains (83). These findings provided an explanation for the diminished protective ability of currently available S. Typhi vaccines for humans, which are derived from S. Typhi Ty2, an rpoS mutant (48). RASVs with arabinose-regulated expression of fur, rpoS, or phoPQ are attenuated and immunogenic. The presence of an arabinose-regulated fur cassette is an integral component of S. Typhi and S. Typhimurium RASVs with regulated delayed attenuation that have been used to deliver the S. pneumoniae Ag PspA to humans and mice (76, 79, 81, 82, 102, 111).

An S. Typhimurium RASV strain displaying regulated delayed attenuation and regulated delayed Ag synthesis phenotypes, and producing the S. pneumoniae Ag PspA, was completely safe when administered orally to newborn mice on the day of birth, as well as 2, 4, and 7 d after birth. That is, doses of ∼5 × 10⁸ CFU S. Typhimurium RASV χ9558(pYA4088) given to baby mice on any of those days did not cause disease symptoms, impair subsequent growth of the mice, or result in the death of any of the immunized mouse pups throughout a 6-wk monitoring period (86). When female mice were immunized with this RASV 2 wk prior to breeding, their offspring exhibited reduced levels of colonization by the RASV if the pups were inoculated 4 or 7 d after birth but enhanced colonization if the pups were inoculated on day 0 (86). Pups born to immunized mothers and then immunized at 7 or 21 d after birth produced higher titers of mucosal (IgA) and systemic (IgG) PspA-specific Abs than pups born to nonimmunized naive mothers and immunized at day 4 or 7 (86). When the pups were challenged i.p. with S. pneumoniae WU2, the pups born to immunized mothers exhibited significantly better protection (40% [day 7–immunized pups] to 50% [day 21–immunized pups] survival compared with 10% [day 21] to 11% [day 7] survival of pups born to naive mothers during a 15-d period after challenge) (86). Subsequent experiments with S. Typhi RASVs with the regulated delayed-attenuation and regulated delayed-Ag synthesis phenotypes that also carried pYA4088, and thus produced S. pneumoniae PspA, gave similar results (82). In these studies, immunization of mothers resulted in quicker Ab responses in their pups, with higher Ab levels detected by 2 wk after immunization of the pups (82). Higher levels of protection against i.p. challenge with S. pneumoniae WU2 were observed in pups born of S. Typhi RASV–immunized mothers compared with pups born of naive mothers in these experiments as well. Thus, immunization of mothers with either S. Typhimurium or S. Typhi RASVs confers protection upon pups soon after birth, and the presence of maternal Abs does not adversely affect subsequent immunization of offspring, which further enhances Ab and cytokine responses, as well as protection to challenge (82, 86).

RASVs are designed to deliver Ags from heterologous pathogens to the immunized host to elicit immune responses that will protect the individual from infection by the pathogen. Genes or sequences encoding protective Ags are generally inserted into plasmid vectors within the RASV, and several approaches have been developed for delivery of the protective Ags.

In many RASV systems, protective Ags are delivered using bacterial secretion systems. Although the type 1 secretion system associated with hemolysin secretion in Escherichia coli was one of the first means investigated for secretion of protective Ags (112, 113), it is inferior to other means that were subsequently developed. A more attractive approach was the use of the Salmonella type 3 secretion system (T3SS) (114), which enables secretion of Ags out of the bacterial cell and translocates them into the cytosol of host cells, where they are processed in the proteasome for class I presentation to augment induction of CD8-dependent immune responses (114). In early studies using this system, Ags with T cell epitopes were fused to SPI-1 T3SS effector proteins (96, 114–116); however, more recently, investigators have used effectors of the SPI-2 T3SS for Ag delivery and have found that the effector SseF seemed to work best (98). Many of the mucosally administered RASVs stimulate Ab production (at mucosal sites and systemically) and cellular immunity, and there is a significant need to better understand these induced cellular immunities. In mice, following invasion through M cells, Salmonella is delivered to intraepithelial dendritic cells, which deliver Salmonella to the lamina propria and to mesenteric lymph nodes. There is evidence that localization of Ag to the lamina propria activates B1 cells that produce polyreactive low-affinity IgA Abs, whereas Ags delivered to mesenteric lymph nodes activate B2 cells, which produce high-affinity Ag-specific IgA Abs (117).

Another important means for delivering protective Ags is via the type 2 secretion system (T2SS). Fusing signal sequences from T2SS proteins to protective Ags has resulted in increased induction of Abs to the protective Ags. Xin et al. (99) investigated the use of several T2SS signal peptides and showed that the signal sequence of β-lactamase was most efficient for the export of protective Ags, resulting in the strongest protective responses against pathogen (S. pneumoniae) challenge. Kang et al. (118) used the β-lactamase signal sequence to deliver the S. pneumoniae PspA protective Ag out of the cytosol into the periplasm of the RASV. In this study, 15–20% of the PspA Ag was found in the supernatant fluid, and Ab titers to the PspA Ag were as high as the Ab titers to Salmonella LPS and outer membrane proteins. Mice immunized with the RASV delivering the secreted PspA Ag were also significantly better protected against S. pneumoniae challenge than mice immunized with an RASV in which PspA remained in the cytosol (119). In this study, which also compared Ab titers generated when PspA remained in the cytosol of the RASV strain with those generated when PspA was transported out of the RASV by β-lactamase T2SS, the Ab titers were ∼100 times greater when PspA was secreted (119). For some Ags, inclusion of the N-terminal and C-terminal signal sequences of β-lactamase has led to better secretion and higher Ab titers (99, 120, 121). Secretion of rAgs into the periplasm can also yield Ag-containing outer membrane vesicles that are highly immunogenic (122).

A number of studies have demonstrated that lysis of the RASV to release protective Ag(s) is a superior means of Ag delivery, especially to induce mucosal immunity and cell-mediated immunity, as the result of delivery of Ags to host cell cytosol in ΔsifA strains (see below), in RASVs and in other recombinant attenuated bacterial vaccines (reviewed in Refs. 23, 37). Nearly all lysis in vivo systems are based on blocking synthesis of unique constituents of peptidoglycan in the bacterial vaccine strain, either by deleting genes for their synthesis or by regulating synthesis of enzymes in the peptidoglycan biosynthetic pathway, resulting in regulated delayed cessation of peptidoglycan synthesis. The consequence is cell wall–less death of the RASV strain and release of any protective Ags present in the RASV cells at the time of lysis (Fig. 1). Arabinose-dependent regulation of expression of the asdA and murA genes on a regulated delayed lysis plasmid was the approach used by Kong et al. (83) in the design of an RASV strain with the regulated delayed-lysis in vivo phenotype. The asdA and murA genes encode enzymes in biosynthetic pathways for unique constituents of the peptidoglycan cell wall in Salmonella and many other bacteria (42, 43, 123). A second feature of the lysis plasmid designed by Kong et al. (83) is its capability to synthesize antisense mRNA for the asdA and murA genes to completely prevent their expression in the absence of arabinose. Because Salmonella is impermeable to the substrate for the murA gene product, the chromosomal murA gene is under araC P_araBAD regulation in this RASV lysis strain. This strain also possesses mutations to block synthesis of colanic acid, a substance produced by bacterial cells to protect them during cell wall–less death. Inclusion of a relA mutation in the strain serves to uncouple cell growth from a dependence on protein synthesis, thereby contributing to complete lysis of the cells. This regulated delayed-lysis in vivo RASV system has been used by investigators in six studies to elicit better immune responses to protective Ags from several pathogens than were achieved using nonlysis strains to deliver the same Ags (73, 83, 92–96). Juárez-Rodríguez et al. (94) placed genes encoding three protective Ags of M. tuberculosis (a fusion of the three Ags) on a regulated delayed lysis plasmid, with one Ag delivered by T2SS and two delivered by T3SS, so that the Ags were secreted continuously by either of two means prior to cell lysis, when a bolus of synthesized Ags was released. As discussed above, delivery of M. tuberculosis Ags by the lysis RASVs resulted in higher levels of Abs to the individual Ags in immunized mice and better protection against challenge compared with mice immunized with nonlysis RASVs, where the Ags were delivered by T2SS and T3SS alone (94, 96).

A consequence of the lysis in vivo system is that the lysing bacteria release endotoxin that could lead to increased inflammation, a cytokine storm, or sepsis, depending on the relative sensitivity of the host immunized. The regulated delayed-lysis in vivo strains have been further altered by inclusion of a deletion–insertion mutation to modify the RASV LPS lipid A synthesis, so that the strains synthesize monophosphorylated lipid A to significantly reduce inflammatory responses to lipid A endotoxin (88). Monophosphorylated lipid A acts as a potent adjuvant by interacting with TLR4–MD2–CD14 via the TRIF-dependent pathway so that it contributes to immunogenicity (124–129) but is not reactogenic, because it does not interact by the MyD88 pathway (125, 127, 129). Thus, these RASVs are fully capable of recruiting innate immune responses, but they also exhibit complete biological containment by cell lysis to preclude persistence of the RASV in vivo or survival of the RASV if excreted (73, 83, 88–95). For animal hosts that are more tolerant to endotoxin, a ΔmsbB mutation can be inserted to lessen the inflammatory response (44, 85, 130, 131). Inclusion of ΔmsbB and ΔsopB mutations reduces inflammation in the intestinal mucosa without diminishing immunogenicity (44, 80, 82, 85, 130).

Another feature to improve these RASVs is inclusion of the ΔsifA mutation in the strain, which enables the RASV to escape the endosome (Salmonella-containing vacuole) after invasion of host cells. SifA has also been shown to be one of the SPI-2 T3SS effectors that is involved in inhibition of Ag presentation, by altering loading of Ag-derived peptides onto MHC class II complexes (132). Endosome escape is an important attribute for enhancing induction of cellular immunity in that, after lysis of the RASV, the Salmonella-synthesized protective Ags are released into the cytoplasm and presented to the proteasome for MHC class I presentation and induction of CD8-dependent immunities (92). RASVs that lyse in the cytoplasm and cause decreased apoptosis/pyroptosis are also ideal for delivery of DNA vaccines encoding Ags that require posttranslational modification by the immunized host to confer protective immunity to viral and parasite Ags (73).

In some cases, the ability to display protective Ags on the surface of the RASVs may invoke improved immune responses. Several investigators have reported using fusions of protective Ags to autotransporters, such as AIDA-1 (133) and HgbA (134), for surface display and presentation. Other investigators have used fusions of Ags to Salmonella outer membrane proteins (135), but such fusions have sometimes proven to be toxic to the RASV, necessitating modifications that led to decreased Ag synthesis but still induced much higher serum Abs (IgG and IgA) and higher vaginal IgA to the Porphyromonas gingivalis HagB Ag than were obtained using an isogenic vaccine strain in which HagB remained in the cytoplasm (135). This group further demonstrated that the higher the level of Ag synthesis (which correlated with decreased fitness for the RASV), the more rapid was the induction of immune responses, especially mucosal responses (136). There is much evidence that secretion (119) or surface display of protective Ags on RASVs and other recombinant attenuated bacterial vaccines results in higher immune responses and protective immunity (133, 134, 137–145). Sizemore et al. (146) determined that a ΔphoPQ-attenuated S. Typhimurium strain induced much higher Ab titers against Yersinia pestis Ags F1 and LcrV when these Ags were retained in the cytoplasm rather than when they were secreted. However, Morton et al. (137) demonstrated that a Salmonella Typhi vaccine expressing Y. pestis F1 on its surface induced strong IgG responses and protection against Y. pestis challenge in mice. Optimal induction of Ab responses via secretion or retention in the bacterial cytoplasm may depend upon characteristics of the protective Ags (37, 121, 146) and whether they are synthesized constitutively, which can reduce vaccine immunogenicity, or by a regulated delayed synthesis process that enhances immunogenicity.

Most vaccines elicit better immune responses if they are composed of multiple protective Ags; this can be accomplished by including several plasmids encoding multiple protective Ags (147) or complex plasmids encoding several Ags as an operon (93, 94, 96) or with multiple genes that are independently regulated (148–150). However, the presence of such complex plasmids can result in genetic instability due to inter- and intraplasmidic recombination, which can be resolved by incorporating deletions of the recF and recJ genes in the RASV chromosome (151). The recF and recJ genes encode proteins that facilitate recombination between plasmids in the same bacterium (RecF) or within a single plasmid (RecJ). When these genes are deleted, neither the virulence nor the colonizing ability of the RASV is compromised (151).

An alternative approach is to use a mixture of RASV strains, with one RASV producing two or more Ags and the second strain producing two or more different Ags. In this approach, there is less burden on the individual RASVs in terms of the number of heterologous Ags needing to be synthesized, but there is a larger repertoire of Ags present in the vaccine mixture. Xin et al. (147), using RASVs producing S. pneumoniae Ags, and Jiang et al. (93), using RASVs producing Clostridium perfringens Ags, showed that immunizing with a mixture of two strains, each delivering a single Ag, gave the same results (Ab titers and protection against challenge) as a single RASV strain delivering both Ags.

Great strides have been made in developing RASVs for delivery of protective Ags from a diversity of pathogenic bacteria, viruses, and parasites. Recent improvements in technologies have enabled development of multiple platforms that can be refined to enhance the desired types of immune responses and to deliver protective Ags by means to preclude or enable posttranslational modification to ensure induction of protective immune responses. Additional improvements that would enhance Ag presentation by deletion of genes encoding SPI-2 T3SS effectors or that would enhance targeting of the RASVs to specific niches in the immunized individuals may further enhance the protective efficacy of RASVs. Thoughtful construction of RASVs has led to vaccines that are safe for newborns, infants, and pregnant mice (81, 86), with the added benefit that immunization of mothers-to-be enhanced, rather than impaired, immunization of neonates (86). Moreover, several studies have demonstrated that prior immunization and the presence of existing Abs to Salmonella LPS do not impair immune responses to heterologous Ags (152, 153), thus enabling reuse of RASVs producing heterologous Ags from more than one pathogen. Aspects of RASVs that remain to be fully explored are the longevity of protective immune responses elicited to Ags produced by these vaccines and the composition and quality of memory T cell populations produced after RASV immunization. The primary objective of vaccine development research is to construct vaccines to safely induce long-term protective immunity against infection. Thus far, results with RASVs are strongly encouraging for attainment of the ultimate objective of delivering vaccine constructs that will be safe and efficacious in protecting animals and humans against infectious disease agents. In using RASVs, ascertaining immunological correlates for achieved protective immunity is difficult, because all three branches of the immune system are activated to collectively contribute to protection.

We thank Shifeng Wang for help in assembling this article.

R.C. and J.E.C.-C. are founders of Curtiss Healthcare, Inc., which has licensed vaccine technologies developed by their research groups. Therefore, they may benefit financially as a result of research reported in this publication.