Rational Considerations about Development of Live Attenuated Y. pestis Vaccines

Abstract

The risk of plague as a bioweapon has prompted increasing research efforts to develop plague vaccines due to its extreme virulence and the ease of its transmission. Subunit vaccines that contain F1 and LcrV antigens of Y. pestis have been tested for safety and immunogenicity, but doubts have been raised about whether subunit vaccines that engender antibody responses will protect against pneumonic plague, which requires both humeral and cellular immune responses for protection. The live, attenuated vaccine EV76, a pgm locus deficient Y. pestis strain, has been used for a long time in the Former Soviet Union and some Asian countries, but is not commercially available in the US and Europe due to safety concerns. However, the live attenuated Y. pestis vaccines are still considered to be the most effective way to prevent plague. In this review, we present our opinions about rationally creating live, safe and immunogenic Y. pestis vaccines with potential use for human based on established researches.

INTRODUCTION

Y. pestis, a Gram-negative bacillus, is the causative agent of plague, primarily infects rats and wild rodents, and is generally transmitted to humans via the bite of infected fleas [1]. Natural reservoirs of Y. pestis include rodents, squirrels, prairie dogs and marmots. Large reservoirs of Y. pestis still exist on all major inhabited continents, except Europe and Australia ¹ and it still remains a serious public health threat in those regions [1, 2]. Plague was responsible for at least 3 great pandemics and killed nearly 200 million people [1]. Infection by Y. pestis can result in three forms of the disease, bubonic, septicemic and/or pneumonic plague. One of them, pneumonic plague is highly contagious and easily transmitted person to person through airborne droplets, resulting in a rapid onset of disease and a mortality rate approaching 100% if treatment is delayed more than 24 h post-exposure [1, 3–7]. Since 1954, the number of confirmed plague cases recorded by WHO that occur worldwide has stabilized to approximately 2000 cases annually [3]. Plague is still endemic throughout the world resulting in sporadic infections that are reported as due to contact with wild rodents, lagomorphs, rural rats, and/or their fleas. Especially, Y. pestis is endemic in transmission by fleas from infected to susceptible rodents throughout the western United States [8]. Thus it remains a serious public health threat. Additionally, the recent emergence of multiple antibiotic resistant strains poses potential therapeutic and prophylactic problems [9–12].

Y. pestis is generally considered to be a great potential biological weapon [3], since the organism with high infectivity has the capacity of aerosol transmission and can be easily obtained from any of the numerous and widely dispersed animal reservoirs of plague [1]. Additionally, Y. pestis is easily genetically manipulated to create strains with specific engineered traits, such as constructing Y. pestis strains resistant to multiple antibiotics often used to treat plague patients. For longer-term protection and countering drug resistance, vaccination is believed to be an optimal choice [13, 14]. But, there is currently no licensed vaccine for human use in the United States. In regard to safety, subunit vaccines based on the F1 capsule that inhibits macrophage phagocytosis to Y. pestis [15] and/or LcrV that sits at the tip of the type III secretion system (TTSS) needle of Y. pestis [16] have obvious advantages over killed whole-cell vaccines and live attenuated Y. pestis vaccines [17]. However, mice vaccinated with F1 antigen alone fail to gain protection against F1-negative Y. pestis infection [18]. Our results indicated that replacement of wild-type lcrV with modified lcrV (lcrV2345) which resulted in E33Q, E34Q, K42Q, E204Q and E205Q amino acid substitutions, did not alter its virulence and dissemination propertied compared to the wild-type Y. pestis [19]. However, immunization with a recombinant attenuated Salmonella vaccine strain synthesizing the modified LcrV did not provide any protection against subcutaneous challenge with virulent Y. pestis CO92 [20]. So concerns about the virulence of F1⁻ Y. pestis strain being the same as the virulent F1⁺ Y. pestis strain [18, 21] and polymorphisms of LcrV [22] that may influence protective efficacy of subunit vaccines based on F1 and LcrV were raised. Additionally, subunit vaccines based on F1 and/or LcrV failed to fully protect African green monkeys against pneumonic plague and their efficiency to protect humans has not been fully demonstrated [23–25]. Research about improving efficacy of the subunit vaccines by using different adjuvants are continuing. Readers are also referred to recent reviews to know about the trend of plague vaccine development [13, 14, 26–29]. Here, we only focus on arguing about how to improve live attenuated Y. pestis vaccines against plague through rational considerations.

LIVE ATTENUATED Y. PESTIS VACCINE

A plague vaccine based on live attenuated Y. pestis provides the theoretical advantage of simultaneously priming against many antigens, thereby greatly enhancing the likelihood of broad-based protection. Research toward the development of new, improved live-attenuated vaccines should continue and be strongly encouraged [13, 29]. In the past, live attenuated strains were generated by selection, rather than precise genetic manipulation, thus raising concerns about their genetic composition and stability. The live EV76 vaccine is an apparent pgm mutant that has been used in some countries [27]. However, a concern is that the EV76 vaccine strain can cause disease in primates, then raising questions about its suitability as a human vaccine [30]. The latest review has summarized the progress of live attenuated Y. pestis vaccine development [31], so we do not describe the details about each construction. We just present our opinions about rational design of live Y. pestis vaccine based on recent research.

INCREASING RECOGNITION BY INNATE IMMUNITY

The innate immune system is our first line of defense against invading bacteria by early recognition of the invader and has a crucial function in antibacterial defense by triggering inflammatory responses that prevent the spread of infection and suppress bacterial growth [32]. Early innate immune responses slow the progress of infection of many pathogens, which allows for adaptive immune responses to develop. However Y. pestis, the causative agent of plague, with exceptional ability to evade or suppress innate responses can cause the death of the host before adaptive responses become effective [33].

Lipid A, the hydrophobic anchor of lipopolysaccharide (LPS), is a glucosamine-based saccharolipid that makes up the outer monolayers of the outer membranes of Gram-negative bacteria. LPS is also known as endotoxin because of its ability to induce toxic inflammatory responses [34, 35]. Many of the immune-activating abilities of LPS can be attributed to lipid A [36]. Sensing of lipid A by the human immune system is critical fohe onset of immune responses and clearance of Gram-negative bacteria infections. LPS activates cells via Toll-like receptor 4 (TLR4) and MD-2 on the cell surface [37–40], but the process of activation is dependent upon the structure of lipid A [41]. Y. pestis adopts a different type of acylation of lipid A to evade the innate immune system by synthesizing tetra-acylated lipid A with poor TLR4-stimulating activity at mammalian temperatures (37°C), whereas it synthesizes hexa-acylated lipid A, a potent TLR4 agonist, in flea vectors at temperatures of 22°C–28°C [33, 42–46]. This is accomplished due to the temperature-regulated expression of a key gene in the acylation pathway, lpxP, which results in hexa-acylated lipid A at low temperature. At 37°C, the body temperature of mammalian hosts, lpxP is not expressed, resulting in synthesis of tetra-acylated lipid A, which is not recognized by TLR4 [47]. Montminy et al. demonstrated that a fully virulent Y. pestis strain engineered to produce hexa-acylated lipid A at 37°C by constitutive expression of the Escherichia coli lpxL gene from a multicopy plasmid is highly attenuated by subcutaneous administration [33]. Based on this work, we introduced the E. coli lpxL gene into the chromosome of Y. pestis strain KIM6+(pCD1Ap) to construct the strains χ10015(pCD1Ap) (ΔlpxP32::P_lpxLlpxL) (Table 1) as a vaccine which provides greater genetic stability than plasmid expression [46]. Note that the pCD1 plasmid (also referred to as pYV in other Yersinia species) encodes LcrV and other Yersinia virulence associated proteins. The χ10015(pCD1Ap) strain has very similar characteristics with Y. pestis KIM1001-pLpxL (Table 1) made by Montminy, et al., such as high attenuation through subcutaneous (s.c.) administration and the same protective efficacy against high dose virulent Y. pestis challenge by the s.c. route [33, 47]. Nevertheless, our observations indicated that the χ10015(pCD1Ap) strain caused mice to be very sick before recovery after s.c. immunization and also caused injection sites of mice to fester and turn into hard and solid lumps. This strain also retained virulence via intranasal infection [46]. The histopathological analysis at 48-hour post-inoculation showed that mice intranasally infected with χ10015(pCD1Ap) developed more severe lung lesions which are similar to mice intranasally infected with KIM5 (Pgm⁻) strain (see supplemental materials) [48]. In our previous research, the crp single mutant of Y. pestis KIM strain, χ10010 (pCD1Ap) (Δcrp) (Table 1), had similar growth attribute as the Δcrp S. Typhimurium that grew more slowly than Crp⁺ strains [49, 50], while the mutant strain χ10017(pCD1Ap) (ΔP_crp21::TT araC P_BAD crp) (Table 1) in which crp expression is dependent on the presence of arabinose grew at the same rate as wild-type Y. pestis strain when supplemented with 0.05% arabinose. χ10017(pCD1Ap) was attenuated in mice when no arabinose was available in mice tissue and presented better protection against pneumonic plague than the crp single mutant strain, χ10010(pCD1Ap) [50]. Our results demonstrate that regulated delayed crp expression is an effective strategy to attenuate Y. pestis while retaining strong immunogenicity, leading to protection against plague [50]. Thus, we exploited our observations to reduce the residual virulence by combining a mutation for arabinose-regulated crp expression into the χ10015(pCD1Ap) strain to result in χ10030(pCD1Ap) (ΔlpxP32::P_lpxL lpxL ΔP_crp21::TT araC P_BAD crp) (Table 1), a strain producing hexa-acylated lipid A at 37°C and expressing crp dependent on the presence of arabinose [46]. Our results demonstrated that the LD₅₀s of χ10030(pCD1Ap) by s.c. and intranasal (i.n.) inoculation were raised more than 1.5 × 10⁷ and 3.4 × 10⁴-fold, respectively, in Swiss Webster mice, compared to the wild-type virulent Y. pestis KIM6+(pCD1Ap) strain. The LD₅₀ of χ10030(pCD1Ap) by s.c. infection was similar to that of χ10015(pCD1Ap), but the LD₅₀ of χ10030(pCD1Ap) by i.n. inoculation was around 3 log higher than that of χ10015 (pCD1Ap) (Table 1). Both s.c. and i.n. immunization with strain χ10030(pCD1Ap) induced very significant protection against both bubonic and pneumonic plague with minimal reactogenicity in mice [46]. These attributes are consistent with our goal of designing a live safe Y. pestis vaccine.

Table 1.

Live attenuated Y. pestis strains as vaccines against plague.

Y. pestis mutant	Genotype	LD50	Reference
Y. pestis KIM1001-pLpxL	pgm+ pCD1+, pMT+, pPCP+, pBR322-lpxL	>107 CFU by s.c. infection in C57BL/6 mice	[33]
Y. pestis KIM5 (also designated as Y. pestis KIM D27)	pgm− pCD1+, pMT+, pPCP+	>107CFU by s.c. infection and ~104CFU by i.n. infection in C57BL/6 mice	[83, 103]
Y. pestis KIM 10	pgm+ pCD1−, pMT+, pPCP−	Totally avirulent	[79]
Y. pestis KIM D27-pLpxL	pgm− pCD1+, pMT+, pPCP+, pBR322-lpxL	>107 CFU by s.c. infection and ~107CFU by i.n. infection in C57BL/6 mice	[84]
χ10010 (pCD1Ap)	Δcrp	>1 × 107 CFU by s.c. infection and >1 × 104 CFU by i.n. infection in Swiss Webster mice	[50]
χ10013 (pCD1Ap)	ΔlpxP	<10 CFU by s.c. infection and ~100CFU by i.n. infection in Swiss Webster mice	[48]
χ10015 (pCD1Ap)	ΔlpxP::PlpxL lpxL	~107 CFU by s.c. infection and 2.7 × 103 CFU by i.n. infection in Swiss Webster mice	[46]
χ10017 (pCD1Ap)	ΔPcrp TT araC PBAD crp	~4 × 105CFU by s.c. infection and and >1 × 104 CFU by i.n. infection in Swiss Webster mice	[50]
χ10027 (pCD1Ap)	ΔlpxP::PlpxLlpxL ΔlacI23::Plpp lpxE	2.3 × 104by s.c. infection and 7.8 × 104CFU by i.n. infection in Swiss Webster mice	[48]
χ10028 (pCD1Ap)	pgm− ΔlpxP::PlpxL lpxL	~108 CFU by s.c. infection and 5 × 103 CFU by i.n. infection in female Swiss Webster mice	[46]
χ10030 (pCD1Ap)	ΔPcrp TT araC PBAD crp ΔlpxP::PlpxL lpxL	>108 CFU by s.c. infection and >106 CFU by i.n. infection in female Swiss Webster mice	[46]

Note: Y. pestis KIM5+, LD₅₀ (s.c.) < 10 CFU, LD₅₀ (i.n.) ~100CFU ⁴⁶; CFU: Colony forming units; LD₅₀: 50% lethal dose; s.c.: Subcutaneous; i.n.: Intranasal.

Inactivation of the lpxM gene in EV NIIEG (pgm-) resulted in the synthesis of penta-acylated LPS at 25°C which might reduce inflammatory response [45, 51, 52]. This lpxM mutant has been evaluated in outbred mice, BALB/c mice and guinea pigs by the subcutaneous route, displaying improved characteristics as a live vaccine, such as decreased endotoxic activity and overall reactogenicity, and enhanced protective immunity compared with the parental vaccine strain EV NIIEG [53]. However, deletion of lpxM had no noticeable influence on the virulence of wild-type strain Y. pestis 231 [51, 54]. Further research demonstrated that the lpxM mutation reduced virulence and enhanced immunity of the ΔlpxM Y. pestis EV was associated with these pleiotropic changes and not just to changes in the lipid A acylation, moreover this mutation at least did not change pla expression in ΔlpxM Y. pestis 231 strain [54]. We infer that combining lpxM mutation into EV NIIEG strain reducing virulence and enhancing immunogenicity might be a matter of bacterial itself which contains undefined chromosomal mutations in EV NIIEG strain due to multiple in vitro passages. Our recent results demonstrated that the χ10013(pCD1Ap) (ΔlpxP32) (Table 1), a lpxP defined deletion in Y. pestis KIM strain only synthesizing tetra-acylated lipid A at both 26°C and 37°C had the same virulence as the wild-type Y. pestis strain. However χ10027(pCD1Ap) (ΔlpxP32::P_lpxLlpxL ΔlacI23::P_lpp lpxE) (Table 1), which expresses codon optimized lpxE from Francisella novicida in χ10015(pCD1Ap) to remove 3′-phosphate, could reduce its inflammatory response to different cell lines but elevated its virulence comparing with its parent strain (χ10015(pCD1Ap)) in mice [48]. Thus, results described above suggested reducing inflammatory responses through artificially synthesizing low activity lipid A in wild-type Y. pestis was not an optimal choice.

Gram-negative flagellin, a Toll-like receptor 5 (TLR5) agonist, is a potent inducer of innate immune effectors such as cytokines and nitric oxide. Similar to other TLR agonists that have adjuvant activity, such as lipopolysaccharide (LPS) and CpG oligonucleotides, flagellin promotes the maturation of dendritic cells as well as their migration to lymph nodes [55, 56]. Type I interferons and Tumor necrosis factor-alpha (TNF-α) have been demonstrated to promote the survival, maturation, and migration of dendritic cells to secondary lymphoid organs [57–59]. Honko et al. showed that flagellin is an effective adjuvant for immunization against lethal respiratory challenge with Y. pestis [60]. Research indicated that motility for Yersinia enterocolitica and Yersinia pseudotuberculosis is temperature-regulated between the narrow range of 30°C (Mot+, motility) and 37°C (Mot-, loss of motility) [61]. In contrast, Y. pestis has been classed as a non-motile organism since it was first isolated in 1894 [61]. Before the genomic sequences were completed for Y. pestis strains CO92 and KIM, Minnich’s group used Y. enterocolitica flagellar genes as probes for Southern blotting against Y. pestis DNA under high stringency hybridization conditions. The results showed that all probes used could hybridize with Y. pestis DNA. RT-PCR also confirmed that transcriptions of flhD, fliA and flgM were detectable from RNA isolated from Y. pestis KIM6 grown at 25°C and 37°C. However, no transcript was detectable for flagellin genes at either temperature, which was consistent with a non-motile phenotype [61].

Y. pestis has thus lost flagellar biosynthetic capacity, even though it has the requisite flagellar genes. DNA sequence analysis shows that a single T insertion generating a frame-shift mutation was identified in flhD, which results in shutting down the entire flagellar regulon and at least three other deletions are present in flagellar genes. Loss of expression of flhDC in this organism renders downstream genes silent and susceptible to further decay and permanent loss due to the improbability of repairing 4 lesions sequentially. DNA microarrays were employed to study the evolutionary genomics of the Y. pestis CO92 strain and 10 Y. pseudotuberculosis strains also demonstrated that the YPO0704–YPO0747 operon (putative flhDC operon) is highly unlikely to be functional in Y. pestis due to multiple frame-shift mutations [62–64]. Constitutive expression of flagellin in Y. enterocolitica completely attenuates its virulence [61]. This deficiency of flagellin is necessary to Yersinia since flagellin, if expressed, can be recognized and transported from the Ysc TTSS (abbreviation in line 57) along with Yops, which might have a dramatic attenuating effect in vivo through induction of innate immunity. Thus, either modification of lipid A or permanent loss of flagellin synthesis in Y. pestis reflect strategies evolved by Y. pestis to evade recognition by the innate immune system of the mammalian host [61, 65]. Based on this knowledge, we may engineer some live attenuated Y. pestis strains to constitutively synthesize flagellin to achieve high attenuation and induce greater immunogenicity.

DEVELOPING LIVE ATTENUATED VACCINES BASED ON FULLY VIRULENT Y. PESTIS STRAINS

In contrast to fully virulent strains of Y. pestis, EV vaccine strains or other pgm-deficient Y. pestis strains are highly attenuated by s.c. administration [66]. Fetherston, et al. [67] first described a 102-kb unstable chromosomal region [68–70] termed the pigmentation (pgm) locus in Y. pestis. Sequence analysis confirms that the 102-kb unstable pgm locus is composed of two distinct parts: the pigmentation segment and a high-pathogenicity island (HPI), which carries virulence genes involved in iron acquisition (yersiniabactin biosynthetic gene cluster) [71]. The pigmentation segment (~68-kb region) includes the hms (hemin storage) locus [72], which confers a pigmented phenotype on colonies grown on Congo red-agar plates. Pujol, et al. showed that the ripA gene in the pigmentation segment is required for replication of Y. pestis in interferon γ-activated macrophages [73]. This locus is also important for transmission of Y. pestis by the flea vector [74, 75]. The ~35-kb Yersinia HPI [71] encompasses 11 ybt genes required for biosynthesis and iron-scavenging via the siderophore yersiniabactin (Ybt), including psn, the structural gene for the pesticin receptor (an outer-membrane protein also involved in bacteriocin sensitivity and used as a protective antigen [20]). The pgm locus deletes spontaneously at a frequency of 10⁻⁵ [76], probably by homologous recombination between its two flanking insertion element (IS) IS100 copies [67]. Importantly, HPI genes are conserved in Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica biotype 1B strains, all of which are not only pathogenic for humans but able to disseminate in other mammalian hosts. High-pathogenicity Y. enterocolitica strains of biotype 1B are inherently lethal for laboratory animals, while low-pathogenicity strains of biotypes 2 to 5 are naturally devoid of the HPI and do not kill mice at low doses [77]. Injection of the animals with iron and/or an exogenous siderophore (Desferal) prior to the bacterial challenge did not significantly change the virulence of the high-pathogenicity strains but increased the virulence of the strains of biotypes 2 to 5, by providing them with the iron molecules necessary for their in vivo growth and dissemination [78]. Further, the virulence of the high-pathogenicity strain Ye8081 (LD₅₀ of 3.1 × 10² CFU) was not modified by the iron treatment. While, the LD₅₀ of the mutant strain Ye8081H-.1 (deletion of the HPI, LD₅₀ >10⁶ CFU) decreased to 2 × 10⁵ CFU upon iron treatment. These results suggest that the reduced virulence observed upon deletion of the 140-kb chromosomal region is partly attributable to the loss of the HPI in Ye8081 [77].

Quenee, et al. demonstrated that immunization of mice with the pigmented strain KIM 10 (lacking lcrV) (Table 1) can generate protective immunity against a bubonic plague challenge with the Y. pestis F1-deficient variants, whereas the Y. pestis KIM D27 strain (pgm-) with a deletion in caf1 cannot (Table 1) [79]. A possible interpretation they gave about this result was that protective antigens encoded by HPI or the pigmentation segment might provide protection against plague [79]. Mice inoculated by oral gavage with a deoxy-adenosine methylase (dam) mutant of Y. pseudotuberculosis IP32953, which lacked the pYV virulence plasmid (encoding lcrV) but had the HPI genes, were protected against bubonic plague challenge with fully virulent Y. pestis strain GB [80]. Since all Y. pseudotuberculosis strains lack caf operon which is specific for most of Y. pestis strains [18, 81] and Y. pseudotuberculosis IP32953 lacks lcrV, the observed immunity is thought to be provided by antigens (such as antigens coded by HPI genes) that are conserved between Y. pseudotuberculosis and Y. pestis [80]. Earlier work, using intravenous inoculation of pYV virulence plasmid-cured Y. pseudotuberculosis into mice, also generated immune protection against s.c. challenge with Y. pestis 6/69 M [82].

Smiley’s group reported that a pgm-deficient Y. pestis strain D27 harboring a multicopy plasmid encoding E. coli LpxL (D27-pLpxL) constitutively produced hexa-acylated forms of LPS. Strain D27 harboring empty plasmid pBR322 (D27-pBR322 strain) retained virulence when administered by i.n. route to wild-type C57BL/6 mice. The median lethal dose (MLD) for D27-pBR322 was 6 × 10³ CFU, which is nearly identical to that reported previously for D27 [83]. In striking contrast, no mice succumbed to i.n. infection with strain D27-pLpxL, even at doses as high as 1 × 10⁷ CFU [84]. We also introduced the E. coli lpxL gene into the chromosomes of the Pgm negative Y. pestis strain KIM6 (pCD1Ap) to construct the strains χ10028(pCD1Ap) (pgm-ΔlpxP::P_lpxL lpxL) as a vaccine [46]. However, our results indicated that χ10028(pCD1Ap) exhibited high attenuation by the s.c. route (LD₅₀, ~108 CFU), but not by the i.n. route (LD₅₀, 5 × 10³ CFU) in Swiss Webster mice (Table 1) [46]. We don’t know whether this discrepancy is due to the strain of mice used. However, immunization with χ10028(pCD1Ap) for Swiss Webster mice did not provide better protection against bubonic and pneumonic plague than χ10015(pCD1Ap) (ΔlpxP::P_lpxL lpxL) and χ10030(pCD1Ap) (ΔlpxP32::P_lpxL lpxL ΔP_crp21::TT araC P_BAD crp) strains based on a fully virulent strain (pgm+) (Fig. 1) [46].

Fig. (1).

Comparison of protection efficiency among mice immunized KIM5 (Pgm⁻), χ10015(pCD1Ap) or χ10028(pCD1Ap) against Y. pestis KIM5+ challenge. Swiss Webster mice (20 mice per group) were vaccinated s.c. with 2.5×10⁷ CFU of Y. pestis KIM5 (Pgm⁻), 2.2 × 10⁶ CFU of χ10015(pCD1Ap), 1.4 × 10⁷ CFU of χ10028(pCD1Ap), 1.4 × 10⁷ CFU of χ10030(pCD1Ap) or PBS as negative control and half mice per group were challenged 35 days later with 3.57 × 10⁷ CFU of Y. pestis KIM5+ via the s.c. route (A) and with 1.24 × 10⁴ CFU of Y. pestis KIM5+ via the i.n. route (B), respectively. In panel A and B, survival of mice immunized all live attenuated Y. pestis strains was significantly higher than PBS controls in all experiments (P<0.001); Survival of mice immunized with χ10015(pCD1Ap) or χ10030(pCD1Ap) was significantly greater than that of mice immunized with Y. pestis KIM5 (Pgm⁻) or χ10028(pCD1Ap) (P<0.01). A Log-rank test was used for analysis of the survival curves. This experiment was repeated twice times with consistent results.

The EV76 derivates, EV NIIEG strain, has been used as a live plague vaccine in the Union of Soviet Socialist Republics (USSR) and is still used in the countries of the Former Soviet Union (FSU) and some Asian countries [27, 85]. Nevertheless, a single dose of the EV NIIEG live vaccine conferred a prompt (day 7 post-vaccination) and pronounced immunity in vaccinees lasting for 10–12 months against bubonic and, to some extent, pneumonic plague [14, 85]. Yang’s group also posed hypothesis that the lack of the pgm locus might affect survival ability of EV76 in humans efficiently, limit its replication and dissemination and result in insufficient contact with immune cells [86], which might develop effectively adaptive immunity. Thus, the fully virulent Y. pestis strain as an origin to develop live efficacious vaccines might be an optimal choice.

ALTERING IMMUNE RESPONSE TO SPECIFIC ANTIGENS OF Y. PESTIS TO ENHANCE PROTECTIVE IMMUNITY

So far, development of subunit vaccines against Y. pestis has aimed primarily to elicit humoral immunity [See reference 13, 24 and 25]. The virulence factors encoded on virulent plasmid pCD1 of Y. pestis disrupt cellular immunity which implies that cellular immunity must be detrimental to Y. pestis replication in vivo [87]. Researches did demonstrate that cellular immunity played an important role in protection against plague [73, 83, 88–91]. Thus, vaccines that preestablish a capacity to rapidly activate Y. pestis-specific cellular immune mechanisms should aid defense against this deadly pathogen [87]. Humoral immunity can aid the clearance of extracellular bacterial infection that might disable prompt cellular immunity to Y. pestis, in turn enabling cellular immunity to efficiently eradicate intracellular bacteria. Based on established studies, plague vaccines should stimulate both humoral and cellular immune responses against specific antigens (including LcrV or F1) of Y. pestis.

Studies demonstrated that Chinese-origin rhesus macaques immunized with EV76 primed a higher anti-F1 IgG titer but an almost undetectable titer to LcrV antigen [17] and mice immunized with the Y. pestis ΔsmpB-ssrA mutant via either the i.n. or intravenous route produced high level of anti-F1 antibody response, but no detectable level of antibodies against LcrV [92], which are consistent with other studies of animals immunized with the EV76 or KWC vaccine [79, 93–98]. Recently, Yang’s group [72] investigated the humoral and memory cellular immune responses in 65 plague patients recovered from Y. pestis infection during 16 years after infection using protein microarray and enzyme-linked immunosorbent spot assays. The study indicated that the seroprevalence to the F1 antigen in all recovered patients is 78.5%. In patients infected more than a decade, the antibody-positive rate still remains 69.5% [86]. The positive occurrence of the F1 antibody in patients was high, but approximately 30% of the recovered patients remained sero-negative to both F1 and LcrV antigens. Moreover, the amount of antibodies in some patients was very low [86].

Yang’s group also looked at the cellular responses in plague recovered patients. The memory T cell responses to recombinant F1 and LcrV antigens were studied utilizing 7 plague patients who recovered from plague 4 or 6 years before, as well as 4 healthy controls from areas where plague is non-endemic. Although the peripheral blood mononuclear cells (PBMCs) from the patients that were stimulated with LcrV and F1 proteins produced IFN-γ, the level of IFN-γ was not significantly different compared with these of the controls [86]. Li, et al. indicated that mice immunized with the EV76 vaccine strain did not induce T cell responses to the F1 antigen [99]. The Smiley group also reported that mice vaccinated with live attenuated Y. pestis KIM5 (pgm-) (Table 1) generated the CD4 and CD8 T cells that synergistically conferred protection against plague, but T cells from those vaccinated mice could not recognize F1, LcrV and all pCD1/pPCP1 encoded proteins [87]. The results suggested that F1 and LcrV might not be the dominant T cell antigens [86, 87].

Thus, the questions are raised that why mice immunized subunit vaccines including LcrV prime dominant antibody responses providing considerable protection against plague [26, 100, 101], but mice infected or immunized live Y. pestis can’t induce the high level of antibody response and specific T cell responses to LcrV [17, 92]? One possibility might be that Y. pestis intentionally avoids eliciting both humoral and cellular immunity to LcrV, since the immune responses might be more efficient to restrain Y. pestis replication. Another possibility is that immune cells phagocytizing Y. pestis may not effectively present LcrV antigen to immune cells through MHC class I or class II pathway which will produce specific IFN-γ or high level of antibody to this antigen, although no direct evidences are demonstrated yet. Based on this hypothesis, whether the immune response to the specific antigen (such as LcrV) can be altered by a different way for antigen presentation? Our preliminary data showed that mice immunized with a live attenuated Y. pseudotuberculosis strain harboring plasmid (pYA5199, yopE_Nt138-lcrV) synthesizing fusion protein YopE(1-138aa)-LcrV delivered by the type three secretion system (T3SS) primed high titers of anti-LcrV antibody. Moreover, the splenic cells from the immunized mice cultured with full length LcrV protein produced high levels of IFN-γ, while immunization with the strain harboring an empty plasmid pYA3332 (no antigen delivered) did not (Sun, et al., in preparation). Mice orally immunized with the attenuated Y. pseudotuberculosis strain harboring pYA5199 were afforded higher-level protection (83% survival) against i.n. challenge with ~130 LD₅₀ of Y. pestis KIM6+ (pCD1Ap) (LD₅₀, ~100) than that of the strain harboring pYA3332 (40% survival) (Sun, et al., in preparation). Our unpublished data suggested that the LcrV delivered by T3SS could stimulate both specific humoral and cellular immune responses, and increase defense against pneumonic challenge (Sun, et al., in preparation). So this strategy might be combined to certain live attenuated Y. pestis strains to enhance both humoral and cellular immunity against plague.

SUMMARY

The opinions argued here are based on the knowledge from physiology, pathogenicity and vaccinology of Y. pestis conducted in the past several decades. Since F1 and LcrV may be non-dominant T cell antigens for the long-term defense against plague in humans, it can be questioned whether it is wise to rely on only using F1 and LcrV antigens as the basis for a human vaccine, even though subunit vaccines composed of the two antigens provide good protection against plague in some animal models [See reference 13, 24 and 25]. Researches also have shown that vaccination with live attenuated Y. pestis primes both CD4+ and CD8+ T cells that may recognize a set of antigens distinct from the LcrV and F1 antigens that elicit protective humoral immunity [87, 102]. There should be other proteins that play roles in stimulating cellular protective responses. So we may dig out antigens stimulating cellular response against Y. pestis through Omics (including genomics, proteomics, transcriptomics, and metabolomics) to provide basis of vaccine development.

The expected live Y. pestis vaccines which should be based on the properties described above will provide the theoretical advantage of simultaneously priming immunity against many antigens, thereby reducing the likelihood of antigen circumvention by clever terrorists [28]. Therefore, we recognize the importance of continuing research through raising recognition by innate immunity and altering immune response by a different way for presenting specific antigens, finally toward the development of new, improved live-attenuated Y. pestis vaccines.