The subcutaneous inoculation of pH 6 antigen mutants of Yersinia pestis does not affect virulence and immune response in mice

The extraordinarily high virulence of Y. pestis (Lorange et al., 2005) is determined by a wide spectrum of its features, initially termed as determinants of virulence (Burrows, 1957, 1963). The classic determinants include the ability of bacterial cells to absorb exogenous dyes and haemin (Pgm⁺), dependence on growth at 37 °C using Ca²⁺ ions present in the medium (Ca^–), synthesis of V and W antigens, synthesis of murine toxin and capsule antigen F1 (Tox⁺ and Fra⁺), synthesis of pesticin concomitant with fibrinolytic and coagulase activities (Pst⁺, Fb⁺ and Cg⁺), purine independence or the ability to produce endogenous purines (Pur⁺). After five decades of research and deliberation stemming from the initial proposal of virulence determinants, some of them, such as W antigen, pesticin and the ability to produce endogenous purines, are no longer recognized as pathogenicity factors (Anisimov, 2002a, b; Brubaker, 2006; Perry & Fetherston, 1997). Recently, a number of additional virulence factors have been identified. They include secreted Yersinia outer proteins (Yops) (Cornelis, 2002), adhesion pili (pH 6 antigen, termed PsaA) (Lindler & Tall, 1993; Vodop'ianov & Mishan'kin, 1985; Vodop'ianov et al., 1990) and six additional chaperone/usher systems (Felek et al., 2007): YadA-like non-pilus adhesin encoded by bicistronic operon yadBC (Forman et al., 2008), lipopolysaccharide (Bengoechea et al., 1998; Knirel et al., 2007; Oyston et al., 2003; Porat et al., 1995), murein (or Braun) lipoprotein Lpp (Sha et al., 2008) and others (Anisimov, 2002a, b; Brubaker, 2006; Perry & Fetherston, 1997).

pH 6 antigen was initially described in 1961 as an antigen synthesized by Y. pestis at the temperature close to body temperature of mammals (35–41 °C) and acidic pH values (5.8–6.0) close to the pH of abscesses or phagolysosomes in macrophages (Ben-Efraim et al., 1961). Recently, it was shown that growth in human plasma also upregulated transcription of psaE and psaF, suggesting a role for pH 6 antigen in the septicaemic phase of plague (Chauvaux et al., 2007). More recently, gene expression studies demonstrated highly persistent expression of the psaA gene from 4 to 48 h after intranasal infection in mouse lungs compared with the spleen and liver, indicating possible involvement of this antigen in the pathogenesis of pneumonic plague (Liu et al., in press). Subunits of pH 6 antigen, which have a molecular mass of 15 kDa, assemble on the surface of Y. pestis and Yersinia pseudotuberculosis into homopolymer macromolecular complexes termed adhesion pili (Lindler & Tall, 1993; Vodop'ianov & Mishan'kin, 1985; Vodop'ianov et al., 1990), forming together a capsule-like structure similar to that formed by Y. pestis F1 capsular antigen (Anisimov, 1999; Cherepanov et al., 1998).

In addition, pH 6 antigen possesses adhesion activity toward erythrocytes (Ben-Efraim et al., 1961; Bichowsky-Slomnicki & Ben-Efraim, 1963; Stepanshina et al., 1993) and mammalian epithelial cells (Galván et al., 2007; Isberg, 1989a, b; Liu et al., 2006; Yang et al., 1996) due to the binding of phosphatidylcholine (Galván et al., 2007) or glycosphingolipids (Payne et al., 1998). It is known to prevent phagocytosis (Huang & Lindler, 2004; Stepanshina et al., 1993), probably through masking the bacterial surface by binding apolipoprotein B-containing lipoproteins from human plasma (Makoveichuk et al., 2003), phosphatidylcholine of pulmonary surfactant (Galván et al., 2007) and/or human IgG1, IgG2 and IgG3 subclasses (Zav'yalov et al., 1996) or other extracellular matrix components, including fibronectin and mucin (Vodop'ianov et al., 1993). In early studies, pH 6 antigen was shown to be cytotoxic for peritoneal (Bichowsky-Slomnicki & Ben-Efraim, 1963) and alveolar (Stepanshina et al., 1993) macrophages. However, high-purity preparations of pH 6 antigen were not cytotoxic for eukaryotic cells (Bakhteeva et al., 2007). Subcutaneous (s.c.) injection of pH 6 antigen to rabbits (1000 µg), guinea pigs (625 µg) and mice (100 µg) did not lead to death (Stepanshina et al., 1993).

The synthesis of pH 6 antigen is encoded by the psa operon which has a structure similar to other operons coding for pilus adhesins that are conveyed to the bacterial surface by the chaperone/usher secretion systems (Felek et al., 2007; Hultgren et al., 1993; Iriarte & Cornelis, 1995; Karlyshev et al., 1994; Thanassi et al., 1998). The psa operon consists of psaE and psaF, two regulatory genes responsible for temperature- (37 °C) and pH- (5.8–6.0) dependent transcription regulation, psaA, the structural gene of pH 6 antigen subunit, and psaB and psaC, coding for periplasmic chaperone and molecular usher, respectively (Lindler et al., 1990; Yang et al., 1996). Expression of the psa operon is positively regulated by binding of the global transcription factor RovA to the psaE and psaA promoter regions (Cathelyn et al., 2006), while Fur protein acts as a repressor of psa genes (Zhou et al., 2006). A ΔpsaA mutant of Y. pestis wild-type strain CO92 had a significant dissemination defect after s.c. infection, but only slight attenuation by the pneumonic disease model (Cathelyn et al., 2006). Mutations in psaE or psaA of attenuated Pgm^– Y. pestis strain KIM5 caused a 200-fold reduction of virulence in retroorbitally infected mice (Lindler et al., 1990), while psaF mutants of wild-type strain 231 were completely avirulent for mice and guinea pigs after s.c. challenge (Panfertsev et al., 1991). These reports suggest that further studies involving the use of mutants with complete loss of the psa operon are required to clarify the role of the pH 6 antigen in the virulence of Y. pestis.

The contradictory data on the role of pH 6 antigen in Y. pestis virulence might be a result of different strain backgrounds and/or methodological differences in the generation of such strains. It was previously suggested (Domaradski, 1987) that transfers of genetic information are more frequent in the case of atypical forms of certain bacteria. In the case of such an explanation, one should speak not of a decrease in virulence of the cells under the influence of plasmids but of a change in the composition of the population, caused by the accumulation (selection) of recombinants with an initially low virulence or entirely avirulent. The selection of clones that have retained virulence at the level of the wild-type parent strains requires an animal passage. In evaluating the virulence of cultures of agents of infectious diseases subjected to experimental manipulations or stored for long periods under laboratory conditions, a preliminary stage of animal passage is necessary to purify the microbial population from avirulent segregants (in the case of Y. pestis, these are bacteria with the phenotypes Lcr^–, Pgm^–, etc.) (Anisimov, 1999).

In this study, we generated ΔpsaA and ΔpsaEFABC mutants that were deficient in the ability to produce pH 6 antigen. We performed their complementation in trans using recombinant plasmids coding for the complete cluster of psa genes or the psaABC locus. Recombinant bacteria were passaged through mice and the effect of s.c. inoculation on the virulence of isogenic Y. pestis strains in both naïve and pH 6 antigen-immunized mice was determined.

Table 1. Strains and plasmids used and constructed in this study More detailed information on biovar-subspecies interrelations and geographical locations of the natural plague foci have been discussed previously (Anisimov et al., 2004).

Animal passage. To select for virulent Y. pestis subcultures, a group of four mice was challenged subcutaneously with 0.2 ml aliquots of strains 231 and I-1996 or their derivatives at a concentration of approximately 10⁸ c.f.u. The animals that were first to succumb to infection were subjected to necropsy and one bacteriologic loop of the brain tissues was suspended in 1 ml 0.9 % NaCl. In the second round of animal passage, four mice were challenged subcutaneously with 0.1 ml of this suspension containing approximately 200 c.f.u. Y. pestis. The cultures isolated from the animals that succumbed to early infection in the second passage were used in subsequent experiments.

Growth of bacteria. For mutagenesis, complementation and animal challenge experiments, Y. pestis strains were grown at 28 °C for 48 h on brain heart infusion (BHI; HiMedia Laboratories) supplemented with 2 % agar at pH 7.2. For testing pH 6 antigen production, bacteria were cultured at 37 °C for 48 h on BHI agar at pH 5.8. For isolation of preparative amounts of pH 6 antigen, BHI, pH 5.8, supplemented with 0.5 % yeast extract (Difco) (Lindler et al., 1990) was used. BHI agar with 5 % sucrose was used for the selection of Y. pestis recombinants containing the suicide vector pCVD442 (Donnenberg & Kaper, 1991). E. coli strain S17-1 λpir and its derivatives were grown at 37 °C for 24 h in Luria–Bertani broth, pH 7.2 (Miller, 1972), or on the same medium supplemented with 2 % agar. Growth media were supplemented, as needed, with ampicillin (100 µg ml^–1), kanamycin (40 µg ml^–1), chloramphenicol (10 µg ml^–1) or polymyxin B (25 µg ml^–1).

Modifications of animal passage procedures used for mutant derivatives of strains 231 and I-1996 are shown below.

Mutagenesis. Deletions in the psa operon in Y. pestis vaccine strain EV were achieved by allelic exchange using a kanamycin resistance (kan) cassette with short flanking homologous regions of the Y. pestis target DNA (Datsenko & Wanner, 2000). The kan cassettes were obtained by PCR amplification using plasmid pUTKm as template and primers PsaAkm1/PsaAkm2 to generate EVΔpsaA and PsaEkm1/PsaCkm2 to generate EVΔpsaEFABC (Supplementary Table S1, available in JMM Online). The PCR products were introduced by electroporation into Y. pestis EVpKD46, as described previously (Conchas & Carniel, 1990). Recombinant colonies were selected on kanamycin BHI agar plates. Correct insertion of the antibiotic resistance cassette was verified by PCR and the inability to produce pH 6 antigen.

Deletions in the psa operon in Y. pestis virulent strains were achieved using the suicide vector pCVD442 (Donnenberg & Kaper, 1991) in order to prevent possible aerosolization of highly virulent Y. pestis that may happen during electroporation, which is necessary for the method described above. Mutant alleles of ΔpsaA and ΔpsaEFABC were amplified from chromosomal DNA of Y. pestis strains EVΔpsaA and EVΔpsaEFABC, respectively, with primer pairs PsaAkm3/PsaAkm4 for ΔpsaA and PsaEkm3/PsaCkm4 for ΔpsaEFABC, and the fragments containing the mutated genes were ligated into the SmaI site of suicide vector pCVD442 (Donnenberg & Kaper, 1991). The ligated plasmids were electroporated into E. coli S17-1 λpir cells to form pCVD442ΔpsaA : : kan and pCVD442ΔpsaEFABC : : kan, these were introduced into wild-type Y. pestis 231 or I-1996 from E. coli S17-1 λpir by conjugation using polymyxin B for counterselection, to produce kanamycin-, ampicillin- and polymixin B-resistant (Km^RAp^RPm^R) merodiploid transconjugants. The Km^RAp^RPm^R transconjugants selected for each plasmid/strain combination were plated onto BHI agar with 5 % sucrose and grown at 28 °C for 2 days. The resultant sucrose-resistant (Suc^R) Km^R ampicillin-sensitive (Ap^S) colonies were screened by using PCR with the corresponding primer pair and by serological screening for the inability to produce pH 6 antigen. Km^RAp^SpH 6^– Y. pestis psa double-crossover mutants (Table 1) for each plasmid/strain combination were pooled and 10⁶ c.f.u. of each culture were used for intraperitoneal (i.p.) challenge of four mice. The cultures on antibiotic-free BHI agar isolated from spleens of animals that succumbed to infection were used for a second i.p. challenge of four mice (dose 10⁴ c.f.u.). The strains isolated from the second round of animal passage were checked for the inability to produce pH 6 antigen and used for further studies.

PCR verification. Three PCRs were used to show that all mutants had the correct structure. The common test primers for kan were k1 and k2 (Supplementary Table S1). Two reactions were performed using flanking locus-specific primers with the respective common test primer (k1, k2) to test for new junction fragments. A third reaction was carried out with the flanking locus-specific primers to verify the loss of the parental (non-mutant) fragment and the presence of the new mutant-specific fragment. Control colonies were always tested in parallel.

Complementation of knockout mutants. Complementation was performed with the plasmid pJG428 [Ap^R tetracycline-resistant (Tc^R)] containing the complete operon psaEFABC from Y. pestis EV (Makoveichuk et al., 2003) cloned in the cosmid pHC79 (Hohn & Collins, 1980). A SauI–ClaI-fragment of the plasmid pJG428 containing locus psaABC was ligated into the vector pBluescript II KS(+) (Stratagene), which was previously digested with SauI and ClaI. The resulting construct was digested with PvuI and ligated to an 800 kb fragment from plasmid pBR325 (Prentki et al., 1981) containing the chloramphenicol resistance (cat) gene. The ligated plasmid was electroporated into E. coli DH5α cells to form pIG924Cm [chloramphenicol-resistant (Cm^R)].

Plasmids pJG428 and pIG924Cm were introduced into Y. pestis mutants by electroporation (Conchas & Carniel, 1990). Transformants of mutant derivatives of Y. pestis strains 231 and I-1996 for each plasmid/strain combination were combined and used for the challenge of four mice (10⁶ c.f.u. per animal). The bacterial cultures, which were isolated from spleens of animals on nutrient media with appropriate antibiotics, were used for a second round of infection of four new mice (10⁴ c.f.u. per animal). The bacterial cultures isolated after the second round of passage were tested clonally for the presence of antibiotic resistance and ability to produce pH 6 antigen. pH 6⁺ clones of each strain were pooled and used for virulence testing.

Estimation of pH 6 antigen production. pH 6 antigen production was estimated by double diffusion analysis in agar (Öuchterlony, 1949), Western blot (Yang et al., 1996), standard ELISA (Ngo & Leshoff, 1985) and/or dot-ELISA (Chandrika et al., 1998), using rabbit antiserum directed against Y. pestis pH 6 antigen (kindly provided by T. B. Kravchenko).

Isolation and purification of pH 6 antigen. Highly purified pH 6 antigen was prepared by the ammonium sulfate method (Makoveichuk et al., 2003) from culture supernatants of Y. pestis strain KM218 after incubation at pH 5.8 at 37 °C for 48 h.

Animal immunization and virulence challenge. Ninety-six of 224 mice were immunized with pH 6 antigen diluted in 0.1 ml PBS (pH 7.2), containing 10 µg pH 6 antigen, which was administered at a single s.c. site on the back of the animals. After 21 days, the animals were boosted with an identical dose at the same inoculation site. pH 6 antigen antibody titres were measured 26 days after the second pH 6 antigen immunization in four mice by ELISA, using 200 ng of Y. pestis-derived pH 6 antigen. The animals were anaesthetized by carbonic gas inhalation prior to the collection of a terminal blood sample by cardiac puncture. Twenty-eight days after the pH 6 antigen booster dose, six immunized and eight naïve animal groups were subcutaneously (in the interior thigh) administered with serial 10-fold dilutions of Y. pestis (10³ to 1 c.f.u., four mice for one dose) of 2-day agar cultures grown at 28 °C. Humane end points were strictly observed. Animals that succumbed to infection were sacrificed and examined bacteriologically. The remaining animals were observed for 21 days. The animals that survived were humanely killed by carbonic gas inhalation. The LD₅₀ and 95 % confidence intervals were determined according to the method of Kärber (Finney, 1978). Mortality times were recorded and the mean times to death were calculated, taking into account all challenge doses for each strain.

For colonization/dissemination analysis, 10 mice were infected for each time point with the wild-type strain or its ΔpsaEFABC variant by the s.c. route in the interior thigh (10² c.f.u.). At various times after infection (24, 36, 48 or 72 h), mice were euthanized by carbonic gas inhalation and superficial inguinal lymph nodes and spleens were removed and weighed. The bacterial load for each organ was determined by plating dilutions of the macerated tissues onto BHI agar plates; these were reported as c.f.u. (g tissue)^–1. Infections were repeated in at least two independent experiments.

Conflicting results were generated by testing the virulence of Y. pestis strains carrying mutations in different genes of the psa operon when using different routes of infection and involving parent strains differing in virulence and geographical origin (Cathelyn et al., 2006; Lindler et al., 1990; Panfertsev et al., 1991). In the present study, two highly virulent biovar antiqua strains, 231 and I-1996, were used. The former was previously used in similar experiments (Panfertsev et al., 1991) and the nucleotide sequence of its psa operon (GenBank accession number EF079883) was shown to be identical to those from the other completely sequenced Y. pestis strains ().

We performed mutagenesis with suicide vectors pCVD442ΔpsaA : : kan and pCVD442ΔpsaEFABC : : kan to generate I-1996ΔpsaA, I-1996ΔpsaEFABC and 231ΔpsaEFABC knockout mutants. Immunochemical tests, double diffusion analysis in agar gel, Western blot and ELISA with rabbit anti-pH 6 antigen sera indicated that they did not produce pH 6 antigen (Figs 1, 2 and 3).

Animal passage of Y. pestis pH 6^– mutants and complemented strains
After the first round of passage of Y. pestis 231ΔpsaEFABCpJG428, the Ap^R and pH 6⁺ cultures were isolated from 50 % (two of four), and after the second round from 100 %, of the animals that succumbed to infection. In contrast, we failed to isolate the Ap^R Y. pestis clones from the animals infected with I-1996ΔpsaEFABCpJG428. These results are in agreement with the suggestion that transfer of genetic material might be more frequent in atypical forms of Y. pestis with reduced virulence (Domaradski, 1987).

After the first passage of Y. pestis 231ΔpsaEFABCpIG924Cm and I-1996ΔpsaEFABCpIG924Cm, 100 % of the isolated cultures were Cm^R. The second round of passage in naïve mice did not eliminate the plasmid either; 100 % of the isolated cultures of both strains were Cm^R and pH 6⁺.

Interestingly, after the second round of animal passage of complemented strains that was performed without antibiotic treatment, elimination of recombinant plasmids carrying the psa operon was not detected. Previously, it was shown that plasmids containing the ColE1 replicon were not able to maintain stability in Y. pestis recipient cells without selective pressure (Drozdov et al., 1995). Similarly, 94–96 % of the complemented mutants isolated from animals in antibiotic-free BHI broth at 37 °C lost pJG428 or pIG924Cm plasmids (data not shown). A similar stabilization of plasmid pFS1 coding for another Y. pestis pilus adhesin, capsular antigen F1, which was also constructed using pHC79, was observed in Salmonella enterica bv. Typhimurium strain SL3261pFS1 after passage in mice (Gremyakova, 2004). Moreover, in our experiments, analysis of bacterial colonization/dissemination in infected animals indicated that wild-type bacteria replicated more rapidly in lymph nodes compared with their pH 6^– derivative, but this was not the case in the spleen. Hence, selection in the mammalian host for the pH 6⁺ or F1⁺ clones offers a selective advantage for bacterial cells producing pilus adhesins and was the reason why virulence of all recombinant strains obtained in this study was estimated without antibiotic treatment of experimentally infected animals. This decision was supported by isolation of clones carrying pJG428 or pIG924Cm plasmids, coding for pH 6 antigen, from the organs of all the mice that succumbed to infection.

Virulence of Y. pestis strains differing in ability to produce pH 6 antigen
All pH 6-immunized animals possessed high antibody titres to pH 6 antigen: 1/5120–1/10 240 (mean antibody titres=1/8960). Comparative study of the virulence of Y. pestis mutant strains using subcutaneously challenged mice did not reveal differences in their LD₅₀ (Table 2). The average survival time of mice that succumbed to infection with strains 231 and I-1996 or their isogenic derivatives did not differ from each other. Analysis of Y. pestis cultures isolated from animals that succumbed to infection indicated that pJG428 or pIG924Cm were stably maintained in vivo: 100 % of the isolated clones were Ap^R pH 6⁺ or Cm^R pH 6⁺, respectively.

Table 2. LD50 and mean survival time in mice infected subcutaneously with Y. pestis strains ND, Not determined.

Table 3. Mean weight of organs from subcutaneously infected mice and bacterial loads in these organs at different times after infection Values shown are means±SD. Values in bold type within the same column are significantly different (P<0.05) from each other, as calculated by Student's t test.

The data presented in this paper on the role of pH 6 antigen in full virulence of Y. pestis are seemingly in conflict with the results of earlier investigations (Cathelyn et al., 2006; Lindler et al., 1990; Panfertsev et al., 1991). Thus, studies of the attenuated Pgm^– strain KIM5 (Lindler et al., 1990) allow us to speculate that the 200-fold decrease in virulence of pH 6^– derivative in intravenous (i.v.) mouse infection might have been caused by the low virulence of the parent strain used. In addition, this strain was only able to cause lethal systemic infection following i.v. challenge. In our opinion, accumulation of mutations, which potentially reduce the ability of the wild-type strain to multiply in the host organism, may be compensated to some extent by the presence of a large number of pathogenicity factors in Y. pestis. Stepwise elimination of these factors can no longer compensate their loss at the expense of overproduction of the residual virulence factors and become apparent as a noticeable decrease of virulence.

It is reported that an approximately 80-fold drop in LD₅₀ and reduced dissemination/colonization of spleens and lungs in mice after s.c. infection with ΔrovA mutant of Y. pestis wild-type strain CO92 correlated with the decrease in expression of the psa operon (Cathelyn et al., 2006). The ΔpsaA mutation in CO92 resulted in a similar dissemination defect after s.c. challenge. These data are not shown in this paper.

We can propose several explanations for the case of the striking differences in virulence of pH 6^– mutants generated from strain 231. We can not rule out that Panfertsev et al. (1991) used a non-animal-passaged laboratory subcultivated clone of the initial strain 231 in their experiments. Due to repeated reseedings on nutrient media, it is possible that the bacterial population may have acquired unidentified mutations (modifications) which potentially reduced the ability of the strain to survive in the host organism. The loss of one more pathogenicity factor, pH 6 antigen, caused the complete loss of subcutaneous virulence in mice.

Comparing the results obtained using Y. pestis strains of different origin, one should consider the significance of individual pathogenicity factors in the virulence of various Y. pestis intraspecies groups (biovars, subspecies, ecotypes, etc.) for various animal species (Anisimov, 2002a). These strain differences result from microevolutionary adaptation of geographically separated Y. pestis in specific rodent species and they are characterized by significant individual metabolic traits. Taking this into account, it may be necessary to include in our experiments all of the previously studied parent strains, which exhibit different levels of attenuation after introduction of mutations into the psa operon. However, the current laws, regulations and restrictions in force on international exchange of highly pathogenic microbes made it practically impossible to use all the desired strains in this study.

Previous experiments on active immunization of laboratory animals with preparations of pH 6 antigen provide evidence of the absence of its protective properties in the presence of a pronounced ability to induce antibody formation (Gremyakova, 2004; Titball & Williamson, 2001; Vodop'ianov et al., 1995) and are in accordance with our current results. We speculate that the absence of protective activity is associated with peculiarities in the pH regulation of production of this antigen. Synthesis of pH 6 antigen by Y. pestis cells inside phagolysosomes of the macrophages or in the centre of necrotically degenerated tissues of abscesses (Ben-Efraim et al., 1961) should make pH 6⁺ bacteria unable to interact with antibodies and immunocompetent cells. Accordingly, the induction of an immune response is like a blank shot, directing the immune system toward an unachievable goal and exhausting its resources. If this is true, pH-independent production of pH 6 antigen should make Y. pestis ΔpsaEFABCpIG924Cm variants susceptible to killing by immunity induced with pH 6 antigen. However, high antibody titres to pH 6 antigen did not correlate with protection even in the group of mice challenged with the strains producing this antigen permanently. These results may explain the absence of reports on isolation of pH 6^– strains from natural plague foci and experimentally infected animals. F1^– strains are not rare, due to the high protective activity of F1 antigen which leads to selection of non-capsulated Y. pestis derivatives in immunized or surviving plague-infected animals (Anisimov, 2002b; Anisimov et al., 2004; Friedlander et al., 1995; Perry & Fetherston, 1997). Thus, antiphagocytic activity, together with the presence of antigen on the bacterial cell surface, cannot guarantee the same protective potency of similar microbial antigens.

Concluding remarks
The main outcome of our investigation is the finding that the loss of the ability to produce pH 6 antigen did not influence virulence of ΔpsaA or ΔpsaEFABC mutants of Y. pestis in the case of s.c. challenge of naïve and immune mice. This argues against the usefulness of using pH 6 antigen as a molecular target for plague prophylaxis and therapy. However, pH 6 antigen, or most likely genes from the psa operon, may be useful targets for laboratory diagnosis of infections caused by Y. pestis and/or Y. pseudotuberculosis.