ABSTRACT
The cryptic plasmid is essential for Chlamydia muridarum dissemination from the genital tract to the gastrointestinal (GI) tract. Following intravaginal inoculation, a C. muridarum strain deficient in plasmid-encoded pGP3 or pGP4 but not pGP5, pGP7, or pGP8 failed to spread to the mouse gastrointestinal tract, although mice infected with these strains developed productive genital tract infections. pGP3- or pGP4-deficient strains also failed to colonize the gastrointestinal tract when delivered intragastrically. pGP4 regulates pGP3, while pGP3 does not affect pGP4 expression, indicating that pGP3 is critical for C. muridarum colonization of the gastrointestinal tract. Mutants deficient in GlgA, a chromosome-encoded protein regulated by pGP4, also consistently colonized the mouse gastrointestinal tract. Interestingly, C. muridarum colonization of the gastrointestinal tract positively correlated with pathogenicity in the upper genital tract. pGP3-deficient C. muridarum strains did not induce hydrosalpinx or spread to the GI tract even when delivered to the oviduct by intrabursal inoculation. Thus, the current study not only has revealed that pGP3 is a novel chlamydial colonization factor in the gastrointestinal tract but also has laid a foundation for investigating the significance of gastrointestinal Chlamydia.
INTRODUCTION
Sexually transmitted Chlamydia trachomatis infections sometimes ascend to the upper genital tract (GT), where they can cause tubal pathology and infertility (1–3). However, the mechanisms of this ascension remain unclear. The murine model of Chlamydia muridarum infection (4–6) has been used for studying C. trachomatis pathogenesis (7–12) and has revealed numerous chlamydial (13–17) and host (7, 11, 18–24) factors required for chlamydial induction of hydrosalpinx (6, 16, 25, 26). For example, the cryptic plasmid (16, 17) or plasmid-encoded pGP3 (13) is essential for C. muridarum hydrosalpinx induction in mice. However, the mechanism by which the plasmid or plasmid-encoded pGP3 promotes C. muridarum pathogenicity in the mouse upper genital tract remains unknown.
The C. muridarum plasmid (pMoPn/pCM) (27, 28) encodes 8 putative proteins, designated plasmid-encoded glycoproteins 1 to 8 (pGP1 to pGP8, respectively) (27, 29). pGP1, -2, and -6 are required for plasmid maintenance (30–32); thus, the generation of mutant plasmids that do not express these proteins has been difficult. Nevertheless, C. muridarum mutants deficient in pGP3, -4, -5, -7, or -8 are viable (30). Studies with these mutants have revealed that pGP4 is a master regulator for plasmid-borne genes, such as pgp3, and chromosomal genes involved in glycogen synthesis, including glgA (31, 32), while pGP5 may function as a negative regulator of the same chlamydial genes (30). A plasmid-deficient C. muridarum strain had a reduced ability to elicit hydrosalpinx and tubal inflammation (16, 17). These phenotypes were similar in C. muridarum strains that contained plasmids that were deficient only in pGP3 or pGP4 (13, 30). Since pGP4 regulates pGP3 but pGP3 does not affect pGP4 expression (30–32), it was concluded that pGP3 is the major virulence factor responsible for plasmid-dependent pathogenicity in the mouse upper genital tract. Although pGP5 negatively regulates chlamydial plasmid and chromosomal genes in cell culture (30), pGP5 deficiency reduced C. muridarum pathogenicity in some mouse strains (33).
Despite the significant roles of the plasmid and pGP3 in C. muridarum pathogenesis in the upper genital tract, neither is essential for C. muridarum colonization of the mouse lower genital tract (13, 16, 17, 33, 34). Plasmid-free C. trachomatis isolates have been isolated from the human urogenital tract (35, 36). More importantly, C. trachomatis serovar D colonization of the genital tracts of nonhuman primates was not impacted by the loss of the plasmid (37). These observations suggest that the plasmid may be dispensable in the genital tract.
C. trachomatis and C. muridarum also infect the gastrointestinal (GI) tracts of humans (38–41) and mice (42, 43), respectively. When C. muridarum was experimentally introduced into multiple mouse mucosae, it readily colonized the GI tract (43). We showed that C. muridarum could spread from the genital tract to the GI tract and establish long-lasting infections at the latter site (42). This spread still occurred in singly housed mice wearing collars that prevented coprophagy (42, 43). Furthermore, C. muridarum survived in the blood, and the hematogenously delivered C. muridarum established long-lasting colonization only of the GI tract (44). These observations together indicated that C. muridarum could spread from the mouse genital tract into the GI tract via a hematogenous route. In contrast, autoinoculation of the mouse genital tract from the GI tract was not detected in mice that had stable GI tract infections for 70 days (45). Since C. muridarum infection in the GI tract is nonpathogenic (42–47), C. muridarum may have evolved as a commensal organism in this niche. Consistent with this, the plasmid is more important for C. muridarum to colonize the GI tract than to infect the genital tract (48). Identifying the plasmid-encoded or regulated factors responsible for the plasmid-dependent C. muridarum colonization in the GI tract may provide critical information for further investigating the mechanisms by which the plasmid promotes C. muridarum fitness.
Here, we report that plasmid-encoded pGP3 is essential for C. muridarum spread from the genital tract to and colonization of the mouse GI tract. In contrast, pGP4-regulated GlgA was not essential for C. muridarum to colonize the GI tract. Furthermore, the lack of GI tract colonization by pGP3-deficient C. muridarum correlated with its attenuated pathogenicity in the upper genital tract. Thus, the current study has both identified pGP3 to be a novel chlamydial colonization factor in the gastrointestinal tract and laid a foundation for further investigating the significance of the gastrointestinal Chlamydia.
RESULTS
Plasmid-encoded pGP3 is required for C. muridarum to spread from the mouse genital tract to the GI tract.C. muridarum dissemination from the mouse genital tract to the GI tract (42) is regulated by the cryptic plasmid (48). We mapped the plasmid-encoded proteins responsible for this by comparing the ability of C. muridarum strains deficient in individual plasmid-encoded proteins (C. muridarum with a complete plasmid [CM-pGFP], C. muridarum with a premature stop codon in the pgp3 gene [CM-pgp3S], C. muridarum with a premature stop codon in the pgp4 gene [CM-pgp4S], C. muridarum with a premature stop codon in the pgp5 gene [CM-pgp5S], and C. muridarum with a premature stop codon in the pgp7 gene [CM-pgp7S], as well as C. muridarum with an in-frame deletion of the pgp8 gene [CM-pgp8D]) to disseminate to the GI tract following intravaginal inoculation. All of the organisms robustly infected CBA/J mice and were shed from the lower genital tracts of CBA/J mice following intravaginal inoculation (Fig. 1), consistent with previous reports (13, 16, 17, 33). Abundant organisms were detected in rectal swab specimens from the CM-pGFP-infected animals, whereas the numbers of plasmid-free (pf) C. muridarum (CMUT3-pf) organisms were much lower (48). In contrast, no CM-pgp3S or CM-pgp4S (deficient in pGP3 or pGP4, respectively) organisms were detected in rectal swab specimens. Since pGP4 regulates pGP3 while pGP3 does not affect pGP4 expression, the observations presented above indicated that pGP3 is critical for C. muridarum spread from the genital tract to the GI tract. Why CM-pgp3S was more defective than CMUT3-pf is unclear. It is also worth noting that CM-pgp5S seemed to behave differently from the other C. muridarum clones. CM-pgp5S developed significantly reduced genital and rectal shedding courses, while the other mutant clones maintained a genital tract live organism shedding course as robust as that of the wild-type control, CM-pGFP. pGP3 was also required for dissemination from the GT to the GI tract in more resistant C57BL/6J mice (Fig. 2), arguing that the role of pGP3 is not related to the inherent susceptibility of the mice to the pathogen (6).
Comparison of the spreading of C. muridarum organisms with or without a deficiency in plasmid or plasmid-borne genes from CBA/J mouse genital tracts to gastrointestinal tracts following intravaginal inoculation. Wild-type (wt) or plasmid-competent C. muridarum organisms (CM-pGFP, n = 5) (a and a1), C. muridarum organisms deficient in plasmid (plasmid-free [pf] clone CMUT3-pf, n = 5) (b and b1), or C. muridarum organisms carrying plasmid deficient in pGP3 (CM-pgp3S, n = 5) (c and c1), pGP4 (CM-pgp4S, n = 5) (d and d1), pGP5 (CM-pgp5S, n = 5 [one mouse died on day 42]) (e and e1), pGP7 (CM-pgp7S, n = 5 [one mouse died on day 56]) (f and f1), or pGP8 (CM-pgp8D, n = 5) (g and g1) were intravaginally inoculated into female CBA/J mice at 2 × 105 inclusion forming units (IFU) per mouse. At various time points postinoculation, as indicated along the x axis, both vaginal (a to g) and rectal (a1 and g1) swab specimens were taken for the titration of live organisms, and the number of recovered live organisms was expressed as the log10 number of IFU per swab specimen, as displayed along the y axis. Note that although all C. muridarum organisms (a to d, f, and g), with the exception of the pGP5-deficient C. muridarum organisms (e), displayed similar live organism shedding courses from the genital tracts, the plasmid-free C. muridarum organisms or organisms deficient in the plasmid-encoded pGP3, pGP4, or pGP5 but not pGP7 or pGP8 developed a significantly reduced shedding course from the same mouse gastrointestinal tracts (*, P < 0.05, area under the curve, Wilcoxon rank-sum test). The pGP3- or pGP4-deficient C. muridarum organisms completely failed to colonize the gastrointestinal tracts.
Comparison of spreading of C. muridarum organisms with or without plasmid-encoded pGP3 to the gastrointestinal tracts of C57BL/6J mice following intravaginal inoculation. The experiments were carried out and the results are presented as described in the Fig. 1 legend. The CM-pGFP (wild type, n = 5) (a and a1) or the CM-pgp3S (n = 5) (b and b1) organisms were each inoculated into 5 C57BL/6J mice. Although both groups developed robust live organism shedding courses, the CM-pgp3S-infected mice failed to shed any live organisms from the gastrointestinal tract, confirming the findings described in the legend to Fig. 1. *, P < 0.05 (Wilcoxon rank-sum test, area under the curve).
The plasmid-encoded pGP3 is required for C. muridarum to colonize the mouse GI tract.To test whether the lack of spreading from the genital tract to the GI tract by pGP3-deficient C. muridarum was due to failure to colonize the GI tract, we compared the ability of the same strains described in Fig. 1 to colonize the GI tract following intragastric inoculation (Fig. 3). When female C57BL/6J mice were intragastrically inoculated with 2 × 105 inclusion-forming units (IFU) per mouse, no CM-pgp3S or CM-pgp4S live organisms were detected in the rectal swab specimens, indicating that these strains could not colonize the GI tract. CM-pgp3S still failed to colonize the GI tract even when the intragastric inoculation dose was increased to 1 × 107 IFU. Similar to the intravaginal infection model, CMUT3-pf organisms colonized the GI tract better than CM-pgp3S organisms. Rectal shedding of CM-pgp5S and CM-pgp8D was also reduced compared to that of CM-pGFP, suggesting that pGP5 and pGP8 contribute to C. muridarum fitness in the C57BL/6J mouse GI tract, although their importance in the GT is less clear (13, 33). Nevertheless, pGP3 was a critical plasmid-encoded factor for C. muridarum colonization of the mouse GI tract in both the intragastric and intravaginal infection models. It is worth noting that the inability of CM-pgp4S to colonize the GI tract may reflect the fact that other chromosomal genes regulated by pGP4, in addition to pGP3, might contribute to colonization of the GI tract.
Comparison of C. muridarum organisms with or without a deficiency in individual plasmid-encoded proteins for colonization of the mouse gastrointestinal tract. C. muridarum organisms with the complete plasmid (CM-pGFP) (a [n = 5] and h [n = 3]), without a plasmid (CMUT3-pf, n = 5) (b), or with a plasmid deficient in pGP3 (CM-pgp3S, n = 4) (c and i), pGP4 (CM-pgp4S, n = 5) (d), pGP5 (CM-pgp5S, n = 5) (e), pGP7 (CM-pgp7S, n = 5) (f), or pGP8 (CM-pgp8D, n = 5) (g) were inoculated intragastrically into female C57BL/6J mice at 2 × 105 (a to g) or 1 × 107 (h and i) IFU per mouse. At various time points postinoculation, as indicated along the x axis, both vaginal and rectal swab specimens were taken for the titration of C. muridarum live organisms. The number of live organisms recovered from the rectal swab specimens was expressed as the log10 number of IFU per swab specimen, as displayed along the y axis (no significant numbers of live organisms were recovered from the vaginal swab specimens; data not shown). Note that C. muridarum organisms deficient in pGP3 or pGP4 failed to develop any significant shedding from the mouse GI tract.
The pGP4-regulated GlgA is not essential for C. muridarum colonization in the mouse GI tract.GlgA expression is positively regulated by pGP4 (30–32), suggesting that a lack of GlgA expression (in addition to the reduced amount of pGP3, as demonstrated above) may also contribute to the reduced colonization of the GI tract by CM-pgp4S organisms. GlgA-deficient C. muridarum mutants produce fewer infectious progeny in cell culture (49), suggesting that GlgA may be a significant C. muridarum colonization factor in animals. Thus, we next evaluated the effect of chlamydial GlgA on C. muridarum colonization in the mouse GI tract. As shown in Fig. 4, both C. muridarum GlgA mutants developed significant IFU titers in rectal swab specimens, indicating that both mutants were still capable of colonizing the mouse GI tract. Since both clones contained a glgA gene mutation and lacked glycogen synthesis (see Fig. S1 in the supplemental material), this observation suggests that GlgA or glycogen synthesis is not required for chlamydial colonization in the GI tract. This conclusion is consistent with the hypothesis that pGP3 but not the pGP4-regulated GlgA is a primary determinant of C. muridarum fitness in the mouse GI tract. Nevertheless, compared to wild-type C. muridarum, both mutant clones had lower IFU titers, suggesting that GlgA or TC0215, a conserved hypothetical protein that contains a type III secretion system translocator arabinose efflux permease domain, or both GlgA and TC0215 may promote C. muridarum fitness in the mouse GI tract. This is because both mutant clones contain mutations in these two genes (Table S2).
Comparison of C. muridarum organisms with or without a deficiency in GlgA. C. muridarum organisms with GlgA (wild-type strain CM-G13.32.1, n = 5) (a) or without GlgA, C. muridarum P1D1 (n = 4) (b) and P3B4 (n = 3) (c), were inoculated intragastrically into female C57BL/6J mice at 2 × 105 IFU per mouse. At various time points postinoculation, as indicated along the x axis, both vaginal and rectal swab specimens were taken for the titration of C. muridarum live organisms. The number of live organisms recovered from the rectal swab specimens was expressed as the log10 number of IFU per swab, as displayed along the y axis (no significant numbers of live organisms were recovered from the vaginal swab specimens; data not shown). Note that C. muridarum organisms deficient in GlgA developed significantly reduced live organism shedding from the mouse GI tract. Nevertheless, all GlgA-deficient C. muridarum organisms were able to maintain significant levels of long-term colonization in the mouse GI tracts. *, P < 0.05 (Wilcoxon rank-sum test, area under the curve).
Plasmid gene-dependent C. muridarum colonization in the GI tract correlates with C. muridarum pathogenicity in the upper genital tract.The above-described experiments demonstrated that the plasmid-encoded proteins, particularly pGP3, promote C. muridarum GI tract colonization. The next question is whether colonization of the GI tract by C. muridarum can impact C. muridarum pathogenicity in the upper genital tract, which is a relevant question here, since pGP3 is a known C. muridarum virulence factor in the mouse upper genital tract (13). We then monitored the hydrosalpinx pathology in the upper genital tracts of CBA/J mice infected intravaginally with C. muridarum strains with or without a deficiency in the plasmid or individual plasmid genes (Fig. 5). We found that C. muridarum strains deficient in either the entire plasmid or the single protein pGP3 or pGP4 were significantly attenuated in inducing hydrosalpinx. When the relationships between C. muridarum pathogenicity in the upper genital tract and live organism shedding from the genital tract or the GI tract were analyzed (Fig. 6 and Table S1), we found that there was no correlation between the hydrosalpinx scores and the live organism shedding titers in the vaginal swab specimens (collected from the genital tracts). Spearman's correlation coefficient for the vaginal swab specimens (defined as Rv) was −0.1773, with the P value being 0.3251. Instead, a strong positive correlation between the hydrosalpinx scores and the live organism shedding titers in the rectal swab specimens (collected from GI tracts) was found, with Spearman's correlation coefficient (defined as Rr) being 0.4153 and the P value being 0.0163. This positive correlation between the upper genital tract pathology and GI tract colonization (Rr = 0.4153) is distinct from the slight negative correlation between the upper genital tract pathology and genital tract infection (Rv = −0.1773, P < 0.05). Thus, we can conclude that the plasmid gene-dependent C. muridarum pathogenicity in the upper genital tract correlates better with the plasmid gene-dependent C. muridarum colonization in the GI tract than with the C. muridarum infection in the genital tract. This correlation was further validated when CM-pgp3S was compared with CM-pGFP for both hydrosalpinx induction and live organism shedding after the organisms were directly delivered into the oviduct via intrabursal inoculation (Fig. 7). Although CM-pgp3S was able to infect the genital tract after it was inoculated into the oviduct, it was significantly attenuated in inducing hydrosalpinx. At the same time, the oviduct CM-pgp3S organisms were still unable to spread into the GI tract. Thus, even after intrabursal inoculation, the pGP3-dependent pathogenicity in the upper genital tract still correlated with the pGP3-dependent spreading to the GI tract. These observations together suggest that the pGP3-dependent spreading of C. muridarum into the GI tract may contribute to the pGP3-dependent pathogenicity in the upper genital tract.
Comparison of abilities of C. muridarum organisms with or without a deficiency in individual plasmid-encoded proteins to induce hydrosalpinx. The same CBA/J mice described in the Fig. 1 legend were sacrificed on day 56 after infection for observation of hydrosalpinx pathology (as described in the Materials and Methods section). One representative genital tract image from each group was chosen. The entire genital tract is displayed, with the vagina on the left and the oviduct/ovaries on the right in the images column. Oviducts with hydrosalpinges are marked with white arrows. The oviduct/ovary from each side is magnified in the panels on the right. The hydrosalpinx scores are indicated in the magnified panels. Hydrosalpinx was scored for severity, and mice with hydrosalpinges were counted for calculation of the incidence rate in a given group. Note that C. muridarum either free of plasmid or deficient in pGP3 or pGP4 significantly reduced the hydrosalpinx severity (*, P < 0.05, Wilcoxon rank-sum test) (b, c, and d), and the plasmid-free organisms also significantly reduced the hydrosalpinx rate (*, P < 0.05, Fisher's exact test) (b).
Comparison of the correlations of plasmid gene-dependent C. muridarum pathogenicity in the upper genital tract with live organism shedding from the genital versus GI tracts. The hydrosalpinx score for each CBA/J mouse, as described in the Fig. 5 legend (x axis), was correlated with live organism shedding from either the genital tract (log10 total number of IFU from vaginal swab specimens collected from a mouse over time, y axis) (top) or the GI tract (log10 total number of IFU from rectal swab specimens per mouse, y axis) (bottom) using Spearman's correlation formula as described in the Materials and Method section or Table S2 in the supplemental material. The total number of IFU from each mouse was obtained by adding the numbers of IFU from all time points observed. The correlation coefficient between the hydrosalpinx scores and the number of vaginal IFU was defined as Rv, which was −0.1773, with the P value being 0.3251, while the correlation coefficient between the same hydrosalpinx scores and number of IFU in the rectal swab specimens was defined as Rr, which was 0.4153, with the P value being 0.0163 (two-tailed), suggesting a positive correlation between the hydrosalpinx scores and rectal C. muridarum colonization. Furthermore, the positive correlation between hydrosalpinx and the number of IFU in the rectal swab specimens was significantly stronger than that between hydrosalpinx and the number of IFU in the vaginal swab specimens (*, P = 0.016). Nv and Nr, number of vaginal swab and rectal swab specimens, respectively.
Comparison of C. muridarum organisms with or without a deficiency in pGP3 for their ability to induce hydrosalpinx in the upper genital tract and spread to the gastrointestinal tract following intrabursal inoculation. C. muridarum organisms with the complete plasmid (CM-pGFP, n = 6) (a) or deficient in pGP3 (CM-pgp3S, n = 5) (b) were inoculated intrabursally into the oviducts of C57BL/6J mice at 2 × 105 IFU per mouse. At various time points postinoculation, as indicated along the x axis, both vaginal (a and b) and rectal (a1 and b1) swab specimens were taken for titration of C. muridarum live organisms. The numbers of live organisms recovered from the swab specimens were expressed as the log10 number of IFU per swab specimen, as displayed along the y axis. On day 63, all mice were sacrificed for observation of the hydrosalpinx pathology, as listed on the right. Note that C. muridarum organisms deficient in pGP3 were significantly attenuated in their ability to induce pathology in the upper genital tract and failed to spread to the GI tract. *, P < 0.05 (Wilcoxon rank-sum test, area under the curve, for shedding courses; hydrosalpinx score for pathology); #, P > 0.05 (Wilcoxon rank-sum test, area under the curve, for shedding courses and Fisher's exact test for hydrosalpinx rates; the lack of a significant difference was probably due to the limited sample size).
DISCUSSION
Since Chlamydia was detected in the GI tract, questions on the mechanisms by which Chlamydia colonizes the gut or the significance of Chlamydia colonization of the gut have caught the attention of chlamydiologists. Based on our recent report that the chlamydial cryptic plasmid is more important for C. muridarum to colonize the GI tract than to infect the genital tract (48), we now report that it is the plasmid-encoded pGP3 that is mainly responsible for the plasmid-dependent phenotype. Furthermore, we have also found that the plasmid gene-dependent chlamydial colonization in the GI tract correlates with the plasmid gene-dependent chlamydial pathogenicity in the upper genital tract. First, C. muridarum strains with or without a deficiency in plasmid or individual plasmid-encoded proteins developed similar levels of live organism shedding from the mouse lower genital tract following intravaginal inoculation, suggesting that all organisms evaluated in the current study maintained similar infectious titers in the mouse lower genital tract. However, a deficiency in plasmid or the plasmid-encoded pGP3 or pGP4 but not pGP5, pGP7, or pGP8 significantly reduced the spreading of the genital tract C. muridarum bacteria to the GI tract, indicating that the deficiency in pGP3 or pGP4 sufficiently phenocopied the plasmid deficiency. Since pGP4 is a positive regulator of pGP3 expression while pGP3 does not affect pGP4 expression (13, 30–33), we can conclude that pGP3 is mainly responsible for the plasmid-dependent spreading. Second, after intragastric inoculation, the same pGP3- or pGP4-deficient C. muridarum strain failed to colonize in the GI tract, suggesting that pGP3 was mainly responsible for mediating plasmid-dependent colonization in the GI tract. Third, two independent C. muridarum mutants deficient in GlgA, a chromosomal protein that is regulated by pGP4, were still able to colonize the mouse gastrointestinal tract, although at levels significantly lower than the level of a wild-type C. muridarum strain. The reduced numbers of IFU recovered may suggest that GlgA can promote C. muridarum fitness in the mouse GI tract. Glycogen synthesis and utilization have been shown to play an important role in maintaining gut bacterial colonization (50, 51). However, since each of the 2 mutants also carried multiple other mutations in their genomes, the reduced colonization by the mutants may have been caused by mutations in non-GlgA proteins. Thus, more experiments in future studies are required to define the genetic basis of the reduced colonization by these mutants. Finally, when the plasmid gene-dependent C. muridarum colonization in the GI tract was compared with the C. muridarum infection in the lower genital tract for correlation with the C. muridarum pathogenicity in the upper genital tract in the same mice, we found that the former had a significantly stronger positive correlation than the latter. The positive correlation of C. muridarum colonization in the GI tract with C. muridarum pathogenicity in the upper genital tract was further strengthened by the observation that when CM-pgp3S was directly inoculated into the oviduct, the oviduct organisms still failed to induce hydrosalpinx and were unable to spread to the GI tract. Thus, the current study not only has revealed that pGP3 is a novel plasmid-encoded colonization factor for improving C. muridarum fitness in the GI tract but also has laid a foundation for further investigating how GI tract C. muridarum may affect the pathogenicity of C. muridarum in the upper genital tract.
We noticed that the defect both in C. muridarum spread from the mouse genital tract to the GI tract and in C. muridarum colonization in the GI tract was greater with the deficiency of pGP3 alone than with the deficiency of the entire plasmid. The plasmid-free CMUT3-pf organisms were still able to spread to and colonize in the GI tract following intravaginal and intragastric inoculation, respectively, although the levels were significantly reduced compared to those of the wild-type control organisms (see Fig. 1 and 3 of the current study and reference 48). However, CM-pgp3S, which was deficient only in the pGP3 protein, completely failed to spread to and colonize in the GI tract, which was still true even when the intragastric inoculation dose was increased by 50-fold. Since CM-pgp3S developed a live organism shedding course as robust as CMUT3-pf did in the genital tract following intravaginal inoculation (Fig. 1) (13), the complete lack of spread to or colonization in the GI tract by CM-pgp3S was unlikely due to its general deficiency in infecting animal tissues. The plasmid-encoded pGP5 has been shown to negatively regulate genes in both the chlamydial plasmid and chromosome (30–32). The excessive deficiency of CM-pgp3S in colonizing the GI tract may be due to the repression of GI tract colonization factors by the pGP5 remaining in CM-pgp3S but no longer remaining in CMUT3-pf. However, the deficiency in pGP5 (CM-pgp5S clone) did not increase the level of colonization in the GI tract (Fig. 3). The pGP5 deficiency alone may not be enough to increase colonization factors in animals, although the pGP5-repressed genes were enhanced in CM-pgp5S in a cell culture system (30). CM-pgp5S was attenuated in infecting the genital tract and spreading to and colonizing the GI tract (Fig. 1 and 3) as well as in inducing hydrosalpinx in some mouse strains (13, 16, 17, 33, 34). These observations suggest that pGP5 deficiency is sufficient for altering the fitness of C. muridarum in the GI tract and reducing C. muridarum invasiveness in the genital tract. The deficiency in pGP8 seemed to also alter the fitness of C. muridarum in the mouse GI tract, but the phenotype of pGP8 deficiency in the cell culture system was not clear (30, 33). Apparently, more experiments are required to further define the precise mechanisms by which each plasmid-encoded protein impacts the fitness biology of C. muridarum in the GI tract. Despite the phenotypic discrepancies described above, the general conclusion that pGP3 plays a critical role in C. muridarum colonization in the GI tract remains strong.
The next question is how pGP3 promotes C. muridarum colonization in the GI tract. Since CM-pgp3S was able to infect the genital tract but completely lacked the ability to colonize the GI tract, the GI colonization system may provide a better platform for mapping the mechanisms by which pGP3 promotes C. muridarum infectivity in mice. C. muridarum long-term colonization in the GI tract has been localized to the cecum/colon (42, 45, 52). Thus, to achieve long-term colonization in the GI tract after an intragastric inoculation, C. muridarum needs to survive various host barriers, including gastric acid, the small intestinal defense mechanisms, and colonic microbiota competition. The question is how pGP3 aids C. muridarum to overcome these host barriers. pGP3 is a stable trimer in a cell-free system (53) or inside cells during chlamydial infection (54, 55), and it is both associated with the chlamydial organism outer membrane and secreted into the host cytosol (56). The outer membrane-associated pGP3 may promote C. muridarum resistance to gastric acid, while the secreted pGP3 may neutralize the antimicrobial factors in the small intestine. The latter is supported by a previous finding that pGP3 bound to the host antimicrobial cathelicidin peptide LL-37 (human) or CRAMP (mouse) and neutralized its antichlamydial activity (57). The trimer structure of the pGP3 C-terminal trimerization domain has been shown to be similar to the trimer structure of tumor necrosis factor alpha (TNF-α) family members (53), suggesting that the secreted pGP3 may be able to modulate host inflammatory pathways mediated by TNF to promote C. muridarum colonization in the GI tract. It is also possible for pGP3 to target other bacteria or bacterial products to create a niche for C. muridarum to maintain long-term colonization in the colon. Chlamydia is known to take up bacterial metabolites for its own biosynthesis (58–62). The cecum/colon is full of bacterial metabolites. The question is whether pGP3 can promote C. muridarum uptake of bacterial metabolites for biosynthesis. Testing of the hypotheses mentioned above is under way.
Regardless of how pGP3 promotes C. muridarum fitness in the gut, the finding that the plasmid-encoded pGP3 regulates C. muridarum fitness in the GI tract has provided structural evidence demonstrating that C. muridarum may be a commensal species in the mouse GI tract. First, the cryptic plasmid or plasmid-encoded proteins are more important for C. muridarum to colonize the GI tract than for it to infect the genital tract, suggesting that C. muridarum may have acquired the plasmid for adaptation to the GI tract. Second, C. muridarum can establish long-lasting colonization only in the GI tract (44), although it readily infects extra-GI tract mucosal tissues (43). C. muridarum organisms are often eliminated from the extra-GI tract tissues by the host defense since C. muridarum has not fully adapted to the extra-GI tract tissues. Third, like normal commensal bacterial species in the gut, C. muridarum seems to be nonpathogenic in the GI tract (42, 43, 45–47), while C. muridarum often induces inflammatory pathologies in extra-GI tract tissues. That is why C. muridarum infection in the genital tract or airway but not the gut has frequently been used as a disease model (4, 63, 64). Finally, some gut bacterial species have been shown to exacerbate pathologies in non-GI tract tissues, such as joint tissues (65). The question is whether the nonpathogenic colonization of C. muridarum in the GI tract can promote pathologies in the genital tract tissues with ongoing or prior C. muridarum infection. There seems to be a better correlation of C. muridarum pathogenicity in the upper genital tract with C. muridarum colonization in the GI tract than with C. muridarum colonization in the genital tract (Fig. 6), suggesting a role of the GI tract chlamydial organisms in promoting chlamydial pathogenicity in the upper genital tract. An effort is under way to evaluate the role of GI tract chlamydial colonization in genital tract pathogenesis and to further reveal the mechanism by which GI tract chlamydial colonization potentially promotes genital tract pathogenesis.
Finally, it is worth noting that the C. muridarum infection in mice is different from C. trachomatis infection in humans. Thus, great caution should be taken when applying knowledge learned from the murine system to human chlamydial infection. Although C. muridarum infection in the mouse genital tract can induce upper genital tract pathologies and complications similar to those observed in C. trachomatis-infected women (1–3), C. muridarum is unlikely transmitted between mice sexually via the genital tracts. Instead, oral-fecal transmission is more likely. Thus, the GI tract mucosal tissue may serve as the primary selection pressure for C. muridarum to evolve, which may explain why the plasmid or plasmid-encoded proteins are more important for C. muridarum to colonize the GI tract than for it to colonize the genital tract. It is not clear whether the virulence ability acquired by C. muridarum during its adaptation to the GI tract also affords C. muridarum the ability to overcome the cervical barrier for colonizing the upper genital tract. However, C. trachomatis is known to transmit sexually between humans, while transmission by the oral-fecal route is less likely. Thus, the finding of the requirement of the plasmid for C. muridarum to colonize the GI tract may not necessarily be applicable to C. trachomatis infections in humans. Nevertheless, C. trachomatis has been detected in the GI tracts of humans (38–41, 66) and can infect human enteroendocrine cells (67). Women practicing oral-anal sex can introduce C. trachomatis into their GI tracts (39, 41). Thus, both the human genital and GI tract mucosae may provide selection pressure for C. trachomatis adaptation. It will be interesting to investigate whether the plasmid or plasmid-encoded pGP3 is also more important for C. trachomatis to colonize the human GI tract than for it to infect the human genital tract.
MATERIALS AND METHODS
Chlamydia muridarum organisms.Chlamydia muridarum strain Nigg3 (GenBank accession number CP009760.1) (14, 15, 48) organisms were propagated and purified in HeLa cells (human cervical carcinoma epithelial cells; catalog number CCL2; ATCC) as described previously (42, 68). A plasmid-deficient clone (designated CMUT3.G5) was derived from Nigg3 as described previously (13, 30). The pGFP:CM plasmid (69) with or without installation of a premature stop codon in the pgp3, pgp4, pgp5, or pgp7 gene or an in-frame deletion of the pgp8 gene was transformed into the plasmid-deficient CMUT3.G5 clone to produce plasmid-competent (designated CMUT3G5-pGFP or CM-pGFP), pGP3-deficient (CMUT3G5-Pgp3S or CM-pgp3S), pGP4-deficient (CM-pgp4S), pGP5-deficient (CM-pgp5S), pGP7-deficient (CM-pgp7S), or pGP8-deficient (CM-pgp8D) C. muridarum organisms as described previously (13, 30).
Two C. muridarum glycogen synthase A (GlgA) mutants were used in the current study. These mutants were generated by mutagenizing C. muridarum in McCoy cells with ethyl methanesulfonate (EMS; catalog number M0880; Sigma-Aldrich, St. Louis, MO) (70) and screening for mutants that exhibited reduced glycogen staining in vitro. Whole-genome sequencing revealed that two mutant isolates, designated P1D1 and P3B4, had different mutations that would cause amino acid substitutions in conserved residues of GlgA (see Table S2 in the supplemental material). The glgA genotypes of these mutants were C1055T for P1D1 and G1157A for P3B4. Since the mutants were produced by EMS mutagenesis, they also each contained additional unique mutations outside glgA. In vitro experiments have shown an attenuated ability of the glgA mutants to infect HeLa cells by about 3-fold compared to the ability of wild-type C. muridarum (49) and a lack of glycogen synthesis by these mutants (Fig. S1). All C. muridarum organisms mentioned above were propagated and purified as elementary bodies (EBs). Aliquots of the purified EBs were stored at −80°C until use. A wild-type clone isolated from Nigg3, G13.32.1, was used as a control for these mutants.
Mouse infection.Six- to 7-week-old female C57BL/6J mice (stock number 000664) and CBA/J mice (stock number 000656) (both from The Jackson Laboratory, Inc., Bar Harbor, ME) were inoculated with purified C. muridarum EBs at 2 × 105 or 1 × 107 inclusion-forming units (IFU) via different routes as described previously (6, 45) and as indicated above and below for the individual experiments. For genital tract infection via intravaginal and intrabursal inoculations, at 5 days prior to the inoculation, mice were subcutaneously injected with 2.5 mg of colloidal depot medroxyprogesterone (Depo-Provera; Pharmacia and Upjohn LLC, Kalamazoo, MI) suspended in sterile phosphate-buffered saline (PBS). For intravaginal inoculation (14, 15), the EB inoculum in 10 μl sucrose-phosphate-glutamate (SPG) buffer (220 mM sucrose, 12.5 mM phosphate, 4 mM l-glutamic acid, pH 7.5) was delivered into the mouse vagina using a 20-μl micropipette tip. For intrabursal inoculation (71), the mice were anesthetized and placed dorsal side up on a sterile gauze pad with the mouse head facing away. A small incision was made at the dorsomedial position and directly above the ovarian fat pad. After the ovarian fat pad was gently pulled out, the ovary was positioned to allow insertion of a needle (30-gauge removable needles; catalog number 7803-07; Hamilton) into the oviduct tubule. When the needle was inserted into the proper position, it was visible under the bursa. The plunger of the syringe (catalog number 7654-01; Hamilton) was gently pushed to inject the 10 μl of inoculum, after which the needle was quickly removed and the puncture site was gently sealed. Successful injections were indicated by slight distention of the bursa. Finally, the reproductive tract and fat pad were gently reinserted into the peritoneal cavity and the body wall was closed and sutured (Reli sutures; catalog number SK683; Reli, Busan, South Korea). After recovery, the mice were returned to their cages for normal care.
For intragastric inoculation, the mouse was restrained to keep the animal immobile and in a position where the head was slightly tilted back to keep the neck in a straight line for easier insertion of the gavage needle. A 22-gauge, 25-mm-long, 2-mm-diameter, stainless steel, ball-tipped gavage needle (Fisher Scientific, Fair Lawn, NJ) attached to a 1-ml tuberculin syringe was used to deliver the chlamydial inoculum in 100 μl SPG to each mouse stomach. Before insertion of the needle, the length from the tip of nose to the last rib of the mouse was measured. This was done to ensure that the needle traversed the esophagus into the stomach without causing asphyxiation (too short) or perforating the stomach (too long). The gavage needle was inserted into the mouth at either side at about a 45-degree angle in the direction that was chosen. Once inside, the needle was slid toward the back of the throat while maintaining the position of the mouse's head tilted back. The mice were observed carefully after the gavage to ensure that no fluid escaped from the mouth or nose. After inoculation, the mice were monitored for vaginal and rectal live organism shedding or sacrificed for evaluation of the pathology in the upper genital tract.
Animal experiments were performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Research Council (72). The protocol was approved by the Committee on the Ethics of Laboratory Animal Experiments of the University of Texas Health Science Center at San Antonio.
Titration of chlamydial organisms recovered from swab specimens.For the monitoring of organism shedding, vaginal and rectal swab specimens were taken on day 3 postinoculation and weekly thereafter. To quantitate the chlamydial organisms, each swab was soaked in 0.5 ml of SPG and vortexed with glass beads to release the chlamydial organisms into the supernatants for titration of serial dilutions on HeLa cell monolayers in duplicate using an immunofluorescence assay as described previously (16, 26, 44, 48). The total number of IFU per swab was converted into the log10 number of bacteria for calculation of the mean and standard deviation for mice from the same group at each time point.
Mouse genital tract gross pathology evaluation.Upon euthanasia, mouse genital tracts were excised and the gross pathology of the hydrosalpinx was documented by high-resolution digital photography. Hydrosalpinx was further scored according to an ordinal scale, where 0 indicates no hydrosalpinx, 1 indicates hydrosalpinx that is observable only under magnification, 2 indicates visible hydrosalpinx smaller than the ovary, 3 indicates hydrosalpinx roughly equal to the size of the ovary, and 4 indicates hydrosalpinx larger than the ovary. The bilateral hydrosalpinx severity was calculated for each mouse as the summed scores for the left and right oviducts. Hydrosalpinx incidence was calculated as the number of mice with a bilateral score of 1 or higher divided by the number of mice in the group.
Immunofluorescence assay.The immunofluorescence assay used for titrating live organisms was described previously (68, 73). Briefly, HeLa cells grown on coverslips were fixed with paraformaldehyde (Sigma) and permeabilized with saponin (Sigma). After washing and blocking, the specimens were subjected to a combination of antibody and chemical staining. Hoechst (blue; Sigma) was used to visualize the nuclear DNA. A rabbit antichlamydial antibody (raised by immunization with C. muridarum EBs; data not shown) plus a goat anti-rabbit IgG conjugated with Cy2 (green; Jackson Immuno Research Laboratories, Inc., West Grove, PA) was used to visualize chlamydial inclusions. The cell samples after immunolabeling were used for counting the inclusions under an Olympus AX-70 fluorescence microscope equipped with multiple filter sets (Olympus, Melville, NY).
Statistical analyses.The number of organisms, expressed as the number of IFU, and the hydrosalpinx scores were compared between groups using the Wilcoxon rank-sum test. The incidences of hydrosalpinx between groups were evaluated using Fisher's exact probability test (http://vassarstats.net/tab2x2.html). Correlations of chlamydial pathogenicity in the upper genital tract with the number of chlamydial IFU recovered from vaginal and rectal swab specimens were analyzed by calculating Spearman rank-order correlation coefficients (http://vassarstats.net/corr_rank.html). Furthermore, the significance of the difference between two correlation coefficients was also calculated (http://vassarstats.net/rdiff.html).
ACKNOWLEDGMENT
This work was supported in part by grants (to G. Zhong) from the U.S. National Institutes of Health.
FOOTNOTES
- Received 14 June 2017.
- Returned for modification 10 July 2017.
- Accepted 4 October 2017.
- Accepted manuscript posted online 16 October 2017.
Supplemental material for this article may be found at https://doi.org/10.1128/IAI.00429-17.
- Copyright © 2017 American Society for Microbiology.
