Chlamydiae are obligate intracellular bacterial pathogens that exhibit a broad range of host tropism. Differences in host tropism between Chlamydia species have been linked to host variations in IFN-γ-mediated immune responses. In mouse cells, IFN-γ can effectively restrict growth of the human pathogen Chlamydia trachomatis but fails to control growth of the closely related mouse pathogen Chlamydia muridarum. The ability of mouse cells to resist C. trachomatis replication is largely dependent on the induction of a family of IFN-γ-inducible GTPases called immunity-related GTPases or IRGs. In this study we demonstrate that C. muridarum can specifically evade IRG-mediated host resistance. It has previously been suggested that C. muridarum inactivates the IRG protein Irga6 (Iigp1) to dampen the murine immune response. However, we show that Irga6 is dispensable for the control of C. trachomatis replication. Instead, an effective IFN-γ response to C. trachomatis requires the IRG proteins Irgm1 (Lrg47), Irgm3 (Igtp), and Irgb10. Ectopic expression of Irgb10 in the absence of IFN-γ is sufficient to reduce intracellular growth of C. trachomatis but fails to restrict growth of C. muridarum, indicating that C. muridarum can specifically evade Irgb10-driven host responses. Importantly, we find that Irgb10 protein intimately associates with inclusions harboring C. trachomatis but is absent from inclusions formed by C. muridarum. These data suggest that C. muridarum has evolved a mechanism to escape the murine IFN-γ response by restricting access of Irgb10 and possibly other IRG proteins to the inclusion.
Chlamydia trachomatis is an obligate intracellular bacterial pathogen that is the etiologic agent of prevalent human infections causing serious public health problems (1). Ocular infection with C. trachomatis is the leading cause of preventable blindness worldwide (2). Urogenital tract infection with C. trachomatis is the most common bacterial sexually transmitted disease in the United States and can lead to pelvic inflammatory disease, ectopic pregnancies, and infertility (3). The development of a vaccine against C. trachomatis infection has been complicated by the inability to genetically manipulate C. trachomatis and by the absence of a small animal model of C. trachomatis infection that adequately recapitulates human disease (4, 5, 6). A more thorough understanding the immune response to C. trachomatis in mice is a prerequisite for the development of a mouse model that better mimics human C. trachomatis infections.
All Chlamydia genera undergo a biphasic developmental cycle. Infection is initiated by the attachment of the highly infectious, metabolically inert elementary body to host cells. The elementary body subsequently facilitates its own uptake into a membrane-bound vesicle termed an inclusion (7). The inclusion evades fusion with the lysosome and the elementary body rapidly differentiates into a metabolically active reticulate body that replicates by binary fission within the protected environment of the inclusion (8). At the end of the developmental cycle the reticulate bodies differentiate back into elementary bodies, which subsequently egress from the host cell to initiate new rounds of infection.
The common cell biology of different Chlamydia sp. is likely to be based on the near synteny of the different Chlamydia genomes (9, 10, 11). For instance, the human pathogen C. trachomatis and the mouse pathogen Chlamydia muridarum share >99% of all predicted open reading frames and genetically diverge primarily in a small hypervariable region of the chromosome termed the plasticity zone (9). However, despite their extensive sequence homology and common cell biology, these two Chlamydia species display distinct host tropism (6).
Different aspects of host-pathogen interactions determine host tropism, including the adaption of pathogens to host-specific immune responses (5). A pivotal element of mammalian immunity consists of responses elicited by the cytokine IFN-γ. Accumulating evidence suggests that the ability of distinct Chlamydia sp. to circumvent host-specific IFN-γ responses plays an important role in determining host tropism (5, 12, 13, 14, 15).
In human cells, an important mediator of cell-autonomous resistance to C. trachomatis is the induction of the enzyme IDO by IFN-γ. IDO catabolizes intracellular pools of tryptophan, thereby starving the pathogen of this essential amino acid (16, 17). Chlamydiae deprived of tryptophan enter a nonreplicating, persistent state (18, 19). However, genital C. trachomatis strains but not C. muridarum can overcome the IDO-dependent growth restriction. It has been found that C. trachomatis strains isolated from the human urogenital tract express tryptophan synthase, an enzyme capable of using exogenous indole for the synthesis of tryptophan (14, 20). Supplementing indole can restore the capacity of C. trachomatis genital isolates to replicate in IFN-γ-treated human cells (20). The ability to synthesize tryptophan may allow C. trachomatis strains to escape IFN-γ-mediated suppression in the genital tract by using indole provided by the local microbial flora. In contrast, C. muridarum lacks a functional tryptophan synthase and supplementing indole does not restore bacterial growth in IFN-γ-treated human cells (9, 14).
In mouse cells, growth of C. trachomatis and C. muridarum is also differentially affected by IFN-γ treatment. In most murine cell lines, IFN-γ treatment drastically reduces growth of C. trachomatis but not C. muridarum (13, 14). A family of IFN-γ-inducible p47 GTPases termed immunity-related GTPases (IRGs)3 has been implicated in IFN-γ-mediated suppression of C. trachomatis growth in mouse cells (13, 21). We have previously shown that mouse embryonic fibroblasts (MEFs) depleted of Irgb10 mRNA by RNA interference (RNAi) or carrying a targeted gene deletion of Irgm3 are defective in IFN-γ-mediated resistance to C. trachomatis (21). In a separate study, Nelson et al. showed that transfection with RNAi oligonucleotides targeting Irga6 led to a 2- to 3-fold increase in recoverable inclusion-forming units in IFN-γ activated mouse epithelial cells (13). The authors suggested that the gene TC438 encoded in the plasticity zone of C. muridarum but absent from the C. trachomatis genome is responsible for immune evasion by C. muridarum in mice. TC438 encodes a large protein with homology to the UDP-glycosyltransferase portion of the large clostridial toxin and the Yersinia virulence factor YopT. The Yersinia effector YopT acts as a cysteine protease to remove the lipid modification from Rho GTPases at a cleavage site that is just N-terminal from the cysteine linked to the prenyl group (22). This cleavage releases the GTPase from the membrane and inactivates it. Nelson et al. suggested that TC438 inactivates the GTPase Irga6 by a similar proteolytic mechanism (13).
To test the hypothesis that C. muridarum interferes with Irga6 function, we first set out to confirm the role for Irga6 as a host factor that conveys resistance to C. trachomatis growth. Surprisingly, we found that mice carrying a gene deletion of Irga6 were as resistant to C. trachomatis infections as coisogenic C57BL/6 (B6) mice both in a systemic infection model in vivo and in IFN-γ-treated cells ex vivo. Instead, we found that the IRG family members Irgb10, Irgm1, and Irgm3 are required for resistance to C. trachomatis. We demonstrate that ectopic expression of Irgb10 in the absence of a general IFN-γ response is sufficient to reduce bacterial yield of C. trachomatis but not C. muridarum. The differential effect of Irgb10 expression on Chlamydia growth correlated with the distinct subcellular localization of Irgb10 in cells infected with C. trachomatis as compared with those infected with C. muridarum. Specifically, we show that Irgb10 protein decorates an inclusion formed by C. trachomatis, whereas Irgb10 is absent from C. muridarum inclusions. Therefore, we suggest a model in which Irgb10 is required to localize to the inclusion to limit bacterial replication and propose that C. muridarum has evolved a mechanism to restrict access of IRG proteins to its inclusion.
Materials and Methods
All mice were maintained and bred under specific pathogen-free conditions. All experiments were approved by the Institutional Animal Care and Use Committee of Harvard Medical School (Boston, MA). Control B6 and B6.129S2-Irf1 mice were obtained from The Jackson Laboratory. The targeted gene deletions of Irgm1 and Irgm3 have been described previously (23, 24). Irgm1−/− and Irgm3−/− mice used in this study were backcrossed for >10 generations to B6. The Irgm1−/− mice used in this study carry the B6 alleles of the IRG genes, Irgm2, Irgm3, and Irgb10. The Irgm3−/− mice carry the 129 alleles of Irgm2 and Irgb10 (data not shown). Irga6−/− mice have also been described previously (25) and were generated in B6-derived (Bruce4) embryonic stem cells. The third exon encoding the whole open reading frame of Irga6 was flanked by LoxP sites and deleted in vivo by crossing to Cre-deleter mice (26). A congenic mouse strain that carries a small 129 genomic interval encompassing the Ctrq3 locus (B6.Ctrq3129) was generated by taking advantage of the fact that most mouse knockout strains have been made using 129-derived embryonic stem cell lines and subsequently backcrossed to B6 mice, thus creating 129 congenic mice on a B6 background. To make the B6.Ctrq3129 strain, the congenic mouse strain B6.129S2-Irf1 carrying a targeted gene deletion in Irf1 on chromosome 11 was further backcrossed to B6 to generate mice that had lost the Irf1 knockout allele but still carried an interval of 129-derived DNA on chromosome 11 surrounding the Ctrq3 locus. These mice were then intercrossed to generate mice homozygous for the Ctrq3129 allele. The 129 genomic interval of the B6.Ctrq3129 mice encompasses 2.1 Mb of DNA containing the 129 alleles of Irgm2 and Irgb10 and is flanked by the markers D11Zbh3 and D11Zbh5 (data not shown).
Genetic markers and genotyping
Genotyping of the knockout alleles for Irgm1 and Irgm3 was performed as described previously (23, 24). Simple sequence length polymorphism markers were used to determine the boundaries of congenic intervals as described previously (21). Primer sequences for marker D11Zbh3 has previously been reported (21). The primer sequences for marker D11Zbh5 are as follows: D11Zbh5 forward, 5′-AATGGAGGTCCAGTGTTACTGC-3′; D11Zbh5 reverse, 5′-TTGAACAGAATCAACAGCAACC-3′.
Chlamydia strains and evaluation of bacterial yield
C. trachomatis serovar L2 434/Bu and C. muridarum were propagated in McCoy cells and purified as described (27, 28). To quantify the bacterial load in Chlamydia-infected cells and in the spleens of infected animals, a previously described quantitative PCR assay was applied (27). Briefly, total nucleic acid from infected cells or spleen homogenates was prepared using the QIAamp DNA mini kit from Qiagen. Chlamydia 16S DNA and mouse GAPDH DNA content of individual samples was then quantified by quantitative PCR on an ABI 7000 sequence detection system using primer pairs and dual-labeled probes. Standard curves were generated from known amounts of Chlamydia and mouse DNA, and these curves were used to calculate the amount (in picograms) of Chlamydia DNA per unit weight (in micrograms) of mouse DNA in the samples. For in vitro experiments, bacterial yield is shown relative to a control sample in arbitrary units with the control being equal to 1.
Unless stated otherwise, cells were treated with 100 U/ml rIFN-γ (Invitrogen) for 15 h before infection or left untreated. Cells were infected with Chlamydia at a multiplicity of infection (MOI) of 1 in SPG buffer (220 mM sucrose, 12.5 mM phosphate, and 4 mM l-glutamic acid (pH 7.5)) by centrifugation at 1928 × g for 1 h at 37°C and then returned to standard medium. At 29 h postinfection (hpi) cells were harvested and bacterial yield was determined as described above. All experiments were conducted in triplicate. Mice, 6- to 8-wk old, had 107 inclusion-forming units of C. trachomatis L2 in 200 μl of SPG buffer injected into the tail vein. At least three mice per genotype were used for each experiment. At 29 hpi mice were sacrificed and the splenic bacterial load was determined as previously described (27).
Cell culture, Abs, Irgb10-GFP, and microscopy
MEFs were generated from the indicated mouse strains as previously described (27). Cell lines were cultured according to the standard protocols provided by the American Tissue Culture Collection. Retroviral transduction of MEFs was conducted using a mouse stem cell virus vector. For immunofluorescence studies, cells were grown and infected on glass coverslips. Cells were fixed in methanol for 5 min, washed in PBS, and blocked with 5% goat serum. For C. trachomatis infections, cells were stained with a mouse mAb against the C. trachomatis major outer membrane protein (MOMP; Accurate Chemical and Scientific). For C. muridarum infections, cells were stained with a mouse mAb to Chlamydia LPS (RDI-Fitzgerald) or a mouse monoclonal anti-MOMP Ab specific for C. muridarum (29). An affinity-purified polyclonal rabbit anti-Irgb10 Ab was generated against the C-terminal peptide LKKKVFLQDSVDSE of Irgb10 and used to detect Irgb10 by microscopy. The following Abs were used to stain for additional IRG proteins: rabbit anti-Irga6 165 (30), goat anti-Irgm1 A19 (Santa Cruz Biotechnology), mouse anti-Irgm1 (31), mouse anti-Irgm3 I68120 (BD-Transduction Labs), and rabbit anti-Irgm3 (32). Primary Abs were visualized by staining with secondary FITC- or rhodamine red X-conjugated donkey anti-rabbit IgG or anti-mouse IgG (Jackson ImmunoResearch Laboratories). To determine the localization of Irgb10 in C. trachomatis/C. muridarum coinfections, fixed cells were labeled with rabbit anti-Irgb10 Ab and mouse anti-MOMP Ab specific for C. muridarum (29). Primary Abs were visualized by staining with secondary rhodamine red X-conjugated donkey anti-rabbit IgG and either Cy5- or 7-amino-4-methylcoumarin-3-acetic acid (ACMA)-conjugated donkey anti-mouse IgG. Subsequently, cells were stained with FITC-coupled mouse anti-MOMP Ab specific for C. trachomatis. To visualize Irgb10 protein directly, the B6 cDNA of Irgb10 was cloned into the vector pEGFP-N2 (Invitrogen). The generated vector encodes Irgb10 with a C-terminal GFP moiety and was transfected into cells using FuGene6 (Invitrogen) according to the manufacturer’s protocol. All epifluorescent images were acquired with a Nikon Eclipse TE2000-U microscope and Plan Fluor ELWD ×60 objective. Images were saved as TIFF files and imported into Adobe Illustrator for labeling.
All comparisons were evaluated for statistical significance through the use of unpaired two-tailed t tests. Where it appeared necessary to highlight significant differences between data points, the level of significance is depicted by ∗ for p ≤ 0.05 and ∗∗ for p ≤ 0.005.
IFN-γ activated MEFs fail to restrict growth of C. muridarum
IFN-γ treatment of most murine cells activates a highly effective antibacterial response to C. trachomatis but fails to restrict growth of C. muridarum. We and others have investigated whether IRG proteins are required to restrict the growth of C. trachomatis in mouse cells (13, 21). It has been suggested that C. muridarum targets and inactivates IRG proteins to overcome the murine IFN-γ response (13); however, no experimental evidence has been provided to support this hypothesis. To investigate the mechanism by which C. muridarum evades cell-autonomous immune restriction, we first tested whether C. muridarum can escape the IFN-γ response in MEFs, a cell type that requires the IRG proteins Irgb10 and Irgm3 for resistance to C. trachomatis (21). As we have shown previously, B6 MEFs treated with 100 U/ml IFN-γ became highly resistant to C. trachomatis growth (21, 27). Bacterial yield from IFN-γ-treated B6 MEFs was ∼20-fold lower than the yield from untreated cells at 29 hpi (Fig. 1). In contrast, C. muridarum growth was impervious to IFN-γ treatment, even at high cytokine concentrations of 103 U/ml (Fig. 1). MEFs therefore can be used as a model system to study immune evasion by C. muridarum.
Irga6 is not required for the immune response to C. trachomatis
Studies by Nelson et al. showed that an RNAi oligonucleotide duplex targeting Irga6 partially reverted IFN-γ-induced growth inhibition of C. trachomatis in mouse oviduct epithelial cells (13). The authors therefore suggested that C. muridarum blocks the function of Irga6 protein to overcome the cell-autonomous immune restriction imposed by IFN-γ activation (13). In this study, we used a recently described mouse strain carrying a Irga6 knockout allele in a pure B6 genetic background (25). Surprisingly, we found that Irga6−/− MEFs were not defective but more efficient in restricting growth of C. trachomatis compared with B6 control MEFs in IFN-γ-treated but not untreated cells (Fig. 2,A). Moreover, in systemically infected mice the splenic C. trachomatis loads of Irga6−/− mice were not significantly different from the bacterial yields in B6 mice (Fig. 2 B) These data collectively show that Irga6 is not required for the immune response to C. trachomatis and that the Irga6 protein is unlikely to be the main target of the immune evasion mechanism used by C. muridarum.
Irgm3 and Irgm1 are required for complete immune restriction of C. trachomatis growth
Previously, we mapped a small 1.3-Mb genetic interval derived from mouse strain C3H/HeJ that conveys susceptibility to C. trachomatis replication both in MEFs and in a systemic infection model (21, 27). This locus, termed Ctrq3, contains the IRG genes Irgb10, Irgm2, and Irgm3 (21, 33). We have previously shown that IFN-γ-treated Irgm3−/− MEFs are more susceptible to C. trachomatis replication than B6 MEFs (21). Similarly, a recent study found that Irgm3−/− mice are more susceptible to Chlamydia psittaci infections (33). Because the Irgm3 knockout allele was generated in 129-derived embryonic stem cells (24), Irgm3−/− mice on a B6 genetic background carry an Irgm3 knockout allele that is flanked by 129-derived DNA encompassing the 129 alleles of Irgm2 and Irgb10 (data not shown). Miyairi et al. raised the interesting possibility that the phenotype of the Irgm3−/− mice was due to the 129 alleles of Irgm2 and/or Irgb10 rather than the targeted gene deletion of Irgm3 itself (33). To test this hypothesis, we generated a congenic mouse strain that carries a 2.1-Mb 129-derived DNA interval encompassing the genes Irgb10, Irgm2, and Irgm3 on a B6 genetic background (B6.Ctrq3129). IFN-γ-treated MEFs derived from the congenic B6.Ctrq3129 mice allow ∼3-fold more C. trachomatis replication than B6 MEFs (Fig. 3). This result strongly suggests that both the C3H and the 129 alleles of Ctrq3 confer increased susceptibility to C. trachomatis infections compared with the B6 Ctrq3 allele, although we currently do not know whether the same polymorphisms are responsible for the increased susceptibility conveyed by the 129 and C3H alleles of Ctrq3. Importantly, IFN-γ-treated Irgm3−/− MEFs were nearly 10-fold more susceptible to C. trachomatis growth than B6.Ctrq3129 MEFs, indicating that the Irgm3 knockout allele itself also conveys increased susceptibility to C. trachomatis infections. We next tested the role of Irgm3 in controlling systemic C. trachomatis infections. Similar to the results obtained in MEFs, Irgm3−/− mice were significantly more susceptible than B6.Ctrq3129 mice, and both Irgm3−/− and B6.Ctrq3129 mice displayed deficiencies in their early immune responses to C. trachomatis compared with B6 mice (Fig. 3). Collectively, these results indicate a pivotal role of Irgm3 in the innate immunity of mice to C. trachomatis.
Based on the GTPase motif sequence, Irgm1 and Irgm2 are the closest homologues of Irgm3 and, therefore, are also attractive candidates as mediators of host resistance to C. trachomatis (34). An Irgm2−/− mouse has not been reported to exist, but Irgm1−/− mice have been shown to display decreased resistance against all intracellular bacterial pathogens tested to date (23, 35, 36, 37, 38, 39, 40). The Irgm1−/− mice used in our experiments have been backcrossed for 12 generations and carry the B6 allele of Ctrq3 (data not shown). Both in IFN-γ activated MEFs and in systemically infected mice Irgm1−/− mice displayed a 5- to 6-fold increase in C. trachomatis yield at 29 hpi (Fig. 4). These data establish that, similar to Irgm3, Irgm1 is also important in resistance to infections with C. trachomatis.
Ectopic expression of Irgb10 restricts growth of C. trachomatis but not C. muridarum
Twenty-three IRG genes exist in the mouse genome and the majority of these genes have been shown to be IFN-γ inducible (34). The data shown above and in our previous work demonstrate that at least three IRG proteins are required for an efficient early immune response to C. trachomatis (21, 27). However, in the human genome only two IRG genes exist, neither of which are induced upon IFN-γ stimulation (34). Because IFN-γ-mediated induction of IRG genes is only seen in mouse and not in human cells, we hypothesized that one or more IRG genes are the primary targets of the immune evasion mechanism used by C. muridarum. To test this hypothesis, we first attempted to restrict Chlamydia growth by ectopic expression of individual IRG proteins in the absence of IFN-γ stimulation. We found that ectopic expression of either Irgm1 or Irgm3 alone had no discernible effect on C. trachomatis growth (data not shown). However, expression of Irgb10 in MEFs (Fig. 5,A) or 3T3 cells (data not shown) resulted in a relatively small but reproducible 2- to 3-fold reduction in C. trachomatis yield at 29 hpi. Importantly, expression of Irgb10 had no effect on C. muridarum growth (Fig. 5 A). These data show that in the absence of other effects that might be induced by IFN-γ, Irgb10 expression alone is sufficient to moderately reduce the growth of C. trachomatis but not C. muridarum.
Irgb10 and Irga6 associate with inclusions formed by C. trachomatis but not by C. muridarum
Most IRG proteins examined to date have been found to localize to diverse membrane compartments and to associate with the pathogen-containing vacuole of organisms like Toxoplasma gondii and Mycobacterium tuberculosis (25, 30, 31, 37, 41, 42, 43, 44). Results from these studies imply that localization to the pathogen-containing vacuole is essential for the antimicrobial effector function of at least a subset of IRG proteins. We therefore examined the subcellular localization of Irgb10 in C. trachomatis-infected cells using an Irgb10 protein fused to GFP at its C terminus. In C. trachomatis-infected cells, we found Irgb10-GFP intimately associated with the inclusion (Fig. 5,B). In contrast, Irgb10-GFP was absent from C. muridarum-containing inclusions (Fig. 5,B). Noticeably, ectopic expression of Irgb10-GFP led to the formation of large Irgb10-GFP aggregates in uninfected cells (Fig. 5,B). Similar to the effect of IFN-γ on aggregates formed by ectopically expressed Irga6 (38), stimulation of MEFs with IFN-γ led to the disappearance of Irgb10-GFP aggregates and dispersion of Irgb10-GFP throughout the cell (data not shown). These data indicate that the subcellular localization of Irgb10 is regulated by additional IFN-γ-inducible factors. To determine the subcellular localization of Irgb10 during the IFN-γ response, we labeled for endogenous Irgb10 using a polyclonal Ab directed against the very C-terminal peptide of Irgb10. IFN-γ-activated, uninfected cells displayed weak staining with anti-Irgb10 that could not be clearly associated with any particular intracellular structure (Fig. 6,A and data not shown). However, in C. trachomatis-infected cells the localization of Irgb10 to the inclusion could be observed as early as 10 hpi and throughout the course of the infection (Fig. 6,A and data not shown). At 20 and 24 hpi, 30 - 40% of the C. trachomatis inclusions stained positive for Irgb10 (Fig. 6,C and Fig. 7). In contrast, after examining >1,000 inclusions we never found Irgb10 to be associated with C. muridarum (Fig. 6 and data not shown).
To determine whether C. muridarum restricts the access of additional IRG proteins to its inclusions, we analyzed the subcellular localization of the IRG proteins Irgm1, Irgm3, and Irga6 in C. trachomatis- and C. muridarum-infected cells. For Irgm1 and Irgm3 we were not able to detect colocalization with inclusions formed by either Chlamydia species at multiple time points postinfection (7, 16, 20, and 24 hpi) using two distinct Abs for each protein (Fig. 6,C and data not shown). The IRG protein Irga6, in contrast, associated with C. trachomatis at a frequency similar to that of Irgb10 but was absent from C. muridarum inclusions (Fig. 6, B and C). These results demonstrate that Irgb10 and Irga6, two IRG proteins strongly associated with C. trachomatis inclusions, fail to accumulate at inclusions formed by C. muridarum.
Coinfection with C. muridarum reduces the number of Irgb10-positive C. trachomatis inclusions
It has recently been reported that coinfection with C. muridarum protects C. trachomatis inclusions from IFN-γ-mediated growth restriction in mouse cells (45). To examine whether the inhibition of Irgb10 function may be involved in the rescue of C. trachomatis growth by C. muridarum, we tested whether coinfection with C. muridarum also restricts the access of Irgb10 to C. trachomatis inclusions. To specifically identify inclusions harboring C. trachomatis, we stained cells with a species-specific anti-MOMP Ab that detects C. trachomatis but not C. muridarum. We observed that coinfection with C. muridarum significantly decreased the number of Irgb10-positive C. trachomatis inclusions in a dose-dependent manner (Fig. 7,A). To determine whether C. muridarum must reside in the same cell as C. trachomatis to restrict the access of Irgb10 to the C. trachomatis inclusion, we triple-stained coinfected cells with anti-Irgb10 and two anti-MOMP Abs that distinguish between the two Chlamydia species. Our data show that colocalization of Irgb10 with C. trachomatis inclusions is exclusively reduced in cells harboring both Chlamydia strains. The most dramatic reduction in Irgb10 colocalization is seen with C. trachomatis inclusions that have fused with C. muridarum inclusions (Fig. 7 B). Collectively, these data suggest that C. muridarum can actively block the acquisition of Irgb10 protein to Chlamydia inclusions and thus avoid an Irgb10-mediated antibacterial response.
Bacterial pathogens have evolved a large repertoire of mechanisms to subvert both innate and adaptive immune reactions (46, 47), including the evasion of the NO synthase-dependent effector branch of the IFN-γ response (48). In this report we demonstrate that bacterial pathogens can directly counteract the IRG-dependent arm of the IFN-γ response. Specifically, we show that the mouse pathogen C. muridarum can evade an antimicrobial response driven by the IRG protein Irgb10.
It has previously been reported that C. muridarum but not C. trachomatis is resistant to the cell-autonomous antibacterial effect of IFN-γ in mouse cells, and it has been suggested that the specific inhibition of the IRG protein Irga6 by C. muridarum is the underlying cause for the resistance (13). However, no conclusive experimental evidence to support this model has been provided. In our study we used Irga6 knockout mice to show that Irga6 is not required for the IFN-γ response to C. trachomatis in fibroblasts and splenocytes and thus conclude that C. muridarum is unlikely to evade the IFN-γ response in mouse cells through the inhibition of Irga6. The discrepancy of our results with the previous report by Nelson et al. (13) may indicate a unique requirement for Irga6 in resistance to C. trachomatis in epithelial cells. Alternatively, technical differences between the two studies could explain the divergent findings; whereas we analyzed knockout mice, Nelson et al. based their conclusions on the effect of a single RNAi oligonucleotide pair targeting Irga6 (13), an experimental approach that frequently produces off-target effects (49, 50).
Surprisingly, we found that the absence of Irga6 enhances the IFN-γ response to C. trachomatis in MEFs. At least two distinct but not mutually exclusive scenarios may explain the observed phenotype. Irga6 has previously been shown to multimerize (51), and it has been suggested that IRG proteins may engage in direct heteromeric interactions with one another (38). Accordingly, Irgb10 and Irga6 could be competing for binding partners (in particular, other IRG proteins) and Irgb10 may acquire these common binding partners more effectively in Irga6−/− MEFs. Additionally, our data show that Irga6 can access the inclusion, even though Irga6 is not required for the restriction of C. trachomatis growth by IFN-γ. Consequently, the absence of Irga6 may allow Irgb10 to access the inclusion more effectively. Although our data suggest that Irga6−/− and B6 mice are similarly susceptible to systemic infections with C. trachomatis, a more careful analysis of the in vivo phenotype (including the analysis of additional time points and bacterial load in other organs) is required to ultimately resolve this issue.
We have previously shown that another IRG protein, Irgb10, plays a central role in IFN-γ-dependent resistance to C. trachomatis in mice (21, 27). In this report we make the striking observation that Irgb10 colocalizes with C. trachomatis inclusions but is not detectable at C. muridarum inclusions. Similarly, Irga6 associates with C. trachomatis but not C. muridarum inclusions. We therefore suggest that C. muridarum escapes the IRG-mediated antibacterial immune response by disallowing Irgb10 and possibly other IRGs access to its inclusion.
In contrast to Irgb10 and Irga6, we did not observe any association of Irgm1 and Irgm3 with Chlamydia inclusions. Analogous to our findings, it has previously been reported that Irgm1 does not localize to the parasitophorous vacuole harboring T. gondii, although Irgm1 is required for growth restriction of this parasite (31). Collectively, these observations suggest that the localization of Irgm proteins to pathogen-containing vacuoles is not essential for this subclass of IRG proteins to act as resistance factors.
Two mechanistically distinct models could explain the differential localization of Irgb10 in C. trachomatis- and C. muridarum-infected cells. In the first model, Irgb10 protein recognizes properties specific to C. trachomatis inclusions. In support of this model, inclusions from C. trachomatis and C. muridarum have been reported to show some differences in their protein composition. For instance, the small Ras-like GTPase Rab6 is associated with C. trachomatis but not C. muridarum inclusions (52). However, we do not favor this model because identical IRG proteins appear to localize to vastly different vacuoles surrounding different intracellular pathogens (25, 37, 38, 41). This also appears to hold true for Irgb10, which not only localizes to the C. trachomatis inclusion but to at least one more vacuole engulfing a pathogen, namely the parasitophorous vacuole of T. gondii (J. C. Howard, unpublished data). Moreover, inclusions containing C. trachomatis can evade targeting by Irgb10 in cells coinfected with C. muridarum. These data strongly suggest that C. muridarum actively inhibits the localization of Irgb10 to the Chlamydia inclusion.
We therefore favor a second model in which C. muridarum actively constrains access of Irgb10 to its inclusion, either by removing Irgb10 from the inclusion membrane or by preemptively blocking trafficking of Irgb10 to the inclusion. Mechanistically, inhibition of Irgb10 membrane targeting could be achieved through degradation of this host resistance factor by a bacterial virulence factor. However, we did not observe any reduction in protein levels of the IRG proteins Irgb10, Irga6, Irgm1, or Irgm3 in C. muridarum-infected cells (J. Coers, unpublished data). Alternatively, C. muridarum may modify the biochemical properties of Irgb10 required for membrane binding. Interestingly, the N-terminal peptide of Irgb10, MGQSSSKPDAKAHNMASS-, constitutes a highly probable site for myristoylation and, based on the structure-function analysis of another myristoylated IRG protein, Irga6, this lipid modification is most likely required for membrane targeting (30). Although no eukaryotic or bacterial demyristoylation enzymes are known to exist, changes in protein conformation can sequester the myristate moiety in a hydrophobic pocket within the protein to cause membrane detachment, a mechanism referred to as the “myristoyl switch” (53). The best described example for myristoyl switching is the regulation of membrane binding of the GTPase ADP ribosylation factor-1 (Arf-1) through alternative binding of GDP and GTP (54, 55). A similar myristoyl switch can easily be envisioned for myristoylated IRG proteins that are known to bind both GDP and GTP (51). Future studies should explore the effect of C. muridarum infections on the ratio of GTP- and GDP-bound forms of Irgb10 and Irga6 in host cells. Lastly, it is also conceivable that a bacterial effector molecule derived from C. muridarum targets not Irgb10 directly but a separate host factor required for localization of Irgb10 to the Chlamydia inclusion.
An important unanswered question is the identity of the bacterial effector molecule(s) required for the evasion of the IRG-mediated antimicrobial host responses by C. muridarum. Because C. muridarum blocks Irgb10 localization to C. trachomatis inclusions only in coinfected cells, we suggest that C. muridarum secrets an inhibitory factor directly into the host cell (and not into the medium). Although C. muridarum can inhibit Irgb10 accumulation at C. trachomatis inclusions in trans (between two inclusions in the same cell), an efficient reduction in the number of Irgb10+ C. trachomatis inclusions requires the two Chlamydia species to reside in the same inclusion. To explain these results, we propose the existence of a factor secreted by C. muridarum into the cytosol and the subsequent association of this factor with the inclusion membrane. If two inclusions are in close proximity to one another, this putative factor may associate with a C. trachomatis inclusion in trans and inhibit Irgb10 localization to the C. trachomatis inclusion. The rare occurrence of Irgb10+ fused inclusions (containing C. muridarum and C. trachomatis) could be a consequence of C. muridarum inclusions fusing with C. trachomatis inclusions that have already acquired Irgb10.
It has previously been proposed that a cytotoxin containing a domain with sequence homology to the Yersinia-derived virulence factor YopT may be the bacterial factor responsible for immune evasion (13). This hypothesis was based on the argument that Yersinia-derived YopT binds to and proteolytically inactivates prenylated Rho GTPases and that the GTPase Irga6 is the only IRG containing a C-terminal peptide sequence (-CLRN) that is reminiscent of a CaaX isoprenylation sequence (13). The authors therefore suggested that the Chlamydia YopT homologue TC438 could inactivate Irga6 through cleavage of the lipidic moiety and thus confer resistance to the murine IFN-γ response (13). However, this curious hypothesis was constructed in the absence of any experimental evidence for prenylation of Irga6 and failed to consider that the arginine in the penultimate position of the -CLRN sequence made prenylation of Irga6 unlikely (56). Furthermore, we have shown here that inactivation of Irga6 cannot be sufficient to achieve immune evasion. Moreover, despite carrying a YopT homologue in its genome, another Chlamydia strain, Chlamydia caviae, is sensitive to IFN-γ treatment in mouse cells (45). Collectively, these data refute the model proposed by Nelson et al (13). Although it cannot be excluded that the Chlamydia YopT homologue plays a role in immune evasion, we need to consider other C. muridarum genes as the mediators of immune escape from the murine IFN-γ response.
Members of the IRG family are key mediators of host resistance to a large number of intracellular bacterial and protozoan pathogens in mice (36, 38, 40). At least 23 IRG genes are found in the mouse genome (34). In contrast, humans possess only two IRG genes: IRGC, which is exclusively expressed in the male gonad similarly as its mouse ortholog, and IRGM, which is a truncated, noninducible ortholog of the mouse Irgm subclass (34). Surprisingly, human IRGM, despite its truncation and lack of IFN-mediated induction, has been implicated in cell-autonomous resistance to Mycobacteria and shown to play a role in the induction of autophagy (57). However, the relative subtlety of the observed phenotype of IRGM-depleted human cells compared with the astonishing phenotypes of the IRG knockout mice suggests that IRG-mediated immunity is of far less importance in humans than it is in mice. Because functional IRG genes appear to be the norm for mammalian genomes (34), the lack of an IFN-inducible, IRG-mediated immune response in primates is surprising and raises questions concerning the evolution of the IRG family. It is currently unclear whether other molecules or distinct pathways in humans substitute functionally for the IRG-dependent branch of the IFN-γ response. Remarkably, the IRG protein Irgm1 has recently been shown to play a role as a negative regulator of proinflammatory signaling in mice (58), and at the same time a separate study showed significant association of a single nucleotide polymorphism in human IRGM with the autoinflammatory syndrome Crohn’s disease (59). Therefore, it is conceivable that the anti-inflammatory function of Irgm1 has been preserved in humans while the antimicrobial function of IRG proteins is largely absent. The apparent difference in the IFN-γ responses of mice and men is another example for the remarkable divergence in the immune responses of two relatively closely related host species (60). This variability in host immune responses creates a formidable obstacle for pathogens to cross from one host species to another and may be an important factor in shaping the host tropism of pathogens.
We thank Kerry McAuliffe for technical assistance and David Gondek and Bethany Mingle for critical reading of the manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health Grants AI062827 (to M.N.S.), AI57831 (to G.A.T.), a Veteran Affairs Merit Review Grant (to G.A.T.), and the SFB670 Grant “Cell Autonomous Immunity” (to J.C.H.). J.C. was supported by a research fellowship from the Deutsche Forschungsgemeinschaft and by the Charles A. King Trust Postdoctoral Research Fellowship.
Abbreviations used in this paper: IRG, immunity-related GTPase; B6, C57BL/6; hpi, hours postinfection; MEF, mouse embryonic fibroblast; MOI, multiplicity of infection; MOMP, major outer membrane protein; RNAi, RNA interference.