Perforin Is Detrimental to Controllinγ C. muridarum Replication In Vitro, but Not In Vivo

CD4 T cells are critical for clearing experimental Chlamydia muridarum genital tract infections. Two independent in vitro CD4 T cell mechanisms have been identified for terminating Chlamydia replication in epithelial cells. One mechanism, requiring IFN-γ and T cell-epithelial cell contact, terminates infection by triggering epithelial production of nitric oxide to chlamydiacidal levels; the second is dependent on T cell degranulation. We recently demonstrated that there are two independent in vivo clearance mechanisms singly sufficient for clearing genital tract infections within six weeks; one dependent on iNOS, the other on Plac8. Redundant genital tract clearance mechanisms bring into question negative results in single-gene knockout mice. Two groups have shown that perforin-knockout mice were not compromised in their ability to clear C. muridarum genital tract infections. Because cell lysis would be detrimental to epithelial nitric oxide production we hypothesized that perforin was not critical for iNOS-dependent clearance, but posited that perforin could play a role in Plac8-dependent clearance. We tested whether the Plac8-dependent clearance was perforin-dependent by pharmacologically inhibiting iNOS in perforin-knockout mice. In vitro we found that perforin was detrimental to iNOS-dependent CD4 T cell termination of Chlamydia replication in epithelial cells. In vivo, unexpectedly, clearance in perforin knockout mice was delayed to the end of week 7 regardless of iNOS status. The discordant in vitro/in vivo results suggest that the perforin’s contribution to bacterial clearance in vivo is not though enhancing CD4 T cell termination of Chlamydia replication in epithelial cells, but likely via a mechanism independent of T cell-epithelial cell interactions.


Introduction
Chlamydia trachomatis genital tract infections have been, and continue to be, the most common bacterial sexually transmitted infections in Western societies [1]. Antibiotic-based public health measures have not decreased the incidence of infection and may be partially counterproductive, reducing Chlamydia immunopathology but also herd immunity [2][3][4]. Based on this epidemiologic reality many public health officials and sexually transmitted disease healthcare providers believe that a Chlamydia vaccine is the only intervention likely to reduce the incidence of infections and their associated monetary and personal costs.
Critical to a vaccine effort is defining the relevant host defense mechanism for protective immunity; i.e. the immunologic goal of vaccination. This is especially important for Chlamydia vaccines as early attempts with crude whole Chlamydia vaccines were associated with poor protection and excess immunopathology in humans and experimental primate models [5]. Investigations into Chlamydia genital tract protective immunity in the C. muridarum model have shown that CD4 T cells are critical to bacterial clearance [6], while CD8 T cells have been associated with immunopathology [7,8].
Identification of an effector cell type does not equate a mechanism of protective immunity. The first identified mechanism for CD4 T cell-mediated termination of Chlamydia replication in epithelial cells was described in 1996. In vitro Chlamydia-specific Th1 cells terminated replication by inducing epithelial nitric oxide production to chlamydiacidal levels through a combination of IFN-c and T cell-epithelial cell contact [9,10]. However in vivo studies were discordant as mice deficient in inducible nitric oxide synthetase (iNOS) resolved C. muridarum genital tract infections [11,12]; suggesting that nitric oxide was either irrelevant to bacterial clearance or redundant in the setting of a second independent clearance mechanism.
We recently showed that clearance of C. muridarum from the genital tracts of C57BL/6 mice during 8 weeks following infection was dependent on having either a functional iNOS or Plac8 mechanism. Mice singly deficient nitric oxide production (treated with iNOS inhibitor N-monomethyl-L-arginine, MLA) or deficient in Plac8 (genetic knockout) cleared C. muridarum genital tract infections at ,6 weeks compared to ,4 weeks for wild type mice. Dual deficient mice, genetically-deficient in Plac8 and pharmacologically-deficient in iNOS activity (MLA treated) were effectively unable to clear a genital tract infection over 8 weeks with 11 of 12 mice still shedding an average of 1300 IFU/mouse on day 56. However, both iNOS-dependent (nitric oxide) and Plac8-dependent mechanisms were required for sterilizing immunity as viable C. muridarum could be recovered from genital tracts of Plac8 knockout mice that had previously cleared a primary infection by delayed treatment with MLA [13]. This result mirrored that of Ramsey et al who showed that viable C. muridarum could be recovered from iNOS-knockout but not wild type mice that had previously cleared a genital tract infection when they were treated with the lymphocyte toxin cyclophosphamide [14].
Two previous studies have shown that perforin knockout mice were not compromised in their ability to clear C. muridarum genital tract infections. One study showed a trend toward less oviduct pathology in perforin knockout mice [15], the other a statistically significant reduction [8]. Because perforin knockout mice have an intact iNOS-clearance mechanism, it is possible that a role for perforin in Plac8-dependent clearance was masked by the iNOSdependent clearance mechanism. With similar reasoning, we hypothesized that if perforin was necessary for the Plac8dependent mechanism, then a perforin knockout mouse treated with MLA would not clear a C. muridarum infection over 8 weeks.
Unexpectedly we found that perforin knockout mice were compromised in their ability to clear C. muridarum genital tract infections, and that the role of perforin in clearance is not likely through enhancing CD4 T cell termination of Chlamydia replication by either of the two known in vitro mechanisms dependent on iNOS [10] and T cell degranulation [16].

Materials and Methods
Mice 4-5 week old female C57BL/6J and C57BL/6-Prf1 tm1Sdz/J mice were purchased from The Jackson Laboratories (Bar Harbor, MA). All mice were housed in Indiana University School of Medicine specific-pathogen-free facilities (SPF). All animal experiments were approved by the AAALAC-accredited Indiana University School of Medicine Animal Care and Use Committee performed in accordance with the United States Animal Welfare Act and the National Research Council's Guide for the Care and Use of Laboratory Animals. Mice were lightly anesthetized with isoflurane for infections and vaginal swabs to minimize stress and discomfort.

Genital Tract Infections
One week prior to infection mice were treated with 2.5 mg of medroxyprogesterone delivered subcutaneously (Depo-Provera, Pfizer Pharmaceuticals, New York, NY). Lightly anesthetized mice were infected vaginally with 5610 4 inclusion forming units (IFU) of C. muridarum biovar Nigg in 10 ml of SPG buffer. Mice were swabbed Tue-Thu to monitor the clearance of C. muridarum from the genital tract. Mice had access to water ad libitum, without or with 50 mM N-monomethyl-L-arginine (EMD4biosciences; Billerica, MA). Water was supplemented as needed and exchanged entirely for fresh solutions twice a week for the duration of the experiment as in the previous publication [13]. Power analysis based on our previous Plac8 MLA publication [13] and a previous perforin knockout mouse study [15] showed that experimental groups of 5 mice had 97-99% likelihood to detect significant differences (p value ,0.05) between groups if the assessment was done in the day 40-56 window post infection.

Pathology Scoring
A simple scoring system was utilized to assess macroscopic genital tract pathology [13]. Hydrosalpinx was scored as 0, 1, or 2 reflecting no, unilateral, or bilateral hydrosalpinx respectively. Mice have a bi-fed uterus (2 uterine horns) that was similarly scored 0, 1, or 2 for hydro-uterus. The maximum pathology score for an individual mouse was 4.

Epithelial Cell Infections/Chlamydia Replication/Titration/ Microscopy
C57epi.1 cell monolayers in 48 well plates were untreated or treated with IFN-c (10 gg/ml) for 10 h prior to infection. Wells were infected with 3 IFU per cell. After addition of C. muridarum the plates were spun at 3006g for 30 min. 4 h post infection the inoculums were removed and CD4 T cell clones were added in RPMI CM. 32 h post infection, wells were scraped, harvested with an equal volume of sucrose-phosphate-glutamate acid buffer (SPG buffer) and stored at 280uC until C. muridarum titers were determined on McCoy cell monolayers using an anti-Chlamydia LPS antibody (generous gift of Bobbi Van Der Pol and Jim Williams IUSOM) and FITC-labeled goat anti-mouse IgG (Rockland Immunochemicals; Gilbertsville, PA) as previously described [18]. For the DIC microscopy, arbitrarily selected polyclonal T cell populations, #5 (wild type) and #15 (perforindeficient), frozen at the third passage were thawed/expanded then used under identical conditions as the Chlamydia replication/ titration experiments except that instead of harvesting the monolayers at 32 h post infection the wells were fixed with 10% formalin and photographed using a Nikon Diaphot 200 inverted microscope equipped with a Diagnostic Instruments spot color camera using the plan20/0.4 DIC lens.

Statistical Analysis
Summary figures for each experimental investigation are presented means and standard deviations (SD). Figure legends indicate the number of mice or independent T cell populations per data point in the experiment. Student's two-tailed t-tests and Wilcoxon non-parametric tests, depending on data distribution, were used to assess significance of experimental data. Homogeneity of variances was assessed using a folded F-test. Statistical analysis of Chlamydia shedding was performed using repeated measures analysis of variance to test for differences between groups over time. All data were verified to meet analytic assumptions. Analyses were performed using SAS 9.3 (SAS Institute, Cary, NC). iNOS had no effect on clearance of C. muridarum from the genital tract of perforin knockout mice. Wild type C57BL/6J (10 mice; 5 untreated, 5 treated with MLA) and perforin knockout (10 mice; 5 untreated, 5 treated with MLA) female mice were treated with medroxyprogesterone, then infected vaginally one week later with 5610 4 IFU of C. muridarum. For MLA treatment groups the drinking water contained 50 mM MLA (iNOS inhibitor) one day before infection through day 56 post-infection. Genital tract shedding was monitored through day 56 post-infection. The 20 mice in the experiment were inoculated on the same day and monitored in parallel for the duration of the experiment. A) Shedding for C57BL/6J mice. B) Shedding for perforin knockout mice. C) Combined data. D) Combined data for % culture positive mice over 8 weeks. Differences in the clearance kinetics for C57BL/6J versus perforin knockout mice for the last 5 weeks of the experiment (last 4 logs of clearance, highlighted). Differences between the perforin knockout mice, without or with MLA treatment, and untreated wild type mice were statistically significant with an overall p value ,0.0001. ** = p value ,0.005 comparing wild type and perforin knockout mice at initial data point (day 5). {{{ = p value ,0.0005 comparing untreated wild type mice to wild type mice treated with MLA at initial data point (day 5). doi:10.1371/journal.pone.0063340.g001

Perforin is not Likely Required for the Plac8-dependent Clearance Mechanism
We tested the role of perforin in Plac8-dependent clearance by treating perforin knockout mice with the iNOS inhibitor MLA and infecting them vaginally with C. muridarum. In our previous study Plac8 knockout mice treated with MLA were effectively unable to clear a C. muridarum genital tract infection over 8 weeks, with 11 of 12 mice continuing to shed C. muridarum with an average intensity of 1300 IFU/swab; all wild type C57BL/6 mice treated with MLA, 11 of 11, cleared the infection by day 40. If the Plac8dependent mechanism were perforin-dependent then perforin knockout mice treated with MLA would have a similarly compromised ability to clear C. muridarum from the genital tract.
C57BL/6J (H-2 b ) and perforin knockout mice (H-2 b ) were treated with medroxyprogesterone to synchronize estrous, then infected vaginally with 5610 4 IFU of C. muridarum, without or with addition of iNOS inhibitor MLA to the drinking water. Mice were swabbed twice weekly through 56 days, then assessed for genital tract pathology. Recovered IFUs from genital swabs were titered on McCoy monolayers (Fig. 1). Three of five wild type mice cleared the infection by day 28; the remaining two mice by day 33 (next sample collection). Addition of MLA to the drinking water of wild type mice lead to delay in clearance to day 40, the same day as our previous study. Perforin knockout mice cleared infections on Day 49, and neither the intensity of shedding nor time to clearance were affected by MLA treatment. Wild type and perforin knockout mice, under either treatment condition, were culture negative during the 8 th week of the experiment. Perforin knockout mice without or with MLA treatment had higher IFU shedding on initial measurement (day 5) and delayed clearance of C. muridarum from the genital tract compared to untreated wild type mice. At the end of experiment all mice were sacrificed and histopathology was scored. As seen in our previous study, treating wild type C57BL/6 mice with MLA did not change the severity of reproductive tract pathology (data not shown). Perforin knockout mice trended toward less pathology than wild type mice (50% less; p value = 0.17) consistent with previously published data [8,15].

Delayed Clearance in the Perforin Knockout Mice is not Due to T cell Polarization
Spleens were harvested from all 20 mice on Day 56 to determine whether there was any difference in cytokine polarization related to the absence of perforin or treatment with MLA. Splenocytes were mock-activated (SPG buffer) and activated in vitro with UV-inactivated C. muridarum (stock in SPG buffer); culture supernatants collected at 72 h for cytokine analysis (Fig. 2). Protective immunity in experimental Chlamydia vaccine studies has previously been shown to been dependent on Th1 T cells [19] making IFN-c and TNF-a [20]. IL-10 SNPs have been associated with susceptibility to Chlamydia-mediated tubal infertility in humans [21]. The Th2 cytokine IL-13 has been found to be detrimental in C. muridarum pulmonary infections [22]. There was no significant difference in IFN-c, TNF-a, IL-10, or IL-13 between any of the mice or treatment conditions. MLA had no effect on cytokine polarization. Cytokine polarization does not explain the delayed clearance in perforin knockout mice.

Perforin is Detrimental to the iNOS-dependent Clearance Mechanism in vitro
Immune T cells from the 5 untreated wild type and 5 untreated perforin knockout mice in the experiment shown in figure 1, harvested on day 56, were separately expanded in vitro for 2 passages using UV-inactivated C. muridarum and supporting cytokines (see materials and methods). The resulting Chlamydia-specific polyclonal T cell populations were 99% CD4 T cells (Fig. 3). We tested the ability of these polyclonal populations to terminate C. muridarum replication in oviduct epithelial cell line C57epi.1, untreated or pretreated with IFN-c, in the absence or presence of the iNOS inhibitor MLA. Epithelial cells were infected with 3 IFU per cell; 4 h later polyclonal T cells were added to the wells at an effector to target ratio of 1.25 to 1 (250 k T cells to ,200 K epithelial cells). 32 h post infection supernatants were collected, and the wells harvested by scraping in SPG buffer to determine the extent of C. muridarum replication; IFU recovered were quantified on McCoy monolayers (Fig. 4). The amount of IFN-c produced by T cells under each experimental condition was determined by ELISA (Fig. 5). Though not quantifiable, examination of the wells under phase contrast microscopy showed an obvious relative preservation of the epithelial cell monolayer at 32 h post infection in the wells containing perforin-deficient T cells compared to wild type T cell wells (Fig. 6), consistent with absence of the perforin the principal effector molecule for T cell-mediated cytolysis. Because there were more visible inclusions but fewer recovered IFU from the wells containing perforin-knockout T cells, it is reasonable to suppose that many EB within those visible inclusions were dead. As seen in our previous study [16], overnight treatment of the C57epi.1 epithelial monolayer prior to infection reduced replication by ,4-fold. In untreated monolayers, perforin-deficient CD4 T cells were 5 times more efficient at terminating C. muridarum replication in epithelial cells than were wild type T cells. That difference in efficiency was blunted somewhat by pretreating the epithelial cell monolayers with IFN-c, boosting epithelial nitric oxide production and MHC class II expression [17]. Both wild type and perforin-deficient early passage polyclonal T cell populations used the iNOS-dependent mechanism to terminate  figure 1 (20 independent splenocyte populations) activated with UV-inactivated C. muridarum. 3610 6 splenocytes were activated with 7.5610 6 UV-inactivated C. muridarum (multiplicity of 2.5 IFU per splenocyte) or mock-activated with equivalent volume of SPG buffer (con = control) in triplicate wells. Supernatants were collected at 72 h and analyzed by ELISA. Data representing each mouse strain/treatment condition were aggregated and compared to the aggregated data for the untreated C57BL/6J mice. Comparisons were made for each cytokine. There were no statistically significant differences in cytokine polarization between wild type and perforin knockout mice with or without MLA treatment. doi:10.1371/journal.pone.0063340.g002 replication as addition of MLA to the experimental wells essentially abrogated their ability to terminate Chlamydia replication. In the presence of 1 mM MLA both polyclonal T cell populations reduced recovered IFU by only 2-fold. This result suggests that CD4 T cells capable of iNOS-independent termination of replication were present in numbers below that necessary to efficiently control replication in the presence of MLA. Consistent with, or in addition to, their greater efficiency in terminating replication, perforin-deficient T cells made ,2-fold more IFN-c under all experimental conditions (see Fig. 5).

Discussion
Studies of protective immunity against Chlamydia infections of the genital tract have been confounded by the existence of three independent and singly sufficient mechanisms for clearing C. muridarum (reviewed in [23]). The first mechanism discovered is dependent on T cell-epithelial cell contact and IFN-c2mediated up regulation of epithelial nitric oxide production to chlamydiacidal levels [10]. The second mechanism described is T cellindependent once established [24]; it is dependent on Chlamydiaspecific antibodies [25] and likely Fc receptors [26]. The third mechanism discovered is dependent on Plac8 [13] and likely CD4 T cell degranulation [16].
Based on the previous work of others showing no difference in clearance between C57BL/6 and perforin knockout mice [8,15], and our work showing a dramatically diminished ability to clear C. muridarum in Plac8 knockout mice treated with MLA [13], we sought to definitively test whether the Plac8-dependent mechanism was dependent on perforin by comparing clearance in perforin knockout mice without and with iNOS inhibitor treatment. Our experiments show that perforin knockout mice had delayed clearance compared to wild type mice, and that MLA treatment had no effect on clearance in the perforin knockout mice. Delayed clearance in perforin knockout mice could not be attributed to a difference in cytokine polarization and was not associated with greater immunopathology.
The finding of delayed clearance in perforin knockout mice differs between Perry et al [15] and our current study, likely reflecting the inoculating dose; 1500 IFU per mouse and 50,000 IFU per mouse respectively. In support of that possibility, Perry et al saw no difference in wild type and perforin knockout mice genital tract IFU at their initial time point; in our study there was 6-fold higher IFU in the perforin knockout mice than wild type mice on day 5. Our results also differ from those of Murthy et al. [8] who used an inoculating dose of 50,000 IFU, but treated mice with medroxyprogesterone on day T-10 and day T-3 prior to infection. In their model, IFU recovered on day five are ,20,000/swab while our recovered IFU on day 5 are ,2,000,000/swab. It is likely that the higher effective inoculating dose in our experimental model (based on inoculum size or hormonal manipulation) revealed defects in clearance not seen in the two other models.
The data in figure 4 show that perforin is detrimental to the iNOS-dependent mechanism for terminating Chlamydia replication in epithelial cells, either because cytolysis of the epithelial target decreases T cell activation and IFN-c production, or more likely, inserting perforin channels (punching holes) in the epithelial membrane lowers the effective concentration or duration of exposure of Chlamydia to epithelial cell-produced intracellular nitric oxide. 5-fold more efficient replication termination by perforindeficient T cells occurred in the setting of diminished cytolysis (see Fig. 6). More intact epithelial cells with visible inclusions were seen in perforin-deficient T cell wells, but fewer viable IFU were recovered from them suggesting that EB within those inclusions were dead; a counter intuitive result for conventional intracellular pathogen host defense paradigms. Because the IFN-c levels were low (1-2 gg/ml), and only increased ,2-fold in perforin-deficient experimental wells, we postulate that the 5-fold greater potency of perforin-deficient T cells reflects a detrimental effect of perforin on functional levels of microbicide within infected cells. In support of that postulate, we have previously shown that addition of IFN-c to our reproductive tract epithelial cell lines 4 or more hours after infection had no effect on C. muridarum replication or on the replication of the much more IFN-c sensitive human C. trachomatis serovars D and L2 [27]. In these current experiments we added the T cells to the infected monolayers 4 h post-infection. We have also previously shown in 4 of 4 Chlamydia-specific T cell clones that the earliest recognition of untreated infected epithelial cells as measured by IFN-c production is between 15-18 h post-infection [17,28]. In our current experiment, IFN-c in meaningful levels were likely non-existent until ,20 h into the experiment, a time point at which the large majority of C. muridarum replication is already complete in C57epi.1 cells [17]. As a working model we propose that CD4 T cell-mediated termination of Chlamydiareplication in epithelial cells occurs via a toxic bag type mechanism. T cells form conjugates with infected epithelial cells, activating targeted cells to make nitric oxide; punching holes in the plasma membrane and/or lysing the epithelial cell is detrimental to this mechanism (Fig. 7).
Perforin-deficient mice treated with MLA did not behave like Plac8-deficient mice treated with MLA. Our previous experience was that Plac8 knockout mice treated with MLA could not clear C. muridarum within 8 weeks, with substantial levels (1300 IFU/swab) of ongoing IFU shedding at the end of week 8, day 56. All perforin-deficient mice, untreated or treated with MLA, cleared their infection by day 49 with no measurable IFU shedding during the 8 th week of the experiment. Because untreated perforin knockout mice unexpectedly differed from C57BL/6 mice, we could not performed the planned statistical analysis comparing perforin-deficient mice to perforin-deficient mice treated with MLA to assess whether perforin was critical to the Plac8-dependent mechanism. However, it is clear from the data here that there was no difference in clearance between a perforin knockout mouse and a perforin knockout mouse treated with MLA. Had the Plac8dependent mechanism been dependent on perforin, we would have expected MLA-treated perforin knockout mice to shed Chlamydia throughout the 8 th week of the experiment; they did not. The delayed clearance in untreated or MLA-treated perforin knockout mice compared to wild type mice is unlikely to represent a compromised Plac8-dependent clearance mechanism that is strongly associated with the in vitro degranulation-dependent termination mechanism [13,16]. Perforin's contribution to bacterial clearance is not likely occurring through enhancing CD4 T cell termination of Chlamydia replication in epithelial cells as it does not appear to relevant to the Plac8-dependent mechanism and is detrimental to the iNOS-dependent mechanism.
T cells, mast cells, eosinophils, basophils, CD4negative NKT cells and NK cells express perforin. NK cells recognize and lyse infected epithelial cells [29,30], but NK cells by themselves have only a modest ability to terminate replication in epithelial cells in vitro (,2-fold) [30]. However, cytolysis of infected cells by any means would expose intracellular Chlamydia EB to extracellular defense mechanisms including complement. Chlamydia trachomatis EB activate the alternative complement pathway and C3b covalently binds to MOMP [31], and early components of complement upstream of the membrane attack complex neutralize Chlamydia EB in vitro [32]. While existing data argue against complement being singly sufficient for clearing genital tract infections, as T cell deficient mice sufficient in complement cannot clear infections, complement could logically contribute to    substantial reductions in viable extracellular EB in genital tract secretions. A recent study in a murine C. psittaci model showing a major contribution of C3 to protective immunity [33] raises the question of whether previous unpublished negative data based on cobra venom depletion of the complement membrane attack complex referred to in Williams et al [34] sufficiently addressed the role of early complement components in innate Chlamydia immunity.

Conclusion
During experimental C. muridarum genital tract infections perforin plays a complex role. Others have shown that it contributes to the immunopathogenesis of scarring [8]. Here we show that perforin contributes to bacterial clearance, likely by a mechanism that does not involve CD4 T cell interactions with infected epithelial cells.