Waddlia chondrophila induces systemic infection, organ pathology, and elicits Th1-associated humoral immunity in a murine model of genital infection

Waddlia chondrophila is a known bovine abortigenic Chlamydia-related bacterium that has been associated with adverse pregnancy outcomes in human. However, there is a lack of knowledge regarding how W. chondrophila infection spreads, its ability to elicit an immune response and induce pathology. A murine model of genital infection was developed to investigate the pathogenicity and immune response associated with a W. chondrophila infection. Genital inoculation of the bacterial agent resulted in a dose-dependent infection that spread to lumbar lymph nodes and successively to spleen and liver. Bacterial-induced pathology peaked on day 14, characterized by leukocyte infiltration (uterine horn, liver, and spleen), necrosis (liver) and extramedullary hematopoiesis (spleen). Immunohistochemistry demonstrated the presence of a large number of W. chondrophila in the spleen on day 14. Robust IgG titers were detected by day 14 and remained high until day 52. IgG isotypes consisted of high IgG2a, moderate IgG3 and no detectable IgG1, indicating a Th1-associated immune response. This study provides the first evidence that W. chondrophila genital infection is capable of inducing a systemic infection that spreads to major organs, induces uterus, spleen, and liver pathology and elicits a Th1-skewed humoral response. This new animal model will help our understanding of the mechanisms related to intracellular bacteria-induced miscarriages, the most frequent complication of pregnancy that affects one in four women.


INTRODUCTION
Approximately 25% of women will experience a miscarriage in their lives, representing the most frequent complication in early pregnancies. Although the arrays of diagnostic tests and possible therapeutic interventions in the management of miscarriage have grown significantly, a cause is identified in only 50% of the cases (Carrington et al., 2005;Baud and Greub, 2011). Genetic factors are the most frequent causes, being identified in about 25% of cases (Regan and Rai, 2000;Rai and Regan, 2006). The proportion of miscarriages attributed to infection is approximately 10-15% (Penta et al., 2003;Baud et al., 2008), but this proportion might be underestimated. Intracellular pathogens, which are not detected on routine culture media, might thus represent possible agents of miscarriages with unknown etiology (Baud and Greub, 2011). Chlamydia trachomatis infections have been associated with miscarriages, ectopic pregnancies, and neonatal pneumonia (Vincent et al., 1998;Baud and Greub, 2011). Likewise, Waddlia chondrophila has been recently associated with adverse pregnancy outcomes both in humans and animals (Baud et al., 2007(Baud et al., , 2008Hornung et al., 2015). W. chondrophila belongs to the Chlamydiales order (Rurangirwa et al., 1999;de Barsy and Greub, 2013) and was originally identified in aborted bovine fetuses from the United States (Dilbeck et al., 1990) and Germany (Henning et al., 2002). Similarly to C. trachomatis, W. chondrophila is an obligate intracellular bacterium and exhibits a biphasic lifecycle (de Barsy and Greub, 2013). In vitro studies have demonstrated the ability of W. chondrophila to infect a variety of cell lines, including endometrial cells and primary human macrophages (Goy et al., 2008;Croxatto and Greub, 2010;Kebbi-Beghdadi et al., 2011a,b), supporting its role in miscarriage. Moreover, recent studies reported a higher seroprevalence of anti-W. chondrophila immunoglobulin G found in patients with miscarriage compared to control groups (Baud et al., 2007. Several animal models were developed for the study of genital infection of bacteria belonging to the Chlamydia genus, thus helping the comprehension of pathologies caused by these bacteria (Vasilevsky et al., 2014). At the contrary, the lack of experimental data and animal models of infection limit our understanding of the disease caused by W. chondrophila. For this reason, we developed a female mouse model to investigate the nature of W. chondrophila genital infection in order to study the pathology and the immune system response induced by this emerging human pathogen.

Infection Procedure
All animal experiments were approved by the "Office Vétérinaire du Canton de Vaud, " Lausanne, Switzerland (authorizations n • 2090.0 and 2090.1) and performed according to our institutional guidelines for animal experiments.
For qPCR analysis, two individual experiments with n = 3 and 5 (total n = 8) for each condition, including a mock control (n = 4), were performed. For microimmunofluorescence analysis, two individual experiments with five infected mice were analyzed for each condition (total n = 10) and 9 mock controls (n = 9). Evaluation of pathological outcomes and severity scores of W. chondrophila infection were done on four mice (n = 4) and 4 mock controls (n = 4). Time course experiments for the assessment of IgG response and determination of IgG isotypes, included a total of 12 infected mice (n = 12) and 12 mock controls (n = 12).

Organ Collection
Organs were collected at days 7, 14, and 21 post-infection to measure bacterial burden and subsequently frozen at −80 • C. For liver, spleen and muscle, a small section (∼2 mm 2 ) was used for DNA extraction and qPCR. The entire lymph node was used for qPCR. Organ samples used for McCoy cell co-culture were utilized immediately after harvest (i.e., not frozen).

Quantitative RT-PCR
DNA was extracted using the Wizard SV Genomic DNA purification system (Promega, Wisconsin, USA). All samples were tested with a 16S rRNA W. chondrophila specific real-time qPCR as previously described .

Immunohistochemistry
Tissue samples or whole organs were fixed in 4% buffered paraformaldehyde (Paraformaldehyde, Electron Microscopy Sciences, PA, USA) and embedded in paraffin. Embedded samples were cut in 5-µm-thick sections on a microtome. Paraffin was removed by two xylene (Sigma-Aldrich Chemie) washes and samples were rehydrated using a graded ethanol series (100, 95, 80, and 70%) and finally rehydrated in PBS. Rehydrated sections were blocked using PBS supplemented with 10% goat normal serum (Interchim, Montluçon, France), 5% BSA and 0.3% triton X-100 (Applichem, Darmstadt, Germany) for 1-h at room temperature (RT) in a wet chamber. After three washes for 5 min each in PBS, 50 µl of blocking solution supplemented with rabbit anti-W. chondrophila polyclonal antibody (Kebbi-Beghdadi et al., 2011a, diluted 1/1000) was added to the slides in a wet chamber and incubated for 1-h at RT. Negative control was performed with the preimmunization serum from the rabbit used to produce the anti-W. chondrophila polyclonal antibody. After three washing steps for 5 min each in PBS, the secondary antibody (goat antirabbit Alexa Fluor 488-conjugated, diluted at 1/1000 in blocking solution, Molecular Probes/Life Technologies), and DAPI (4 ′ ,6-diamidino-2-phenylindole, diluted at 1/1000 in blocking solution, Molecular Probes/Life Technologies) were added and incubated for 1-h at RT in a wet chamber protected from light. After three final washes with PBS and one with deionized water, slides were mounted using Mowiol 4-88 (Sigma-Aldrich Chemie), dried overnight and subsequently used for microscopy analysis.

Assessment of W. chondrophila Viability from Organs in McCoy Cells
Spleen and liver samples (∼2 mm 2 ) were collected and cell suspensions were made by mechanical disruption and passage through a 70 µm nylon cell strainer (Falcon Corning, New York, USA) with 2 ml Dulbecco's Modified Eagle Medium supplemented with sodium pyruvate and glutamine (GE Healthcare, Glattbrugg, Zurich, Switzerland) and 10% Fetal bovine serum (FBS; Connectorate AG, Dietikon, Switzerland). Cell suspensions were transferred to a 50 ml conical tube containing glass beads and vortexed to crack open cells. 100 µl of cell suspension were transferred to a 24-well plate containing McCoy cells at confluence in DMEM supplemented medium, immediately centrifuged at 1790 RCF for 10 min and placed in an incubator (37 • C, 5% CO 2 ) for 3 h. After incubation, 50 µg/ml of gentamycin (Bioconcept, Allschwil, Switzerland) and 30 µg/ml of ampicillin (Ampicillin sodium salt, Sigma-Aldrich Chemie, Buchs, Switzerland) were added to the samples in order to inhibit growth of extracellular and other intracellular bacteria, respectively, as W. chondrophila is resistant to ampicillin at >32 µg/ml due to the presence of a β-lactamase Bertelli et al., 2010). Viability was assessed by qPCR and by light microscopy examination of McCoy cells 3 h or 7 days post-inoculation.

Histopathological Analysis
Uterine horns, liver, and spleen from mock and infected mice were removed at indicated time points postinfection (p.i.) and fixed in 4% buffered formalin and embedded in paraffin for histopathological analysis. The tissue sections were stained with Hematoxylin & Eosin (H&E) at the Mouse Pathology Facility (University of Lausanne, Switzerland) and sent for morphologic evaluation, in a blind fashion, to a board certified veterinary pathologist (P.A.). Pathologic changes were scored according to severity on a numerical scale ranging from 0 to 4, where 0 indicates absence of lesions and 4 the highest grade.

Statistical Analysis
Data were analyzed using the GraphPad Prism 6 software (GraphPad Software Inc., San Diego USA). When necessary, groups were compared using the Mann-Whitney rank sum test. Differences with a P-value of <0.05 were considered statistically significant. The level of significance is depicted as * P < 0.05; * * P < 0.01. Data are presented as means ± standard error of mean (SEM).

W. chondrophila Infectious Dose and Heat-inactivation
We first investigated whether a dose response or heat inactivation (HI) modulate pathogenicity of W. chondrophila and humoral immunity in a mouse genital infection model. Mice were inoculated via the uterine horns with HI bacteria (10 9 ) and decreasing amounts (10 9 , 10 7 , 10 5 ) of live bacteria and sacrificed on day 14 postinfection (p.i.; Figure 1 and Supplementary Figure 1). Bacterial load in organs was then assessed by specific W. chondrophila qPCR. Infection with live W. chondrophila resulted in high bacterial loads in genital but also in non-genital organs including spleen, lumbar lymph nodes and liver. Decrease of the inoculum from 10 9 to 10 7 resulted in a bacterial load reduction of >9000-fold in cervix/vagina and >300-fold in the uterine horns and non-genital organs ( Figure 1A). Furthermore, bacterial burden in mice infected with 10 9 HI W. chondrophila was also severely reduced in all organs compared to mice infected with 10 9 live bacteria ( Figure 1A). Surprisingly, we were not able to detect any bacteria in organ samples from mice that were infected with 10 5 live W. chondrophila ( Figure 1A).
Presence of W. chondrophila in organ sections of mice infected with 10 9 bacteria was confirmed using immunohistochemistry.  Analysis of the IgG response, showed strong anti-Waddlia titers in mice genitally challenged with 10 9 and 10 7 live bacteria ( Figure 1C). Mean humoral response of mice challenged with HI bacteria was below the cut-off dilution of 1:32 and a total abrogation of anti-Waddlia response was observed for mice infected with 10 5 live bacteria (Figure 1C and Supplementary Table 1). Because of the high bacterial burden and robust titers from 10 9 -infected mice, we used this dose for all subsequent experiments in this study.

W. chondrophila Genital Infection Results in High Bacterial Burden in Genital and
Non-genital Organs qPCR analysis on uterine horns and cervix/vagina samples demonstrated a high bacterial burden on days 7, 14, and 21 p.i. (Figure 2A). Infection also spread to lumbar lymph nodes, spleen, and liver ( Figure 2B), but bacteria were never identified in blood (data not shown). To assess whether the bacteria in those organs were able to replicate, spleen, and liver homogenates were incubated with McCoy cells for 7 days, but no replication was observed (data not shown). Vaginal shedding was resolved 7 days post-infection (<10 1 copies/µl), and replicating bacteria were still present at days 2 and 4 post-infection (data not shown).
However, day 21 had the highest percentage of mice that demonstrated salpingitis (25%). Mock mice and mice infected with 10 9 HI W. chondrophila did not demonstrate overt organ pathologic changes.
To characterize microscopic lesions induced by W. chondrophila infection, liver, spleen and uterine horns from mice 14 days p.i. from both 10 9 infected and mock challenged mice were analyzed (Figures 3A-I). Liver from infected mice exhibited leukocytes infiltration: mostly neutrophils and macrophages (the latter increasing over time) coalescing into histiocytic nodules or granulomas (Figures 3B,C) and variable sized areas of coagulative hepatocellular necrosis. 75% of mice exhibited macrophage infiltration on day 7, 50% on day 14, and 75% on day 21 in the liver (Supplementary Table 2). In spleens of infected mice, both the white and red pulps were replaced by multifocal histiocytic nodules (Figure 3E, arrows). Extramedullary hematopoiesis (i.e., hematopoiesis that occurs in organs other than bone marrow, EMH) was quite prominent as shown in Figure 3E (red arrows). Increased number of apoptotic cells was observed in the red and white pulp ( Figure 3F). Histiocytes and neutrophils were present in the endometrium of infected mice (Figure 3H arrows), but often they were found also within the myometrium and the serosa. Necrotic deciduomas were observed in the lumen of infected animals only ( Figure 3I). Liver, spleen, and uterine horn pathology severity scoring for necrosis, leukocyte infiltration and EMH were assessed on days 7, 14, and 21 p.i. (Figure 4). Figures 4A,B show that necrosis is more severe in the liver than in the spleen at days 7, 14, and 21 p.i. On day 14, incidence of liver changes such as necrosis and neutrophilic infiltration was 100% in infected mice (Supplementary Table 2). Severity median scores (Figure 4) demonstrated the trend of the W. chondrophilaassociated lesions in infected mice reaching a peak on day 14 and decreasing on day 21. A significant difference was also seen when compared to mock-infected mice (p < 0.005 at day 14 for leukocyte infiltration in all organs, liver necrosis and spleen EMH).

W. chondrophila Genital Infection Elicits a Robust IgG and Predominately Th1-associated Antibody Response
Analysis of anti-W. chondrophila antibody response demonstrated that specific IgG antibodies were present by day 14, peaked on day 28, and slightly declined by day 42 (Figure 5A and Supplementary Table 2). Furthermore, elevated titers were present for at least 6 weeks. Mice challenged with mock did not elicit anti-Waddlia specific antibodies. Figure 5B demonstrates that the IgG response elicited by W. chondrophila was mainly comprised of the IgG2a isotype and to a lesser extent of the IgG3 isotype, indicating a Th1-associated humoral response.

DISCUSSION
W. chondrophila is an intracellular Chlamydia-related bacterium that has been strongly implicated in adverse pregnancy outcomes in women (Baud et al., 2007Hornung et al., 2015). However, little is known how genital W. chondrophila infection spreads, elicits an immune response, and induces pathology.
Chlamydia mouse infection models have demonstrated that Chlamydia muridarum is a natural pathogen of mice whereas C. trachomatis is only infectious in mice if infected in high doses directly in the uterine horn (reviewed in Vasilevsky et al., 2014). Since, this is the first reported instance of a W. chondrophila genital mouse infection model and since W. chondrophila is closely related to C. trachomatis, we used a semi-rigid cannula per vaginam to pass the cervix and injected live freshly cultured W. chondrophila into the uterine horn to increase the chances of a progressive genital infection. We first wanted to establish the optimal dose for a progressive bacterial infection and investigate whether decreasing amounts of live or HI W. chondrophila modulated the bacterial load in the organs and the antibody production. We determined that 10 9 live W. chondrophila was the optimal dose for a genital infection, due to the induction of a pathological state and the generation of a robust antibody response. Lack of IgG titers in mice infected with 10 5 bacteria might be explained by the fact that this infectious dose may not recruit enough dendritic cells (DCs) to initiate an IgG specific humoral response. Indeed, this hypothesis is supported by a study conducted by Maxion et al. where a decrease of C. muridarum infection resulted in a decrease of CD11c + DC recruitment (Maxion et al., 2004). Heat inactivation often denatures protein epitopes, which may affect protein binding onto MHC class I/II on T cells and subsequent production of important Th1-associated cytokines (Yu et al., 2011;Ding et al., 2014). Since, MHC class II binding of proteins on antigen presenting cells and subsequent presentation to T cells is critical for initiating a Th1 humoral response, this may be a plausible explanation for the low IgG titers in mice that were immunized with HI W. chondrophila. Mice challenged with W. chondrophila were used to investigate the bacterial burden in uterine horns, lumbar lymph nodes, spleen, and liver over a 3-week time period. High bacterial loads were observed throughout the experiment in the organs. The main limitation of the qPCR analysis used in our experiments is the fact that this technique detects W. chondrophila DNA, thus indirectly representing bacteria. Therefore, our data might indicate that bacterial DNA is present in the organs but not as intact EBs outside cells or RBs inside cells. In addition, we might also be measuring bacterial DNA that is present inside immune cells such as neutrophils, macrophages, or DCs. Spleen and liver homogenates were incubated with McCoy cells in order to assess whether these tissues contained W. chondrophila that were able to replicate. However, the clear evidence of bacteria by immunohistochemistry (day 14 p.i.) and the absence of replication in McCoy cells suggest that W. chondrophila remains in a non-replicating persistent form. The same has been observed for aberrant bodies of C. trachomatis when T cell-mediated immune response is induced and interferon (IFNγ) is secreted (Baud et al., 2008). Since we demonstrated a high Th1-mediated immune response, IFN-γ levels might delay or abort differentiation into replicative (RB) and infectious (EB) stages.
None of the mock or infected mice lost weight or showed any overt signs of distress, highlighting the asymptomatic nature of a W. chondrophila genital infection. However, a high percentage of mice exhibited macroscopic lesions with an overall highest frequency at day 14l. Mock challenged mice did not show any evidence of gross changes at necropsy, including absence of granulomas in examined organs. The multiorgan involvement in infected mice at both macro-and histopathologic level points toward a dissemination from the uterus to tissues. The predominance of neutrophils followed by macrophages in the inflammatory infiltrate is consistent with the bacterial origin of the insult (Kim et al., 2008). Moreover, formation of granulomas has been previously associated with the presence of intracellular pathogens (Pettengill et al., 2009). The shift of the predominant leukocytic cell type over time correlates with the kinetics of the infection and inflammatory process, which evolves from a subacute (7 days) to a more chronic stage. Necrotic deciduomas were observed in the lumen of infected animals only, but the significance of such lesion is not clear. Occurrence of deciduomas is described in the literature in rodents as hyatrogenic effect of inoculation or intrauterine luminal delivery of substances or trauma/luminal manipulations in pseudopregnant animals (Finn and Keen, 1963;Hirabayashi et al., 1999). Another common cause is administration of sexual hormones for estrous cycle manipulation such as pseudopregnancy induction. As mock mice did receive the same manipulations and treatments, except for infection with W. chondrophila, a pathogenic correlation between the lesion and the etiologic agent cannot be excluded. In the present study all deciduoma were necrotic and undergoing degeneration. Apoptosis and regression is reported as normal event at about 10 days in literature (Hirabayashi et al., 1999). However, a direct effect of W. chondrophila infection can be hypothesized as similar bacteria, such as C. trachomatis, can infect decidual cells and lead to adverse pregnancy outcome (de la Torre et al., 2009). Validation of this hypothesis and dismissal of spontaneous degeneration of the decidual reaction would need further investigation.
A granulomatous inflammation is a known response to infection caused by intracellular bacterial organisms (Pettengill et al., 2009). The nodular arrangement of histiocytes mirrors the gross raised gray-white foci in the liver and in the spleen and is characterized by a Th-1 type immune response, which in the present study is further corroborated by the immunological data in our study.
Mice genitally infected with live W. chondrophila elicited a robust long-lasting systemic IgG humoral response that consisted mainly of IgG2a and IgG3 isotypes, which is characterized as Th1-associated response in mice (Lin et al., 2004). Interestingly, our results are in complete accordance with a study conducted by Morrison and Morrison where C57BL/6 mice genitally infected with C. muridarum demonstrated similar pattern (Morrison and Morrison, 2005).
In this study, we have developed the first W. chondrophila genital infection mouse model. Our results demonstrated that W. chondrophila genital infection resulted in organ pathology and a systemic bacterial burden that spread to cervix/vagina, lumbar lymph nodes, and spleen. The humoral immune response was IgG-mediated containing mainly IgG2a subtypes, thus suggesting Th1-skewed immunity.

AUTHOR CONTRIBUTIONS
SV, JG, GG, and DB conceived and designed the experiments. SV, JG, LP, and MS performed the experiments. AP performed all the pathological interpretations. SV, JG, MS, and DB analyzed the data. SV and MS wrote the paper.

FUNDING
This work was supported by the Department of Obstetrics and Gynecology, Maternity, Lausanne, Switzerland and by the SNSF grant number 310030_156169/1 attributed to DB. DB is also supported by the "Fondation Leenaards" through the "Bourse pour la relève académique, " by the "Fondation Divesa, " and by the "Loterie Romande."

ACKNOWLEDGMENTS
We thank Anne Ammerdorffer for critical review of the manuscript and the Mouse Pathology Facility for processing of tissue samples.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fcimb. 2015.00076