Persistent Mycoplasma genitalium Infection of Human Endocervical Epithelial Cells Elicits Chronic Inflammatory Cytokine Secretion

Abstract

Infection with Mycoplasma genitalium has been associated with male and female urogenital disease syndromes, including urethritis, cervicitis, pelvic inflammatory disease (PID), and tubal factor infertility. Basic investigations of mucosal cytotoxicity, microbial persistence, and host immune responses are imperative to understanding these inflammatory urogenital syndromes, particularly in females, considering the potential severity of upper tract infections. Here, we report that M. genitalium can establish long-term infection of human endocervical epithelial cells that results in chronic inflammatory cytokine secretion and increased responsiveness to secondary Toll-like receptor (TLR) stimulation. Using a novel quantitative PCR assay, M. genitalium was shown to replicate from 0 to 80 days postinoculation (p.i.), during which at most time points the median ratio of M. genitalium organisms to host cells was ≤10, indicating that low organism burdens are capable of eliciting chronic inflammation in endocervical epithelial cells. This observation is consistent with clinical findings in women. Persistently secreted cytokines predominately consisted of potent chemotactic and/or activating factors for phagocytes, including interleukin-8 (IL-8), monocyte chemotactic protein 1 (MCP-1), and macrophage inflammatory protein 1β (MIP-1β). Despite persistent cytokine elaboration, no host cell cytotoxicity was observed except with superphysiologic loads of M. genitalium, suggesting that persistent infection occurs with minimal direct damage to the epithelium. However, it is hypothesized that chronic chemokine secretion with leukocyte trafficking to the epithelium could lead to significant inflammatory sequelae. Therefore, persistent M. genitalium infection could have important consequences for acquisition and/or pathogenesis of other sexually transmitted infections (STIs) and perhaps explain the positive associations between this organism and human immunodeficiency virus (HIV) shedding.

INTRODUCTION

Mycoplasma genitalium is a morphologically small and genetically reduced bacterium implicated in sexually transmitted reproductive tract disease in men and women. Strong clinical associations exist between M. genitalium infection and male nongonococcal urethritis (NGU) (reviewed in reference 15), and, more recently, genital infections have also been correlated with lower and upper reproductive tract inflammation in women (reviewed in references 25 and 30). Although the target cell(s) of M. genitalium has yet to be defined, significant associations exist between M. genitalium and cervicitis, pelvic inflammatory disease (PID), and tubal infertility (23, 25, 30). Among female reproductive tract tissues, M. genitalium DNA has been detected in specimens from the vagina, endocervix, endometrium, and the fallopian tubes, suggesting that, after sexual transmission, the organism can disseminate from the lower to the upper tract. Similarly, outbred mice inoculated vaginally showed time-dependent ascension of M. genitalium from the vagina to the cervix, followed by detection in the uterus, fallopian tubes, and ovaries (27). Therefore, the cervix appears to be either a primary or transient target of infection. Considering the clinical associations of M. genitalium with urogenital inflammation, it remains imperative to investigate the basic mechanisms of mucosal infection and disease induced by this organism.

Among male and female reproductive tract syndromes for which M. genitalium has been implicated, all could be attributed to long-term colonization of the urogenital tract, which we refer to here as “persistence.” Consistent with the ability to survive long-term in urogenital tissues and similar to other sexually acquired urogenital pathogens, it is hypothesized that M. genitalium has evolved specific mechanisms to evade the host immune system. Indeed, persistent cervical infection has been observed in several clinical studies (4, 6, 10, 13) and has been associated with chronic symptoms (4). Persistent infection by M. genitalium is likely mediated, at least in part, by recombinational variation of genes encoding surface-exposed antigens (13, 14, 22) and through intracellular localization (2, 7, 17, 26, 31). Despite these mechanisms for avoiding the host's immune system, M. genitalium appears to induce a significant inflammatory response, as demonstrated with epidemiologic associations with urogenital disease (25, 30) and the observation that human ecto- and endocervical epithelial cells respond to acute infection with proinflammatory cytokine secretion (26). A link to human immunodeficiency virus (HIV) infection has been observed (29) since cervicitis caused by M. genitalium occurs more often in HIV-positive subjects (20) and, importantly, since M. genitalium persists longer in these women (6). In addition, high M. genitalium burden is correlated with increased HIV shedding from the cervix (24). Collectively, understanding the nature of endocervical immune responses to M. genitalium infection is important for grasping not only this organism's pathogenic potential but also its apparent role as a cofactor for susceptibility and/or severity of HIV disease.

Despite the clinical relevance, basic aspects of persistent infection by M. genitalium have not been studied. Persistent colonization of urogenital mucosae could have significant consequences, including long-term inflammation leading to epithelial damage and potentially enhancing susceptibility to other sexually transmitted infections (STIs). Our results describe several key aspects of bacterial and host dynamics of persistent endocervical infection in a two-dimensional (2-D) cell culture model. Together, the findings provide insight into the observed clinical syndromes of M. genitalium-infected women and offer a potential explanation for the observed epidemiologic associations with HIV infection.

MATERIALS AND METHODS

Axenic propagation of M. genitalium.

Mycoplasma genitalium type strain G37 (ATCC 33530) was grown in Friis FB medium (19) as described previously (26). Bacterial growth was monitored by the formation of characteristic adherent microcolonies and a pH-mediated color change of the broth medium. M. genitalium was harvested at the peak of viability during the log phase of growth (27) by scraping and centrifugation (at 2,000 × g for 15 min at 25° C) of adherent microcolonies. The pellet was rinsed twice with phosphate-buffered saline (PBS), resuspended in cell culture medium, and then vigorously pipetted to dissociate the microcolonies prior to inoculation of human endocervical epithelial cells. To inactivate M. genitalium, viable organisms were harvested during the log phase of growth and then incubated at 80°C for 5 min. Loss of bacterial viability was verified by a 14-day absence of M. genitalium growth at 37°C in Friis FB medium.

Cell culture, cytotoxicity, and persistence assays.

Two representative primary-like epithelial cell lines derived from the human endocervix were utilized in this study: (i) the well-characterized A2EN cell line (passage number 46) (9) and (ii) the low-passage-number ShEN101 line (passage number 7). Both the A2EN and ShEN101 cell lines were derived from human endocervical tissues excised from discarded hysterectomy material as described previously (9). The A2EN cells were maintained in 150-cm² T-flasks with keratinocyte serum-free medium (KSFM; Invitrogen, Carlsbad, CA) supplemented with bovine pituitary extract (50 mg/liter), recombinant epidermal growth factor (5 μg/liter), CaCl₂ (44.1 mg/liter), penicillin G (100 U/ml), and streptomycin sulfate (100 μg/ml). The low-passage-number ShEN101 cell line was maintained similarly in Clonetics KGM-2 keratinocyte growth medium-2 (Lonza, Walkersville, MD) supplemented with CaCl₂ (44.1 mg/liter), penicillin G (100 U/ml), streptomycin sulfate (100 μg/ml), and 1% fetal bovine serum (FBS). All cultured cells were verified to be free of common contaminating mycoplasmas as determined by a MycoSensor quantitative PCR (qPCR) assay kit (Agilent, Cedar Creek, TX).

Epithelial cell lysis due to acute or persistent M. genitalium infection was defined by leakage of cytosolic lactate dehydrogenase (LDH) in culture supernatants. LDH was detected using a Cytotoxicity Detection Kit Plus (Roche Applied Science, Indianapolis, IN) following the manufacturer's suggested protocol. Percent cytotoxicity was determined by dividing the mean background-subtracted experimental values from values obtained after complete lysis of an equal number of cells processed in parallel. The value obtained after complete lysis of cells represents the maximum releasable LDH and corresponds to 100% of cells.

For each study, naïve A2EN and ShEN101 cells were inoculated in 150-cm² T-flasks with a defined inoculum ranging from 2 to 200 mycoplasmas per cell or with medium alone in separate control flasks. After inoculation, M. genitalium organisms were quantified from cultures of infected cells using a TaqMan quantitative PCR (qPCR) assay as described below. Specifically, organisms were quantified from the cell-associated fraction (representing attached and intracellular organisms) of the culture. Cell-associated organisms were obtained from cell pellets after the culture supernatant was decanted and the samples were washed once in PBS, trypsinized, and resuspended in 2 ml of fresh medium. This procedure was performed when A2EN and ShEN101 cells reached 100% confluence and was included as part of the standard “splitting” of the cultures to prevent overcrowding of the cells. DNA was extracted from a 0.1-ml aliquot of resuspended cells (Qiagen DNeasy, Valencia, CA) and stored at −20° C for qPCR.

PCR quantification of M. genitalium.

To enumerate M. genitalium organisms in experimental samples, we developed and optimized an internally controlled real-time qPCR assay targeting a 92-bp region of the MG190 (mgpA) gene. Using Primer Express (version 3.0; Applied Biosystems, Inc., Foster City, CA.), two primers and one 6-carboxyfluorescein (FAM)-labeled TaqMan probe were designed based on a highly conserved region of mgpA determined by nucleotide alignment of more than 30 M. genitalium clinical specimens from the United States and Europe (21; also L. Ma, unpublished data). The PCR setup was as follows: 12.5 μl of TaqMan Universal PCR Master Mix (Applied Biosystems, Carlsbad, CA), 1.0 μM forward primer (190F, 5′-GAACTGAGGAGTAATGGGATTAATGTC-3′), 1.0 μl of reverse primer (190R3, 5′-TTAGTAATGATCGCTCCACTTGC-3′), 0.25 μM MG190 TaqMan probe (MG190P, 5′-AGATATAGCCATTAAGTATGGTGGG-3′), 0.25 μM internal positive control (IPC) probe (5′-CCCGCGAAATTAATACG-3′), 150 copies of the IPC construct (see below), and 5 μl of template.

The IPC was constructed initially using the pET28a(+) plasmid (Novagen/EMD Biosciences, Darmstadt, Germany) as a PCR template. An 86-bp fragment corresponding to nucleotides 343 to 428 of pET28a(+) was amplified using two primers, each containing a sequence specific for pET28a(+) at the 3′ end and the sequence of either qPCR primer 190F or 190R3 at the 5′ end. The resulting IPC construct (136 bp) contained an 86-bp pET28a(+) fragment in the center bound by sequences of primers 190F and 190R3. The IPC probe sequence was complementary to nucleotides 380 to 396 of the pET28a(+) plasmid and was synthesized with a 5′ VIC label and 3′ minor groove binder (MGB) quencher. This design allowed the IPC to be amplified with the same primers as the M. genitalium target DNA but detected by a different probe.

Real-time PCR was performed using a 7500 Real-Time PCR System (Applied Biosystems, Carlsbad, CA) with two-step cycling parameters as follows: 50°C for 2 min and then an initial denaturation at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. For M. genitalium quantification, a standard curve was generated using 10-fold serial dilutions of a plasmid containing the MG190 gene cloned from the G37 type strain (21). Each PCR run included triplicate wells of the MG190 standard dilutions that had been prequantified using spectrophotometry (260 nm), and the experimental samples were run in duplicate or triplicate. The lower limit of detection was empirically determined to be 3 genome copies per reaction.

The specificity of the qPCR assay was tested using the following genomic DNA specimens (approximately 10⁵ genome equivalents per reaction), which were kindly provided by Jørgen S. Jensen (16): Acholeplasma laidlawii (strain AT), Mycoplasma alvi (Isley), Mycoplasma faucium (DC 333T), Mycoplasma fermentans (GT and S38), Mycoplasma gallisepticum (15302), Mycoplasma hyorhinis (GDL), Mycoplasma iowae (695), Mycoplasma lipophilum, Mycoplasma penetrans (GTU), Mycoplasma pirum (Zeus), and Mycoplasma primatum (Navel). The following bacterial species were purchased from the American Type Culture Collection (ATCC, Manassas, VA): Mycoplasma arginini (ATCC 23838), Mycoplasma hominis (ATCC 23114), Mycoplasma pneumoniae (ATCC 15531), Mycoplasma salivarium (ATCC 23064), Ureaplasma parvum (ATCC 700970), Ureaplasma urealyticum (ATCC 27814), and Neisseria gonorrhoeae (ATCC 700825). In addition, we obtained human genomic DNA from GenScript (Piscataway, NJ) and Chlamydia trachomatis DNA (8) for a comprehensive panel of specificity testing. The qPCR assay showed no cross-reactivity with human DNA or DNA from other common urogenital microbes.

Quantification of secreted cytokines and chemokines.

Persistently infected and mock-inoculated cells were maintained in 150-cm² T-flasks as described above. In order to evaluate the profile and magnitude of cytokine elaboration, persistently infected or mock-inoculated cells (A2EN and ShEN101 collected at 14, 36, and 60 days postinoculation [p.i.]) were first trypsinized and then centrifuged at 250 × g for 5 min. Each cell type was then resuspended in fresh medium, counted in triplicate using a hemacytometer, and seeded at equal densities (5 × 10⁴ cells/well; five replicate wells) into a 96-well plate. For all experiments except as noted below, secreted cytokines were quantified from culture supernatants after 48 h of incubation using a Milliplex MAP human cytokine/chemokine immunoassay (Millipore, Billerica, MA) and the analytes granulocyte-macrophage colony-stimulating factor (GM-CSF), gamma interferon (IFN-γ), interleukin-10 (IL-10), IL-12 (p70), IL-13, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, monocyte chemotactic protein 1 (MCP-1), and tumor necrosis factor alpha (TNF-α). For poly(I·C) stimulation studies, we utilized a Bio-Plex Pro human cytokine immunoassay (Bio-Rad, Hercules, CA) with the following analytes: IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12 (p70), IL-13, IL-17, granulocyte colony-stimulating factor (G-CSF), GM-CSF, IFN-γ, MCP-1, macrophage inflammatory protein 1β (MIP-1β), and TNF-α. Cytokines not presented in the results tables were either not significantly increased at any time point in the study, were below the limit of detection, or were outside the limits of reliable extrapolation from the standard curve run in parallel. In separate studies, the Milliplex and Bio-Plex assays have been validated in side-by-side comparisons, and, as the kits are based upon Luminex beads and chemistry (Luminex Corp., Austin, TX), no significant differences in cytokine values were observed between the kits (data not shown). Careful attention was exercised to collect only culture supernatants to ensure that only secreted and not intracellular cytokines were measured.

Antibiotic treatment and responsiveness to secondary Toll-like receptor (TLR) stimulation.

To address whether intra- and/or extracellular mycoplasmas elicit cytokine secretion and to determine whether single-dose antibiotic killing of M. genitalium infection would ablate cytokine secretion, infected and mock-inoculated cells were exposed to gentamicin (200 μg/ml) for 2 h or to azithromycin (8 μg/ml in culture medium) for 2 days. After antibiotic treatment, fresh medium without antibiotic was added to the cells, which was followed by an additional 14 days of incubation. After cells were counted and seeded at 1 × 10⁵ cells per well (96-well plate), cytokines were measured from culture supernatants of M. genitalium-infected cells (48 h after seeding) at 2 and 14 days following azithromycin treatment. Mock-infected cells treated with gentamicin or azithromycin were processed in parallel as controls. Bacterial viability was assessed in A2EN cultures by seeding cells into Friis FB medium and then monitoring for a pH-mediated color change or for evidence of adherent mycoplasmas on the cells or cultureware. Efficacy of antibiotic treatment was verified by an absence of these signs for up to 14 days after treated cells were seeded into Friis FB medium. Persistently infected cells not treated with antibiotics were used as a positive control and consistently yielded pH-mediated color changes within 5 days of seeding into Friis medium.

To evaluate the responsiveness of persistently infected cervical epithelial cells to secondary innate stimulation, infected A2EN cells (36 days p.i.) and mock-inoculated cells maintained in parallel were counted and seeded into triplicate wells of a 96-well plate as described above (1 × 10⁵ cells/well). Six hours after the seeding step, infected and mock-inoculated cells were exposed to the synthetic TLR3 agonist poly(I·C) (100 μg/ml in culture medium) (Invivogen, San Diego, CA) or to an equal volume of culture medium as a negative control. After 48 h of incubation, secreted cytokines and chemokines were quantified from culture supernatants as described above.

Statistical analyses.

One-way analysis of variance (ANOVA) followed by Dunnett's posttest (Prism, version 4.0; GraphPad, San Diego, CA) was used to calculate differences in LDH levels from supernatants of endocervical cells with different amounts of M. genitalium inocula. In addition, ANOVA with Dunnett's posttest was also used for comparison of M. genitalium replication kinetics among A2EN and ShEN101 cells and for multiple comparisons among treatment groups in the poly(I·C) studies. We used a Student t test to evaluate differences in IL-8 secretion levels following azithromycin or gentamicin treatment of persistently infected cells or in comparisons of the secretion levels of individual cytokines from a single cell type to levels of mock-inoculated cells. Unless otherwise noted, significance was indicated at a P value of <0.05 for all statistical comparisons.

RESULTS

Acute cytotoxic effects of M. genitalium infection are dose dependent.

In order to determine whether M. genitalium causes epithelial cytotoxicity, serial dilutions of mid-log-phase organisms (1 × 10⁵ to 1 × 10⁷ genome equivalents per well; multiplicity of infection [MOI] of 2 to 200) were inoculated onto monolayers of A2EN and ShEN101 endocervical epithelial cells. With the highest inoculum (MOI of 200), as early as 2 days p.i. rounding of A2EN and ShEN101 cells was observed, ultimately leading to significant detachment of cells between 4 and 14 days p.i. (Fig. 1; results for the A2EN cells are presented but were indistinguishable from those in ShEN101 cells). No overt cellular changes were observed at an MOI of 20 (Fig. 1), but by 14 days p.i., A2EN and ShEN101 monolayers infected at an MOI of 200 contained between 2.5- to 3-fold fewer cells than cultures infected at an MOI of 20. These findings highlight the modest cell lysis, detachment, or growth inhibition of human endocervical epithelial cells induced by high M. genitalium inocula. Compared to mock-inoculated cells, no microscopic or quantitative differences were observed in cultures of endocervical epithelial cells inoculated with M. genitalium at an MOI of 2 (data not shown).

Fig 1.

Microscopic cytotoxicity in response to M. genitalium G37 infection of human endocervical epithelial cells. Immortalized human endocervical epithelial cell lines (A2EN and ShEN101), which were derived from two different donors, were inoculated with escalating doses of M. genitalium (MOIs of 2 to 200) and then observed via light microscopy for gross signs of monolayer disruption. Inoculation of 200 organisms per cell resulted in rounding of A2EN and ShEN101 cells by 2 days (d) p.i. at the highest inoculum (MOI of 200). Continued deterioration of the cell monolayer occurred between 4 and 14 days p.i. when significant detachment of cells was observed at the highest inoculum (MOI of 200). No obvious microscopic signs of monolayer disruption were observed at an MOI of 20 or at an MOI of 2 (data not shown) compared to mock-infected cells at any time point from 2 to 14 days p.i. Findings from the low-passage-number ShEN101 cells were indistinguishable from A2EN and omitted for the sake of simplicity.

We next determined whether M. genitalium exposure results in acute host cell lysis by measuring lactate dehydrogenase (LDH) from A2EN and ShEN101 supernatants at 24 to 72 h p.i. In both A2EN and ShEN101 cells, no significant increases in LDH activity were observed at 24 h p.i. at an MOI of 2, 20, or 200 (Fig. 2A and B) (P > 0.05, ANOVA). However, by 48 h p.i., M. genitalium elicited a dose-dependent release of LDH, which was significant at an MOI of 200 in both A2EN and ShEN101 cells compared to levels in mock-inoculated cells (Fig. 2A and B, respectively) (P < 0.05, ANOVA). Thus, the loss of cells over time in the cultures inoculated at MOIs of 20 and 200 as described above likely resulted from cytotoxicity. In both A2EN and ShEN101 cells, no significant LDH secretion was observed after inoculation of M. genitalium at an MOI of 5 or 10 (data not shown). Moreover, after heat inactivation, M. genitalium failed to elicit LDH release from either endocervical cell type, indicating that bacterial viability is required for the observed cytotoxicity (Fig. 2A and B).

Fig 2.

Acute cytotoxicity of M. genitalium to endocervical epithelial cells A2EN (A) and ShEN101 (B) was dose dependent. Human endocervical epithelial cells were exposed to various inocula of M. genitalium G37 (black fill) or an equal volume of culture medium as a negative control (no fill). After 24 or 48 h of infection, LDH was measured from culture supernatants as a marker for epithelial cell lysis. Heat-inactivated organisms (gray fill) served as controls to determine whether viability was necessary for immune activation. A statistically significant LDH increase relative to controls is noted by an asterisk at a P value of <0.05 (ANOVA).

Establishment of persistent M. genitalium infection of human endocervical epithelial cells.

Based on the data above, we used an MOI of 10 to avoid induction of cytotoxicity at 48 h p.i. At this MOI M. genitalium replication was relatively slow in A2EN and ShEN101 cultures (Fig. 3A), achieving only a 10-fold increase in titer at 36 days p.i. Cytotoxicity was not was observed by light microscopy from 0 to 60 days p.i. in either persistently infected cell line. Based on three independent experiments for each cell type, M. genitalium replication increased significantly from 36 days to 70 to 80 days p.i., with mean titers increasing from 1.6 × 10⁷ to 8.5 × 10⁹ organisms per ml during this time period. PCR quantification of M. genitalium from the cellular and supernatant fractions showed consistently increased loads in the cellular fraction (Fig. 3B) up to 36 days p.i. In parallel with robust increases in M. genitalium titers, by 70 days p.i., cell rounding was pronounced in infected A2EN and ShEN101 cultures, affecting more than 80% of cells. Given the level of host cell detachment and/or lysis evident at this time, M. genitalium titers were increased in culture supernatants at 60 and 80 days p.i., resulting in reversal of the ratio of cellular to supernatant organisms (Fig. 3B).

Fig 3.

Replication of M. genitalium in human endocervical epithelial cell cultures. (A) After inoculation of A2EN or ShEN101 cells with M. genitalium G37 (MOI of 10), bacterial replication was monitored by a qPCR assay, as described in Materials and Methods, from 0 to 80 days p.i. (B) Differences in qPCR titers are expressed as the mean ratio (±standard error of the mean) of the cellular to supernatant fraction at each time point from three independent studies.

Between the two human endocervix-derived epithelial cell lines, the results were remarkably consistent, with no significant differences in M. genitalium titers observed at any time point from 0 to 80 days p.i. (Fig. 3A) (P > 0.05, Student's t test). Titers were adjusted to account for the number of cells lost due to splitting of the cultures, but, overall, mean M. genitalium titers increased over 500-fold in A2EN cultures and over 5,000-fold in ShEN101 cells from inoculation to cessation of the study at 85 days p.i. As described in Materials and Methods, the reported titers are from the cellular fraction of persistently infected cell cultures, which represent a combination of both adherent and intracellular organisms.

Long-term infection elicits persistent inflammation with minimal cytotoxicity.

After inoculation at an MOI of 10, we first confirmed acute secretion (48 h p.i.) of IL-6, IL-8, G-CSF, GM-CSF, and MCP-1 in response to M. genitalium from human A2EN and ShEN101 cells (see previously published data in reference 26; also data not shown). Next, to investigate the long-term cytokine responses to persistent infection, culture supernatants were collected at 14, 36, and 60 days p.i. as described above in Materials and Methods. Human A2EN and ShEN101 endocervical epithelial cells infected with M. genitalium showed significantly increased secretion of IL-7 and IL-8 at 14 days p.i., followed by marked but insignificant secretion of IL-6 and significantly increased levels of IL-7 and IL-8 at 36 days p.i. (P < 0.05, ANOVA) (data not shown). By 60 days p.i., in A2EN and ShEN101 cells M. genitalium-induced cytokines included IL-6, IL-7, IL-8, GM-CSF, and MCP-1 (Table 1) (P < 0.05, ANOVA). No significant increases in the other tested cytokines were observed at any time point (see the complete list of analytes in Materials and Methods). Although the intensity of cytokine elaboration varied between both endocervical cell types, the profiles of response to M. genitalium were very similar between both cell types, and thus the A2EN line was used for the remainder of the studies.

Table 1.

Profile of cytokine and chemokine secretion from human endocervical epithelial cells persistently infected with M. genitalium

Cytokine	Amt of cytokine (pg/ml) ina:
	A2EN cells	ShEN101 cells
IL-6	47.54 ± 4.6 (4.22)	20.57 ± 2.5 (3.77)
IL-7	27.37 ± 0.9 (3.34)	1.96 ± 0.35 (16.64)
IL-8	3,079 ± 72.9 (4.14)	647.8 ± 19.0 (1.37)
GM-CSF	232.7 ± 9.8 (20.72)	7.0 ± 0.7 (2.16)
MCP-1	35.35 ± 1.5 (25.18)	32.39 ± 2.3 (75.32)

^aHuman endocervical epithelial cells (A2EN and ShEN101) derived from two donors were inoculated with M. genitalium G37 (MOI of 10). After 60 days of persistent infection, culture supernatants were collected to quantify secreted cytokines. Values are expressed as the means ± standard errors of the means, and fold increase compared to mock-inoculated cells is shown in parentheses. The presented values were obtained after normalization to the mean of three replicate experiments for mock- and M. genitalium-inoculated cells. All cytokines listed above were significantly increased relative to mock-inoculated cells (P < 0.05; Student's t test).

In order to determine the relative contributions of intra- and extracellular organisms to the observed cytokine secretion, in separate studies and after 36 days of infection, cells were treated with either gentamicin (2-h exposure at 200 μg/ml) or azithromycin (2-day exposure at 8 μg/ml), and then cytokine secretion was measured 2 and 14 days after treatment. Significant secretion of IL-6 and IL-8 was observed 2 days after gentamicin or azithromycin treatment relative to antibiotic-treated, mock-infected cells (Fig. 4A shows results for IL-8) (P < 0.05, Student's t test). Fourteen days after antibiotic treatment, cytokine elaboration had returned to baseline levels in persistently infected, azithromycin-treated cells but remained significantly increased in gentamicin-treated cells (P < 0.005, Student's t test).

Fig 4.

Azithromycin but not gentamicin treatment ablated endocervical cytokine responses from persistently infected cells. In order to determine whether killing of M. genitalium would eliminate the observed inflammatory response, persistently infected A2EN cells (36 days p.i.) were exposed to either 200 μg/ml of gentamicin (Gm) for 2 h to kill extracellular organisms or 8 μg/ml of azithromycin (Az) for 2 days to kill both intra- and extracellular organisms. As controls, antibiotic-treated mock-inoculated cells were processed in parallel. IL-8 secretion was measured from culture supernatants 2 (A) and 14 days (B) after inoculation. Data are presented as pg/ml of secreted IL-8 ± standard errors of the means. As determined using a Student t test, significant differences versus appropriate mock-inoculated controls are noted as follows: *, P < 0.05; **, P < 0.005.

Persistent M. genitalium infection through day 60 of culture did not result in significant microscopic changes in cell morphology. To confirm that cytotoxicity was not occurring during this time, we measured LDH secretion from A2EN cells and found no significant differences between mock-infected and persistently infected cells at 14, 36, or 60 days p.i. (Fig. 5) (P > 0.05, Student's t test). However, as described above, cytotoxicity was observed between days 70 and 80 p.i., as indicated by cell rounding and detachment in cells persistently infected with M. genitalium. The observed cytotoxicity was concurrent with rapid M. genitalium replication, resulting in a more than 500-fold increase in titer during this 10-day period (Fig. 3A). Due to the substantial microscopically documented cytotoxicity, it was not possible to accurately quantify and seed viable cells for LDH comparisons between 70 and 85 days p.i.

Fig 5.

Endocervical epithelial cytotoxicity occurs only with high M. genitalium loads. In three independent replicate experiments, we tested the ability of M. genitalium to elicit epithelial cell toxicity as measured by LDH secretion at 14 and 36 days after inoculation. Marked differences in cell numbers and morphologically abnormal cells prevented accurate comparisons of LDH secretion after 60 days p.i. As a positive control and for calculation of percent cytotoxicity, an equal number of cells were detergent lysed to determine the maximum releasable LDH.

Persistently infected endocervical epithelial cells are responsive to exogenous TLR stimulation.

Since we observed that long-term infection elicits persistent proinflammatory cytokine secretion, we next determined whether persistently infected cells maintain sensitivity to secondary innate stimulation. In two separate independent studies, persistently infected A2EN endocervical epithelial cells secreted significant levels of IL-7 (22.3 ± 0.7 pg/ml) and IL-8 (11356 ± 87 pg/ml) at 36 days p.i. relative to mock-inoculated cells (P < 0.05, ANOVA). In these studies, levels of IL-6 (74.0 ± 2.1 pg/ml) were not significantly increased by M. genitalium infection (P > 0.05), but mean IL-6 levels were more than 5-fold higher in supernatants of persistently infected cells than in mock-inoculated cells (14.5 ± 1.0 pg/ml). Poly(I·C) treatment of mock-inoculated A2EN cells induced significant secretion of IL-1β, IL-6, IL-8, G-CSF, and MIP-1β relative to nontreated cells (Table 2) (ANOVA, P < 0.05). Next, we observed that M. genitalium infection enhanced sensitivity to poly(I·C) stimulation since several cytokines were significantly increased compared to sensitivity in persistently infected cells not treated with poly(I·C) (Table 2). In short, poly(I·C) treatment of persistently infected cells resulted in synergistic enhancement of the TLR3 agonist's effect on the cells. It is noteworthy that M. genitalium infection alone was not sufficient to elicit G-CSF and MIP-1β secretion at any time point, and this shows that secretion of these two cytokines in response to poly(I·C) is also synergistically enhanced.

Table 2.

Cytokine responses to TLR3 stimulation using human A2EN endocervical epithelial cells persistently infected with M. genitalium

Cytokine	Amt of cytokine (pg/ml) by treatmenta
	Mock infection	With poly(I · C)
		Mock infection	G37 infection
IL-1β	14.1 ± 0.4	47.2 ± 3.2*	47.1 ± 1.5
IL-6	14.5 ± 1.0	476.2 ± 0.9*	1,212 ± 46.8**
IL-7	11.7 ± 0.5	12.8 ± 0.3	19.6 ± 2.0**
IL-8	4,339 ± 122	16,445 ± 2,722*	26,406 ± 2,135**
G-CSF	40.0 ± 3.1	235 ± 7.3*	307 ± 11.2**
MIP-1β	21.1 ± 5.2	874 ± 16.7*	2,090 ± 91.4**

^aHuman endocervical epithelial cells (A2EN) were inoculated with M. genitalium G37 (MOI of 10). Mock-inoculated cells were processed in parallel. After 36 days of infection, cells were exposed to the TLR3 agonist poly(I · C) or an equal volume of culture medium as a negative control. Culture supernatants were collected to quantify secreted cytokines. Presented data are from a single representative experiment among three independent studies that were statistically indistinguishable. Values are expressed as the means ± standard errors of the means.

^*, P < 0.05 for mock infections and poly(I · C) treatment versus mock infection alone;

^*, P < 0.05 for G37 infection and poly(I · C) treatment versus mock infection and poly(I · C) (ANOVA).

DISCUSSION

This work represents the first investigation of host immune responses to persistent M. genitalium infection of female reproductive tract epithelial cells. In order to accomplish this study, we first defined the acute organism burden necessary to elicit detectable cytotoxicity as determined microscopically and by LDH release in human endocervical epithelial cells (Fig. 1). Inoculation with approximately 20 organisms per cell elicited detectable LDH secretion, but not until approximately 200 M. genitalium organisms per cell were added did we observe a significant level of cell lysis (Fig. 2). We then inoculated cells at an MOI below this threshold in order not to overtly affect cell viability and not to allow establishment of persistent infection. With an inoculum of 10 organisms per cell, M. genitalium replicated relatively slowly over a 36-day period, followed by gradual increases in titer from 36 to 60 days p.i. Because we used a PCR-based measure of M. genitalium titers, it is possible that the doubling rate was masked due to nonreplicating and/or dead organisms. Nonetheless, after 60 days of infection, rapid M. genitalium replication was associated with visible microscopic cytotoxicity, suggesting that high microbial loads are required to elicit epithelial lysis as was observed using high MOIs for initial cell monolayer inoculation experiments. This conclusion is supported by the fact that significant LDH release was observed only after acute M. genitalium infection at the highest MOI (Fig. 2). In vitro endocervical cytokine responses to M. genitalium have been observed as early as 6 h p.i. (26), again suggesting that epithelial cells play an important role in rapid innate signaling soon after mucosal exposure. Here, we showed that persistent M. genitalium infection provoked endocervical cells to secrete significant levels of proinflammatory cytokines and chemokines over time, a novel and clinically relevant finding.

Unfortunately, very limited clinical data exist regarding the organism burden of M. genitalium infection in women, and none of it is standardized in any relevant way to host cells in vivo (3, 5, 32). Moreover, previous studies did not attempt to assess the relationship of organism burden to the host inflammatory response. In our in vitro studies, considering all time points at 0 to 80 days p.i., the median ratio (range) of M. genitalium organisms to host cells was 8 (4 to 5,126) in ShEN101 cells and 10 (1 to 1,501) in A2EN cells using quantitative PCR. Although we did not visualize the level of parasitization on a cellular level, the observed ratios of bacteria per cell suggest that low organism burdens can elicit chronic inflammation in this model of persistence. Collectively, in human endocervical epithelial cells, we observed a relatively low-level infection that was characterized by chronic inflammation with minimal impact on host cell integrity. Future clinical studies will be important for understanding how organism burden relates to inflammatory urogenital disease in women.

Clinically, persistent M. genitalium infection appears to be very common (4, 6, 10, 13) though it is unclear that the endocervix is the primary site of infection. The organism has been associated with cervicitis in several clinical studies, but conclusive evidence is lacking since several studies have found no correlation with disease (reviewed in references 23, 25, and 30). The disparate findings for associations of M. genitalium with cervicitis are likely attributable to population heterogeneity among studies and different criteria for clinical diagnosis. Ultimately, a universal case definition for cervicitis would significantly enhance our understanding of M. genitalium disease in women and better direct antimicrobial therapy. With the understanding that additional studies are needed to adequately address endocervical organism burdens, our in vitro studies demonstrate that cytotoxicity is directly correlated with the number of mycoplasmas. Importantly, host cytokine responses occur with low organism burdens, and we hypothesize that this allows the organism to persist in the mucosa with minimal direct epithelial toxicity.

Chronic cytokine and chemokine secretion could explain how M. genitalium burden in the endocervix is significantly associated with HIV shedding (24). We hypothesize that the chronic inflammatory state induced by M. genitalium infection enhances HIV susceptibility and localized shedding due to recruitment of susceptible macrophages and T cells to the mucosa. Although epithelial toxicity appears to be minimal, in fact the profile of cytokine secretion by persistently infected endocervical epithelial cells included several key proinflammatory mediators for localized inflammation that are consistently increased in both acute (26) and, as we show here, persistent infection. IL-8 is a potent chemotactic molecule for neutrophils and has been positively associated with HIV shedding in cervicovaginal lavages (28). Therefore, persistent M. genitalium infection could have significant consequences for HIV shedding and/or acquisition of primary virus infection at the cervical mucosa. Cytologic profiling of leukocytes in cervical specimens from M. genitalium-positive subjects compared to those without infection would be a useful in vivo approach to test the hypothesis that M. genitalium enhances HIV infection through induction of a chronic host inflammatory response.

The observation that M. genitalium infection enhances epithelial sensitivity to TLR agonists suggests that infected subjects might be hyperresponsive to exogenous innate immune stimulation induced by other STIs including bacterial vaginosis. This is demonstrated by an additive response pattern for several cytokines (Table 2). For example, MIP-1β and G-CSF were not increased by persistent infection alone, but when persistently infected cells were stimulated with poly(I·C), the effect was significantly greater than that measured in uninfected cells treated with poly(I·C). In short, not only does persistent M. genitalium infection result in inflammatory cytokine secretion, but also infected endocervical epithelia appear to have heightened responses to other stimuli. Since the endocervix is the primary site of infection by C. trachomatis and is also colonized by N. gonorrhoeae, modulated immune function in endocervical tissues could be important, considering the high prevalence of these pathogens.

While informative for investigating epithelial persistence, the 2-D epithelial cell culture system we employed is not without limitations. First, the endocervix is a simple columnar, mucous-producing epithelium whose function is not fully reproduced in 2-D cultures because they lack directional polarity (1). In lieu of more advanced three-dimensional models, the current studies employed standard 2-D methods, which did not allow continual separation of intra- and extracellular mycoplasmas but facilitated long-term culture. Based on our study design, the reported 60-day responses to persistent infection represent stimulation from both adherent and intracellular organisms. If extracellular organisms are killed with gentamicin, persistent IL-8 secretion remains for at least 14 days after treatment, but this is not the case if both intra- and extracellular organisms are killed with azithromycin. These findings suggest that intracellular organisms alone are capable of eliciting cytokine secretion, but intra- and extracellular stimulation together leads to a more robust host response. It is well established that M. genitalium can occupy intracellular niches in urogenital epithelia (3, 26, 31), but the life cycle of M. genitalium in the context of human reproductive tract infections is not fully understood. It is possible that mucosal infections in humans are predominately intracellular in nature due to cell-mediated and antibody immune pressures, but focused clinical investigations are necessary to address this issue. In vitro modeling of persistent intracellular infection in the absence of viable extracellular organisms is difficult because gentamicin, although good for short-term studies, slowly enters eukaryotic cells (11) and gradually begins to kill intracellular organisms. We took a reductionist approach to determine how M. genitalium interacts specifically with human endocervical epithelial cells, but this does not consider the full complement of cells or extracellular milieu present in the endocervical mucosa. Considering the chemotactic profile of cytokines secreted in response to M. genitalium infection, it is likely that phagocytic cells would be present in the epithelium during persistent infection.

Collectively, as a recognized cause of persistent human infections associated with chronic inflammation, M. genitalium is a prevalent and important urogenital pathogen of men and women. We still have much to learn about the life cycle of M. genitalium and the pathological effect of persistence, but it is now evident that this organism is a likely cause of reproductive tract inflammation. Understanding the mechanisms of infection and persistence are of utmost importance considering that the current treatment recommendations for male NGU and cervicitis may not be optimally effective for M. genitalium infections. Probably due to rising resistance to macrolide antibiotics (30), single-dose azithromycin for M. genitalium has been associated with NGU treatment failure (4). In fact, single-dose azithromycin appears to induce drug resistance through mutations in the 23S rRNA gene (12, 18). Genomic recombination of mgpB (MG191 gene) and mgpC (MG192 gene) adhesins with noncoding MgPar regions exemplifies how M. genitalium has adapted to a persistent lifestyle in humans since this antigenic variability is likely in response to host immune pressure (13, 14, 22). Clearly, continued investigation of M. genitalium pathogenesis is imperative for a fuller understanding of human urogenital disease and for the design of better strategies to prevent complications.