The multifaceted role of oestrogen in enhancing Chlamydia trachomatis infection in polarized human endometrial epithelial cells

Authors


E-mail: pbwyrick@etsu.edu; Tel. (+1) 423 439 8079; Fax (+1) 423 439 8044.

Summary

The oestrogen receptor (ER) α−β+ HEC-1B and the ERα+β+ Ishikawa (IK) cell lines were investigated to dissect the effects of oestrogen exposure on several parameters of Chlamydia trachomatis infection. Antibody blockage of ERα or ERβ alone or simultaneously significantly decreased C. trachomatis infectivity (45–68%). Addition of the ERβ antagonist, tamoxifen, to IK or HEC-1B prior to or after chlamydial infection caused a 30–90% decrease in infectivity, the latter due to disrupted eukaryotic organelles. In vivo, endometrial glandular epithelial cells are stimulated by hormonally influenced stromal signals. Accordingly, chlamydial infectivity was significantly increased by 27% and 21% in IK and HEC-1B cells co-cultured with SHT-290 stromal cells exposed to oestrogen. Endometrial stromal cell/epithelial cell co-culture revealed indirect effects of oestrogen on phosphorylation of extracellular signal-regulated kinase and calcium-dependant phospholipase A2 and significantly increased production of interleukin (IL)-8 and IL-6 in both uninfected and chlamydiae-infected epithelial cells. These results indicate that oestrogen and its receptors play multiple roles in chlamydial infection: (i) membrane oestrogen receptors (mERs) aid in chlamydial entry into host cells, and (ii) mER signalling may contribute to inclusion development during infection. Additionally, enhancement of chlamydial infection is affected by hormonally influenced stromal signals in conjunction with direct oestrogen stimulation of the human epithelia.

Introduction

Over the past several decades, research has emerged that connects increased levels of the hormones, oestrogen and progesterone, with increased occurrence and/or severity of sexually transmitted infections (STI). Oestrogen has the potential to stimulate gene transcription in the causative agent of genital warts and cervical cancer, human papillomavirus, possibly leading to persistent viral infection (Sonnex, 1998; Shew et al., 2002). Trichomonas vaginalis possesses oestrogen binding sites and high oestrogen levels in animal models correspond to increased infectivity of T. vaginalis (Sonnex, 1998). Infection with herpes simplex virus type 2 is enhanced under high progesterone exposure (Sonnex, 1998). Conversely, in the mouse corneal model of latent herpes infection, viral shedding can be reactivated following exposure to oestrogen (17-β-oestradiol) (Vicetti Miguel et al., 2010). Additionally, gonococcal genital infections occur more frequently prior to and during the time of menstruation when oestrogen levels rise (Sweet et al., 1986; Sonnex, 1998). Given these reports, it is important that the underlying mechanisms of hormonal influence on sexually transmitted pathogens be elucidated.

The most common bacterial sexually transmitted pathogen worldwide is the Gram-negative, obligate intracellular organism Chlamydia trachomatis. Chlamydial infections, although often asymptomatic, can predispose a female patient to chronic disease manifestations, such as pelvic inflammatory disease and tubal infertility (Peipert, 2003). In the 1980s, it became apparent that C. trachomatis infections were also subject to the influence of female reproductive hormones. Data accumulated from clinicaltrials suggested that detection of chlamydial infection in female patients during the oestrogen phase of the menstrual cycle was highly predictive of upper genital tract complications, such as salpingitis (Sweet et al., 1986). Elegant studies of chlamydial infection in the guinea pig model by Rank and colleagues indicated increased intensity and duration of inflammation, fibrosis and tubal dilation when chlamydiae migrated into the upper genital tract during high oestradiol levels (Rank et al., 1982; 1993; Rank and Sanders, 1992). Guseva et al. reported that both luminal and glandular epithelial cells, harvested from the cervix, uterus and uterine horns of mature female swine and grown ex vivo, were significantly more susceptible to Chlamydia suis S45 in the proliferative, oestrogen-dominant phase versus the secretory, progesterone-dominant phase (Guseva et al., 2003). In vitro, exposure to oestradiol increased infectivity of C. trachomatis serovar K and L1 by 50–60% in HeLa cells (Bose and Goswami, 1986). Subsequently, Wyrick et al. were the first to show that C. trachomatis serovar E attachment to and infection of explanted, primary human endometrial epithelial cells were greater when the tissues were collected during the oestrogen-dominant phase as opposed to the progesterone-dominant phase (Maslow et al., 1988; Wyrick et al., 1989; 1993).

It has become increasingly recognized that hormones released from the ovaries have both direct and indirect effects on human endometrial epithelial cells. Several indirect effects have been demonstrated to come from effector molecules, primarily growth factors, released from stromal cells surrounding the endometrial glands in the uterine basalis and functionalis, which contribute to the stimulation of growth and maturation of the uterine epithelial cells (Cunha et al., 1985). Chlamydiae, in the form of persistent reticulate bodies (RB), are postulated to inhabit the undifferentiated epithelial cells at the base of the uterine glands. As the infected epithelial cells proliferate and mature, chlamydiae progress through their developmental cycle, and are transported upward, such that infectious elementary bodies (EB) may infect freshly differentiated luminal epithelial cells repopulating the endometrial lining previously shed in mensus. Thus, C. trachomatis creates a reservoir of infection that can be propagated with each menstrual cycle (Richmond, 1985).

Several possibilities exist to explain the observed effects of oestrogen on chlamydial infection: (i) membrane oestrogen receptors (mERs) or proteins involved in the oestrogen receptor (ER) complex on the eukaryotic plasma membrane could serve as receptors for chlamydial entry, (ii) the oestrogen-induced physiological state of the genital epithelia may provide an environment that is conducive for chlamydial growth, and/or (iii) ERs, themselves, are important to chlamydial entry and the physiological response of the host, both direct and indirect, to this hormone is vital to the developmental cycle of C. trachomatis. To investigate these critical issues, the effects of oestrogen exposure on several parameters of C. trachomatis serovar E infection in vitro in the endometrial epithelial cell lines, HEC-1B and Ishikawa (IK), were compared in the presence or absence of co-culture with endometrial stromal cells.

Results

Hormone responsiveness of IK and HEC-1B cell lines

Hormone responsiveness to oestrogen has been defined as (i) the rapid translocation of ERα from the cytoplasm to the nucleus in the presence of oestrogen, and (ii) an upregulation of progesterone receptor (PGR) and transforming growth factor alpha (TGFα) mRNAs following exposure to oestrogen (Hall et al., 2001). Immunofluorescence analysis demonstrated that ERα began translocating from the cytoplasm of IK cells to the nucleus within 30 min to 1 h of exposure to oestrogen (E2; Fig. 1A). Translocation of ERβ occurred within 30 s of oestrogen exposure in HEC-1B cells (data not shown). Transcripts for both PGR and TGFα were also significantly increased (P < 0.05) in E2-exposed IK cells (Fig. 1B, white striped bars) when compared with transcript levels in unexposed controls (Fig. 1B, white bars). Conversely, HEC-1Bs indicated no increase in TGFα transcripts following E2 exposure (Fig. 1B, black striped bars). Transcripts for both ERα and ERβ were present in IK cells at constant levels in the presence or absence of oestrogen. As expected, data from RT-PCR analysis confirmed the observations that HEC-1Bs express ERβ, but do not express ERα (Guseva et al., 2005) or PGR as evidenced by the lack of transcripts (Fig. S1A). For comparison, reported ER-negative ductal breast cancer cell lines were also tested, which C. trachomatis serovar E could infect, albeit at very low levels, i.e. 10–20% infectivity in HCC1806 cells. However, very low amounts of both ERα and ERβ transcripts were detected by real-time RT-PCR analysis and ERβ protein was found on the surface of HCC1806 cells by flow cytometry and Western blot analyses (Guseva et al., 2005). Immunofluorescence staining of IKs and HEC-1Bs revealed the presence of both ERα and ERβ on the surface of IK cells but only ERβ was found on the surface of HEC-1Bs, again confirming the absence of ERα in this cell line (Fig. S1B). Unfortunately, the reported ERα-/β- double knockout IK cell line, which we obtained from two different laboratories, expressed both ERα and ERβ by qRT-PCR as well as by immunofluorescent staining and Western blot analyses using newly developed reagents and a selection of eight monoclonal or polyclonal anti-ERα and anti-ERβ antibodies obtained from four biotechnology companies. Thus, the ERα+β+ IK cell line remains responsive to stimulation by oestrogen, according to these particular markers of hormone responsiveness, whereas the ERα−β+ HEC-1B cell line does not respond to oestrogen in the same manner, purposely providing the opportunity to analyse the effects of oestrogen on chlamydial genital infection in two distinct endometrial epithelial cell lines.

Figure 1.

Hormone responsiveness of IK and HEC-1B cell lines. The response of IK and HEC-1B cell lines to oestrogen exposure was measured by (A) IFA analysis of the movement of ERα from the cytoplasm to the nucleus and (B) expression of PGR and TGFα by quantitative RT-PCR.
A. IK cells were grown in oestrogen-free medium for 72 h and then exposed to 10−8 M E2 for 48 h. Replicate cell monolayers were fixed, permeabilized and immunostained at 0, 24 and 48 h post oestrogen exposure with antibodies generated against ERα and Alexa Fluor 488 second-affinity antibody (ABY).
B. Expression of PGR and TGFα in polarized IK (white bars) and HEC-1B (black bars) cells grown in oestrogen-free medium (E2−, solid bars) or in medium containing 10−8 M E2 (striped bars) for 48 h. Total RNA was subjected to quantitative RT-PCR using primers specific for PGR and TGFα. Amplification of PGR and TGFα was normalized to the housekeeping gene β2M. Asterisks (*) represent a significant difference (P ≤ 0.05) between E2− and 10−8 M E2 samples for each cell line.

Chlamydial EB associate with ERβ on the surface of HEC-1B cells

As an initial step in determining the role of ERs in chlamydial attachment, the interaction of infectious EB with HEC-1B cells was tested. Immunoelectron microscopy analyses demonstrated that gold-conjugated second-affinity antibodies reacted with primary anti-ERβ antibodies localized at or near the apical surface of polarized HEC-1B cells as well as with attached serovar E EB (Fig. 2A arrows and B). Since an ER homologue is not apparent in the chlamydial genome, residual host cell debris may account for the binding of some gold-labelled antibodies to the percoll-purified EB. Two-dimensional gel analysis of the apical membrane of polarized HEC-1B cells, obtained by apical lift, confirmed the surface presence of ERβ (Fig. 2C, circles). As shown previously by Davis et al. (2002), ER-specific chaperonin protein disulfide isomerase (PDI) was also on the surface of HEC-1B cells (Fig. 2C, square). In addition, surface exposure of ERβ was demonstrated by flow cytometry examination of HEC-1B cells (Fig. 2D). The weak ERα signal revealed by flow cytometry was likely due to the presence of a truncated, possibly inactive form of ERα, as indicated by a smaller band occasionally present in some Western blot analyses or the possibility of some cross-reacting antibody epitopes. Multiple attempts to knock down/knock out ERβ by siRNA/shRNA interference were unsuccessful in HEC-1B. While ERβ message could be reduced as much as 85–98% and resulted in a corresponding decrease in 35S-labelled EB attachment, protein levels were decreased but still detectable, suggesting an increased half-life of mERβ. Finally, the presence of ERβ antibody prior to and during EB inoculation onto HEC-1B cells reduced EB infectivity significantly by 60% (P < 0.05; see below). Taken together, these data suggest that direct binding of EB to ERβ could be involved in attachment. Alternatively, since ERβ antibodies also associated with EB (Fig. 2B), EB may have a surface adhesin that binds oestrogen which, in turn, binds to ERβ to facilitate EB attachment to the epithelial surface.

Figure 2.

Chlamydial EB associate with ERβ on the surface of HEC-1B cells. Analysis of the involvement of plasma membrane-associated ERβ in C. trachomatis serovar E attachment/infectivity.
A and B. Immunoelectron microscopy of Lowicryl samples of EB inoculated onto polarized HEC-1B cells for 30 min and exposed to gold-conjugated anti-ERβ polyclonal antibodies.
C and D. Confirmation of the surface location of ERβ on HEC-1B cells by (C) Western blot of 2-D gels of apical membrane-lifted components – circles designate ERβ and the square designates the positive control PDI, and (D) flow cytometric analysis of formaldehyde-fixed HEC-1B cells immunostained with anti-ERα or anti-ERβ and Alexa Fluor 488 second-affinity antibody (IgG).
Magnifications: (A) ×34 000; (B) ×49 500.

Blocking of ERs with antibodies against ERα and/or ERβ decreased chlamydial infectivity

Given the observation that ERβ-directed antibodies associate with EB on the surface of HEC-1Bs, the effect of blocking ERs with ER-specific antibodies on the epithelial cell surface prior to chlamydial infection was investigated. Studies have demonstrated that blocking cell surface PDI with anti-PDI antibodies causes chlamydial entry and infectivity to decrease; therefore, we used PDI-specific antibodies as a positive control (Davis et al., 2002; Abromaitis and Stephens, 2009). Prior incubation of IK and HEC-1B cells with selected antibodies resulted in a significant reduction in the per cent infectivity of serovar E in both cell lines. Per cent infectivity in IK cells was significantly reduced (P < 0.05) by the anti-PDI (55%), anti-ERα (45%), anti-ERβ (51%) antibodies and the antibody cocktail containing anti-ERα and anti-ERβ antibodies (CKTL; 55%) (Fig. 3, white bars). Likewise, per cent infectivity in HEC-1B cells was significantly reduced by anti-PDI (45%), anti-ERβ (60%) and the anti-ERα/anti-ERβ antibody CKTL (68%) (Fig. 3, black bars). Anti-ERα antibodies only slightly reduced the chlamydial infectivity (20%, not significant), in agreement with the findings that HEC-1Bs lack a functional ERα (Fig. 3, black bars). These data suggest that antibody blocking of the epithelial ligand-binding domain of the mERs or their PDI adapter complex resulted in significant decreases in chlamydial infectivity, implying a role for ERs and PDI in facilitating entry of C. trachomatis serovar E EB into uterine epithelial host cells.

Figure 3.

Blocking of ERs with antibodies against ERα and/or ERβ decreases chlamydial infectivity. Monolayers of IK (white bars) and HEC-1B (black bars) were exposed to anti-IgG (negative control), anti-PDI (positive control), anti-ERα, anti-ERβ antibodies or a cocktail of the anti-ERα and anti-ERβ antibodies (CKTL) prior to infection with C. trachomatis inoculum, also containing the appropriate antibodies. Infected monolayers were incubated for 48 h before harvest for infectivity assay analysis. Per cent infectivity was determined by averaging the percentage of epithelial cells that developed an inclusion in 10 reticule fields (Zeiss Axiovert S100, 400× magnification) from three individual replicate samples. Asterisks (*) indicate a significant difference (P ≤ 0.05) in per cent infectivity between the experimental condition and the anti-IgG-exposed controls.

Tamoxifen negatively effects chlamydial infectivity and development

The clinically relevant ER antagonist tamoxifen (Tx) was also examined to determine its effect on chlamydial attachment and entry. In IK or HEC-1B cells exposed to Tx prior to infection, chlamydial infectivity dropped by 30% and 57%, respectively, when compared with unexposed, infected controls (Fig. 4A). In addition to a decrease in the number of inclusions present, it was also observed that the inclusions present in Tx-exposed epithelia appeared smaller and denser than the inclusions in the unexposed controls. These data, in conjunction with the antibody blocking experiments, add supportive evidence that ERs are important for chlamydial attachment/entry into epithelial cells. The observation that inclusion size seemed to be smaller in Tx-exposed cells further implied that functional ERs are important for inclusion development following entry into the host, possibly due to stimulation of particular cellular signalling pathways.

Figure 4.

The oestrogen receptor inhibitor, tamoxifen (Tx), negatively affects chlamydial infectivity.
A. Polarized IK (white bars) or HEC-1B (black bars) cells were exposed to 10−4 M Tx for 2 h before infection with C. trachomatis; the infected samples were harvested for infectivity assays at 48 hpi.
B. Polarized IK (white bars) or HEC-1B (black bars) cells were infected with serovar E EB. At 6, 12 and 24 hpi, 3 × 10−5 M (IK) or 6 × 10−5 M (HEC-1B) Tx was added to the culture medium. All samples were harvested for infectivity assays at 48 hpi.
Asterisks (*) indicate a significant difference (P ≤ 0.05) in per cent infectivity between the Tx-exposed and unexposed (Tx−) controls.

The hypothesis that oestrogen-stimulated physiological effects on genital epithelial cells are required for proper development of chlamydial inclusions was also examined using non-cytotoxic doses of Tx. Serovar E-infected, polarized IK or HEC-1B cells were exposed to 3 × 10−5 M (IK) or 6 × 10−5 M (HEC-1B) Tx by the addition of Tx to the culture medium at 6, 12 or 24 hpi, for a total exposure time of 42, 36 or 24 h respectively. Once again, chlamydial infectivity was lower in the samples exposed to Tx for both cell lines. In IK cells, infectivity was significantly decreased by 90%, 71% and 44% when Tx was added to the cells at 6, 12, or 24 hpi respectively (Fig. 4B, white bars). Likewise in HEC-1B cells, infectivity was significantly decreased by 60%, 32% and 41% when Tx was added to the cells at 6, 12, or 24 hpi respectively (Fig. 4B, black bars).

The dramatic effect of Tx exposure following chlamydial entry was also demonstrated with TEM analysis. Tx exposure in uninfected or C. trachomatis-infected HEC-1B and IK cells led to some distinct alterations in morphology of the eukaryotic cellular organelles: (i) the stacked Golgi complex was disrupted (Fig. 5A) compared with chlamydiae-infected Tx-unexposed control samples (Fig. 5C), (ii) the endoplasmic reticulum became dilated and lacked ribosome association (Fig. 5D and *G), (iii) mitochondria lost their characteristic oblong shape and cristae (Fig. 5E) and became long thin structures, often appearing as beaded or ‘dumbbell’-shaped (Fig. 5F and G, arrowhead); and (iv) numerous stress vacuoles were also observed in HEC-1B and IK cells following Tx exposure (Fig. 5A, B, G and I). Interestingly, some chlamydial inclusions appeared to develop normally (Fig. 5A) whereas others contained enlarged RB, suggesting hindrance of the developmental cycle (Fig. 5B). Several epithelial cells had vesicles that contained single RB or transitional forms also indicating developmental interruption (Fig. 5G and H). Lastly, some inclusions were damaged, allowing basal escape of chlamydiae versus the normal apical mode of release for serovar E (Fig. 5I). Given the results from these experiments, it can be concluded that preventing ER functions before or after entry of chlamydial EB interferes with normal epithelial cell physiology and, thus, the chlamydial infectious cycle, substantiating that mERs play a role both in entry of C. trachomatis into epithelial cells and inclusion development.

Figure 5.

Effect of tamoxifen on chlamydiae-infected endometrial epithelial cells. Polarized HEC-1B (A–E) or IK (F–I) cells were infected with C. trachomatis serovar E and then exposed to tamoxifen for 6 (B, D, E, G, H), 12 (A, F) or 24 (I) h and processed at 48 hpi for high-resolution examination by TEM.
A and B. (A) Some epithelial cells contained normal chlamydial inclusions despite Golgi complex disruption, whereas (B) abnormal, enlarged RB were found in other inclusions.
C and D. (C) Control HEC-1B cell displaying normal stacked Golgi complex and grey tubular ribosome-studded endoplasmic reticulum ultrastructure versus (D) tamoxifen-exposed HEC-1B cell with dilated endoplasmic reticulum without associated ribosomes (white linear structures).
E and F. (E) Normal-appearing mitochondria with cristae are shown versus (F) shrunken, beaded mitochondria in some tamoxifen-exposed cells.
G and H. (G) Forty-two hour tamoxifen-exposed IK cell displaying stress vacuoles, an altered mitochondrion (m; arrowhead) and dilated endoplasmic reticulum (*); chlamydial development can be thwarted early as indicated by lack of RB development (arrow), even after (H) EB-to-RB transition (arrow).
I. Infected IK cell exposed to tamoxifen at 24 hpi shows a mid-developmental cycle chlamydial inclusion whose inclusion membrane has been damaged, resulting in premature escape of RB (arrow) at the basal domain.
Magnifications: (A) ×16 400; (B) ×11 700; (C) ×43 700; (D) ×11 700; (E) ×123 500; (F) ×27 000; (G) ×11 700; (H) ×16 400; and (I) ×7140.

The effect of oestrogen exposure on C. trachomatis infectivity in IK and HEC-1B cell lines in the presence and absence of immortalized stromal cells

The Kaufman and Lessey laboratories developed a co-culture system in which IK cells were cultured with immortalized human endometrial stromal SHT-290 cells (Arnold et al., 2001; 2002; Barbier et al., 2005). In their model, polarized epithelial cells were cultured on extracellular matrix (ECM)-coated filters above, but not touching, the stromal cells. On hormone addition, secreted stromal effectors interacted basally with the epithelial cells in paracrine fashion. The IK cells responded with increased proliferation and an enhanced hormone responsiveness that more closely mimicked in vivo conditions (Arnold et al., 2001). The model substantiates that simply polarizing genital epithelial cell lines in vitro is not sufficient to identify some important aspects of genital epithelial cell hormone responsiveness. Indeed, neither epithelial cell numbers nor C. trachomatis infectivity was significantly different in polarized IK (Fig. S2, white bars) or HEC-1B (Fig. S2, black bars) cells cultivated in the absence (Fig. S2, solid bars) or presence of 10−8 M E2 (Fig. S2, striped bars). Therefore, the co-culture model (Fig. 6A and B) was adopted to more accurately investigate the important indirect effects, i.e. stromal-induced paracrine effects, of oestrogen on C. trachomatis serovar E infections in IK and HEC-1B cells.

Figure 6.

Endometrial epithelial cell/stromal cell co-culture model.
A. During the menstrual cycle, epithelial cells present in the glands of the endometrium must differentiate, proliferate and migrate up the glands to the surface of the endometrium to re-establish the luminal epithelial lining shed during menses. This process is directly stimulated and regulated by ovarian-released hormones. Stromal cells surrounding the endometrial glands also respond to hormonal signals by releasing effector molecules that cross the basement membrane and contribute to the stimulation of growth and maturation of epithelial cells. Persistent chlamydial RB are postulated to inhabit the undifferentiated epithelial cells at the base of the endometrial glands. As the infected epithelial cells mature and proliferate, developing RB/EB are transported upward to an area of freshly differentiated luminal epithelial cells. By establishing an infection in the endometrial glandular epithelial cells, C. trachomatis creates a reservoir of infection that can be propagated with each menstrual cycle.
B. Diagram of the in vitro epithelial cell/stromal cell co-culture model. Endometrial epithelial cells (IK or HEC-1B) are polarized on ECM-coated filters. The filter inserts are hung suspended over tissue culture cluster wells containing stromal cell monolayers (SHT-290) growing on the bottom of the well. The epithelial and stromal cells are co-cultured in medium with or without oestrogen (10−8 M E2).

Expression of both PGR and TGFα mRNA was significantly increased in IK cells following oestrogen exposure during co-culture with SHT-290s, while expression of both ERα and ERβ remained constant in the presence or absence of oestrogen in IK cells during stromal cell co-culture. These data indicate that the IK cell line remains responsive to E2 during co-culture with SHT-290s. Conversely, neither PGR nor TGFα expression was increased in HEC-1B samples during stromal cell co-culture, again suggesting that HEC-1Bs are not responsive to oestrogen via these specific measurements of hormone responsiveness.

Although not always significant, an increasing trend in the number of IK cells present following co-culture with SHT-290s (IK/SHT-290) in the presence of E2 was observed (Fig. 7A, white striped bars). Likewise, the number of HEC-1Bs increased during co-culture with SHT-290s (HEC-1B/SHT-290) following E2 exposure (Fig. 7A, black striped bars) compared with HEC-1B/SHT-290 samples without E2, verifying the proliferative effect of oestrogen during co-culture. In control experiments, no proliferation of IK or HEC-1B cells was observed with 10−8M E2 in the absence of SHT-290s, as reported by Arnold et al. (2001; 2002).

Figure 7.

Effect of oestrogen exposure on C. trachomatis infection in IK and HEC-1B during co-culture. Following oestrogen starvation, polarized IK and HEC-1B cultures were moved into wells containing confluent monolayers of SHT-290 cells. Additional IK and HEC-1B cultures and the co-cultures (IK/SHT-290 and HEC-1B/SHT-290) were replenished with E2− or 10−8 M E2 medium.
A. Epithelial cells were counted to determine the amount of cell proliferation for IK (white bars) or HEC-1B (black bars) following 48 h in the absence (E2−, solid bars) or presence (10−8 M E2, striped bars) of oestrogen. Asterisks (*) indicate a significant difference (P ≤ 0.05) in cell counts between oestrogen-exposed and unexposed samples for the respective cell line. The average of 12 determinations per experimental condition is presented.
B. Chlamydiae-infected epithelial samples were harvested at 48 hpi for infectivity assays. Per cent infectivity data were determined as previously described. Asterisks (*) indicate a significant difference (P ≤ 0.05) between the oestrogen-exposed and the unexposed samples for the respective cell line.
C. Epithelial cells from uninfected (UnI) and chlamydiae-infected (Ct) IK, HEC-1B, IK/SHT-290 and HEC-1B/SHT-290 ± E2 were harvested at 24 and 48 hpi for Western blot analysis. Western blots were probed with antibodies against total cPLA2 (cPLA2), phosphorylated cPLA2 (pcPLA2), total ERK (ERK) and phosphorylated ERK (pERK). An example of two experimental repeats is presented.

Infectivity assays demonstrated that chlamydial infection was significantly increased in both IK/SHT-290 and HEC-1B/SHT-290 samples exposed to 10−8 M E2 by 27% and 21% respectively (Fig. 7B, white and black striped bars) compared with equivalent infections in IK/SHT-290 and HEC-1B/SHT-290 samples cultured in the absence of E2 (Fig. 7B, white and black solid bars). Increased chlamydial infection in IK/SHT-290 and HEC-1B/SHT-290 samples exposed to E2 was also observed in immunofluorescence micrographs of infected IK and HEC-1B monolayers stained for detection of chlamydial inclusions (Fig. S3). We confirmed that the EB do not normally escape the IK or HEC-1B cells basally nor migrate through the 0.2 µm filter pore channels nor infect the stromal cells below.

Based upon these results, it appeared that stromal cells released soluble factors in response to oestrogen exposure that enhanced chlamydial infectivity. To confirm this hypothesis, polarized IK or HEC-1B cultures were incubated in conditioned medium from SHT-290 cells cultured in the presence or absence of E2 and infected with serovar E. Per cent infectivity was significantly increased in IK (13%) and HEC-1B (25%) following incubation with oestrogen-exposed conditioned medium for 48 h prior to chlamydial infection. Additionally, per cent infectivity was significantly increased in IK (15%) following incubation with oestrogen-exposed conditioned medium for 24 h prior to chlamydial infection (data not shown). Shotgun proteomic analyses of basal supernatants from HEC-1B/SHT-290 samples detected proteins known to be involved in tissue remodelling, including secreted-phosphoprotein-1 (Table S1 and Fig. S4A), although this growth factor seems to play a more significant role during pregnancy and implantation when progesterone concentrations are increased (Kim et al., 2010).

Collectively, these data show that oestrogen-induced stromal effectors released during co-culture stimulated the epithelial cells in a manner that enhanced chlamydial infection in both IKs and HEC-1Bs. Interestingly, HEC-1Bs, while seemingly unresponsive to direct oestrogen stimulation, were indirectly responsive to oestrogen via hormone-induced stromal cell factors. Since an increase in per cent infectivity without SHT-290 co-culture was not observed with either IKs or HEC-1Bs, these data strongly support that enhanced chlamydial glandular infection during estrous phase is likely mediated in a paracrine manner through hormone stimulation of the endometrial stroma.

Effect of oestrogen exposure on cellular signalling events in polarized IK and HEC-1B cells in the presence or absence of stromal cell co-culture

Data from multiple laboratories show that selective activation of members of the mitogen-activated protein kinase (MAPK) pathway can be stimulated during infection with C. trachomatis L2 (Su et al., 2004; Fukuda et al., 2005; Buchholz and Stephens, 2007; 2008; Vignola et al., 2010). Additionally, activation of mERs stimulate intracellular signalling events, including activation of the MAPK pathway (Hall et al., 2001; Levin, 2003; 2009; Revankar et al., 2005). Inflammatory molecules, such as the antimicrobial peptide secretory leucocyte protease inhibitor (SLPI), and cytokines are hormonally regulated in the female reproductive tract (FRT). Furthermore, the inflammatory cytokines IL-6 and IL-8 are produced following chlamydial infection in vitro (Quayle, 2002; Beagley and Gockel, 2003; Su et al., 2004; Wira et al., 2005; 2010; Fahey et al., 2008). Given these observations, stimulation of the MAPK pathway and production of inflammatory mediators during chlamydial infection in IK and HEC-1B cells with or without SHT-290 co-culture and in the presence or absence of oestrogen were analysed.

Proteomics analysis revealed increased abundance of MAPK in oestrogen-exposed HEC-1B cell lysates from co-cultured samples (Table S2 and Fig. S4B). Phosphorylation of two downstream members of the MAPK pathway, the extracellular signal-regulated kinase (ERK) and the calcium-dependant phospholipase (cPLA2) were examined by Western blot analyses. Overall, no global trends in expression or phosphorylation of cPLA2 or ERK were observed with oestrogen exposure during epithelial cell culture alone or with co-culture (Fig. 7C). However, phosphorylation of ERK (Fig. 7C, pERK) was decreased in oestrogen-exposed IK samples versus controls in both uninfected and infected samples, whereas co-culture ± oestrogen appeared to increase pERK expression in the IK/SHT-290 samples. Phosphorylation of cPLA2 in both IK and HEC-1B samples resulted in a 200 kDa band as opposed to the expected 100 kDa band that has been reported in HeLa cell lysates. It is possible that the observed 200 kDa protein in IK and HEC-1B cells represents a dimer of pcPLA2 produced in these cell lines. A 100 kDa band was observed in HeLa cell lysate controls probed with the same antibody (data not shown). Examination of the 200 kDa band indicated that pcPLA2 was slightly increased in oestrogen-exposed IK/SHT-290 samples regardless of infection status, with less pcPLA2 production in chlamydiae-infected samples overall (Fig. 7C, pcPLA2). Based on these results, it is possible that during co-culture, oestrogen-induced stromal effectors may stimulate cellular signalling events in IK cells by promoting phosphorylation of ERK and cPLA2 during oestrogen exposure. However, the presence and phosphorylation of ERK and cPLA2 remained consistent in all HEC-1B and HEC-1B/SHT-290 samples, indicating that these pathways may be constitutively expressed in HEC-1B cells.

Release of SLPI from IK and HEC-1B cells was affected by C. trachomatis infection, co-culture with SHT-290s and oestrogen exposure. Overall, significantly more SLPI was released from HEC-1B and HEC-1B/SHT-290 samples when compared with the IK and IK/SHT-290 samples (Fig. 8A and B). In all culture conditions examined, SLPI release was increased with oestrogen exposure in uninfected cells. These results corroborate previous reports on the effect of oestrogen on SLPI production (Fahey et al., 2008). This increase in SLPI production was abolished in C. trachomatis-infected samples, no matter the cell type or co-culture status (Fig. 8A and B), implying that, although oestrogen favours antimicrobial polypeptide release, chlamydial infection has the ability to modulate the epithelial cell response to oestrogen. Both IL-8 and IL-6 concentrations were increased in IK and IK/SHT-290 samples when compared with the cytokine concentration found in HEC-1B and HEC-1B/SHT-290 samples. Only minimal amounts of IL-8 and IL-6 were produced in IK or HEC-1B supernatants compared with supernatants from co-cultures (Fig. 8C–F). In IK/SHT-290 cultures, significantly more IL-8 was released from infected samples compared with uninfected samples, both apically (Fig. 8C) and basally (data not shown). Furthermore, increased IL-8 was released during E2 exposure in infected IK/SHT-290 samples (Fig. 8C, striped bars) versus the absence of E2 (Fig. 8C, solid bars). Conversely, IL-8 production was significantly decreased in both uninfected and C. trachomatis-infected samples apically (Fig. 8D) and basally (data not shown) in HEC-1B/SHT-290. Expression patterns similar to IL-8 release were observed in IK/SHT-290 and HEC-1B/SHT-290, respectively, for IL-6 production, albeit in lower concentrations (Fig. 8E and F). Overall, HEC-1B/SHT-290 cultures produced less cytokines in the presence of oestrogen no matter the infection status. However, C. trachomatis infection enhanced IL-8 and IL-6 production in the absence of oestrogen and, to greater extent, during oestrogen exposure in IK/SHT-290 cultures (Fig. 8C and E). The data presented here indicate that cytokine expression can vary depending on the epithelial cell type, culture condition, infection status and hormone exposure.

Figure 8.

Effect of oestrogen exposure on cytokine production in IK and HEC-1B during co-culture. Replicate polarized IK and HEC-1B cultures were grown in oestrogen-free medium until they reached 75% confluency. Following oestrogen starvation, one set of IK and HEC-1B cultures was moved into wells containing confluent monolayers of SHT-290 cells. The additional IK and HEC-1B cultures and co-cultures (IK/SHT-290 and HEC-1B/SHT-290) were replenished with either oestrogen-free medium (E2−) or medium containing 10−8 M E2. Following 48 h of oestrogen exposure, the epithelial cells were infected with C. trachomatis (Ct) or mock-infected (UnI) and incubated ± E2 for 48 h. Apical and basal supernatants were collected for analysis of SLPI (A and B), IL-8 (C and D) and IL-6 (E and F) concentrations. Significant differences (P ≤ 0.05) in concentrations are denoted as follows: (black circles) 10−8 M E2 versus E2− for each culture condition; (red triangles) UnI versus Ct for each culture condition; or (blue squares) IK or HEC-1B versus IK/SHT-290 or HEC-1B/SHT-290 for the equivalent samples.

Discussion

Numerous studies examining infection of animal models, explanted human and swine endometrial tissues and in vitro tissue culture cells indicate that oestrogen positively influences C. trachomatis infections (Rank et al., 1982; 1993; Moorman et al., 1986; Maslow et al., 1988; Wyrick et al., 1993; Davis et al., 2002; Guseva et al., 2003). Since the question of how oestrogen exerts its effects on chlamydial infection has remained unclear, this study undertook an examination of two common polarized endometrial epithelial cell lines with different mER composition to begin a more detailed evaluation of the specific effects of oestrogen on infection with this major STI pathogen.

Entry of EB into epithelial cells has been postulated to occur primarily by various endocytic mechanisms involving the type III secretion effector, translocated actin-recruiting phosphoprotein (TARP), clathrin-mediated endocytosis and caveolae-mediated entry (Wyrick et al., 1989; Clifton et al., 2004; Dautry-Varsat et al., 2004; Engel, 2004; Hybiske and Stephens, 2007). Interestingly, mERs are located in both clathrin-coated pits and caveolae (Webley et al., 2004; Levin, 2009). Previously, we reported an association with oestrogen-receptor-specific PDI and EB attaching to the surface of HEC-1B cells (Davis et al., 2002). The importance of PDI in chlamydial attachment/entry was further elucidated by Abromaitis and Stephens. Their elegant studies revealed that even though PDI is physically required for EB attachment, it is not the receptor; however, the thio-mediated oxido-reductive enzymatic function of PDI is required for entry (Abromaitis and Stephens, 2009). The current study has demonstrated that functional mERs, either ERα or ERβ or both, are important for chlamydial infection and that the ER complex is important for chlamydial entry into the host epithelial cell, but it cannot be ruled out that other receptors, some reported and others not yet identified, are upregulated by oestrogen exposure and are also involved in chlamydial entry. In any case, there are several mechanisms by which chlamydiae could utilize mERs to stimulate uptake of EB. First, chlamydiae could bind directly to the ER in the ligand-binding region to stimulate uptake of EB. Second, EB could bind a site distal to the ligand-binding domain and, in the presence of oestrogen, hormone binding would trigger uptake of the receptor along with the attached EB. Third, it is likely that EB use another member of the mER complex, an accessory protein such as PDI, to gain access to the host cytosol (Abromaitis and Stephens, 2009). EB also appear to require receptor-activated signalling pathways that are stimulated by oestrogen interaction with mERs to facilitate entry.

In addition to hormone-induced transcription, stimulation of mERs has been shown to activate numerous non-genomic signalling pathways. Stimulation of mERs activates the phosphatidylinositol-3 kinase (PI3K) pathway, which is involved in cellular proliferation (Levin, 2009; 2003). Interaction of the chlamydial type III effector molecule, TARP, with PI3K has also been confirmed (Lane et al., 2008; Mehlitz et al., 2010). Calcium mobilization following oestrogen exposure of tissue cultured cells has also been reported (Hall et al., 2001; Levin, 2009), suggesting a role for intracellular calcium and calcium-activated annexins in chlamydial infectivity (Hybiske and Stephens, 2007).

Oestrogen receptor α and ERβ are comprised of two domains, activating function-1 (AF-1) and AF-2. AF-1 is located at the N-terminus of the protein and is primarily responsible for ligand-dependant transcriptional activation. AF-2 is located at the C-terminus near the ligand-binding domain. While the DNA- and ligand-binding domains of ERα and ERβ are highly conserved, there is significant sequence divergence in both the N-terminus and the C-terminus; therefore, it is thought that these receptors may exhibit some redundancy of function as well as varying functions in vivo. In fact, the AF-1 domain in ERβ has been reported to be less functional with regards to transcriptional activation compared with the AF-1 domain of ERα (Hall et al., 2001; Ramsey et al., 2004). These differences in functional homology between ERα and ERβ may partially account for the observed differences in hormone responsiveness in IK and HEC-1B cell lines, since HEC-1B cells only possess ERβ whereas IK cells have both α and β forms of the receptor (Guseva et al., 2005).

Antibody blocking of ERs and exposure to Tx prior to chlamydial infection caused decreases in chlamydial infectivity. Based upon its interaction with the AF-2 domain, Tx acts as a pure antagonist of ERβ while it can act as a partial agonist of ERα (Levin, 2003). Interestingly, the effect of Tx exposure prior to infection was greater in HEC-1B cells having only ERβ, compared with IK cells that possess both ERα and ERβ. This observation suggests a strong importance for ERβ signalling during EB entry into HEC-1B. Additionally, this observation points to a difference in the domains of ERβ and ERα. Since Tx can partially stimulate ERα, there may have been some signal transduction occurring in the IK cells that led to the difference in infectivity between IK and HEC-1B cells exposed to Tx prior to infection.

The important and somewhat surprising effects of oestrogen on chlamydial infection in the endometrial epithelia were illustrated when Tx was exposed to the host cells after chlamydiae infection. Epithelial cell physiology was impacted in that some morphologically detected disruption of key organelles – endoplasmic reticulum, Golgi complex and mitochondria – occurred; these organelles are documented to be critical for chlamydial nutrient and energy acquisition, i.e. sphingomyelin, cholesterol, glycerophospholipids, lipid droplets, ATP, etc., as recently review by Saka and Valdivia (2010). Chlamydial replication as well as inclusion development and completion of the developmental cycle were varyingly affected depending on the length of Tx exposure. Interestingly, even in some well-developed, more mature inclusions, damage to the inclusion membrane occurred resulting in wrongly oriented basal escape of RB.

Additionally, epithelial ER signalling reduced by Tx exposure post-chlamydiae infection likely affected the MAPK pathway, including phosphorylation of ERK (Hall et al., 2001; Buchholz and Stephens, 2007), modulation of the downstream effector cPLA2 (Su et al., 2004; Vignola et al., 2010), as well as PI3K/AKT signalling and calcium mobilization, all of which have been shown to be essential for chlamydial inclusion development (Majeed et al., 1999; Verbeke et al., 2006; Lane et al., 2008; Mehlitz et al., 2010)

The exciting application of the endometrial stromal cell co-culture system could pave the way for better understanding of the effects of hormones on chlamydial infection in vivo albeit using an in vitro system in which individual parameters can be more easily manipulated. The paracrine effects of oestrogen exposure, perhaps via TGFβ (Cunha et al., 1985; Arnold et al., 2002) and other growth factors, did aid in proliferation and maturation of the epithelial cells and did enhance serovar E infectivity. Increased amounts of IL-8 and IL-6 were also produced by the epithelial cells in the co-culture system, especially in the IK cells, on oestrogen exposure, confirming MAPK activation and ERK phosphorylation (Fukuda et al., 2005; Buchholz and Stephens, 2007; 2008). These data strongly support Stephens' chlamydial cellular pathogenesis paradigm that growth factor-stimulated epithelial cells, acutely or persistently infected with Chlamydia, produce cytokines and chemokines that trigger a foci of inflammatory responses that promote cellular proliferation, tissue remodelling and healing processes; if they persist, scarring can result (Stephens, 2003).

To declare that hormonal regulation of the female reproductive tract is complicated would be a vast understatement and would disgrace the sheer elegance of an organ system that is uniquely designed to protect the body from foreign invasion while allowing for the creation of life. Given the critical nature of the female reproductive hormones on regulation of the menstrual cycle, it comes as no surprise that sexually transmitted pathogens would evolve mechanisms by which to use this environmental scenario to their advantage. For decades, it has been accepted that C. trachomatis infections are influenced by the presence of oestrogen. Based upon collective data originating over many years of research, we conclude that oestrogen and its receptors play dual roles in chlamydial infection: (i) mERs aid in attachment/entry of EB into host cells, and (ii) mER signalling contributes to inclusion development during infection. Additionally, enhancement of chlamydial infection is indirectly but importantly affected by hormonally influenced stromal signals in conjunction with direct oestrogen stimulation of the uterine epithelia.

Experimental procedures

Chlamydia and tissue culture cell lines

Chlamydia trachomatis serovar E (UW5-CX) was cultured in HEC-1B cells grown on Cytodex microcarrier beads (Pharmacia Uppsala, Sweden), harvested by centrifugation and stored in 2SPG [0.02 M phosphate buffer, 0.2 M sucrose, 5 mM glutamine (pH 7.2)] at −80°C (Guseva et al., 2007). The human endometrial carcinoma cell line HEC-1B (ATCC# HTB-113) was maintained in Dulbecco's modified Eagle's medium (DMEM; Gibco) with 10% fetal calf serum (FBS) and 1× Glutamax (Gibco). The endometrial carcinoma cell line IK was obtained from Dr Alison Quayle and S.J. Greene (Louisana State University Health Science Center, New Orleans, LA). IK cells were maintained in phenol-red free DMEM-F12 (Gibco) with 5% FBS and 1× Glutamax (Gibco). The immortalized stromal cell line (SHT-290) was obtained from Dr David G. Kaufman (University of North Carolina School of Medicine, Chapel Hill, NC). SHT-290 cells were maintained in phenol-red free DMEM (Gibco) with 2% Charcoal-stripped FBS (CSFBS) and 1× Glutamax (Gibco).

Transmission electron microscopy

Polarized monolayers of HEC-1B or IK cells, ± infection with C. trachomatis serovar E and ± exposure to oestrogen and/or Tx, were processed and embedded in Epon-araldite (Epon 812, EM Sciences) resin for high contrast or in Lowicryl K4M resin (Polysciences) for immunoelectron microscopy (Giles et al., 2006). Ultrathin sections were cut on a Leica Ultracut UCT microtome. For immunoelectron microscopy, ultrathin sections on gold grids were first blocked with 1% ovalbumin/0.01 M glycine in PBS for 5 min prior to exposure to ERβ primary polyclonal antibody (H-150, Santa Cruz) for 3 h at 37°C and, after washing, subsequently exposed to 15 nm colloidal gold-conjugated secondary affinity antibody (Amersham Biosciences) for 30 min at 37°C. Thin sections were examined in a Philips Tecnai-10 electron microscope (FEI) operated at 80 kV. Control samples included parallel labelling with an irrelevant primary antibody or gold-conjugated secondary antibody alone to determine background cross-reactivity.

Apical lift and 2D-PAGE analysis

Isolation of HEC-1B apical membrane components was performed using the apical lift procedure as described previously (Davis et al., 2002). Briefly, the apical surfaces of polarized HEC-1Bs were labelled at 4°C by biotinylation. Poly-l-lysine-coated glass plates were pressed gently but firmly against the labelled HEC-1B monolayers for 15 s and lifted off, removing labelled apically exposed epithelial cell membrane components. Apical proteins were scraped off the coverslip and concentrated with centrifugation. The biotinylated proteins were resolved with 2D-PAGE analysis and their identity was determined using MALDI-MS performed by the Protein Core Facility, Howard Hughes Medical Institute, Columbia University, New York, NY (Davis et al., 2002).

Flow cytometry

HEC-1B cell pellets (1 × 106 cells) were incubated with 2 µg ml−1 primary polyclonal or monoclonal antibodies against ERα (D-12, Santa Cruz Biotechnology, Santa Cruz, CA) or ERβ (Dako Cytomation, Denmark) for 20 min at room temperature (RT). Control samples were incubated with equivalent amounts of PBS or mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, PA). Following incubation, samples were washed with PBS and fixed with 3.7% formaldehyde for 20 min at RT. HEC-1Bs were washed with PBS and incubated with anti-mouse Alexa Fluor 488 (Molecular Probes) for 30 min at RT, washed with PBS and resuspended in FACS buffer (3% FBS in PBS). The epithelial cells were analysed by flow cytometry using a FACScan (Becton Dickinson and Co., Sunnyvale, CA) with a sample size of 10 000 cells. All flow cytometry data analyses were performed using the CellQuest software system (Becton-Dickinson).

RNA isolation and RT-PCR

Following oestrogen starvation, IK or HEC-1B cells were grown ± oestrogen for 48 h. Total RNA was collected using the RNeasy Kit (Qiagen) as per manufacturer's instructions. RNA quality was examined by optical density (OD) and an Agilent Bioanalyser 2100 (Agilent Technologies). RNA with an OD260/OD280 ratio of > 1.9 and a RIN of > 9.0 was subjected to reverse transcription using the Qiagen Quantitect Reverse Transcription system. Expression of ERα, ERβ, PGR and TGFα mRNA was determined using Quantitect primer assays (Qiagen). For real-time RT-PCR (qRT-PCR), SYBR green-based reactions were set up in a 96-well format containing Platinum qPCR SuperMix-UDG (Invitrogen). Reactions were performed on a Bio-Rad iCycler using the following conditions: incubation for 2 min at 50°C, denaturation for 10 min at 95°C and 40 cycles of 15 s at 95°C, 30 s at 55°C and 30 s at 72°C. Expression of the gene of interest (GoI) was determined by comparing amplification of the GoI in experimental samples to amplification of a serial dilution (100–106 target copies) of a plasmid containing a primer-specific DNA sequence. Amplification of ERα, ERβ, PGR and TGFα was normalized to the housekeeping gene beta-2-microglobulin (β2M). Alternatively, RT-PCR reactions were performed on a DNA Engine (Bio-Rad) using the 5 PRIME Mastermix kit. Amplimers were electrophoresed on a 2% agarose gel along with the amplimer from a positive control plasmid (Pos) containing a primer-specific DNA sequence. Amplimer bands were visualized with ethidium bromide using a G-Box transilluminator (SynGene).

Immunofluorescence staining of ERs

Ishikawa or HEC-1B cells were grown on poly-l-lysine-coated coverslips in 24-well plates to a density of 1 × 106 cells per well. Cell monolayers were fixed with 3.7% formaldehyde in PBS for 20 min at RT. Samples prepared for staining of internal ERs were permeabilized with 0.1% NP-40 for 10 min, incubated with blocking solution (1% BSA, 10% normal donkey serum in PBS) for 1 h at RT and incubated at 4°C overnight with either mouse IgG as a control, anti-ERα monoclonal antibody (D-12, Santa Cruz Biotechnology) or anti-ERβ polyclonal antibody (07-359, Upstate, Lake Placid, NY) at a concentration of 2 µg ml−1 in 0.5% BSA in PBS. Cell monolayers were washed and exposed to anti-mouse Alexa Fluor 488 (Molecular Probes) at 1 µg ml−1 in 0.5% BSA in PBS for 1 h at RT, washed and mounted in Prolong Gold ± DAPI (Invitrogen). Photomicrographs were captured at 320× magnification on a fluorescent Zeiss Axiovert S100 microscope fitted with an AxioCam digital camera.

Infectivity assays

Experimental samples were fixed with methanol. The fixed cells were stained for chlamydial inclusions with Pathfinder anti-MOMP stain (Bio-Rad). The average number of inclusions and host cells was determined for each condition by counting the number of inclusions and epithelial cells present in 10 reticule fields at 400× magnification on a fluorescent Zeiss Axiovert S100 microscope fitted with an AxioCam digital camera. Three replicate samples were analysed for each condition. Values are expressed as the average per cent of cells infected in the field examined.

Antibody blockage of surface-exposed ERs

Ishikawa and HEC-1B cells were cultivated on poly-l-lysine-coated coverslips. The monolayers were washed with PBS and overlaid with 50 µl of the appropriate dilution of antibodies against IgG, PDI, ERα, ERβ, or a cocktail of the anti-ERα and anti-ERβ antibodies. The antibody-exposed samples were incubated for 30 min at 4°C. Following incubation, antibody solutions were removed and the monolayers were overlaid with 50 µl of C. trachomatis inoculum. The appropriate dilution of each individual antibody was added to the inoculum. Samples were incubated at 4°C for 1 h before the inoculum was removed and the cells were replenished with medium. Infected monolayers were incubated for 48 h at 35°C before being harvested for infectivity assays.

Tx exposure

Confluent monolayers of polarized IK or HEC-1B cells were exposed to 10−4 M Tx for 2 h, washed with PBS, infected with C. trachomatis and harvested for infectivity assays at 48 hpi. Alternatively, polarized IK or HEC-1B cells were infected with serovar E EB. At 6, 12 and 24 hpi, 3 × 10−5 M (IK) or 6 × 10−5 M (HEC-1B) Tx was added to the culture medium. All samples were harvested for infectivity assays or TEM at 48 hpi.

C. trachomatis infection of oestrogen-exposed epithelial cells ± stromal cell co-culture

Ishikawa or HEC-1B cells were cultivated on ECM-coated filters (Biocoat) in oestrogen-free medium (phenol-red free DMEM + 2% CSFBS + glutamax) until they were 75% confluent. Following oestrogen starvation, the medium was replenished with the appropriate experimental medium, E2− (oestrogen-free medium) or E2+ (10−8 M E2, 17-β-oestradiol, Sigma Aldrich). For epithelial cell/stromal cell co-culture, immortalized human stromal cells (SHT-290) were plated in six-well plates in oestrogen-free medium and incubated at 37°C until they became confluent. Oestrogen-starved polarized IK or HEC-1B cells on ECM-coated filters were transferred to the wells containing the SHT-290s. Medium on the co-culture samples was replenished withthe appropriate experimental medium as described. The epithelial cells or co-cultures were exposed to oestrogen for 48 h before being harvested for various assays or infected with C. trachomatis.

Following oestrogen exposure, a dilution of serovar E EB in 2SPG storage buffer was overlaid on IK or HEC-1B polarized monolayers for 1 h at 35°C. All infections were performed using an inoculum calculated to infect 80% of the epithelial cells. Uninfected samples were overlaid with an equal volume of 2SPG. Likewise, in co-culture samples, the SHT-290s were overlaid with 2SPG to prevent drying. Following adsorption, the samples were replenished with experimental medium ± E2.

Cell proliferation assays

After 48 h of oestrogen exposure, IK or HEC-1B cultures were collected by trypsinization and fixed in 3.7% formaldehyde. Cell counts were performed on a haemocytometer. The data presented represent an average of six cell number determinations per sample.

Western blots

Ishikawa or HEC-1B cells grown on ECM coated filters were collected by centrifugation. Cell pellets were lysed in SDS sample buffer (62.5 mM Tris-HCl, 2% w/v SDS, 10% glycerol, 50 mM DTT) containing Halt protease cocktail (Thermo Fisher Scientific). Total protein concentration was determined using the RCDC Assay (Bio-Rad). The proteins were electrophoresed on 4–12% Criterion SDS-Tris gels (Bio-Rad) and transferred to PVDF membranes. The blots were blocked with 3% non-fat milk in Tris-buffered saline with 0.01% Tween20 (TBST-MLK) probed with primary antibodies against ERK, phosphor-ERK, cPLA2 or phospho-cPLA2 (Cell Signaling), followed by incubation with goat anti-rabbit HRP-conjugated secondary antibodies at a concentration recommended by the manufacturer. The proteins were detected using Super Signal West Pico chemiluminescent substrate (Thermo Fisher Scientific) and exposed to X-ray film.

ELISA

The concentration of SLPI present in apical supernatants was determined using the Quantikine IL-6 ELISA (R&D Systems) according to the manufacturer's instructions.

Luminex assay

Aliquots of apical or basal supernatants taken from co-culture samples were assayed for the presence of GM-CSF, IL-1β, IL-6, IL-8 and TNF-α using the BioSource Multiplex Bead Immunoassay (Invitrogen) according to the manufacturer's instructions. Samples were UV-irradiated as previously described before the assay was performed so that no infectious EB were present in the samples (Deka et al., 2006). A Luminex 100 instrument (Luminex Corporation, Austin, TX) was used to measure the quantity of each cytokine bound to antibody coupled beads.

Shotgun proteomic analysis

Two-dimensional liquid chromatography tandem mass spectrometry-based proteomics was used to analyse HEC-1B cell lysates and basal supernatants from SHT-290 co-cultures grown with or without oestrogen exposure as described above. Approximately 3 mg of total protein from cell lysates, and approximately 500 µg of protein from supernatants were collected for proteomic analysis. SwellGel Blue Albumin Removal Kit (Pierce) was used to deplete the dominant protein albumin from the culture supernatant, present due to FBS supplementation of the media. Samples were lysed/denatured with 6 M guanidine and reduced with 10 mM DTT at 60°C for 1 h, diluted sixfold with 50 mM Tris with 10 mM CaCl2 (pH 7.6), and digested with 1:100 of sequencing grade trypsin (Promega). Peptides were desalted off-line by C18 solid phase extraction (Waters) prior to a 24 h (cell) or 6 h (supernatant) chromatographic separation, utilizing increasing (0–500 mM) pulses of ammonium acetate followed by 2 h aqueous to organic solvent gradients online with an LTQ-Orbitrap (Thermo Fisher). MS/MS spectra were matched to a human protein database by SEQUEST and subsequently filtered using DTASelect (Yates et al., 1995; Tabb et al., 2002).

Statistical analyses

Statistical analyses were performed using Microsoft Excel. Comparison of means was performed using a two-sample t-test for independent samples. P-values of ≤ 0.05 were considered significant.

Acknowledgements

The authors would like to thank: Dr Brante Sampey for his advice on the subtle nuances of stromal cell growth; Tiffany Ford for her valuable contributions to the cytokine analyses; the Organic and Biological Mass Spectrometry Group at Oak Ridge National Laboratory for access to mass spectrometry instrumentation; the excellent work of the Molecular Biology Core Facility, Department of Pharmacology and the Electron Microscopy Core Facility, Department of Pathology at James H. Quillen College of Medicine and Dr Alison Quayle for many helpful conversations and manuscript suggestions.

Ancillary