Dr C. R. Wira, Department of Physiology, Dartmouth Medical School, Borwell Building, 1 Medical Center Drive, Lebanon, NH, 03756-0001 USA. Email: Charles.R.Wira@Dartmouth.Edu Senior author: Charles R. Wira
Oestradiol-17β (Oe2) stimulates uterine epithelial cell proliferation and is critical for normal uterine differentiation and secretory function. Oe2 can act directly on the epithelium via the epithelial oestrogen receptor (OR) or indirectly via the OR-positive underlying stroma. A primary role for epithelial–stromal interactions has been established for mediating steroid hormone action in the uterus. This study was undertaken to determine the mode of Oe2 action in regulating epithelial cell cytokine release in the uterus. Mouse uterine epithelial and stromal cells were isolated and cultured separately. Transepithelial resistance (TER) was monitored with an EVOM voltohmmeter to determine monolayer polarity and integrity. Epithelial cells grown alone or in coculture with stromal cells were treated with Oe2. Supernatants collected were assayed for transforming growth factor-β (TGF-β) and tumour necrosis factor-α (TNF-α) by bioassay and enzyme-linked immunosorbent assay, respectively. While Oe2 treatment of epithelial cells led to a significant decrease in TER, the amount of TNF-α released was not altered. However, when epithelial cells were cocultured with stromal cells and treated with Oe2, apical TNF-α release was significantly decreased, compared to cells not treated with hormone. As determined by oestrogen receptor antagonist studies, Oe2 primed epithelial cells for the action of the stromal paracrine factor(s). In contrast, TGF-β release by epithelial cells was not affected by Oe2 when grown alone or in the presence of stromal cells. These studies indicate that Oe2 has both direct and indirect effects on the uterine epithelium. While epithelial monolayer integrity is directly influenced by Oe2, TNF-α release in response to Oe2 is dependent on the presence of stromal cells, indicating that paracrine communication is necessary for steroid regulation of some but not all cytokines.
Epithelial cells were at one time considered to act as a physical barrier in the female reproductive tract to separate the host from potentially harmful bacterial and viral pathogens. More recently, however, epithelial cells, under the influence of oestradiol, have been shown to carry out a number of essential functions as part of the innate and adaptive immune systems.1–3 For example, analysis of uterine epithelial cell function has revealed that epithelial cell transport of polymeric immunoglobulin A (IgA) as well as class II-mediated antigen presentation increases in response to oestradiol.4–7 In other studies, oestradiol has been shown to have a stimulatory effect on secretory leucocyte protease inhibitor, a known bactericidal agent8, but an inhibitory effect on the secretion of interleukin-6 by uterine epithelial cells.9
Numerous studies of oestrogen action over the past 30 years have shown that oestradiol acts directly on target cells such as epithelial cells by binding to specific, high affinity receptors that act as coactivators to alter gene transcription.10 While many actions are mediated through oestradiol receptors (OR) in a given target cell, studies by Cunha et al. have led to the conclusion that some of oestradiol's actions on epithelial cells are mediated through underlying stromal cells.11 Using ORα-knockout mice to study the relative roles of epithelial and stromal ORα in Oe2-induced mitogenesis, Cooke et al. demonstrated that epithelial cell proliferation in the uterus is a paracrine event mediated by the OR-positive stromal cells.12
Cytokines are essential components of a complex intercellular communication network that exists in the female reproductive tract.13–15 Recognition of the cyclic-dependent pattern of hormone release during the estrous cycle led to the conclusion that several cytokines are regulated by Oe2 and progesterone.9,15,16 For example, epithelial cells of the uterus produce a number of biologically important cytokines, including tumour necrosis factor-α (TNF-α) and transforming growth factor-β (TGF-β).17 TNF-α and TGF-β message and protein have been shown to be under Oe2 control in the uterus.18,19 Using epithelial cells from the uteri of intact animals led to the conclusion that polarized epithelial cells release both TNF-α and TGF-β in culture.20 While both cytokines were released into the apical and basolateral compartments, TNF-α was preferentially released apically, in contrast to TGF-β which was preferentially released basolaterally. As a part of these studies, we found that coculture of epithelial cells with stromal cells increased epithelial cell electrical integrity measured as transepithelial resistance (TER), decreased both the apical and basolateral release of TNF-α and had no effect on either apical or basolateral release of biologically active TGF-β. When added to the basolateral compartment, conditioned medium from stromal cells affected TNF-α and TER, but not TGF-β, in a similar manner to coculture with stromal cells.20 These studies indicated that uterine stromal cells act via a soluble factor(s) to regulate uterine epithelial cell integrity and secretory function.
Because Oe2 action can act both directly and indirectly on the uterine epithelium, the objective of this study was to determine whether Oe2 has a direct effect on epithelial cell cytokine release and if an Oe2 effect on the epithelium requires the presence of underlying stromal cells. The goals of this study were to: (1) determine whether Oe2 has a direct effect on TNF-α and TGF-β release; (2) evaluate whether the presence of stromal cells influences the release of TNF-α and TGF-β by epithelial cells in response to Oe2; (3) determine whether the effect of Oe2 on TNF-α is receptor mediated; and (4) identify the cell compartment (epithelial versus stromal) through which Oe2 acts to affect TNF-α release by epithelial cells.
Materials and methods
Sexually mature, BALB/c, female mice (9–11 weeks) were obtained from the National Cancer Institute colony at Charles River Laboratories (Kingston, NY). Animals were housed in a constant-temperature room with 12 hr light/dark intervals and allowed food and water ad libitum. For each experiment following sacrifice by CO2, uteri were pooled from 8 to 12 animals at all stages of the oestrous cycle. All procedures involving animals were conducted after approval of the Dartmouth College Institutional Animal Care and Use Committee.
Epithelial cell preparation
To prepare epithelial cells, uteri were removed, slit lengthwise, pooled and incubated with 0·25% trypsin (Sigma, St. Louis, MO)/2·5% pancreatin (Gibco-BRL/Invitrogen, Grand Island, NY) for 60 min at 4° and 60 min at 22°. Following transfer to ice-cold (3°) Hanks' balanced salt solution (HBSS; Gibco-BRL/Invitrogen), digested uteri were vortexed to release sheets of epithelial cells. Uterine tissues were rinsed and vortexed an additional three times and resulting cell suspensions pooled. Epithelial sheets were recovered by passing the cell suspension through a 20 µm nylon mesh (Small Parts Inc, Miami Lakes, FL), collected, and centrifuged (500 g). Epithelial sheets were resuspended in complete medium consisting of Dulbecco's modified Eagle's minimal essential medium (DMEM without phenol red)/Ham's F-12 nutrient mixed 1 : 1 (Gibco/Invitrogen) + 10% charcoal stripped fetal bovine serum (FBS; Hyclone, Logan, UT) supplemented with 20 mm HEPES, 100 µg/ml streptomycin, 100 U/ml penicillin, 2 mm l-glutamine (all from Gibco/Invitrogen). Cell sheets were seeded in the apical compartment of 0·4 µm pore size Falcon (6·4 mm diameter) or 0·4 µm pore size Nunc (10 mm diameter) cell culture inserts (Fisher Scientific, Pittsburgh, PA) coated with Matrigel (without phenol red; Collaborative Biomedical Products, Bedford, MA). Cells were seeded in a volume of 300 µl at a ratio of approximately three to four cell inserts per uterus and incubated at 37° with 5% CO2. Inserts were placed in 24-well tissue culture plates (Falcon/Nunc; Fisher Scientific) containing 850 µl or 500 µl of medium in the basolateral compartment and incubated at 37° with 5% CO2. Throughout each experiment medium was collected from the basolateral and apical chambers and replaced at 48 hr intervals, centrifuged (10 000 g), and stored at −80° until assayed. A minimum of four to six inserts per treatment group was used in each experiment.
Stromal cell preparation
To isolate stromal cells, pooled uteri, following the removal of epithelial cells, were incubated for 30 min at 37° in 0·05% trypsin + 0·02% ethylenediaminetetraacetic acid (EDTA) + 400 µ/ml DNase (Gibco-BRL/Invitrogen). Tissues were dispersed by gentle rubbing in medium on a 40 micron nylon mesh (Small Parts Inc) and the resulting cell suspension was centrifuged (500 g) for 10 min. Stromal cells were resuspended in complete DMEM/Ham's F12 + 10% charcoal stripped FBS and seeded in 1 ml at 5 × 105/ml per well in 24-well plates. Medium was collected from each well, replaced at 48-hr intervals, centrifuged (10 000 g), and stored at −80° until assayed. Purity of the stromal cell preparation was established by immunohistochemistry as previously described.20 Following 4 days in culture, with medium changes at 48 hr intervals to remove non-adherent cells, stromal cells were stained for CD45 (Pharmingen, San Diego, CA). Whereas fresh stromal preparations contain 5–20% leucocytes, stromal cultures at 4 days were devoid of CD45 positive cells. Based on these findings and morphological analysis, we conclude that the stromal cells in culture were 99% fibroblast at the time of epithelial coculture.
Co-culture of epithelial and stromal cells
In experiments involving the coculture of epithelial and stromal cells, epithelial and stromal cells were grown separately to confluence on cells inserts and/or in 24-well plates as described above. Once epithelial cells achieved high TER (>1000 Ω/well), inserts of polarized epithelial cells were transferred to 24-well plates containing stromal cells. Epithelial and stromal cells were not in direct contact in any coculture experiments. Throughout each experiment, medium was collected from the apical compartment and replaced at 48 hr intervals, centrifuged (10 000 g), and stored at −80° until assayed.
Co-culture of epithelial cells with conditioned stromal medium (CSM)
To prepare CSM, stromal cells were isolated and cultured as previously described.20 Briefly, stromal cells were grown in complete medium ± Oe2 (CSM/CSM Oe2) and replaced at 48 hr intervals. CSM used in these experiments was the medium collected from cells between days 2 and 4 of culture. Medium was centrifuged (10 000 g), stored at −80°, diluted 1 : 1 with fresh DMEM/Ham's F12 + 10% stripped FBS and placed in the basolateral compartment of confluent uterine epithelial cells.
TER was monitored as an indication of tight junction formation and epithelial monolayer integrity using a set of Ag:AgCl electrodes and an EVOM™ epithelial voltohmmeter (World Precision Instruments Inc., New Haven, CT). Electrical resistance measurements were taken daily after alcohol sterilization of the electrode probe.
Aliquots (50 µl) of supernatants collected from epithelial cells grown alone, or in coculture with stromal cells/CSM ± Oe2, were assayed by TGF-β bioassay, as previously described.21 Briefly, biologically active (mature) TGF-β was measured using mink lung epithelial cells transfected with a plasminogen activator inhibitor-1 promoter–luciferase construct. Cells were kindly provided by Dr James Gorham (Dartmouth Medical School). This quantitative bioassay is based on the ability of TGF-β to induce plasminogen activator inhibitor-1, resulting in a dose-dependent increase in luciferase activity. Luciferase activity induction in mink lung epithelial cells is specific and sensitive to picogram quantities of TGF-β.22 Cells were thawed, washed, plated at 1 × 105/100 µl in opaque, flat-bottom 96-well plates (USA Scientific, Inc., Ocala, FL), centrifuged (800 g) for 15 s, and allowed to adhere for 3 hr at 37°. Following incubation, cells were centrifuged, and the medium was replaced with 50 µl fresh medium and either 50 µl of serially diluted standards (recombinant human TGF-β1; R & D Systems, Minneapolis, MN) or cell supernatants containing TGF-β. Cells were incubated overnight (17–20 hr), washed two times in HBSS (100 µl), and lysed with 50 µl lysis reagent (Promega Corp., Madison, WI) for 15 min at room temperature. Luciferase activity of lysates was measured by adding luciferase reagent (100 µl; Promega Corp.) to each well and recording illumination for 10 s following a 2 s delay in a Microplate Luminometer model LB96V (EG & G Berthold, Gaithersburg, MD). Supernatants (100 µl) collected from epithelial cells grown alone or in coculture with stromal cells/CSM ± Oe2, were also assayed by TNF-α enzyme-linked immunosorbent assay (ELISA; R & D Systems, Minneapolis, MN). ELISAs were performed according to the commercial kit protocol.
Hormone and antagonist preparation and treatment
Oe2 (Calbiochem, La Jolla, CA) and ICI 182,780, a receptor antagonist of oestradiol (Tocris, Ellisville, MO) were each dissolved in 100% ethanol, evaporated to dryness, and resuspended in DMEM/F12 + 10% stripped FBS to the appropriate concentration. To control for the alcohol present in the steroid and antagonist preparation, an equivalent amount of ethanol was evaporated in flasks used to prepare the control medium. In experiments involving the treatment of polarized epithelial cells with Oe2, epithelial cells were grown to confluence on cells inserts. Once epithelial cells achieved high TER (>2000 Ω/well), medium was removed from the apical and basolateral compartments and replaced with fresh medium alone or medium containing Oe2 or Oe2 and ICI 182,780 at the appropriate concentration. In experiments involving ICI 182,780, the antagonist was added in combination at the same time as Oe2. In all experiments, Oe2 and ICI 182,780 remained in culture throughout the experiment.
The data were calculated as the mean ± standard error of the mean. INSTAT for McIntosh (GraphPad Software, San Diego, CA) was used to perform a one-way repeated measures analysis of variance (anova). systat 9 for Windows (SPSS Science, Chicago, IL) was used to perform a two-way anova, or a two-way repeated measures anova. When an anova indicated that significant differences existed among means, preplanned paired comparisons were made using the Bonferroni method to adjust P-values. A P-value of less than 0·05 was taken as indicative of statistical significance.
Oestradiol effect on epithelial cell TER
To study the direct effect of Oe2 on polarized epithelial cell TER, isolated mouse uterine epithelial cells were grown to high TER on cell inserts (four to six inserts/group) and treated with medium alone or in the presence of Oe2 (10−7m). As seen in Fig. 1, epithelial cells, when exposed to Oe2 for 48 hr, significantly lowered TER values. In other studies, to more fully define the onset of oestrogen inhibition, a time course study (4–48 hr) was carried out in which epithelial cells were incubated with Oe2 (10−7m). Under these conditions, Oe2 inhibition of TER was observed initially at 20 hr and persisted for 48 hr of hormone exposure relative to that seen with epithelial cells in medium alone (data not shown).
Lack of effect of oestradiol on epithelial cell cytokine release
To study the direct effect of Oe2 on the release of cytokines by polarized epithelial cells, isolated mouse uterine epithelial cells were grown to confluence on cell inserts (four to six inserts/group) in medium and treated with hormone prior to cytokine analysis of supernatants from the apical and basolateral chambers. Following high TER readings on day 6 of culture, epithelial cells were exposed to either fresh medium or medium containing oestradiol (10−7m) for 48 hr. As shown in Fig. 2(a), Oe2 had no effect on the amount and directional release (apical versus basolateral) of biologically active TGF-β relative to that seen in controls. Similarly, levels of TNF-α released by epithelial cells grown in control medium and medium containing Oe2 were not different (Fig. 2b). The preferential release patterns of both cytokines (Fig. 2) were maintained following Oe2 treatment, despite our finding that Oe2 decreased TER (data not shown).
Oestradiol treatment in vitro in the presence of stromal cells
We have previously shown that stromal cells influence epithelial cell function, as measured by increases in TER and decreases in TNF-α release by epithelial cells in coculture.20 To examine whether stromal cells mediate the effects of Oe2 on epithelial cell cytokine release, epithelial cells were grown alone or in the presence of stromal cells along with Oe2 (10−8m) in both the apical and basolateral medium. Following 48 hr of treatment, medium was collected from the apical compartment and analysed for TGF-β and TNF-α. As shown in Fig. 3(a), release of TGF-β by epithelial cells was not affected by Oe2, coculture with stromal cells, or incubation with Oe2 in the presence stromal cells. In contrast, TNF-α release was inhibited beyond that seen with stromal cells alone when epithelial cells were incubated with stromal cells and Oe2 (10−8m; Fig. 3b). With stromal cells present, TNFα was inhibited by 25–30%. When Oe2 was present along with stromal cells, inhibition of TNF-α release was approximately 55–60% of control values (epithelial cells alone and epithelial cells incubated with Oe2; Fig. 3b).
Effect of ICI 182,780 on stromal-cell mediated decrease by oestradiol of epithelial TNF-α
To determine whether the effect of Oe2 on epithelial TNF-α release is mediated through oestrogen receptors in stromal cells, polarized epithelial cells were incubated with Oe2 (10−8m) and/or ICI 182,780 (10−6m), a receptor antagonist of oestradiol, placed in both the apical and basolateral compartments. As seen in Fig. 4(a), release of TNF-α was not affected when epithelial cells were incubated with Oe2, ICI 182,780, or Oe2 + ICI 182,780. In contrast, as shown in Fig. 4(b), when added to the apical and basolateral culture media along with Oe2 in the presence of stromal cells, ICI 182,780 blocked the inhibitory effect of Oe2, but not the inhibitory effect of stromal cells alone, on TNF-α release by uterine epithelial cells.
Oestradiol acts to prime epithelial cells for stromal cell action
One concern not addressed by the addition of ICI 182,780 to stromal cells (Fig. 4b) was whether the effect of Oe2 on TNF-α release might be mediated through epithelial cell OR rather than through stromal cells. To eliminate the influence of Oe2 on stromal cells, we used CSM that previously had been shown to decrease TNF-αin vitro.20 In this study, CSM was prepared by incubation of stromal cells in the absence (CSM) or presence of Oe2 (CSM Oe2; 10−7m) for 48 hr. Polarized epithelial cells were incubated either alone (Ec Con) or with Oe2 (Ec Oe2; 10−7m) for 48 hr prior to the addition at time 0 h of CSM or CMS Oe2 to the basolateral chamber. In all cases, following 48 hr of coculture, apical supernatants were collected and analysed for TNF-α. As shown in Fig. 5, Oe2 added to epithelial cells alone has no effect on TNF-α release. Moreover, when epithelial cells were incubated in the presence of CSM, TNF-α release by epithelial cells decreased significantly by approximately 20% relative to that seen with control cells. Unexpectedly, we found that incubation of epithelial cells with Oe2 (Ec Oe2) for 48 hr prior to the addition of CSM had an inhibitory effect on TNF-α release beyond that seen with CSM in the absence of Oe2. Under these conditions, pretreatment with e2 followed by incubation with CSM inhibited TNF-α release by approximately 40% relative to control cells (CON). When CSM from Oe2-treated stromal cells (CSM Oe2) was compared to CSM from stromal cells incubated in the absence of Oe2, no additional inhibition was observed. This finding suggested that Oe2 has no effect on stromal release of the soluble TNF-α mediator, but rather in some way exerted its effect via epithelial cells.
To determine whether the effect of Oe2 was on epithelial cells, epithelial cells were incubated with CSM Oe2 in the presence of ICI 182,780. Our approach in this experiment was to use ICI 182,780 to block epithelial OR and thereby eliminate the possible effect of Oe2 in CSM Oe2 on epithelial cells. As seen in Fig. 5, the presence of ICI 182,780 added along with Oe2 to epithelial cells reversed the inhibitory effect of CSM Oe2 on TNF-α release. Based on these findings, we conclude that the inhibitory effect of Oe2 on CSM-mediated inhibition of TNF-α release by epithelial cells is via the epithelial cells, possibly by affecting CSM mediator-receptor expression.
The results presented demonstrate that Oe2 has no direct effect on the amount or directional release of TGF-β and TNF-α by epithelial cells. While we have previously shown that coculture of epithelial cells with stromal cells decreases epithelial TNF-α release, the findings in this paper indicate that Oe2 treatment in vitro, in the presence of stromal cells, leads to a significant further decrease in epithelial cell TNF-α secretion beyond that seen with stromal cells alone. In contrast to TNF-α, epithelial cell release of biologically active TGF-β is not affected by Oe2 either directly or indirectly via stromal cells. In the presence of stromal cells, ICI 182,780 inhibits the ability of Oe2 to decrease TNF-α release, indicating that the Oe2 inhibition of TNF-α secretion is an oestrogen-receptor mediated event. Finally, these studies demonstrate that Oe2 does not affect stromal release of the soluble TNF-α mediator, but rather exerts its effect via OR in epithelial cells to prime these cells for the action of the stromal paracrine factor.
In the mouse endometrium, both TNF-α and TGF-β mRNA and protein have been shown to be under Oe2 control.18,19 Whereas earlier reports of oestrogen treatment of mice indicated a biphasic effect on TNF-α expression, the recent findings of Hong et al. using microarray analysis indicate that oestradiol given in vivo to ovariectomized mice down regulates the expression of TNF-α mRNA in the uterus within 6–12 hr of hormone treatment.23 Our results extend these in vivo findings by showing that inhibition of TNF-α release by oestradiol under in vitro conditions occurs only in the presence of stromal cells. Our study suggests that underlying stromal cells are needed to mediate the effects of oestrogen in vitro. While there are a small number of studies reporting direct in vitro Oe2 effects, including a decrease in interleukin-6 release by polarized mouse uterine epithelial cells9 and an increase in hyaluronate secretion by rabbit uterine epithelial cells,24 it is now recognized that epithelial responsiveness to Oe2 is complicated and often requires communication with the underlying stroma.25–27
Communication between epithelial cells and underlying stromal cells during the reproductive cycle is essential for both successful reproduction and immune protection. Previously, we demonstrated that antigen presentation by antigen-presenting cells in the uterine stroma is down regulated in response to oestradiol.7 More recently, we found that oestradiol inhibition of stromal cell antigen presentation is mediated through the stimulatory effect of oestradiol on TGF-β production by epithelial cells.28 Our findings in the present study that underlying stromal cell in the uterus affect epithelial cell integrity and TNF-α production supports our working hypothesis that stroma and epithelium work as an integrated unit, each producing factors which regulate the growth, differentiation, and immune function of each other.
Epithelial and stromal cells of the adult mouse uterus express OR, indicating that Oe2 can act directly on each cell compartment.12 Our finding that Oe2 inhibition of TNF-α release by epithelial cells in the presence of stromal cells is OR mediated, leaves open the question as to which cell Oe2 acts through. Studies in the uterus using OR-knockout mice have shown that Oe2-induced epithelial cell proliferation, as well as down regulation of epithelial progesterone receptor (PR) is mediated by OR in underlying stromal cells.12,29 However, Oe2-induced epithelial cell secretory functions in the uterus require both epithelial OR and stromal cell OR.30 Our results indicate that Oe2 acts through uterine epithelial cell OR to prime these cells to be more responsive to soluble stromal factor(s). When epithelial cells were treated with ICI 182,780 to block OR, CSM Oe2 had no effect beyond that seen with CSM alone. Our results suggest that Oe2-induced effects on TNF-α release are mediated through complex interactions in which Oe2 primes epithelial cells to be more responsive to the inhibitory effects of a soluble factor(s) secreted by uterine stromal cells. These findings, however, do not exclude the possibility that Oe2 has separate effects on stromal cells to affect the release of soluble mediator produced. Studies are presently underway to explore this possibility.
What remains to be determined is how Oe2 influences uterine epithelial cells to increase sensitivity to stromal paracrine factors. It is possible that Oe2 acts by augmenting epithelial expression of receptors for the stromal mediator(s), to allow epithelial cells to respond more effectively to the specific factor(s). Numerous growth factor receptors, including epidermal growth factor (EGF) and insulin-like growth factor-1 (IGF-1), which appear to be involved in endometrial physiology, have been identified in both the human and rodent uterus.31,32 Oestradiol up-regulates uterine epithelial cell receptor expression for both EGF32,33 and IGF-I31,34 and been shown to modulate signals of both growth factors through changes in their respective receptor expression.27,32,35 While others have shown that EGF and IGF are associated with increased epithelial proliferation, the EFGR-ligand system also plays a role in epithelial protection and repair, as well as dampening inflammation in both the respiratory and gastrointestinal tracts.36–38 Our finding of decreased uterine epithelial cell TNF-α release in response to stromal cells and Oe2 may be the result of activating such a system.
In contrast to our results involving Oe2 inhibition of TNF-α, TGF-β release by epithelial cells was not affected by Oe2 either directly or indirectly via stromal cells. Previous work in our laboratory has shown that Oe2 given to ovariectomized rats 24 hr prior to killing increases TGF-β production by isolated uterine epithelial cells in culture. One explanation for the lack of an effect might be that our previous studies used rat uterine epithelial cells rather than those from mouse, which might possibly respond differently to Oe2. Alternatively, recognizing that the majority of TGF-β is biologically inactive39 Oe2 may affect activation as well as synthesis. Because our experiments measured only biologically active TGF-β, further studies need to be carried out to look at total TGF-β, as well as the preferential isoform produced in response to Oe2. Takahashi19 demonstrated by immunohistochemistry, in situ hybridization, and Northern RNA analysis that diethylstilboestrol (DES), a synthetic oestrogen, given to mice increases TGF-β mRNA and protein of the three mammalian isoforms TGF-β1, 2, and 3 in the uterus. Following administration of DES, TGF-β3 mRNA increased within 30 min followed by TGF-β1 and TGF-β2 mRNA at 3 hr. In situ hybridization showed that the most pronounced stimulation of TGF-β1, 2 and 3 occurred in the uterine epithelium and immunohistochemistry demonstrated that oestrogen stimulated a prolonged elevation of the proteins for all isoforms. This apparent paradox could be explained by the fact that the assay used in their studies measured total (TGF-β1, 2, and 3) whereas our studies measured biologically active TGF-β. What remains to be determined is whether Oe2 increases biologically active or latent TGF-β. Studies involving isoform specific changes, as well as latent TGF-β production by uterine epithelial cells in response to in vitro Oe2 treatment are currently underway in the laboratory.
The release of TNF-α and TGF-β by uterine epithelial cells at the apical surface seen in our studies raises important questions about the physiological role of these cytokines in the lumen of the female reproductive tract. Others have shown that cytokines and chemokines are present in secretions of the female reproductive tract.40,41 Our finding that TNF-α is released apically suggests that TNF-α might act in a paracrine manner to regulate tissue degradation and reorganization of cells in the uterus and vagina during the reproductive cycle. Alternatively, along with other chemokines such as macrophage inflammatory protein-3α, which is preferentially released apically, chemotactic and bactericidal42,43, TNF-α and/or TGF-β may either synchronize epithelial cells along the reproductive tract for implantation or maintain a level of immune protection in the uterus, cervix and vagina against potential pathogens.
In conclusion, our findings demonstrate that the effects of Oe2 on uterine epithelial cell function are both direct and indirect. While monolayer integrity is directly influenced by Oe2, the effect of Oe2 on epithelial TNF-α release requires epithelial cell priming in order to respond to the paracrine signals from the stroma. These studies suggest that oestradiol regulation of cytokine secretion by epithelial cells in the uterus is most likely mediated through the complex interactions of stromal and epithelial cells that utilize growth factors as immune modulators rather than solely through the direct effect of oestradiol on its target cell. These findings suggest that epithelial–stromal communication, in addition to endocrine balance, is an important determinant in the response of the immune system to potential pathogens.
This work was supported by research grant AI-13541 from NIH.