Effect of Escherichia coli and Lactobacillus rhamnosus on Macrophage Inflammatory Protein 3{alpha}, Tumor Necrosis Factor Alpha, and Transforming Growth Factor {beta} Release by Polarized Rat Uterine Epithelial Cells in Culture

Entry of bacteria from the vagina into the uterus raises thequestion of uterine epithelial cell (UEC) signaling in responseto the presence of bacteria. Our model system helps to definemicrobially elicited UEC basolateral cytokine release, importantin regulating underlying stromal immune cell protection. UECsfrom adult rats were grown in cell culture inserts to establisha confluent polarized monolayer as was determined by transepithelialresistance (TER). Polarized epithelial cell cultures were treatedapically with live or heat-killed Escherichia coli or Lactobacillusrhamnosus prior to collection of basolateral media after 24h of incubation. Coculture of polarized UECs with live E. colihad no effect on epithelial cell TER. In response to exposureto live E. coli, epithelial cell basolateral release of macrophageinflammatory protein 3 ${alpha}$ (MIP3 ${alpha}$ ) and tumor necrosis factor alpha(TNF- ${alpha}$ ) increased at a time when basolateral release of biologicallyactive transforming growth factor ß (TGF-ß)decreased. Incubation of UECs with heat-killed E. coli resultedin an increased basolateral release of MIP3 ${alpha}$ and TNF- ${alpha}$ , withoutaffecting TER or TGF-ß. In contrast to E. coli, liveor heat-killed L. rhamnosus had no effect on TER or cytokinerelease. These studies indicate that polarized rat UECs respondto gram-negative E. coli by releasing the cytokines MIP3 ${alpha}$ andTNF- ${alpha}$ , signals important to both the innate and adaptive immunesystems. These findings suggest that UEC responses to bacteriaare selective and important in initiating and regulating immuneprotection in the female reproductive tract.

INTRODUCTION

Top

Introduction

The immune system in the female reproductive tract has evolvedto meet the unique requirements of procreation, protecting motherand fetus against infection, while at the same time ensuringtolerance of the presence of commensal microbes, allogeneicsperm, and the fetal-placental unit(s) (54). Uterine epithelialcells provide a mechanical barrier between the lumen of theuterus and internal body systems (43). Central to immune protectionin the female reproductive tract, uterine epithelial cell functionsare complex and include antigen presentation (12, 38), immunoglobulinA (IgA) transport (44, 59), the production of antimicrobialagents (13, 55), and the release of cytokines into the lumenand underlying tissues (9, 23, 45, 46, 60).

In women, a healthy vaginal ecosystem is populated by a diversemicrobiota dominated by lactobacilli (4), which periodicallyenter the upper reproductive tract. Movement of sperm alongwith bacteria occurs as a consequence of sexual intercourseas well as peristaltic contractions, which contribute to themovement of vaginal contents through the cervix into the uterusand subsequently to the Fallopian tubes (27, 37; A. K. Parsons,R. A. Cone, and T. R. Moench, presented at Microbicides 2002,Antwerp, Belgium, 12 to 15 May 2002). In recent studies, a radio-opaquedye was placed in the vaginae of women during the menstrualcycle, while on oral contraception, and following menopause.The dyes were detected in both the uterus and the Fallopiantubes within 3 h (Parsons et al., Microbicides 2002). Thesestudies indicate that irrespective of endocrine balance, theupper reproductive tract is routinely exposed to vaginal microfloraas well as to potential pathogens.

In the female genitourinary tract, Escherichia coli, a gram-negativespecies, is a primary source of urogenital infections (10, 11,48), including pyometra (18, 34), uterine abscess (14), pretermpremature rupture of membranes during pregnancy (3), and urinarytract infections (17). Lactobacillus rhamnosus, formerly designatedLactobacillus casei subsp. rhamnosus, a gram-positive, nonmotile,microaerophilic rod, is a commensal which has been isolatedfrom the human gastrointestinal tract (16, 35) and the lowerfemale reproductive tract (33, 42).

Important in the initiation of an immune response, release ofthe chemokine macrophage inflammatory protein 3 ${alpha}$ (MIP3 ${alpha}$ ) has beendemonstrated both in human gut epithelial cells (19, 20, 52)and in the human uterine epithelial cell line HHUA (49). Asthe ligand for the CCR6 receptor, MIP3 ${alpha}$ is chemotactic for immatureCD35⁺ bone marrow cell-derived dendritic cells as well as Bcells and memory T cells (29, 30). Tumor necrosis factor alpha(TNF- ${alpha}$ ), which is produced by uterine stromal and epithelialcells (50), plays a role in the activation of macrophages, granulocytes,and cytotoxic T cells as well as in the maturation of dendriticcells (7). Helping to maintain a balanced response in the immunesystem, transforming growth factor ß (TGF-ß)is generally inhibitory to inflammatory immune responses (28,31). Within the context of the uterus, TGF-ß is associatedwith successful pregnancy (8). For example, changes in normalTNF- ${alpha}$ /TGF-ß ratios are associated with fetal loss (2).In other studies, work in our laboratory has recently shownthat TGF-ß, which is released predominately to thebasolateral compartment by uterine epithelial cells, mediatesthe inhibition of antigen presentation by uterine epithelialcells (58).

The overall goal of this work was to define the pattern of cytokinerelease by polarized rat uterine epithelial cells in responseto the presence of the selected gram-negative and gram-positivebacteria. Specifically, our objectives were (i) to determinewhether MIP3 ${alpha}$ is released by primary uterine epithelial cells,(ii) to ascertain if transepithelial resistance (TER), as ameasure of cell integrity, is affected by the presence of bacteriain this system, and (iii) to establish if differential releaseof MIP3 ${alpha}$ , TNF- ${alpha}$ , and TGF-ß by rat uterine epithelialcells occurs in response to E. coli and L. rhamnosus.

MATERIALS AND METHODS

Top

Materials and Methods

General procedures. Specific-pathogen-free female Lewis rats weighing 125 to 175g (Charles River Breeding Laboratories, Kingston, N.Y.) weremaintained with an alternating 12-h dark-light cycle and freeaccess to rat chow and water. Following sacrifice by decapitation,uteri were recovered for isolation of uterine epithelial cells.All procedures involving animals were conducted following approvalby the Dartmouth College Institutional Animal Care and Use Committee.

Preparation of epithelial cell cultures. Tissues from adult rats (6 to 14 animals/experiment) at randomstages of the reproductive cycle were pooled for each experiment.Uteri from intact rats were removed, rinsed in sterile ice-coldHanks' balanced salt solution (HBSS) (Gibco, Grand Island, N.Y.),weighed, and then transferred to a pancreatin (Gibco)-trypsin(Sigma, St. Louis, Mo.)-DNase (Worthington, Lakewood, N.J.)digest (400 U of DNase/ml of pancreatin, 46,500 U of trypsin/mlof pancreatin, 19.5 ml of pancreatin/g of uterine tissue) aspreviously described (23). Under sterile conditions, uterinetissues in the enzyme digest were cut into fine pieces, transferredto 6-well culture plates, and incubated for 1 h at 4°C,followed by an additional 60 min at room temperature. Followingdigestion, uterine pieces were vortexed prior to passage througha sterile 250-µm-mesh screen. Epithelial cells were recoveredby pouring the resulting suspension through a 20-µm-meshscreen. Following an additional rinse, epithelial cells werecollected and suspended in F12K medium (American Type CultureCollection [ATCC], Manassas, Va.) plus 10% fetal bovine serum,supplemented with 100 µg of streptomycin/ml and 100 Uof penicillin/ml (F12K complete medium), and plated onto growthfactor-reduced Matrigel-coated 10-mm NUNC, 0.4-µm-pore-sizepolycarbonate membrane tissue culture inserts (Nalge Nunc International)at a density of three to four insert wells per rat uterus. Insome experiments, epithelial cells prepared as described abovewere suspended and grown in RPMI 1640 medium (Life TechnologiesInc.) containing 25 mM HEPES supplemented with 10% fetal bovineserum (HyClone Laboratories, Inc., Logan, Utah), 5% NCTC109(BioWhittaker, Inc., Walkersville, Md.), 50 µM 2-mercaptoethanol,100 µg of streptomycin/ml, 100 U of penicillin/ml, and2 mM L-glutamine (RPMI complete medium). Cell cultures preparedin RPMI complete medium were plated on growth factor-reducedMatrigel-coated (BD Biosciences) 0.4-µm-pore-size Falconcell culture inserts (Becton Dickinson Labware). Uterine epithelialcells were cultured at 37°C under 5% CO₂. The developmentof a polarized confluent monolayer of uterine epithelial cellswas monitored by measuring TER on an EVOM Voltohmmeter (WorldPrecision Instruments).

Bacterial preparations. Lyophilized L. rhamnosus (ATCC 7469) and E. coli (ATCC 29839)were obtained from the ATCC. ATCC 29839 is a biosafety levelI (nonpathogenic), motile, F⁺, Str^s, Met^-, Thr⁺ strain. L. rhamnosusand E. coli were reconstituted and grown in sterile de Man,Rogosa, and Sharp (MRS) broth or Trypticase soy broth (TSB),respectively (Difco, Sparks, Md.). Frozen stocks of bacteriawere prepared by placing 700 µl of bacteria in a culturemedium into a sterile cryovial with 400 µl of sterileglycerol. Tubes were inverted to mix the contents and were storedat -80°C. Frozen stocks were used to prepare fresh bacterialcultures for those experiments using live bacteria. To measurebacteria, cultures grown to stationary phase were spun at 500x g for 10 min, resuspended in sterile saline (0.9%), centrifugedas described above, and, following resuspension in saline, placedon ice for 2 h. Bacterial counts were determined by opticaldensity analysis of the bacterial saline suspension followedby serial dilution, plating on agar, and triplicate plate counting.

To limit bacterial replication in experiments using live bacteria,E. coli and L. rhamnosus were grown to stationary phase, rinsedtwice in cold saline, resuspended in saline, and placed on icefor 2 h prior to coculture with rat uterine epithelial cells.In experiments in which heat-killed bacteria were used, L. rhamnosusand E. coli were grown to stationary phase, centrifuged, andrinsed as noted above, suspended in sterile saline, and placedon ice for 2 h. Bacterial suspensions were placed in sterileglass tubes and immersed in a 70°C water bath for 20 min.Following cooling to room temperature, suspensions were aliquotedand frozen at -20°C.

Treatment of epithelial cells with live and heat-killed bacteria. In experiments with live bacteria, epithelial cells were fedwith antibiotic-free complete medium for 24 h prior to bacterialtreatment. At the time of bacterial addition, media were removedfrom the basolateral and apical compartments of polarized epithelialcell cultures. Apical compartments received either a bacterialinoculum (300 µl) or sterile saline (300 µl). Inall cases, basolateral compartments received antibiotic-freecomplete medium (500 µl for NUNC/F12K cultures; 800 µlfor Falcon/RPMI cultures). At the end of a 24-h incubation period,basolateral media were collected, centrifuged at 10,000 x gfor 5 min, and stored at -20°C until analysis. In experimentswith heat-killed microbes, F12K complete medium or RPMI completemedium with 100 µg of streptomycin/ml and 100 U of penicillin/mlwas used throughout the experimental period. Heat-killed bacteriain 100 µl of saline were added to 200 µl of apicalmedium. Control wells were treated with the same proportionof saline and apical medium.

Measurement of MIP3 ${alpha}$ , TNF- ${alpha}$ , and TGF-ß. MIP3 ${alpha}$ and TNF- ${alpha}$ protein levels were determined with an enzyme-linkedimmunosorbent assay (ELISA) kit for rat MIP3 ${alpha}$ or rat TNF- ${alpha}$ (DuoSetELISA Development System; R&D Systems, Minneapolis, Minn.).Levels of TGF-ß were measured by a bioassay usinga mink lung epithelial cell line (MLE) transfected with theplasminogen activator inhibitor-1 (PAI-1) promoter linked toa luciferase reporter gene, as previously described (1, 56).This bioassay is based on the cell line's specific sensitivityto picogram levels of biologically active TGF-ß, whichinduces expression of PAI-1, resulting in dose-dependent luciferaseactivity. Briefly, transfected MLE cells were seeded onto a96-well plate at 10⁵/100 µl of medium/well and spun ona Beckman centrifuge at 1,500 rpm for 15 s. Following a 3-hincubation at 37°C, the medium was removed and replacedwith 50 µl of fresh medium plus 50 µl of a seriallydiluted standard (recombinant human TGF-ß) or culturemedium. Cells were then incubated for an additional 17 h beforebeing washed with HBSS and lysed by adding 50 µl of CellCulture Lysis Reagent (Promega)/well for 15 min. Luciferaseactivity was determined by the response of cell lysates to 100µl of Luciferase Reagent (Promega Corp.) for 10 s/wellin a model LB96V Microplate Luminometer (EG&G Berthold,Gaithersburg, Md.).

Statistics. Data were compared by one-way analysis of variance, followedby a Tukey multiple comparison posttest. Differences with aP value of <0.05 were considered significant.

RESULTS

Top

Results

Cytokine/chemokine release by polarized rat uterine epithelial cells. Studies from our laboratory and others have demonstrated thatrat uterine epithelial cells in culture produce TNF- ${alpha}$ (23) andbiologically active TGF-ß (58) and release these intoboth the apical and basolateral compartments of tissue cultureinserts. Previously, Wira and colleagues reported the purityof uterine epithelial cells prepared by enzymatic digestionprior to filter capture of epithelial sheets (44, 57). As seenin Fig. 1, when epithelial cells are grown to confluence andform tight junctions, as determined by TER, they produce MIP3 ${alpha}$ ,TNF- ${alpha}$ , and TGF-ß. Moreover, when cultured in F12K completemedium in NUNC tissue culture inserts, polarized rat uterineepithelial cells release MIP3 ${alpha}$ and TGF-ß preferentiallyto the basolateral compartment. In contrast, TNF- ${alpha}$ is releasedpreferentially to the apical compartment.

View larger version (27K):
[in this window]
[in a new window]
FIG. 1. Constitutive release of MIP3 ${alpha}$ , TGF-ß, and TNF- ${alpha}$ by polarized uterine epithelial cells. Rat uterine epithelial cells were grown to confluence for 5 to 6 days in F12K complete medium at 37°C in NUNC cell culture inserts. Following a 24-h incubation in fresh medium, apical and basolateral media were collected for analysis of MIP3 ${alpha}$ and TNF- ${alpha}$ by ELISA or of TGF-ß by bioassay. Values shown are mean cytokine levels ± standard errors from five or more wells per group. Results for each cytokine are representative of three or more separate experiments. **, significantly (P < 0.01) different from cytokine level measured in the apical chamber.

Effect of live bacteria on TER in polarized uterine epithelial cells. To explore the effect of bacteria on uterine epithelial cellintegrity, TER was measured prior to treatment and at the endof a 24-h coculture of cells with bacteria placed in the apicalchambers of cell culture inserts. As shown in Fig. 2, when polarizedrat uterine epithelial cells were incubated with 9.8 x 10³ or9.8 x 10⁴ live E. coli bacteria in saline, TER was not significantlydifferent from that of saline-treated control wells. Similarly,when uterine epithelial cells were cocultured with 8.4 x10³or 8.4 x10⁴ live L. rhamnosus organisms in saline, the integrityof the polarized monolayers as determined by TER was not affected.

View larger version (17K):
[in this window]
[in a new window]
FIG. 2. Influence of live E. coli and live L. rhamnosus on rat uterine epithelial cell TER. Polarized uterine epithelial cells were grown in NUNC cell culture inserts. Apical medium was replaced with live E. coli (A) or live L. rhamnosus (B) in saline. Controls were treated apically with sterile saline. Basolateral compartments were fed with antibiotic-free F12K complete medium. TER (expressed in ohms per well) was measured after 24 h of incubation with bacteria (background TER, 180 ${Omega}$ /well). Values shown are mean TERs ± standard errors for groups of three to four wells per treatment group. E. coli results are representative of four separate experiments (n = 4), and L. rhamnosus results are representative of six experiments (n = 6).

Basolateral release of MIP3 ${alpha}$ , TNF- ${alpha}$ , and TGF-ß by polarized uterine epithelial cells in response to bacteria. To determine the effect of bacteria on basolateral release ofthe chemokine MIP3 ${alpha}$ , polarized uterine epithelial cells weregrown to confluence prior to coculture with live E. coli orlive L. rhamnosus. As seen in Fig. 3A, addition of live E. colito the apical chamber increased the release of MIP3 ${alpha}$ in basolateralmedia. In contrast, the presence of live L. rhamnosus in theapical chamber had no effect on basolateral release of MIP3 ${alpha}$ relative to that for controls (Fig. 3B). Measurement of TNF- ${alpha}$ in basolateral media (Fig. 4A) indicated that E. coli significantlyincreased the release of TNF- ${alpha}$ into the basolateral compartment,whereas the presence of L. rhamnosus had no effect on TNF- ${alpha}$ releaserelative to that for control cultures (Fig. 4B). In other experiments(data not shown), E. coli and L. rhamnosus were cocultured withuterine epithelial cells for 24 h prior to measurement of TNF- ${alpha}$ and MIP3 ${alpha}$ . Under these conditions, the presence of L. rhamnosushad no effect on the stimulatory action of E. coli on TNF- ${alpha}$ orMIP3 ${alpha}$ release into the basolateral chamber.

View larger version (15K):
[in this window]
[in a new window]
FIG. 3. Effect of live E. coli on basolateral MIP3 ${alpha}$ release by polarized rat uterine epithelial cells. Primary rat uterine epithelial cells were grown to polarized monolayers in NUNC cell culture inserts and treated apically with live E. coli (A) or live L. rhamnosus (B) in saline. Basolateral compartments received antibiotic-free F12K complete medium at the time of bacterial treatment. Basolateral media were collected after 24 h of incubation. Values shown are means ± standard errors from four wells per group. **, significantly (P < 0.01) different from control (n = 3).

View larger version (14K):
[in this window]
[in a new window]
FIG. 4. Effect of live E. coli in physiologic saline on basolateral release of TNF- ${alpha}$ by rat uterine epithelial cells. Polarized monolayer cultures of rat uterine epithelial cells were grown in RPMI medium in Falcon cell culture inserts prior to apical inoculation with live E. coli (A) or live L. rhamnosus (B) in saline. Basolateral media from a 24-h incubation were analyzed for TNF- ${alpha}$ content. Values shown are means ± standard errors for groups of four wells. **, significantly (P < 0.01) different from controls (n = 5).

TGF-ß must be present in its activated form in orderto bind to its receptor (1). To determine if bacteria affectthe release of biologically active TGF-ß, uterineepithelial cell cultures were grown to confluence in Falconcell culture inserts prior to treatment with bacteria. Conditionedmedia from the basolateral compartments were analyzed by a specificand sensitive bioassay, described in Materials and Methods.In this assay a truncated PAI-1 promoter region, fused to aluciferase reporter gene, results in a TGF-ß dose-dependentincrease in luciferase activity (1). As shown in Fig. 5, thelevel of biologically active TGF-ß decreased whenepithelial cells were treated apically with E. coli. In contrast,L. rhamnosus had no effect on the presence of biologically activeTGF-ß in media from the basolateral compartment.

View larger version (15K):
[in this window]
[in a new window]
FIG. 5. Effect of bacteria on basolateral release of TGF-ß by polarized rat uterine epithelial cells. Uterine epithelial cells cultured in RPMI medium in Falcon cell culture inserts were treated apically with live E. coli (A) or live L. rhamnosus (B) in saline and incubated for 24 h. Following collection of basolateral media, levels of biologically active TGF-ß were determined by bioassay as described in Materials and Methods. Values shown are mean TGF-ß concentrations ± standard errors for groups of three wells. *, significantly different from control. (P < 0.05) (n = 3).

Epithelial cell responses to heat-killed E. coli and L. rhamnosus. Recognizing that live bacteria, placed in the apical chamberof tissue culture inserts, replicate over a 24-h incubationperiod, we prepared heat-killed bacteria in order to controlthe number of bacteria present, as well as to eliminate thepossible effects of bacterial metabolites on cell integrityand cytokine production. As seen in Fig. 6, incubation of ratuterine epithelial cells for 24 h with increasing doses of heat-killedE. coli or heat-killed L. rhamnosus had no effect on TER relativeto that in control wells. In contrast, heat-killed E. coli increasedthe basolateral release of both MIP3 ${alpha}$ (Fig. 7A) and TNF- ${alpha}$ (Fig.8A) in a dose-dependent manner. As seen in Fig. 7B and 8B, inagreement with our findings with live bacteria, heat-killedL. rhamnosus had no effect on the basolateral release of MIP3 ${alpha}$ or TNF- ${alpha}$ by polarized uterine epithelial cells. In other studies(data not shown), we found that basolateral release of TGF-ßby epithelial cells was not affected by treatment with heat-killedE. coli or L. rhamnosus.

View larger version (19K):
[in this window]
[in a new window]
FIG. 6. Influence of heat-killed bacteria on TER of polarized uterine epithelial cell cultures. Polarized rat uterine epithelial cells, grown on NUNC cell culture inserts, were fed with F12K complete medium prior to treatment with heat-killed (HK) E. coli (A) or heat-killed L. rhamnosus (B) in the apical compartment (control wells received medium). TER was measured after a 24-h incubation with bacteria. Values shown are mean TERs ± standard errors for groups of three to four wells per treatment group. For E. coli, n = 5; for L. rhamnosus, n = 3.

View larger version (17K):
[in this window]
[in a new window]
FIG. 7. Basolateral release of MIP3 ${alpha}$ by rat uterine epithelial cells in culture. Polarized uterine epithelial cells, grown in F12K complete medium in NUNC cell culture inserts, were treated apically with either heat-killed (HK) E. coli (A) or heat-killed L. rhamnosus (B). Following a 24-h incubation, basolateral media were harvested and analyzed for the presence of MIP3 ${alpha}$ . Values are means ± standard errors for three wells per group. * and **, significantly different (P < 0.05 and P < 0.01, respectively) from control. For E. coli, n = 5; for L. rhamnosus, n = 3.

View larger version (17K):
[in this window]
[in a new window]
FIG. 8. Effect of heat-killed (HK) bacteria on basolateral release of TNF- ${alpha}$ . Rat uterine epithelial cells, cultured on NUNC cell culture inserts in F12K complete medium, were treated with heat-killed E. coli (A) or L. rhamnosus (B). Basolateral media were removed following a 24-h apical incubation with heat-killed bacteria. Values are means ± standard errors for four wells per group. **, significantly (P < 0.01) different from control (n = 3).

DISCUSSION

Top

Discussion

The studies presented here demonstrate that epithelial cellsfrom rat uteri are responsive to bacterial challenge. We showthat, in response to E. coli exposure at their apical surfaces,epithelial cells release cytokines and chemokines that are knownto enhance immune protection through the recruitment and activationof immune cells. That this response is selective is supportedby our finding that whereas E. coli increases the basolateralrelease of MIP3 ${alpha}$ and TNF- ${alpha}$ , L. rhamnosus has no effect. Thesefindings further indicate that under conditions of maintainedcell integrity, as determined by high TER, uterine epithelialcells, in response to increasing numbers of live and heat-killedE. coli, exhibit dose dependent increases in the basolateralrelease of MIP3 ${alpha}$ and TNF- ${alpha}$ at a time when biologically activeTGF-ß release either is not affected or is decreased.

Polarized uterine epithelial cells release MIP3 ${alpha}$ and TNF- ${alpha}$ intothe culture medium in response to E. coli placed on their apicalsurfaces. To the best of our knowledge, this is the first reportthat MIP3 ${alpha}$ is produced by rat uterine epithelial cells. Thisfinding extends the earlier observation by Sun et al., who reportedthat the human uterine epithelial cell line HHUA produces MIP3 ${alpha}$ (49). Our findings demonstrate that, in addition to providinga barrier to protect underlying cells, uterine epithelial cellslikely play an important role both in recruiting and activatingimmune cells in the female reproductive tract. Others have shownthat MIP3 ${alpha}$ is chemotactic to B cells and memory T cells as wellas to immature bone marrow cell-derived CD35⁺ dendritic cells,all of which express the CCR6 receptor (29, 30). Similarly,TNF- ${alpha}$ plays a central role in the activation of macrophages,granulocytes, and cytotoxic T cells in an acute immune response(15, 39). Using epithelial cells from the gastrointestinal tractand cervical epithelial cells, Kagnoff and colleagues demonstratedthe secretion of interleukin-8, GRO alpha, granulocyte-macrophagecolony-stimulating factor, and interleukin-6 in response toChlamydia infection (40). Our findings that MIP3 ${alpha}$ and TNF- ${alpha}$ arereleased into the basolateral media by uterine epithelial cells,both prior to and following E. coli exposure, extend these findingsand suggest that production of cytokines and chemokines by uterineepithelial cells may, in part, account for the presence of aresident population of immune cells in the uteri of intact,noninfected female rats as well as for the increase in immunecells that occurs following intrauterine bacterial challenge(12, 24, 25).

Our studies demonstrate that, in addition to stimulating chemokineand cytokine release, E. coli exposure results in the inhibitionof release of biologically active TGF-ß by uterineepithelial cells. Production of TGF-ß, an antiinflammatorycytokine (29, 32, 53), by uterine epithelial cells has beenshown by others to be under hormonal control (51). In responseto estrogen treatment, both TGF-ß mRNA expressionand TGF-ß production increase in mouse uterine andvaginal epithelial cells. In other studies, it has been foundthat estradiol-induced synthesis of biologically active TGF-ßmediates the inhibitory effects of estradiol on antigen presentationin the uterus and vagina of the rat (56, 58). Our finding inthe present study, to the best of our knowledge, is the firstdemonstration that E. coli, but not L. rhamnosus, affects uterineepithelial cells so to reduce biologically active TGF-ßrelease at a time when MIP3 ${alpha}$ and TNF- ${alpha}$ production is increased.From the standpoint of immune protection within the reproductivetract, our findings of enhanced MIP3 ${alpha}$ and TNF- ${alpha}$ production, alongwith suppressed TGF-ß production, are consistent withoverall protection of this mucosal surface. On the one hand,immune cell recruitment and activation are stimulated by MIP3 ${alpha}$ and TNF- ${alpha}$ ; on the other, TGF-ß, which is immunosuppressive,is inhibited. The recognition that TNF- ${alpha}$ and TGF-ßare under hormonal control during the reproductive cycle andthroughout pregnancy has led to the identification of a dynamicinteraction between these cytokines that is essential for successfulpregnancy. For example, when the ratio of TGF-ß toTNF- ${alpha}$ is high, successful pregnancy ensues (2). In contrast,when the ratio of TGF-ß to TNF- ${alpha}$ falls, increased fetalabsorption occurs in mice. Our findings in the present studysuggest that in response to E. coli, the ratio of TGF-ßto TNF- ${alpha}$ falls to favor immune protection and maternal survival.What remains to be determined is the effect of sex hormoneson the cytokine responsiveness of uterine epithelial cells tobacterial challenge.

The presence of lactobacilli in the vagina is known to contributeto host immune defense against infection by competitive exclusionof other bacteria and by producing bacteriocins and lactic acid,which are inhibitory to many pathogens (36, 41). Other studieshave documented the importance of hydrogen peroxide (H₂O₂)-producingstrains of lactobacilli in inhibiting microorganisms directlyor by reacting with endogenous peroxidase and chloride to formtoxic radicals (26). In a study of sex workers in Kenya, a predominanceof lactobacilli in vaginal microflora was linked to a decreasedprevalence of gonorrhea, Chlamydia infection, and trichomoniasis(32). Interest in the role of lactobacilli for prevention ofsexually transmitted diseases prompted us to coculture E. coliwith L. rhamnosus in order to determine whether the latter wouldaffect the proinflammatory response of uterine epithelial cellsto E. coli. Unexpectedly, we found that coincubation of E. coliwith L. rhamnosus had no effect on the E. coli-stimulated releaseof MIP3 ${alpha}$ and TNF- ${alpha}$ by epithelial cells. What is suggested fromthis study is that L. rhamnosus is not acting acutely on epithelialcells to alter their responsiveness to E. coli. That L. rhamnosuscan have long-term effects on resistance to disease was suggestedfrom studies in which rats received bladder inoculations ofL. rhamnosus 3 weeks before challenge with E. coli (42). Animalspretreated with L. rhamnosus had fewer bladder infections whenchallenged with E. coli than did carrier-pretreated controls.The absence of lactobacilli in treated bladders at the timeof challenge suggested that the protective effects of lactobacilliwere not related to a direct effect of the lactobacilli on E.coli but were more likely due to underlying influences on themucosal immune system in the urogenital tract. Whether L. rhamnosushas comparable long-term effects on uterine epithelial cellsin our system remains to be determined.

Increases in MIP3 ${alpha}$ and TNF- ${alpha}$ levels following incubation of uterineepithelial cells with live E. coli are comparable to those seenfollowing exposure to heat-killed E. coli. Rather than stimulationoccurring as a consequence of logarithmic bacterial growth,which leads to the buildup of metabolites and toxins, the formationof adherent biofilms, and/or the depletion of epithelial substrates(6), our findings suggest that other factors are involved, sinceheat-killed bacteria are as effective as live bacteria in stimulatingMIP3 ${alpha}$ and TNF- ${alpha}$ release. That MIP3 ${alpha}$ and TNF- ${alpha}$ are released in responseto E. coli but not to L. rhamnosus suggests a level of specificityin which epithelial cells distinguish between potential gram-negativepathogens and gram-positive commensals. In dose-response studies,in which live and heat-killed L. rhamnosus bacteria were tested,no effect on epithelial MIP3 ${alpha}$ or TNF- ${alpha}$ release was observed. Whatremains to be done is to identify the mechanism(s) through whichE. coli exerts its stimulatory effects on MIP3 ${alpha}$ and TNF- ${alpha}$ release.Based on the pioneering studies of Janeway (21), the innateimmune system has been identified as a first line of defensewhich protects at the onset of infectious microbial challenge(22). This system relies on conserved germ line-encoded receptorsand molecules that recognize conserved pathogen-associated molecularpatterns (PAMP) found in groups of microorganisms. The patternrecognition receptors (PRR) of the host that recognize PAMPare expressed on many effector cells, including macrophages,dendritic cells, and, as found more recently, epithelial cells(5). One explanation for the increases in MIP3 ${alpha}$ and TNF- ${alpha}$ releasein response to E. coli seen in our studies is that lipopolysaccharide,which is a constitutively expressed component of the cell wallof E. coli, binds to Toll-like receptors to directly stimulatecytokine and chemokine production. Alternatively, rather thandirectly affecting MIP3 ${alpha}$ , TNF- ${alpha}$ , and TGF-ß release individually,the effects of E. coli may be indirect. This is suggested fromrecent studies showing that TNF- ${alpha}$ stimulates the release of MIP3 ${alpha}$ (20) as well as down-regulating the cell surface expressionof chemokine receptors (47). Studies to identify the mechanism(s)through which uterine epithelial cells respond to E. coli challengeare under way in our laboratory.

In conclusion, these findings indicate that epithelial cellsin the uterus are able to discriminate between potential pathogensand commensal bacteria, which normally reside in the lower reproductivetract and periodically enter the uterus (4, 27). The recognitionthat sexually transmitted disease agents, including chlamydiaeand human immunodeficiency virus type 1, the causative agentof AIDS, rapidly enter the uterus both free and along with spermfollowing sexual intercourse raises important questions aboutthe role of the epithelial cells that line the entire reproductivetract in conferring protection either directly or indirectlyby signaling to underlying immune cells. Our studies indicatethat uterine epithelial cells are responsive to potential pathogens.What remains to be identified is the presence of recognitionreceptors that have evolved to distinguish between differenttypes of bacteria as well as among bacteria, viruses, sperm,and the fetus, all of which are allogeneic. Studies to understandthe mechanisms through which uterine epithelial cells respondto bacterial and viral pathogens, as well as the way in whichthese cells are regulated by hormones and cytokines, are neededto provide the foundation of information essential for the survivaland reproductive health of women.

ACKNOWLEDGMENTS

We gratefully thank Richard Rossoll, Charu Kaushic, and RonTaylor for technical assistance that led to the completion ofthese studies. We also express our appreciation to James Gorhamfor help in setting up the assay to measure TGF-ß.

This work was supported by research grants AI-13541 and AI-51877from NIH.

FOOTNOTES

* Corresponding author. Mailing address: Department of Physiology, Dartmouth Medical School, Borwell Building, 1 Medical Center Dr., Lebanon, NH 03756. Phone: (603) 650-7733. Fax: (603) 650-6130. E-mail: Mardi.A.Crane{at}Dartmouth.edu.

Editor: S. H. E. Kaufmann

REFERENCES

Top