Effects of a gestational level of estradiol on cellular transition, migration, and inflammation in cervical epithelial and stromal cells.

Immortalized endocervical epithelial cells were provided by Dr. Richard Pyles (The University of Texas Medical Branch at Galveston, TX, USA). These cell lines were previously validated to model the lower genital epithelial cells. 31 Endocervical epithelial cells were cultured in keratinocyte serum-free medium (KSFM), a culture medium highly selective for epithelial cells, supplemented with bovine pituitary extract (30 μg/mL), epidermal growth factor (0.1 ng/mL), CaCl 2 (0.4 mM) and primocin (0.5 mg/mL) (ant-pm-1; Invivogen; San Diego, CA) Cervical stromal cells from hysterectomy material from women with benign gynecological conditions were used in this study. Using an IRB-approved protocol (IRB-AAAI0337), human cervical tissue samples were obtained at Columbia University Irving Medical Center from premenopausal, nonpregnant women (< 50 years of age) undergoing total hysterectomy for benign indications. Women with a history of cervical surgery (loop electrosurgical excision procedure/cone biopsy) were excluded. Informed consent with written documentation was obtained from all research participants. Immediately after the hysterectomy specimen was removed from the patient, a cervical tissue sample (full thickness sample about 2 cm long) was collected from the area of the external os and placed in pre-cooled freezing medium (90% fetal bovine serum [FBS; Sigma-Aldrich, Inc.] with 10% dimethyl sulfoxide [Me2SO]). Samples were frozen at −80°C and then sent to The University of Texas Medical Branch at Galveston. Cervical tissues were thawed in a water bath at 37°C. The cervical epithelium was removed and the cervical stromal layer were incubated in 15 mL complete cervical stromal media composed of Dulbecco’s modified Eagle’s medium: Nutrient Mixture F-12 media (DMEM/F12; Mediatech Inc., Manassas, VA) supplemented with 10% FBS, 10% penicillin/streptomycin (Mediatech Inc.), and 10% amphotericin B (Sigma-Aldrich, Inc.), at 37°C, and 5% CO 2 for 10 min. Three successive transfer to 15 mL fresh cervical stromal media. The cervical stromal tissue was then cut into small fragments (1 × 1 mm) and was incubated with collagenase (0.15% in DMEM without serum) at 37°C for 1 h. An equal volume of complete cervical stromal media was then added to neutralize collagenase and was filtered through 70 μm nylon mesh. The cell digest was centrifuged at 3000 × g for 10 min and the cell pellets were grown in complete CSC medium at 37°C, 5% CO 2 until it reached 80% confluence. EEC or CSC were seeded at approximately 80% confluence in an eight-well glass slide and incubated at 37°C with 5% CO 2 for 24 hours (h). Cells were then rinsed with sterile 1× phosphate-buffered saline (PBS) and were subjected to different treatments for 6 days. To test the effects of infection, inflammation and the effects of E2 in EEC and CSC, cells were treated with one of the following: i) control culture medium; ii) 5000 pg/mL E2 (E8875-1G; Sigma-Aldrich, Inc.) (third trimester level) 32 ; iii) 100 ng/mL LPS and 5000 pg/mL E2; and iv) 100 ng/mL LPS, 5000 pg/mL E2 and 1 μg/mL tamoxifen citrate (T) (0999/100; R&D Systems, Inc., Minneapolis, MN). Tamoxifen, an estrogen receptor blocker, was used to block the effects of estrogen in our experiments. All treatments were prepared in normal growth medium and dosed in a constant volume. Control cultures were treated with the same volume of regular culture medium. The doses of LPS, E2, and tamoxifen were based on their well-established and reported doses in human pregnancy and in in vitro models. 32 , 33 EEC or CSC were seeded at approximately 80% confluence in an eight-well glass slide and incubated at 37°C with 5% CO 2 for 24 h. Cells were then rinsed with sterile 1× PBS, and then scratched evenly down the middle of the well, in a straight line, using a 200 μL pipette tip. 34 Triplicate cultures were washed with sterile 1× PBS four times to remove any cell debris and were subjected to the culture conditions described as above. The cells were then incubated at 37°C, 5% CO 2 , and 95% air humidity for 1 h, 18 h or 36 h for EEC, and 1 hr, 24 h or 48 h for CSC. Wound closure was quantified using ImageJ software version 1.51J (NIH, Bethesda, MD; http://imagej.nih.gov/ij ). To check for the presence of estrogen receptors (ESR) and toll-like receptors (TLR) in EEC and CSC, we performed immunocytochemical staining for ESRα (NBP2-61944; Novus Biologicals Inc., Littleton, CO), ESRβ (NB100-81909; Novus Biologicals Inc.), TLR2 (ab9100; Abcam, Cambridge, MA) and TLR4 (ab22048; Abcam). The dilution factor used for primary antibodies was 1:200 for ESRα and ESRβ, and 1:200 for TLR2 and TLR4. Immunocytochemical staining for vimentin (3.7 μL/mL; ab92547; Abcam) and cytokeratin-18 (CK-18) (1 μL/mL; ab668; Abcam) were performed after the scratch assay, and after 6 days exposure with treatment conditions as described above. The dilution factor for primary antibodies was 1:300 for vimentin and 1:500 for CK-18. After each time point, cells were fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X, and blocked with 3% bovine serum albumin in PBS before incubation with primary antibodies overnight at 4°C. This protocol was adequate to prevent the non-specific binding of primary antibodies in our system based on our previous studies. 23 , 34 After washing with PBS, slides were incubated with Alexa Fluor 488- and Alexa Fluor 594-conjugated secondary antibodies (Life Technologies, Carlsbad, CA) diluted 1:400 in 1× PBS for 1 h in the dark. Slides were washed with PBS, treated with NucBlue Fixed Cell Stains ReadyProbes Reagent ( R37606 ; Thermo Fisher Scientific, Watham, MA), and then mounted using Mowiol 4 to 88 mounting medium (475904-100GM-M; Sigma-Aldrich, Inc.). Images of scratch sites were recorded 1 h, 18 h, and 36 h after the initial scratch for EEC; and 1 h, 24 h, and 48 h for CSC. Images used for quantification of CK-18 and vimentin in EEC and CSC were taken 6 days after treatment with LPS, and E2. Bright-field microscopy images were captured using a Nikon Eclipse TS100 microscope (10×) (Nikon, Melville, NY). Three regions of interest per condition were used to determine the overall cell morphologic features. Fluorescence microscopy images were captured using a Keyence BZ-X810 all-in-one fluorescence microscope (20×). Three random regions of interest per field were used to determine red (CK-18) and green (vimentin) fluorescence intensity. Uniform laser settings, brightness, contrast, and collection settings were matched for all images collected. Images were not modified (brightness, contrast, and smoothing) for intensity analysis. ImageJ software version 1.51J (NIH, Bethesda, MD; http://imagej.nih.gov/ij ) was used to measure CK-18 and vimentin staining intensity. Image analysis was conducted in triplicate for all cell experiments. The cell shape index was determined for EEC and CSC cultures by evaluating one frame from each N (total of three images) per treatment for cell circularity using ImageJ software. 35 The shape index was calculated using the following formula: SI = 4π × area/perimeter 2 , which is an established method that was originally reported to determine vascular cell shape. 36 A circle has a shape index of 1, while a straight line has an index of 0. EEC and CSC were treated for 6 days with control medium, LPS, or E2 before lysis in RIPA lysis buffer with freshly added protease and phosphatase inhibitors (0.01%). The cell lysate was collected after scraping the culture plate, and the insoluble material was removed by centrifugation at 10,000 × g for 20 minutes at 4°C. The concentration of protein in each cell lysate was determined using the BCA protein assay kit (Pierce BCA Protein Assay Kit, Thermo Scientific, Waltham, MA, USA). Equal amounts of protein (8 μg) from each sample were loaded onto a 10% SDS-PAGE gel and electrophoresed at 120 V. The resolved proteins were transferred to a polyvinylidene difluoride membrane using the iBlot transfer apparatus (Bio-Rad Laboratories, Hercules, CA, USA). The membranes were blocked in Tris-buffered saline (TBS) containing 0.1% Tween-20 (TBS-T) and 5% skim milk for 2 h at room temperature. Blots were incubated separately with antibodies against, β-actin (Sigma-Aldrich, A5441), E-cadherin (Abcam, ab15148), N-cadherin (Abcam, ab98952), SNAIL (Abcam, ab180714), and vimentin (Abcam, ab92547) at 4°C and shaken overnight. Blots were washed three times with TBS-T and incubated with the appropriate peroxidase-conjugated IgG secondary antibody for 1 h at room temperature. All blots were developed using chemiluminescence reagents ECL Western Blotting Detection System (Amersham Piscataway, NJ, USA), in accordance with the manufacturer’s recommendations, followed by autoradiography. Densitometry was performed to normalize the data for statistical analysis Enzyme-linked immunosorbent assay was performed to determine the levels of active and inhibitor free MMP9 (F9M00, R&D Systems, Inc., Minneapolis, MN) as a marker of inflammation. Standard curves were developed with duplicate samples of known quantities of recombinant proteins that were provided by the manufacturer. Sample concentrations were determined by relating the absorbance values that were obtained to the standard curve by linear regression analysis. Culture medium collected from EEC and CSC 6 days after treatment with LPS and E2 was tested for IL-1β, IL-10, and TNF-α using a MILLIPLEX MAP Kit for Human Cytokine/chemokine Magnetic Bead Panel following the manufacturer’s protocol. Standard curves were developed using duplicate samples of known-quantity recombinant proteins that were provided by the manufacturer. Sample concentrations were determined by relating the absorbance of the samples to the standard curve using linear regression analysis. Concentrations below the assay detection limits (IL-1β = 2.18 pg/mL; IL-10 = 1.74pg/mL) were considered as one-tenth of each value. Data were analyzed for significant differences using GraphPad Prism software version 7 (GraphPad Software, San Diego, CA). The Shapiro-Wilk test for normality was performed to check for the normality of the data. Parametric tests including one-way analysis of variance followed by the Tukey multiple comparison posthoc test and the t-test were used for comparison of the results for normally distributed data. Non-parametric tests, namely the Kruskal-Wallis test with Dunn’s multiple comparison test and the Mann-Whitney U test, were used for comparison of the results for data that were not normally distributed. Statistically significant differences are indicated by p < 0.05.

We first detected the presence of TLR2 and 4 using immunocytochemical staining. TLRs function as LPS sensor and whose activation results in the production of pro-inflammatory cytokines in cells. 37 Both EEC and CSC express TLR2 and TLR4 but the expression of TLR2 was greater than TLR4 ( Fig. S1 ). We then confirmed the effect of LPS, a proxy for infection, in the cellular transition, migration, and inflammation in EEC and CSC. We found out that LPS promoted EMT (increased vimentin/CK-18 ratio, p = 0.001 and decreased cell shape index, p = 0.001), slowed down wound healing ( p = 0.011), increased the IL-1β/IL10 ratio ( p = 0.001), and MMP9 ( p = 0.01) levels expressed by EEC ( Supplementary Table 1 ). LPS did not affect the cellular state, IL-1β/IL10 ratio, or MMP9 levels produced by CSC. However, LPS also slowed down wound healing in CSC ( p = 0.0007) ( Supplementary Table 2 ). These results were part of our study on the role of steroid hormones in the cellular remodeling of the cervix (under review). Estrogen mediates its effect through ESR. 38 We first detected the presence of ESRα and ESRβ in EEC and CSC through immunocytochemical staining. Both EEC and CSC express ESRα and ESRβ in the cytoplasm and the nuclear regions ( Fig. S2 ). We conducted cell shape analysis and immunocytochemical staining for CK-18 (epithelial marker) and vimentin (mesenchymal marker) followed by bright-field and fluorescence microscopy to determine the effect of E2 in EMT in EEC and in CSC (n = 6). In this study, we found that E2 can maintain the innate metastate of EEC. A third trimester level of E2 did not change CK-18 and vimentin expression ( Fig 1A - B ), and cell shape index ( Fig 1C ) remained as seen in control cells (untreated culture conditions). Western blot analysis also showed that E2 did not change the expression of EMT markers (E-cadherin, N-cadherin, and SNAIL ( Fig 1D - G ] and vimentin [ Fig S3A ]) compared to control. E2 also maintained the metastate of EEC even in the presence of an infection stimulus (LPS). This was shown by the unchanged CK-18 and vimentin expression, cell shape index and levels of EMT markers with LPS and E2 co-treatment compared to control. Treatment with tamoxifen, a known E2 receptor blocker, blunted the effects of E2 in EEC and increased vimentin expression in EEC. These results suggest that E2 may help maintain the innate state of EEC under untreated and LPS-treated conditions. Our previous study showed that CSC exclusively expressed vimentin and LPS treatment did not change vimentin levels in CSC. 24 In this study, we checked for the effects of E2 in the cellular transition of CSC. E2 did not alter vimentin expression based on our immunocytochemical staining ( Fig 2A - B ) and western blot analysis ( Fig S3B ). It also did not alter the cell shape index ( Fig 2C ) in CSC. We saw an increasing trend in E-cadherin and decreasing trend in N-cadherin in E2-treated CSC, but this did not reach statistical significance ( Fig 2D - G ). Tamoxifen treatment did not change the cellular state of CSC, indicating that E2 does not affect the cellular transition of CSC under untreated and LPS-treated conditions. To determine how E2 affects the wound healing process in the cervix, we performed an in vitro scratch assay using EEC and CSC (n = 6). It was noted in our previous study that EEC normally closed the scratch within 36 h while the CSC closed the wound within 48 h 24 . Under control conditions, E2 slowed down the wound healing process in EEC (1.6-fold), although this was not statistically significant ( Fig 3A - B ). The scratch in EEC treated with E2 did not completely close even after 36 h. We also observed the same trend with CSC. E2 slowed down the closure of the scratch wound. Co-treatment with E2 and LPS, and treatment with tamoxifen significantly slowed down the wound healing process (control vs LPS + E2: p = 0.006; control vs LPS + E2 + T: p = 0.008) ( Fig 3C - D ). These data show that E2 may not aid and can even slow down the wound healing process of the cervix in in vitro conditions. To check how E2 affects cellular transition in migrating cells, we performed CK-18 and vimentin staining during the in vitro scratch assay (n = 6). The vimentin expression increased while CK-18 expression decreased as EEC migrated to close the scratch wound. After 36 h, the vimentin expression decreased, and CK-18 expression increased to baseline after the scratch wound had completely healed ( Fig 4A ). E2 treatment did not promote MET in EEC after 36 h under untreated or LPS-treated conditions. Compared to the control, EEC treated with E2 alone and co-treatment with LPS and E2 have significantly higher vimentin expression relative to CK-18 at 36 h (control vs E2: p < 0.05; control vs LPS + E2: p < 0.05) ( Fig 4B ). MET plays a role in re-epithelialization and wound closure. 39 , 40 EMT may hinder the attachment of cells to the basement membrane and to each other, thereby preventing closure of the wound. 22 This may partly explain why E2 did not accelerate the wound healing process in EEC. E2 did not accelerate scratch wound closure and did not affect vimentin expression in CSC. After 48 h, E2 treatment alone did not alter the vimentin expression. However, co-treatment of CSC with LPS and E2 led to a 1.6-fold increase in vimentin expression compared to control ( Fig 4C - D ). These data suggest that E2 does not affect cellular transition and does not promote wound healing in CSC. The lack of effect in cellular transition may explain why E2 does not prevent the deleterious effects of an infection stimulus in CSC wound healing. The observed 1.6-fold increase in vimentin expression after 48 h may also explain the incomplete closure of the CSC scratch wound with LPS and E2 co-treatment. To determine the effects of E2 in EEC and CSC inflammatory cytokine production, we collected medium samples and assessed IL-1β, IL-10, and MMP9 levels (n = 5). E2 treatment did not affect the expression of IL-1β and IL-10 by EEC. E2 increased MMP9 production in EEC but not to a statistically significant level. However, co-treatment of E2 and LPS increased IL-10 and MMP9 ( p < 0.05) production compared to control. The levels of inflammatory cytokines and MMP9 did not change significantly in CSC treated with E2. The data show that E2 did not have an anti-inflammatory effect in EEC and CSC and even increased MMP9 production in the presence of an infection stimulus.

Cervical remodeling is involved in maintaining pregnancy until term. Steroid hormones such as P4 and E2 have been reported to influence this process during pregnancy. P4 helps to maintain pregnancy while E2 is more associated with cervical ripening and parturition. However, the exact role of E2 in the cellular transition, migration, and inflammatory response of the cellular components of the cervix are still lacking. In this study, we investigated cellular migration and transition in untreated and LPS-treated human EEC and CSC in vitro and assessed the impact of physiologic E2 in cellular remodeling. We tested cellular remodeling using the scratch assay (wound healing). Our main findings were: 1) E2 maintained the metastate of EEC and the mesenchymal state of CSC under untreated and LPS-treated conditions; 2) E2 did not accelerate EEC wound CSC wound healing process and even slowed it down under LPS-treated conditions; 3) cyclic EMT and MET were involved in EEC and CSC wound healing; 4) E2 increased the vimentin expression relative to CK-18 in migrating and remodeling EEC; 5) E2 did not exert anti-inflammatory effects and increased MMP9 activation in EEC and CSC. Collectively, these results show that E2 is more likely to support EMT and localized inflammation in migrating and remodeling cervical cells, especially in the presence of an infection stimulus. Further in vivo studies are needed to determine the potential contribution of a third trimester level of E2 in initiating and promoting parturition process during pregnancy. We investigated the effect of E2 in the cellular transition of EEC and CSC. A pregnancy level of E2 maintained the innate state of EEC under untreated and LPS-treated conditions. However, it did not promote MET in CSC. This is consistent with our previous finding that a pregnancy level of P4 does not promote cellular transition in CSC. 24 This may suggest that the cervical stroma maintains its mesenchymal phenotype despite the hormonal changes throughout pregnancy. This mechanism may aid in maintaining cervical tissue integrity throughout pregnancy. While E2 is generally associated with EMT in other cell types, the level of E2 during the third trimester of pregnancy was not enough to exert a pro-EMT effect in EEC. 26 , 41 EMT in EEC and CSC may be detrimental to the cervix because it can decrease cervical epithelial barrier integrity and promote inflammation and ECM degradation in the cervical stroma. E2 may function to help maintain the innate state of EEC and CSC to prevent pathologic damage to the cervix during pregnancy and parturition. E2 did not accelerate the wound healing process in either EEC or CSC under untreated and LPS-treated conditions. This is consistent with previous studies which showed that E2 can inhibit migration and decrease proliferation of different cell types such as endothelial cells, colorectal cells, esophageal cells, thymocytes, and skin cancer cells. 42 – 46 E2 did not promote MET even after wound closure in EEC and CSC under untreated and LPS-treated conditions. Our results show that when cellular migration and remodeling is induced via a scratch wound, E2 treatment resulted in a higher level of vimentin in EEC and CSC compared to control. Previous reports regarding the effect of E2 in wound healing remain divisive, as this can be attributed to a tissue-specific effect of E2. 47 E2 can accelerate cutaneous wound healing but it can also negatively regulate epithelial wound healing in other tissues (i.e., the cornea). 25 , 48 The role of E2 in the wound healing process may have implications in the postpartum repair of the cervix. After delivery, E2 levels start to go down to non-pregnant levels. This physiologic drop in E2 levels is necessary to promote postpartum cervical wound repair which ensures normal recovery of cervical tissue integrity and competency. This process may be interrupted if high E2 levels persist after delivery. 8 There is still a lack of clarity on the relationship between E2 and inflammation. Previous studies have shown that E2 may exert both anti- and proinflammatory effects in different tissues. 30 , 49 In this study, we found that E2 did not affect inflammatory cytokines (IL-1β, IL-10, and TNF-α) produced by EEC and CSC but increased the MMP9 levels produced by EEC under untreated and LPS-treated conditions. MMP9 has been shown to be at an increased level in cervicovaginal fluid and maternal serum samples from preterm birth patients. 50 Increased cervicovaginal levels of MMP9 are also associated with the cervical ripening process in the human cervix. 51 , 52 MMP9 degrades type IV collagen, elastin, proteoglycan, and fibronectin, which are important components of the cervical extracellular matrix. 53 There were some limitations in this study. 1) We used only endocervical epithelial cells and cervical stromal cells. Other cellular components of the cervix, such as ectocervical and transformation zone epithelial cells, may have a different responses to E2 treatment in terms of cellular transition, migration, and inflammation. 2) This in vitro set up did not include immune cells, which greatly contribute to the inflammatory response in the cervix. Collectively, our present findings support the role of third trimester levels of E2 in cervical ripening under untreated and LPS-treated conditions. Under untreated conditions, E2 maintained the innate state of EEC and CSC but did not accelerate wound healing or exert anti-inflammatory effects. However, when co-treated with LPS, E2 did not prevent EMT and slowed down the wound healing process in EEC and CSC. It also increased MMP9 production in EEC, which can promote collagen degradation in the cervix and eventually lead to cervical ripening.

Cervical ripening is a process needed for a harmless and successful delivery of the fetus. 1 Changes in the cervical extracellular matrix (ECM) during this process lead to decreased tensile strength of the cervix. 2 – 4 Cervical ripening also involves an inflammatory process characterized by the influx of neutrophils and increased production of inflammatory cytokines such as IL-6 and IL-8. 5 , 6 Cervical ripening can be altered by the loss of progesterone function, infection, and sterile inflammation. 6 , 7 These pathologic processes can eventually lead to premature cervical ripening and preterm birth. Cervical ripening is associated with high levels of estrogen. 8 Estrogen (E2) is one of the major steroid hormones produced during pregnancy. The plasma concentration of E2 increases up to 100-fold during pregnancy and peaks around the third trimester of pregnancy, before the onset of parturition. 9 , 10 This temporal increase in E2 prior to labor has prompted numerous studies investigating the importance of E2 in initiating labor and delivery. Clinically, there is still not enough evidence to support the notion that E2 can be used to hasten cervical ripening and induce labor and delivery. 11 However, several animal and in vitro studies have pointed out the potential role of E2 in promoting cervical ripening, labor, and delivery. E2 can promote uterine contractility and excitability and increase the responsiveness of myometrial cells to uterotonins such as oxytocin. 12 It stimulates prostaglandin release from the decidual cells of fetal membranes. 13 , 14 It also increases levels of pro-collagenases that degrade type I collagen in the cervix. 15 , 16 Aside from these, the expression level of E2 receptor α (ESR1) in the cervix increases during the second half of a normal pregnancy, remains steady until term delivery, and then decreases postpartum. 17 This increased expression level in the cervix is also associated with spontaneous preterm labor and delivery. 18 These studies show that E2 promotes parturition by accelerating biochemical and biophysical processes in the cervix, the fetal membranes, and the uterus. Steroid hormones such as E2 and progesterone (P4) can also affect cellular transitions like epithelial mesenchymal transition (EMT) and inflammation, which are tightly associated processes that sustain each other. 19 , 20 EMT is a biological phenomenon that involves the transition of epithelial cells into mesenchymal cells. 21 This process usually enhances the migratory properties and invasiveness of cells. 22 Cellular transition was shown to be involved in tissues involved in pregnancy and parturition. Reversible EMT and mesenchymal-to-epithelial transition (MET) mediated amnion remodeling and cellular remodeling of the cervix during pregnancy. This cyclic remodeling is influenced by steroid hormones, particularly P4, which promotes MET in amnion epithelial and cervical epithelial cells. 23 , 24 E2 enhances EMT and the migratory potential of cells in different pathologic conditions, such as endometriosis and gynecologic malignancies. 25 , 26 The role of E2 in EMT was further confirmed through studies showing that silencing E2 receptors results in the inhibition of EMT, and suppresses the invasiveness and migratory potential of human breast and endometrial cancer cells. 27 , 28 E2 also affects the inflammatory response in different human cell lines. However, previous studies have provided conflicting results regarding the role of E2 in anti-inflammatory and proinflammatory responses. 29 , 30 Studies on the role of E2 in promoting EMT and inflammation in cervical epithelial and stromal cells are still lacking. Knowledge on how E2 affects cellular transition and inflammation in the cervix can help us better understand cervical remodeling during pregnancy. We hypothesize that E2 can promote EMT and inflammation, which may contribute to cervical ripening. In this study, we documented the role of a gestational level of E2 in the cellular transition, migratory, and inflammatory responses of endocervical epithelial cells and cervical stromal cells.