Decidual Cells Induce Release of Free and Exosome‐Bound Interferon Epsilon From Vaginal Epithelial Cells

ABSTRACT

Problem

We tested the hypothesis that oxidative stress (OS)‐induced inflammatory response in decidual cells (DECs) may transfer and/or trigger the release of interferon epsilon (IFNε)‐positive extracellular vesicles (EVs) from vaginal epithelial cells (VECs) to minimize vaginal disturbances.

Method of Study

VECs were treated for 48 h under the following conditions: (1) standard VEC media, (2) OS‐inducing cigarette smoke extract (CSE); and supernatant from (3) normal/untreated DECs, and (4) CSE‐treated DECs. The concentration of cytoplasmic, secreted, and VEC‐derived EV bound IFNε (n = 3 each) was measured by enzyme‐linked immunosorbent assay (ELISA). EVs were isolated from culture media by cushioned‐density gradient ultracentrifugation and characterized by immunoblotting and nanoparticle tracking analysis.

Results

Induction of OS in VECs with CSE increased intracellular IFNε in VECs (p = 0.0007) but not free or EV‐bound IFNε compared to control VECs. Exposure to conditioned media from untreated and CSE‐treated DECs induced increased intracellular (p = 0.004, p = 0.049) and free IFNε (p = 0.04, p = 0.03) from VECs. VEC‐derived EVs (126 ± 11.8 nm) expressed exosome markers, and did not change in size regardless of the treatment but decreased in number due to exposure to untreated (p = 0.004) and CSE‐treated DECs (p = 0.025) conditioned media. Furthermore, VEC exosomal IFNε increased by 2.6‐fold (p = 0.0001, untreated DECs) and ∼4‐fold (p = 0.041, CSE‐treated DECs) compared to controls.

Conclusions

Mucosal immune defense mediated by IFNε may be an innate response by VECs under OS. This was further evidenced by an overall increase in IFNε due to both physiologic and pathologic impact of decidua on VECs. IFNε may indicate a stress response by VECs or paracrine crosstalk between gestational tissues.

1. Introduction

In humans, improper decidualization due to oxidative stress (OS) can result in improper placentation possibly precipitating preeclampsia [1, 2, 3, 4, 5]. Elevated vaginal fluid (VF) [6] and VF‐derived extracellular vesicles (EVs) enriched with interferon epsilon (IFNε) have been linked to preeclampsia with preterm delivery. IFNε is the only type I IFN constitutively expressed in the epithelial cells of the vagina, cervix, and endometrium [7, 8, 9, 10, 11, 12, 13] that is regulated by reproductive hormones [7, 13, 14] and toll‐like receptors (TLR), and thought to provide baseline protection against common genital tract bacterial and viral infections [7, 8, 15]. However, IFNε is understudied in human pregnancies, and we do not know its involvement in normal physiologic or non‐infectious or sterile inflammatory responses, especially in relation to adverse reproductive outcomes.

OS in gestational tissues induces harmful tissue‐dependent inflammatory responses propagated through bidirectional paracrine signaling across the feto–maternal interface by EVs [16, 17, 18, 19]. Beyond an optimal threshold, such immune responses can disrupt immune tolerance at the feto–maternal interface, rupture the fetal membranes [20, 21, 22], and trigger parturition [23]. For instance, CSE‐induced OS activates p38 mitogen‐activated protein kinases (p38MAPK) and inflammation in decidual (DEC), amnion mesenchymal (AMC), and chorion cells, resulting in senescence in the chorioamnion, but not in the DECs [24]. EVs from DECs or myometrial cells treated with CSE induced increased release of pro‐inflammatory cytokines from amnion epithelial (AEC) and chorion trophoblast (CTC) cells and IL‐10 from the CTCs only [16]. These observations imply that OS can cause maternal cells to secrete EVs carrying inflammatory mediators to other cells in the reproductive tract, ultimately triggering a cell‐dependent fetal placental inflammatory response.

Furthermore, OS induced p38MAPK‐mediated cell death and sterile inflammation in the cellular components of the cervix [25]. The inflammatory signals emanating from the cervical tissues can be transmitted to cells of the feto–maternal interface (DECs, chorion cells, AMC, and AEC) by cervical EVs [19]. Interestingly, overexpression of p38MAPK enhances type 1 IFN gene expression and antiviral activity [26, 27, 28]. However, these intriguing pathophysiological interactions have not been tested in the decidual‐vaginal axis, a major channel to the intrauterine cavity for both exogenous and endogenous stressors, and a readily accessible source of biomarkers for reproductive health and gestational pathologies.

Therefore, we tested the hypothesis that OS‐induced inflammatory response in DECs may transfer and/or trigger the release of IFNε‐positive EVs from vaginal epithelial cells (VECs). We report that both physiologic and OS‐induced inflammatory responses in decidual cells (DECs) appear to transfer or trigger the release of significant levels of free and EV‐bound IFNε from VECs. IFNε may be a responder of VEC stress response or an indicator of crosstalk between gestational tissues via mechanisms that are still not entirely clear. The absence of IFNε or reduction in VECs during pregnancy may indicate improper communication and DEC dysfunction. This study is physiologically relevant as it highlights the potential communication between cells of the upper (decidua) and lower (vaginal epithelium) female reproductive tract (FRT) through paracrine transfer of EVs and their cargos.

2. Materials and Methods

Human DECs were isolated from placentas obtained from non‐laboring women who underwent elective caesarean delivery at term [16, 23, 29, 30]. These cells were previously immortalized using PA317 LXSN 16E6E7 (ATCC CRL‐2203) and SV40 Cell Immortalization Kit, and showed similar morphology, cell‐specific markers, and cell signalling pathway activation as primary cells [30]. The immortalized human VECs were initially donated by Dr. Richard Pyles [31], and validated to model lower genital tract epithelial cells in our lab [29, 32, 33, 34]. These cells were employed to demonstrate OS‐induced inflammatory crosstalk between gestational tissues. A schematic diagram of the experimental protocol is shown in Figure 1.

FIGURE 1.

Experimental protocol for treatment of vaginal epithelial cells with conditioned media from decidual cells undergoing oxidative stress induced by cigarette smoke extract (CSE). ELISA, enzyme‐linked immunosorbent assay; NTA, nanoparticle tracking analysis. Created with BioRender.com.

2.1. Cell Culture

Immortalized DECs were propagated in complete media containing Dulbecco's Modified Eagle Medium/Nutrient Mixture F‐12 (DMEM/F12, Fisher # MT10092CV) enriched with 10% heat inactivated and EV‐free fetal bovine serum (FBS, Gibco # 10437‐028), 1% penicillin/streptomycin (Corning # 30‐001‐CI), and 1% amphotericin B (Sigma # A2942) for 48 h at 37°C and 5% CO₂ until 80% confluence was attained. The DECs were cultured in two T75 flasks – one for induction of OS by treatment with CSE and the other as control (no CSE treatment).

Our group has extensively employed cigarette smoke extract (CSE) to induce non‐infectious OS in gestational tissues, including maternal decidua and fetal membranes [16, 19, 20, 21, 23, 24, 25, 35, 36, 37, 38]. Water soluble CSE was prepared as we previously described [16, 19, 20, 21, 23, 24, 25, 35, 36, 37, 38].

To prepare CSE treatment media for a T75 flask, 160 uL of CSE was added to 8 mL of exosome‐free DMEM/F12 (1:50 dilution). As shown in our previous data, this concentration of CSE is adequate to induce OS without causing cell death [16, 19, 20, 21, 23, 24, 25, 35, 36, 37, 38]. EV‐free DMEM/F12 was used in order to avoid contamination by FBS‐derived EVs. CSE treatment of DECs (pathologic treatment) was performed by incubating the DECs with 8 mL of CSE treatment media for 48 h at 37°C in 5% CO₂, while another DECs was incubated with 8 mL of EV‐free DMEM/F12 without CSE (physiologic treatment). After 48 h, the media was collected from both the pathologic (OS) and physiologic (no OS) treatments and transferred to respective VEC experimental samples in a ratio of 1:1 (i.e., 2.5 mL of untreated or CSE‐treated DEC conditioned media + 2.5 mL of VEC standard media). It is noteworthy that the preparations of DEC undergoing OS (pathologic) and no OS (physiologic) were only used as the primary treatments (stimulants) for this study and not for comparison with the eventual experimental VEC samples below.

Immortalized VECs [31, 32, 33] were grown in complete Keratinocyte Serum‐Free Growth Medium (KSFM) [32, 33] for 48 h in 5% CO₂ at 37°C until ∼80% confluency. A CSE treatment media containing 160 uL of CSE in 8 mL KSFM (1:50) was also prepared. VECs were incubated in T25 flasks for an additional 48 h under the following conditions: (1) 5 mL of standard KSFM media (VEC, no CSE treatment), (2) treated with OS‐inducing CSE (VEC + CSE) in a ratio of 1:1 (2.5 mL of each media as stated above), and exposed to conditioned media from (3) untreated DEC (VEC + DEC), and (4) CSE‐treated DEC (VEC + DEC + CSE). Each group of samples contained three biological replicates (n = 3). That is, for each experimental (VEC + DEC and VEC + DEC + CSE) and control (VEC alone and VEC + CSE) samples, three biological replicates were included. To view overall cell morphology and proliferation, bright‐field microscopy images of the treated and untreated VECs were captured using a Nikon Eclipse TS100 microscope (4×) (Nikon) coupled with the NIS‐Elements D imaging software version 4.30.01 (Build 1021).

2.2. Cell Lysis and Protein Estimation

After 48 h, the culture supernatant (media) from treatment and control VEC samples were aspirated and stored in −80°C immediately without any additional processing. On the other hand, the cells remaining in the culture flask were lysed with 300 µL of lysis buffer containing 1 mL radioimmunoprecipitation assay (RIPA) buffer, supplemented with 10 µL phenylmethylsulfonyl fluoride (PMSF) and 10 µL cocktail of protease and phosphatase inhibitor for protein preservation. Protein concentration was determined by bicinchoninic acid (BCA) protein assay using the Pierce BCA protein assay kit (Thermoscientific # 23225). The experimental protocol is as previously described [33].

2.3. EV Isolation and Characterization From Culture Media

EVs were isolated from the culture media by cushioned‐density gradient ultracentrifugation (C‐DGUC) (Figure 2). To separate the large and small EV particles, the samples were prepared by centrifugation at 200 × g for 10 min at 4°C. After this, the supernatant was aspirated and centrifuged at 3000 × g for 10 min at 4°C. The resultant supernatant was subsequently transferred into a 2 mL microcentrifuge tube and centrifuged at 17000 × g for 15 min at 4°C, and the supernatant was filtered through a 0.2 µm syringe filtered into a 14 × 95 mm polypropylene centrifuge tube (Beckman Coulter # 331374). Five‐hundred microliters of 30% sucrose (cushion) was slowly added to the bottom of the tube without mixing and balanced with sterile 1X phosphate buffered saline (PBS) by weighing the content of the tubes. The filtrate containing the sucrose cushion was ultracentrifuge at 138,000 × g for 2 h at 4°C, after which the EV‐containing sucrose layer was carefully aspirated into a new polypropylene centrifuge tube containing 10 mL sterile 1X PBS and ultracentrifuge again at 138,000 × g for 90 min at 4°C. The resultant supernatant was decanted carefully, and the EVs at the bottom of the tube were resuspended in 100 µL sterile 1X PBS, vortexed briefly, aspirated, and stored in a clean 1.5 mL Eppendorf tube at −80°C. To maximize EV recovery, a second fraction of EVs was collected via a similar method in 50 µL sterile 1X PBS.

FIGURE 2.

Isolation of extracellular vesicles from vaginal epithelial cells and decidual cell culture media by cushioned‐density gradient ultracentrifugation and downstream analysis. BCA, bicinchoninic acid assay; ELISA, enzyme‐linked immunosorbent assay; NTA, nanoparticle tracking analysis; PBS, phosphate buffered saline. Created with BioRender.com.

In line with the minimal information for studies of EVs (MISEV) guidelines on EV characterization [39, 40], EV identity (expression of unique exosome protein markers) in all samples was confirmed by immunoblotting using Exo‐Check Exosome Antibody Arrays (cat# EXORAY210B‐8, SBI) following the manufacturer's instruction. Twenty microliters of EVs were lysed using 2 µL 10X RIPA lysis buffer (provided in the antibody array kit). The lysis mixture was vortexed for 30 s and 3 µL of labelling reagent (that has been equilibrated to room temperature (RT)) was added to the mixture and vortexed again to mix properly. This lysis mixture was subsequently incubated for 30 min at RT with constant gentle mixing. The labelled EV lysates were passed through packed bed columns to remove the excess labelling reagent. The eventual eluted EV lysate was combined with 5 mL Blocking Buffer in a 15 mL conical tube and inverted three times to mix. The mixture was transferred to the ExoCheck array membrane, which was first cleaned with 5 mL distilled water for 2 min and incubated overnight on a shaker at 4°C. The next day, the mixture was carefully decanted and the membrane was washed twice with 5 mL of 1X wash buffer for 5 min per wash while shaking. After the second wash, 5 mL of detection buffer (containing 5 mL of detection reagent A + 1.5 µL of detection reagent B) was added to the membrane and incubated for 30 min at RT on a shaker. After incubation, the detection buffer was washed off the membrane three times with 5 mL of 1X wash buffer for 5 min per wash while shaking. The membranes were developed with Clarity Max Western ECL substrate (cat# 1705062, Bio‐Rad) and the blot was imaged using the Bio‐Rad ChemiDoc imaging system using the signal accumulation method (four images, 120 s).

Furthermore, the size and concentration (particles/mL) were confirmed by nanoparticle tracking analysis (NTA) using the ZetaView PMX‐110 (Particle Metrix, Meerbusch, Germany) and its corresponding software (ZetaView 8.05.14 SP7). All frozen EVs were thawed on ice, and a 1:1000 dilution of the EV samples was made with filtered 1X PBS. Samples were loaded in the ZetaView Nanoparticle Tracking Analyzer, and the number of particles/mL and the size distribution were quantified for each sample. Before cytokine measurement by enzyme‐linked immunosorbent assay (ELISA), the EVs were lysed by mixing 45 µL of EV sample with 5 µL of RIPA lysis buffer, and total (surface and cytoplasmic) protein concentration was determined as described for the cell lysates and culture media.

2.4. ELISA

The level of IFNε (ng/mL) expressed within the treated and untreated VECs (intracellular), secreted into their culture media and packaged in EVs, was measured by ELISA using Human IFNε sandwich ELISA kit (LS Bio # LS‐F11717) according to the manufacturer's instruction. For clarity, we measured IFNε concentration in the culture media and this is separate from that measured in the EVs isolated from the culture media. To ensure equal amount of proteins was utilized for the ELISA in all samples, the protein concentration in the samples was normalized by combining 20 µg (cell lysates and culture media) and 10 µg (EVs) of protein with appropriate volumes of sample diluent (provided in the kit) to make up a final volume of 100 µL. The cell lysates and culture media samples were subsequently diluted 1:10 with the sample diluent as the concentrations of IFNε in the initial undiluted test samples exceeded the detection range of the kit (0.312–20 ng/mL). On the other hand, the EV samples were not diluted due to low total protein (as expected), and the measured IFNε concentrations were within the kit's detection limit.

2.5. Statistical Analyses

All statistical analyses were conducted using GraphPad Prism version 10.3.1 (509). The Shapiro–Wilk test was used to determine if the distribution of variables was normal. We subsequently log‐transformed the data and determined differences between groups using an unpaired t‐test with Welch's correction (Welch's t‐test) for comparison involving two groups. For comparison between the four groups, a one‐way ANOVA with Welch's correction (Welch's ANOVA) was used. Detailed descriptions of the statistical tests used are included in the legends of each figure, and p values <0.05 indicated a statistically significant difference.

3. Results

3.1. Cell Morphology and Proliferation

The VECs had the typical polygonal, cobblestone appearance [31, 41] (Figure 3). No obvious alterations in overall cell morphology were observed after CSE or conditioned media treatment. However, the cell confluency unexpectedly appeared reduced after treatment with conditioned media from DECs with or without CSE (although not explicitly measured). This may be due to cell death from nutrient deprivation as the VECs were grown in 50% of their ideal culture medium (KSFM). However, components of the 50% DEC conditioned media (i.e., secreted endocrine and paracrine mediators) can also affect VEC proliferation (Figure 3).

FIGURE 3.

Bright field microscopy showing vaginal epithelial cell (VEC) morphology and proliferation (n = 3 for each group). VEC: VEC in standard media without CSE treatment; VEC+CSE: VEC treated with CSE treatment; VEC exposed to conditioned media from untreated (VEC+DEC), and CSE‐treated DECs (VEC+DEC+CSE). Scale bar = 100 µm.

3.2. EV Identity, Size, and Particles/mL

EVs derived from the VECs were characterized following the guidelines for minimal information for studies of EVs [39, 40]. To confirm the presence of characteristic exosomal protein markers from the VEC‐derived EVs, we performed antibody arrays (immunoblotting) on four (one from each group) representative samples. The VEC‐derived EVs expressed exosome markers, including the transmembrane/lipid‐bound extracellular: tetraspanins (CD63, CD81), epithelial cell adhesion molecule (EpCAM), intercellular adhesion molecule 1 (ICAM); and cytosolic: apoptosis‐linked gene 2 (ALG‐2)–interacting protein X (ALIX), tumor susceptibility gene 101 (TSG101), Flotillin‐1 (FLOT1), annexin A5 (ANXA5) makers [42] (Figure 4A). Furthermore, the median size of the EVs was 126 ± 11.8 nm (within exosome range) [39, 40, 42], which did not change significantly regardless of treatment (Figure 4B,C and Table 1). However, the conditioned media from untreated and CSE‐treated DECs reduced VEC EV secretion compared to untreated (p = 0.004, p = 0.025) and CSE‐treated VECs (p = 0.007, p = 0.019) (Figure 4D and Table 1). This corresponds to the unexpected reduced cell confluency observed after exposure to DEC conditioned media. As stated above, this unexpected result may be due to cell death from nutrient deprivation.

FIGURE 4.

Characteristics of vaginal epithelial cell‐derived extracellular vesicles (VEC‐EV). (A) Exosomal protein markers. (B) Representative median size and concentration. (C, D) Comparison of VEC‐EV median size and concentration between the samples. VEC: VEC in standard media without cigarette smoke extract (CSE) treatment; VEC+CSE: VEC treated with CSE; VEC exposed to conditioned media from untreated (VEC+DEC), and CSE‐treated DECs (VEC+DEC+CSE). ALIX, apoptosis‐linked gene 2 (ALG‐2)–interacting protein X; ANXA5, annexin A5; CD63, CD81, tetraspanins; EpCAM, epithelial cell adhesion molecule; FLOT1, flotillin‐1; ICAM1, intercellular adhesion molecule 1; TSG101, tumor susceptibility gene 101; GM130 (cis‐Golgi marker) = an expected weak signal of cellular contamination [39, 40, 59, 60]; Positive control = labelled positive control for horseradish peroxidase detection; Blank = background control. Comparison between two (Welch's t‐test) and all (Welch's ANOVA) groups (n = 3 for each group). * p < 0.05, ** p < 0.01.

TABLE 1.

Characteristics of vaginal epithelial cell‐derived extracellular vesicles.

Sample	Median size, nm	Particles/mL
VEC	125.8 ± 9.17	1.27E + 11 ± 9.47E + 10
VEC + CSE	131.0 ± 18.82	1.69E + 11 ± 1.14E + 11
VEC + DEC	118.8 ± 2.56	3.58E + 09 ± 1.60E + 09
VEC + DEC + CSE	122.3 ± 8.75	6.23E + 09 ± 4.80E + 09

Note: Values are presented as mean ± SD of three biological replicates in each sample group. VEC: vaginal epithelial cell in standard media without cigarette smoke extract (CSE) treatment; VEC+CSE: VEC treated with CSE (1:50); VEC exposed to conditioned media from untreated (VEC+DEC), and CSE‐treated decidual cells (VEC+DEC+CSE).

The size and concentration (particles/mL) of EVs in the untreated and CSE‐treated DECs used as the treatments were also ascertained (Table 2). Although these DECs were only used as treatments, they secreted lesser number but similar size of EV particles compared to the VECs.

TABLE 2.

Characteristics of decidual cell‐derived extracellular vesicles and IFNε concentration.

Sample	Size, nm	Particles/mL	IFNε (ng/mL)
DEC + CSE	127.8	7.65E + 09	1.60
DEC	120.1	1.45E + 10	1.93

Note: Pathologic (undergoing oxidative stress, DEC+CSE) and normal (DEC) decidual cells were used merely as stimulants to treat the vaginal epithelial cells (VECs), which were the primary focus of this study. No statistical comparison was performed using these treatment samples. DEC, untreated decidual cells; DEC+CSE, decidual cells treated with cigarette smoke extract (CSE).

3.3. IFNε Concentration

We have observed elevated VF [6] and VF‐derived EVs (under review) in individuals who developed preterm preeclampsia. Therefore, we hypothesized that OS‐induced responses in cells of the upper FRT, such as DECs, transfer or trigger innate IFNε response that can be measured in EVs derived from VECs. VEC EVs may propagate this innate defense response or be used as proxy markers for decidual adverse responses. Interestingly, we observed that despite the reduced cell proliferation and correspondingly lower EV concentration (Figures 3 and 4), treatment with conditioned media from untreated and CSE‐treated DECs induced increased intracellular (p = 0.004, p = 0.049) and free IFNε (p = 0.04, p = 0.03) from VECs (Figure 5A,B). Furthermore, the conditioned media from untreated (2.6‐fold, p = 0.0001) and CSE‐treated DEC (3.9‐fold, p = 0.041) significantly increased the concentration of IFNε in VEC‐derived EVs compared to controls including CSE‐treated VECs (2.4‐fold, p < 0.0001; 3.6‐fold, p = 0.049) (Figure 5C). We also determined whether induction of OS in VECs by direct exposure to CSE will impact IFNε secretion like the conditioned media from untreated and CSE‐treated DECs. Interestingly, we observed that direct treatment of VEC with CSE, which did not affect cell proliferation, significantly increased intracellular IFNε in VEC (p = 0.0007) but not free or EV‐bound IFNε (Figure 5A–C). Because induction of OS in VECs induced increased intracellular IFNε that was neither significantly secreted nor packaged into EVs, we postulate that the VEC‐derived IFNε + EVs we observed after treatment with DEC conditioned media originated from the DEC media or DEC‐derived EVs. Interestingly, the concentrations of IFNε in EVs derived from untreated (1.931 ng/mL) and CSE‐treated DECs (1.601 ng/mL) (which were merely employed as the “stimulants” in this study) (Table 1) were similar to that of VEC treated with DEC conditioned media (1.996 ± 0.09 ng/mL, n = 3). IFNε from the CSE‐treated and untreated DEC cell lysates and culture media were below the detection limit of the assay when the samples were diluted (1:10) like the VEC samples. However, the undiluted CSE‐treated (9.938 ng/mL) and untreated DEC (14.777 ng/mL) cell lysates had measurable IFNε. IFNε was not measured in the undiluted culture media of the DEC samples. These results imply that VECs express more IFNε than DECs, but there is a possible transfer of free and EV‐bound IFNε in the DEC condition media to the VECs during treatment (Figure 5B,C).

FIGURE 5.

Concentration of IFNε: (A) expressed within the vaginal epithelial cells (VECs); (B) secreted into their culture media; (C) packaged in extracellular vesicles (EVs). IFNε concentration are presented as logarithm scale. VEC: VEC in standard media without cigarette smoke extract (CSE) treatment; VEC+CSE: VEC treated with CSE; VEC exposed to conditioned media from untreated (VEC+DEC), and CSE‐treated DECs (VEC+DEC+CSE). Comparison between two (Welch's t‐test) and all (Welch's ANOVA) groups (n = 3 for each group). * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

4. Discussion

EVs can transmit OS‐induced cellular response signals (mediators) in both directions across the feto–maternal interface and from gestational tissues in the upper to the lower FRT and vice versa [16, 17, 18, 19]. These paracrine signals can induce preeclampsia and preterm delivery depending on the state of the cells at the time of their release [16, 17, 18, 19]. We recently found free [6] and EV‐bound (under review) IFNε in VF of preeclamptic individuals. These discoveries inspired us to hypothesize that OS‐induced adverse cellular response in DECs may transfer or induce the production and secretion of IFNε, including VEC‐derived IFNe+ EVs. This may indicate VECs innate immune response to decidual stimuli, or may be a proxy response of communication between the lower and upper FRT. Hence, we treated VECs with conditioned media from DECs in physiologic state or undergoing OS. We observed that conditioned media from DECs in their physiologic or OS state, which generally produced lesser IFNε, can transfer and/or induce a significant increase in intracellular, free, and EV‐bound IFNε from VECs. Whereas direct induction of OS in VECs (without DEC media) only increased intracellular IFNε in VECs without affecting free or EV‐bound IFNε significantly. The EVs secreted by the VECs, which expressed the characteristic EV protein markers, did not change in size regardless of the treatment but decreased significantly in number due to treatment with DEC conditioned media. Interestingly, despite the decrease in EV secretion by the VECs on exposure to DEC conditioned media, the concentration of IFNε in the EVs increased up to 4‐fold above those of untreated and CSE‐treated VECs, supporting the hypothesis of transfer and/or induction of IFNε+ EVs. Together, these observations indicate that VECs can generally produce significant levels of intracellular IFNε in response to stress stimuli. However, DECs, irrespective of their stress status, can transfer or trigger the release of more‐free and EV‐bound IFNε from VECs. The increased free and EV‐bound IFNε secreted by VECs on exposure to DEC condition media may also be an innate immune response to nutrient starvation as those VECs were exposed to only 50% of the ideal VEC growth medium.

We observed (1) VECs produce IFNe and secrete them via EVs, (2) VECs under OS release increased EV‐packaged IFNe, and (3) decidual impact on VECs can cause endogenous stress response causing IFNe release. The VEC response observed can be interpreted in different ways: (1) may indicate the influence of decidua on VECs to generate a response that is protective to the vaginal environment, (2) may be a proxy response of communication between the lower and upper FRT, (3) may be innate immune response to nutrient starvation. We do not believe that the IFNe response observed is a biomarker of decidualization problems or decidual pathology, as normal decidua caused a similar response. However, we conclude that communication between decidua and VEC may produce an innate response to provide mucosal immune protection. One major limitation of our study is that the role of the cervix that separates decidua from the vagina may modify the decidual signaling prior to its impact on vaginal cells. Regardless, if it reaches the vagina, decidual mediators can impact VECs to respond and produce an innate immune response. As it is, our study cannot conclude that this response may be a biomarker of preeclampsia or any other decidualization issues.

As far as we know, this is the first study investigating EV‐mediated immune crosstalk between DECs and VECs. However, we have previously reported the transfer of inflammatory signals from ectocervical cells undergoing OS to cells of the feto–maternal interface by cervical EVs [19]. This CSE‐induced OS propagated by ectocervical EVs led to an increase in granulocyte‐macrophage colony‐stimulating factor (GM‐CSF) and IL‐8 in DECs and TNF‐α in amnion mesenchymal cells [19], which are involved in fetal membrane inflammation (chorioamnionitis), weakening, and rupture [43, 44, 45]. IFNε is constitutively expressed in the FRT, including the epithelium of the endometrium and cervicovaginal space, as well as the myometrium and chorioamniotic membranes [7, 8, 9, 10, 11, 12, 13]. However, it is not known if IFNε is involved in chorioamnionitis and membrane rupture. Similar to IFNε, which can act independently of TLRs, there was no indication that the OS‐associated transfer of inflammatory cervical EVs to DECs and amnion mesenchymal cells we previously observed [19] was mediated by TLRs. However, it is unclear if both processes are mediated by the same mechanism.

Although we previously observed a trend toward increased odds of iatrogenic preterm birth probably due to increased VF IFNε in individuals with preterm preeclampsia, we only observed significant association between highest tertile of IFNε and chorioamnionitis [6]. Chorioamnionitis can occur without verifiable infectious stimulants and pathogens [46, 47]. IFNε expression in cervicovaginal fluid [48] and myometrium [9] is increased in later gestation but not associated with labor [9]. However, an increased amniotic fluid IFNε was observed in individuals who experienced spontaneous preterm labor with intra‐amniotic infection [9]. Although this observation supports the antimicrobial role of IFNε [8, 49], the actual source of the IFNε could not be ascertained as infectious stimulants did not induce IFNε expression by amnion epithelial cells, chorioamnion or cervical cells in vitro [9]. Because intra‐amniotic inflammatory response is multifactorial [9], the increased amniotic fluid IFNε may arise from other undisclosed (perhaps non‐infectious) sources but not amnion epithelial cells. This is plausible as IFNε expression is not significantly influenced by specific microorganisms [7, 11, 12, 50] due to absence of response elements for pattern recognition receptors (PRR) signaling [7], and could be unaltered or decreased in the presence of human papilloma virus [9] and herpes simplex virus infection [48].

Because IFNε is expressed in tissues across the FRT [7, 9, 13], it may be an integral factor of host immunity in these tissues in pregnant [9] and non‐pregnant states. IFNε activates natural killer (NK) cells [51, 52, 53], CD4⁺ and CD8⁺ T cells and B cells [52, 54] and macrophage cytokine/chemokine response [8, 55]. From our observations, we propose that this homeostatic and baseline mucosal immune response mediated by IFNε [8] may include a response to non‐infectious stimuli that induce OS and inflammation in reproductive tissues. This is supported by evidence of significantly elevated intracellular, free, and EV‐bound IFNε on exposure of VECs to media from DECs despite a considerable reduction in proliferation and secretion of EVs by the VECs. That is, normal or stressed DECs appeared to have transferred their inflammatory phenotype (more‐free and EV‐bound IFNε) to the VECs, albeit we are not certain if this was mediated exclusively by DEC‐derived EVs or in conjunction with other elements contained in or without the conditioned media. For example, nutrient starvation (VECs exposed to 50% of ideal growth medium and 50% DEC conditioned media), can induce OS [56, 57, 58]. This could mean that the VEC + DEC cells produced significant amount of free and EV‐bound IFNε in response to starvation‐induced OS. However, there is a possible transfer of DEC‐derived IFNε+ EVs to the VECs during treatment as EVs derived from DECs and VEC + DEC had similar concentration of IFNε. Additionally, IFNε was detected in the undiluted DECs, and direct exposure of VECs to CSE did not affect free and EV‐bound IFNε.

We previously showed that OS did not affect EV size but significantly increased the number of EVs secreted in cervical tissues [19]. In this study, direct induction of OS in VECs did not affect the size and number of EVs released. Moreover, the concentration of EV‐bound or freely secreted IFNε did not change significantly between untreated and CSE‐treated VECs. Only intracellular IFNε increased significantly in response to OS treatment. While further investigation is required, this may imply a unique localized or unpropagated inflammatory response to OS in VECs. Like in our previous study of the cervical‐feto‐maternal tissue immune crosstalk [19], the upward transmission of inflammatory signals from VECs to DECs can be determined in subsequent studies as there may be no direct immune crosstalk between these tissues in the context of local OS.

Since OS induced p38MAPK‐mediated inflammation in DECs [24] and ectocervical cells [19, 25], and promotes type 1 IFN gene expression [26, 27, 28], we hypothesize that the elevated intracellular, free, and EV‐bound IFNε we observed in VECs after treatment with conditioned media from DECs undergoing OS or not could be associated with p38MAPK activation. That is, stress signals from the decidua can be propagated to cells of the lower FRT perhaps through EVs that can increase cytokine (IFNε) release. Further investigation is required to confirm whether detection of IFNε‐mediated inflammatory response in the relatively distant VECs can be used as a proxy maker for an improper or inflammatory decidual response that may lead to adverse reproductive outcomes. The effect of decidualization may not be found or measurable in VECs or VEC‐derived EVs.

From the aforementioned reports from our lab and others [9], it is necessary to delineate the pathological conditions that can induce IFNε expression, transport in EVs and its possible sources. Specifically, future experiments should expose VECs to conditioned media from LPS‐treated DECs and determine whether VEC‐derived IFNε+ EVs differ between sterile and infectious inflammatory response. The reduced cell proliferation and corresponding low EV secretion by VECs after exposure to 50% normal/pathologic DEC conditioned media highly suggest cell death due to nutrient deprivation. However, this did not significantly impact our findings as the protein concentration in each sample was normalized during the ELISA. Interestingly, IFNε levels in these cells and EVs were still significantly higher compared to the control VEC samples that had more cells and EVs. To provide adequate nutrition and minimize cell death, direct treatment of VECs with EVs isolated from DECs could be performed to determine whether DEC‐derived EVs are exclusively responsible for the increase in VEC IFNε. DECs could also be treated with conditioned media from VECs to determine if vaginal inflammation can affect decidualization through IFNε. Not studying cells from the vagina, cervix and decidua as a unit with inclusion of immune cells is another limitation of the current study. These limitations can be overcome by employing our vagina‐cervix‐decidua organ‐on‐a‐chip model bearing culture chambers populated by VECs, cervical epithelial and stromal cells, and DECs that simulates the FRT during pregnancy [29, 34]. To further determine if our model affects decidualization, we could treat the DECs with gestational levels of hormones.

In conclusion, we demonstrate that mucosal immune defense mediated by IFNε seems to be an innate response by VECs under OS. This was further evidenced by an overall increase in free and EV‐bound IFNε due to both physiologic and pathologic impact of decidua on VECs. IFNε may be a marker of stress response by VECs or an indicator of a paracrine crosstalk between gestational tissues via mechanisms that are still not entirely clear. A lack of IFNε or reduction in VECs during pregnancy may indicate improper communication and DECs dysfunction.

Ethics Statement

The authors have nothing to report.

Conflicts of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Amabebe E., Richardson L. S., Kumar A., Menon R., and Taylor B. D., “Decidual Cells Induce Release of Free and Exosome‐Bound Interferon Epsilon From Vaginal Epithelial Cells.” American Journal of Reproductive Immunology 94, no. 3 (2025): e70129. 10.1111/aji.70129

Data Availability Statement

All data underlying this study are available from the corresponding author on reasonable request.