Prostaglandin-induced cervical remodelling in humans in the first trimester is associated with increased expression of specific tight junction, but not gap junction proteins

Background Prostaglandins (PG) are widely employed to induce cervical remodelling (CR) in pregnancy. However, the underlying molecular mechanisms are not fully elucidated. Tight junctions (TJ) and gap junctions (GJ) regulate paracellular and intercellular solute transport respectively but their role in the process of CR remains unexplored. We hypothesized that the synthetic prostaglandin E1 analogue Misoprostol (M), widely used in clinical practice to induce CR, may alter TJ and GJ expression as part of the changes in the extracellular matrix (ECM) associated with remodelling. We investigated the effects of Misoprostol exposure on the expression of cervical TJ (claudins 1, 2, 4, 5, 7 and occludin) and GJ (connexins 43, 30 and 26) in the 1st trimester. Methods Cervical biopsies were obtained from pregnant women and comparisons of TJ and GJ protein expression (by western blotting) and immunolocalisation (laser scanning confocal microscopy) made between those who were administered vaginal Misoprostol (n = 10) and those who were not (n = 5). Results We found that Misoprostol-treated tissue (M+) had higher expression of Claudins 1,2,4,7 and occludin (p < 0.05) than untreated (M-) tissue. Expression levels of Claudins 1, 2 and 4 were positively correlated to interval from Misoprostol treatment to biopsy, whilst occludin was negatively correlated. Misoprostol-treated cervical tissue demonstrated more endothelial claudin-5 and occludin, whilst expression of GJs were unchanged. Conclusion Our observations suggest, for the first time, that increased expression of tight junction proteins may be one of the mechanisms by which Misoprostol induces CR in humans. Further studies are needed to explore if TJ proteins may be therapeutic targets to alter timing of CR in clinical practice.


Background
The mechanisms by which the rigid tubular cervix softens, thins out and dilates (cervical remodelling, CR) during pregnancy and before parturition are not fully understood. Cervical remodelling is pivotal to the clinical course and outcome of labour. Thus, understanding the underlying molecular mechanisms of the process may facilitate the prediction, prevention, or initiation of labour in clinical practice.
Connective tissue predominates the cervix and undergoes most of the remodelling of that organ. Although findings vary with experimental models and species, CR is associated with increased stromal hydration [1], disorganization of collagen bundles [2,3] and altered concentrations of hyaluronic acid and dermatan sulphate [4]. During the final stages of CR, there is increased nitric oxide synthesis [5], with extravasation of leucocytes [6] and activated fibroblasts [7]. A role for the epithelium during CR is suggested by its expression of cyclo-oxygenase (COX)-2, interleukin-8 (IL-8), glucocorticoid receptors [7][8][9] and aquaporin channels [10].
Prostaglandins (PGs) are widely used to ripen the pregnant cervix. Easier mechanical dilation of the softer 'PG-primed' cervix has significantly reduced operative morbidity associated with surgical termination of pregnancy (STOP) [11][12][13]. Misoprostol, a synthetic 15deoxy 16-hydroxy 16-methyl analogue of natural PG-E1 has been the drug of choice due to its effectiveness, low cost and minimal side effects at a dose of 400 micrograms (μg) for 3 hours [11,14]. Human cervical biopsy studies in 1st trimester suggest that exogenous PG induces CR by causing disorganization of collagen, vasodilatation, increasing stromal leukocytes, and stimulating the release of pro-inflammatory cytokines [6,15,16]. However the mechanisms by which vaginally administered PG induces these changes in the extracellular matrix are not fully established.
TJs are sites of intercellular contact in the most apical regions of the lateral membranes of epithelial and endothelial cells, and regulate paracellular water and solute transport. TJs are composed of cytoplasmic scaffolding proteins (zona occludens) and tetra-span transmembrane proteins such as junction adhesion molecules, occludin and claudins [17]. Non-pregnant human cervical ectothelium, which is stratified squamous variety, has been shown to express claudins 1, 2, 4, 7 and occludin [18,19]. Changes in transcripts of cervical claudins 1 and 2 have been described in pregnant mice [20]. As predominant regulators of paracellular water and solute transport [21], it is possible that changes in functional expression of TJs may contribute to CR. Animal studies have shown that TJ may modulate the permeability of cervical epithelium during pregnancy [20]. GJs are also trans-membrane channels connecting the cytoplasm of the adjacent cells. They facilitate free transfer of ions up to 1 kiloDalton (kDa) [22]. Expression of Gap Junction (GJ) proteins, namely Connexins (Cx) 43, 26 and 30 in human cervical ectothelium has been described [23,24], but whether they play a role in CR is unknown. It has been postulated that tissue specific co-expression of various GJ and TJ proteins, by virtue of their selective permeability, may regulate tissue adaptation to various physiological needs [22].
We hypothesized that the PGE1 analogue Misoprostol may alter cervical TJ, and perhaps GJ expression, as part of the process of CR in a time-dependent fashion.

Methods
Two 3 × 3 × 10 mm biopsies were collected trans-vaginally from the anterior lip of the ectocervix from fifteen women prior to STOP under general anaesthetic between 9 and 11 (mean = 10.4) weeks of viable gestations not known to have fetal malformations with Trischler's punch biopsy forceps by a single trained researcher (VG). Ten women received 400 micrograms vaginal Misoprostol (M) with mean interval from treatment to biopsy of 95 min (range 50-140 min) while 5 women did not by choice. The mean ages of the treated and un-treated groups were 24.5 years (range 19-34) and 27 years (range 23-32) respectively. The North Sheffield research ethics committee approved this study (ref-08/H1310/35). All women gave written informed consent. Inclusion criteria were age between 18-45 years, previous vaginal delivery, and viable singleton pregnancy less than 12 weeks. Women with a history of previous cervical surgery, abnormal cervical smears, vaginal bleeding or cervico-vaginal infections within one week of the biopsy were excluded. We studied only parous women in order to: a) achieve some degree of homogeneity in the study groups, and b) ensure that normal cervical function and compliance had been demonstrated by previous successful vaginal delivery. Furthermore, all nulliparous women are given Misoprostol prior to STOP to facilitate surgical dilatation and would have been unsuitable for our studies since we also wanted to investigate cervical tissue not pre-treated with Misoprostol. All samples were studied by both Western blotting (WB) and Immunofluorescence (IF) and representative images and blots have been shown in the figures.

Laser scanning confocal microscopy (LSCM)
Localization of the matrix proteins was studied by LSCM imaging by sequential dual indirect immunofluorescence with Cyanine-3 (Cy-3) and Fluorescin isothiocynate-C (FIT-C) conjugated secondary antibodies. 10 μm thick frozen sections were fixed in 100% ice cold acetone for 10 min. After two 5 min washes, blocking of non-specific antigen sites was carried out in 5% goat normal serum (GNS) and 5% bovine serum albumin (BSA) in 1X PBS at RT for 1 hour. Overnight incubations in primary antibodies (all from Zymed labs, USA) were carried out at 4°C as mentioned in section on western blotting: connexin-43, claudins-2, 4, and 5 at 1:100 dilution; 1:200 dilution for connexin-30, occludin, claudins 1 and 7 with 1:60 dilution for connexin-26. Next day, incubations with Cy-3 conjugated goat anti-rabbit antibody (#111-165-144, Jackson ImmunoResearch, USA) and FIT-C conjugated goat anti-mouse secondary antibody (#115-095-146) were carried out sequentially at 1:400 in dark at RT for 45 min each. Negative controls were incubated in either mouse or rabbit isotype control IgG (#08-6599, #08-6199; Zymed Labs, USA) instead of primary antibodies. Positive controls were treated in exactly similar way throughout. After three 10 min washes in 1X TBS-T, sections were air dried before mounting in a drop of Vectashield hard-set mounting medium with DAPI (Vector Labs, USA). Images were acquired with an LSM 510 NLO inverted microscope (Carl Zeiss, Germany) using 20X/0.8 and 40X/1.3 oil immersion objectives. Images were acquired sequentially to eliminate the risk of cross talk. FIT-C was examined using 488 nm Argon laser excitation, a dual band 488/ 543 dichroic mirror, a NFT545 beam splitter and BP500-530 nm IR emission filter. Cy3 was examined using a 543 nm Helium Neon laser, a dual band 488/ 543 dichroic mirror, a NFT545 beam splitter and BP565-615 nm IR emission filter. Images of DAPI staining were taken using multiphoton laser tuned to 740 nm a KP650 dichroic mirror and 390-465 nm emission filter. Unstained and control specimens were examined using the same microscope settings as those used to acquire images from the antibody labeled tissue sections.

Statistical analysis
Data were analysed using SPSS version 18 (Statistical Package for Social Science, Inc., Chicago, IL) employing descriptive statistics, analysis of variance, T-tests, Mann-Whitney U tests and tests of correlation as were appropriate following assessment of normality of distribution (Shapiro-Wilk's test) and equality of variance between groups (the Levene's test), and corrections for multiple comparisons by post hoc analysis. Data was considered significant for p values less than 0.05. LSCM data was not used for quantification.

Results
All patient samples were studied by both techniquesimmune fluorescence analysis and western blotting. The patterns described were seen in all samples and the confocal images and immunoblots shown are representative of the entire samples.

Effect of Misoprostol on expression of tight junction proteins
There was significantly increased expressions of claudins 1, 2, 4 and 7 (Figure 1a

Vascular expression of claudin-5 and occludin
Claudin-5 was selectively localized to the vascular endothelium (Figure 4a-green) whilst occludin was expressed in the vascular smooth muscle in addition to the squamous ectothelium in the human cervix ( Figure  4a-red). PG treatment appeared to increase expression and also cause co-localization of claudin-5 and occludin in the vascular endothelium (Figure 4b-overlay). Furthermore, the blood vessels in the PG-treated cervix   were consistently larger compared to the un-treated cervix (Figure 4d vs. 4c). Our attempts at immuno-blotting of claudin-5 were unsuccessful.

Effect of Misoprostol treatment on expression of Gap junction proteins
Misoprostol treatment was not associated with significantly altered expression of any of the GJ proteins studied: a trend towards increased expression of Cx-30 in Misoprostol-treated tissue did not attain statistical significance (Figure 5a and 5b). Furthermore, immunelocalisation patterns of these GJs in human cervical ectothelium did not differ with Misoprostol treatment (Figure 5c and 5d).

Discussion
This study reports, for the first time, that cervical remodeling induced by Misoprostol in the first trimester is associated with increased expression of claudins 1, 2, 4, 7 and occludin, as well as increased vascular endothelial expression of claudin-5 with concomitant vasodilatation. Our findings support the hypothesis that cervical TJs may play an important role in Misoprostol-induced cervical remodelling. Taken together with recent reports that PG-induced cervical remodelling involves increased stromal leukocytes and cytokines [27] which can regulate paracellular permeability [28], our observations strongly suggest that altered TJ expression and function may modulate cervical ECM remodeling by altering paracellular permeability.
We found a positive correlation between expression levels of claudins 2 and 4 and the time interval from Misoprostol treatment to biopsy, further supportive of the thesis that Misoprostol administration and TJ expression are linked, perhaps causally. It is tempting to speculate that this temporal change in expression levels may be attributable to changes associated with the signaling cascade involved in cervical remodelling. Clinical trials have shown that the efficacy of vaginal Misoprostol for cervical priming is both dose-and time-dependent [11,14], and is related to time to attain peak plasma concentrations (70-80 min after vaginal administration) [29]. It is plausible that the increase in cervical epithelial TJ expression in response to Misoprostol treatment serves to promote trans-epithelial permeability, increasing local tissue delivery of Misoprostol and its plasma concentrations. Further studies are required to confirm this, and whether TJs play a more direct modulatory role in CR.
Misoprostol significantly increased expression of all the forms of occludin: we consistently found occludin bands at 65 kDa, 50 kDa and between 37-40 kDa in all studied specimens. Previous in vitro studies have attributed such bands between 62 to 82 kDa on SDS-PAGE to phosphorylated forms of occludin [30][31][32]. Whether post-translational modification of occludin by phosphorylation is a mechanism by which Misoprostol treatment alters trans-epithelial resistance and permeability during cervical remodelling is unclear. An alternative hypothesis has been that the 50 kDa and 37-40 kDa isoforms of occludin may reflect rapid occludin turnover during pregnancy [33]. Such breakdown turnover to the 50 kDa isoform has been shown to reduce TJ resistance of cervical epithelium [34], suggesting that low molecular weight forms of occludin, being low resistance pathways, may increase epithelial layer permeability leading to increased stromal hydration. The significance of the inverse correlation of cervical occludin with time of Misoprostol administration is unclear. One explanation is that the maximal effect of Misoprostol on occludin expression is achieved very rapidly compared to claudins 2 and 4, waning thereafter in a time-dependent fashion.
Claudin-5 was selectively localized to the vascular endothelium whilst occludin was expressed mainly in the vascular smooth muscle. Their increased expression and mediation of trans-endothelial migration of T lymphocytes and macrophages [35], together with the reported recruitment by PG of chemotactic leukocytes [16] and mast cells [6], are strongly supportive of important roles for TJ in the regulation of vascular permeability [15,27,36] in the ECM during Misoprostol-induced CR.
Whether Misoprostol-induced nuclear membrane expression of claudins 1 and 4 ( Figure 3f) observed by us could imply increased transcription or interaction with nuclear PGE 1-4 receptors as suggested by others [37] needs further investigation. Misoprostol did not significantly affect the expression of gap junction proteins, suggesting that they are unlikely to play key roles in Misoprostol-induced CR.
The interpretation of our observations should be subject to some caveats. Firstly, we have not demonstrated any significant effect of misoprostol on cervical GJ expression in the limited number of tissue samples studied. However ethical constraints precluded taking further biopsy samples once significant effects had already been demonstrated for TJ expression. We would have needed more than two hundred cervical biopsies to demonstrate a 20% difference in the expression of connexins-43 and 26 between Misoprostol-treated and non-treated cervix with 80% power at the 95% confidence level. It is plausible that a role for GJs in cervical ripening was not detectable by our limited sample sizes and assay techniques, or that GJs may modulate remodelling processes induced by other synthetic or naturally-occurring prostagladin analogues which we have not studied. Secondly, the heterogeneity of expression of different TJs (seen as varying band intensities in western blotting- Figures 1 and 2b) in the Misoprostol-treated group could be partly due to the varied duration of exposure to Misoprostol, as well as individual variations in subject responses [27]. Despite these considerations we still observed significant differences between Misoprostoltreated and untreated cervix, strongly suggesting a true effect of Misoprostol on cervical expression of some TJ proteins. It was not practicable to fix the time interval between administration of Misoprostol and the cervical biopsy as this depended on a highly variable transit time between the wards, the operating theatres and the administration of anaesthetic for the clinical procedure. Recording the variation in time to biopsy from Misoprostol administration enabled us to serendipitously observe a time-dependent change in expression of some of the TJ proteins studied.

Conclusion
We report for the first time that cervical tissue treated with Misoprostol, a synthetic PGE1 analogue, demonstrated increased expression of specific epithelial and endothelial TJ proteins in the pregnant human cervix in a time-dependent fashion. This suggests that signals that regulate trans-epithelial transport in the cervix may act by regulating TJ mediated paracellular permeability, highlighting a potential role for these matrix proteins in modulating cervical remodelling induced by Misoprostol. Further studies are needed to clarify the mechanism of this interaction during tissue remodelling, and to determine whether our observations in first trimester are mirrored during spontaneous ripening at term. These studies are necessary as they may clarify the mechanism of failed initiation of labour using PG analogues, as well as uncover new therapeutic targets.