Soluble E‐cadherin: A marker of genital epithelial disruption

The genital epithelial barrier is a crucial first line of defence against HIV, and epithelial disruption may enhance HIV susceptibility. Assessment of genital epithelial integrity requires biopsies, but their collection is not practical in many research settings. A validated biomarker of genital epithelial barrier integrity would therefore be useful. The purpose of this study was to evaluate soluble E‐cadherin (sE‐cad) as a marker of genital epithelial disruption.

foreskin are lined by a keratinized stratified squamous epithelium. 5 The expression of cell-cell junction proteins varies between these genital epithelia: E-cadherin is localized to the endocervical columnar epithelium and the basal layers of the stratified epithelium in the vagina and ectocervix but found throughout the stratified epithelial layers in the male genital tract. 4,6 Altered expression or function of the cellcell junction proteins can result in epithelial barrier dysfunction and increased permeability. In the context of HIV susceptibility, increased permeability and breakages in the epithelial barrier may enhance virion access to intra-epithelial or submucosal HIV-susceptible target cells and increase susceptibility, whereas an intact epithelium would be expected to decrease susceptibility. 7,8 Epithelial disruption in the genital tract can be caused by topically applied products (e.g., the spermicide nonoxynol-9), inflammatory stimuli (e.g., sexually transmitted infections; STIs), and physical disruption (e.g., receptive vaginal sex or trauma). [9][10][11][12] Disruption of the epithelial barrier is difficult to measure clinically.
Traditional quantification of cell-cell junction proteins by immunofluorescence or immunohistochemistry requires invasive tissue biopsies, and quantification of epithelial barrier integrity by transepithelial electrical resistance can be performed in vitro but not clinically. The epithelial integrity of other mucosal tissues, such as the gastrointestinal tract, can be assessed through the quantification of soluble biomarkers in blood, urine, or stool. [13][14][15][16] For example, elevated serum levels of soluble zonulin, a tight junction protein, are associated with decreased barrier integrity and increased intestinal permeability. 17 However, there are no validated biomarkers that can be used to measure epithelial barrier integrity in the male or female genital tract.
Soluble E-cadherin (sE-cad) is a soluble monomer of E-cadherin, a component of the adherens junction that contributes to the integrity of the epithelial barrier. 18 E-cadherin cleavage and the subsequent release of sE-cad is associated with increased epithelial barrier permeability in the nasal epithelium, gut epithelium, and female genital tract. [19][20][21] However, because proinflammatory stimuli are often used to elicit sE-cad release, this has made it difficult to elucidate whether sE-cad is a direct marker of epithelial disruption or is indicative of a proinflammatory state. To address this question, we physically disrupted endocervical epithelial cells in vitro to validate sE-cad as a marker of epithelial disruption, independent of any antecedent inflammation, and then further assessed sE-cad levels in human genital secretions as an in vivo biomarker of epithelial integrity.

Study design and sample selection
Clinical samples from two previously reported human cohorts were analysed. 23

Immunoassay
Cell culture supernatants were assessed for levels of sE-cadherin, IL-6, IL-1α, and IL-1β in duplicate using an electro-chemiluminescence ELISA platform (Meso Scale Discovery, Rockville, Maryland, USA) and averaged. Cervicovaginal secretions and penile swabs were analyzed for sE-cad using MSD. The lower limit of quantification (LLOQ) values was determined through the standard curve as the lowest detectable value with a coefficient of variation (CV) ≤ 30%.

Image analysis
Tiled images of whole foreskin tissue sections were imaged (20×) using

Statistics
Soluble immune factor levels were normalized through log10transformation for Pearson correlations. Soluble immune factor levels were log10-transformed and compared between physically disrupted and control conditions using multiple t-tests. Paired comparisons of soluble immune factors were performed with the Wilcoxon matchedpairs signed-rank test. All statistical tests were performed with GraphPad Prism (version 9.0.2). All statistical tests were performed as two-tailed analyses, and an alpha-value of .05 was used.

Soluble E-cadherin levels after physical disruption of an endocervical epithelial monolayer
To address whether sE-cad is a direct biomarker of epithelial disruption, we physically disrupted an in vitro endocervical monolayer in the absence of additional inflammatory stimuli, and measured levels of sEcad, IL-1α, IL-1β, and IL-6 in the epithelial supernatant at 5 min, 0.5 h, 1.5 h, and 24 h post-disruption. Eight physical disruptions were created per well with a P200 pipette tip, disrupting and displacing cells, as depicted in Figure 1A. For data transparency, we show all data points from all experiments. All experiments were performed in quadruplicate, and two control wells were excluded from one experiment, due to elevated inflammatory cytokines assumed to be due to contamination. Epithelial trauma increased sE-cad levels within 5 min ( Figure 1B; P < .001, P = .07, P < .001, P < .01), and sE-cad levels remained elevated after epithelial disruption at 0.5 h (P < .01 for all), 1.5 h (P < .05 for all), and 24 h (P < .001, P < .001, P = .052, P < .01). IL-1α and IL-1β were also elevated after epithelial disruption at 5 min (IL-1α: P < .05 for all; IL-1β: P < .01 for all), 0.5 h (IL-1α: P < .01 for all; IL-1β: P < .01 for all), 1.5 h (IL-1α: P < .001 for all; IL-1β: P < .01 for all), and at 24 h (IL-1α: P = .215, P < .001, P < .001, P < .001; IL-1β: P < .01, P < .001, P = .102, P < .01). Physical disruption did not lead to an increase in IL-6 levels at 5 min or 0.5 h, but levels increased by 1.5 h (P < .001 for all) and remained elevated at 24 h (P < .01 for all). In addition, we re-analyzed these data treating each experiment as a single biological replicate and observed similar results. sE-cad was significantly elevated at 5 min (P < .05) and 0.5 h (P < .05) and not significant at 1.5 h (P = .12) and 24 h (P = .38) ( Figure S1). Of note, sE-cad levels increased substantially over time in the supernatant collected from undisrupted wells (P < .001; Figure 1), presumably due to ongoing epithelial remodeling even in the absence of physical disruption.

Soluble E-cadherin levels after physical disruption of penile foreskin epithelial monolayer
We then performed identical experiments in penile foreskin epithelial monolayers to determine whether sE-cad is also a biomarker of epithelial disruption in the male genital tract. All experiments were repeated in triplicate. Epithelial disruption increased sE-cad within 5 min (P < .001, P < .01, P = .09; Figure 2), and sE-cad levels remained elevated at 0.5 h (P < .01, P = .53, P < .01) and 1.5 h (P = .09, P < .05, P < .001). Levels of IL-1α, IL-1β, and IL-6 were not increased after epithelial disruption in the foreskin epithelial monolayer. Similar results were observed when treating each experiment as a single biological replicate; sE-cad was elevated at 5 min, 0.5 h, and 1.5 h following physical disruption (P < .05 for all; Figure S2).

Active production of soluble immune factors in endocervical epithelial cells after physical disruption
Since levels of sE-cad, IL-1α, and IL-1α β were immediately elevated after epithelial trauma and remained high for several hours, we next assessed whether there was ongoing release over time, or only release at the time of the initial physical trauma. Endocervical monolayers were rinsed with PBS and incubated with fresh cell culture media at either 1.5 or 24 h following physical disruption. The cell culture supernatant was sampled 0.5 h afterwards (i.e., at 2 and 24.5 h following physical disruption) and assessed for levels of inflammatory cytokines and sE-cad. sE-cad levels were comparable between the physically disrupted group and the controls at both time points, demonstrating there was no subsequent production after the initial physical disruption. Similar changes were observed in IL-1α and IL-1β levels (Figure 3), and in one experimental repeat the control cells had slightly higher IL-1β at the 24-h time point when compared to control (difference: 1.068 pg/mL, P < .05). However, IL-6 levels remained consistently elevated at 1.5 h (P < .01 for all) and 24 h (P < .001 for all) following physical disruption and media change (Figure 3), demonstrating ongoing cytokine production after the initial epithelial trauma.

Impact of endocervical abrasion on cervicovaginal sE-cad levels in vivo
Endocervical cytobrush sampling is relatively non-invasive to the underlying tissue and commonly used to collect immune cells from the cervix. Since cell sampling can elicit moderate disruption in the endocervical epithelial barrier, we hypothesized that this would also cause an increase in cervicovaginal sE-cad levels in vivo. Endocervical cytobrush samples were obtained from 10 STI-free and bacterial vaginosis (BV)-negative women of reproductive age (see Methods section, above), with collection of cervicovaginal secretions by Softcup at baseline (pre-cytobrush collection) and again 6 h after cytobrush collection; levels of inflammatory cytokines and sE-cad were assessed at both timepoints. The respective baseline measurements were used as a control for each woman, as there is considerable heterogeneity in cytokines levels amongst women due to differences in vaginal microbiome composition. 27 In keeping with our in vitro findings, the cervicovaginal concentration of sE-cad was substantially elevated 6 h following cytobrush sampling (mean difference [+6 h-baseline] = +log 10 0.2432; P < .05; Figure 5). There is no equivalent intervention to endocervical cytobrush sampling in the male genital tract, thus we were unable to address this question in the male genital tract in vivo. , and tissue-bound E-cadherin was quantified by immunofluorescence microscopy in inner foreskin tissues collected during the penile circumcision procedure ( Figure 6A). There was a significant inverse association between the level of sE-cad in coronal sulcus secretions and the density of membrane-bound E-cadherin by both MFI and tissue area coverage ( Figure 6B: r = −.4907, P < .05, Figure 6C: r = −.4127, P < .05).

F I G U R E 5
Impact of endocervical cytobrush collection on cervicovaginal sE-cad concentration. Cervicovaginal secretions were self-collected prior to cytobrush (baseline) and again 6 h after cytobrush sampling (n = 10). The mean of each time point is detected in red, and the mean difference (+6h-baseline) in sE-cad from was + log 10   In previous studies, the addition of inflammatory stimuli to endocervical, ectocervical, nasal, and intestinal epithelia was shown to decrease membrane-bound E-cadherin, increase sE-cad, and to increase epithelial permeability. 16 Physical disruption of endocervical epithelium in vitro also caused an immediate increase in supernatant levels of the cytokines IL-1α and IL-1β. Since these cytokines are stored in the cytosol as pre-formed molecules prior to secretion, 30,31 it is possible that the immediate increase in IL-1α and IL-1β following physical disruption may reflect cytosolic leakage rather than active transcription-dependent production. Conversely, IL-6 is not constitutively produced and stored in the cytosol, which may explain the lack of a change in IL-6 immediately following physical disruption. Since IL-1 pathways can elicit downstream IL-6 production, the delayed increase in IL-6 following physical disruption may be a consequence of the local release of IL-1α and IL-1β from physical disruption. 32 Consistent with our hypothesis that elevated IL-1α and IL-1β post-disruption reflect immediate cytosolic leakage rather than transcription-dependent production, we did not observe ongoing release of IL-1α or IL-1β into cell culture supernatant when the tissue culture media were changed immediately after physical disruption.
In contrast, IL-6 levels in the endocervical monolayer supernatant continued to increase for at least 24 h following physical disruption, suggesting that elevations in this cytokine are due to transcriptiondependent processes rather than cytosolic release. The fact that foreskin monolayer disruption did not induce release of IL-1α or IL-1β (likely due to biological differences between epithelia) and that there was no late IL-6 increase in this model suggests that the delayed IL-6 production seen in endocervical cells may have been stimulated by IL-1-dependent pathways, although confirming this hypothesis will require additional work.
While immortalized cells share morphological properties and antigen markers with their tissues of origin and can serve as useful models to study the effects of mechanical epithelial disruption and inflammation, these models have important limitations. In this study, endocervical epithelial cells were used as the representative female genital tract cell line, due to the complexity of modelling the stratified structure of the ectocervical or vaginal epithelium, 33  This is further supported by our findings in vivo following cytobrush collection, although we were unable to elucidate the detailed time course of sE-cad changes in vivo due to the lack of samples collected immediately (i.e., 5 min) following cytobrush sampling.
In summary, our results validate sE-cad concentration in genital secretions as a soluble biomarker of epithelial disruption and suggest that this may serve as a non-invasive alternative to biopsy for the serial evaluation of genital epithelial integrity in longitudinal studies. Furthermore, the observation that there is immediate release of proinflammatory cytokines IL-1α and IL-1β after mechanical epithelial disruption, with a more delayed production of IL-6, suggests that there is an inextricable relationship between epithelial disruption and inflammation. Future studies are needed to elucidate this relationship in greater detail.

ACKNOWLEDGMENTS
The authors thank all study participants for their involvements and contributions to the study. They are grateful for Elaine Reguera Núñez for in vitro technical assistance and guidance.

CONFLICT OF INTEREST
The authors declare no competing interests.