Divergent roles of estrogen receptor subtypes in regulating estrogen-modulated colonic ion transports and epithelial repair

Although it was described previously for estrogen (E2) regulation of intestinal epithelial Cl− and HCO3− secretion in sex difference, almost nothing is known about the roles of estrogen receptor (ER) subtypes in regulating E2-modulated epithelial ion transports and epithelial restitution. Here, we aimed to investigate ERα and ERβ subtypes in the regulation of E2-modulated colonic epithelial HCO3− and Cl− secretion and epithelial restitution. Through physiological and biochemical studies, in combination of genetic knockdown, we showed that ERα attenuated female colonic Cl− secretion but promoted Ca2+-dependent HCO3− secretion via store-operated calcium entry (SOCE) mechanism in mice. However, ERβ attenuated HCO3− secretion by inhibiting Ca2+via the SOCE and inhibiting cAMP via protein kinases. Moreover, ERα but not ERβ promoted epithelial cell restitution via SOCE/Ca2+ signaling. ERα also enhanced cyclin D1, proliferating cell nuclear antigen, and β-catenin expression in normal human colonic epithelial cells. All ERα-mediated biological effects could be attenuated by its selective antagonist and genetic knockdown. Finally, both ERα and ERβ were expressed in human colonic epithelial cells and mouse colonic tissues. We therefore conclude that E2 modulates complex colonic epithelial HCO3− and Cl− secretion via ER subtype-dependent mechanisms and that ERα is specifically responsible for colonic epithelial regeneration. This study provides novel insights into the molecular mechanisms of how ERα and ERβ subtypes orchestrate functional homeostasis of normal colonic epithelial cells.

Epithelial ion transports are pivotal physiological process in several human organs, such as gastroenterological (GI) tract, respiratory tract, reproductive tract, and skin. Intestinal epithelium either absorbs electrolytes or secretes anions (such as HCO 3 − and Cl − ), providing the driving force for fluid transport to maintain fluid homeostasis in human body (1,2).
The duodenal mucosa not only senses luminal nutrients but also regulates ion transports (particularly HCO 3 − and Cl − secretion), which in turn is important for nutrient absorption and mucosal protection (5,6). It was reported that patients with duodenal ulcer had significantly diminished proximal duodenal HCO 3 − secretion compared with healthy volunteers (7,8), suggesting not only that normal duodenal HCO 3 − secretion is pivotal to mucosal protection but that diminished duodenal HCO 3 − secretion contributes to duodenal ulcer. We previously revealed a sex difference in duodenal HCO 3 − secretion in mice and found the expression and function of estrogen receptors (ER) in murine duodenal epithelium (9). We further demonstrated that estrogen (E 2 ) may protect human duodenum against the acid-induced injury by mediating duodenal HCO 3 − secretion likely via ER activation, which explains the lower incidence of duodenal ulcer in women than age-matched men (10). However, it is largely unknown about the involvement of different ER subtypes (ERα and ERβ) in the process of intestinal epithelial HCO 3 − secretion, let alone the underlying mechanisms of ERα-and ERβmediated HCO 3 − secretion.
Colonic epithelial anion secretion is a well-established physiological process closely linked to overall fluid and electrolyte movement in the colon (11)(12)(13). It is vital to maintain normal colonic HCO 3 − secretion, which loss (such as in diarrhea) may cause not only imbalance of pH values and electrolytes in whole body but also local disruption of colonic environments, such as epithelial barrier and microbiome (11)(12)(13). Therefore, the studies on colonic HCO 3 − secretion may offer an opportunity for improving human GI health.
Unfortunately, it has not been explored if E 2 regulates colonic HCO 3 − secretion via ER activation so far; and if so, what ER subtypes and mechanisms are involved. While E 2 promotes duodenal HCO 3 − secretion (9, 10), it inhibits colonic Cl − secretion in sex difference (14)(15)(16). However, the underlying mechanisms are unclear, and it is even unknown if E 2 inhibition of Cl − secretion is via ER or not. Moreover, integrity and homeostasis of intestinal mucosa are crucial for GI function, which depends upon the balance between mucosal injury and healing (17). Although epithelial cell restitution plays an important role in healing process (17), little is known about the involvement of ER subtypes in colonic mucosal healing. Therefore, in the present study, we hypothesized that ER subtypes may play different roles in the regulation of E 2 -modulated colonic HCO 3 − and Cl − secretion, and we examined the underlying mechanisms of ER subtypes in E 2 -mediated colonic epithelial HCO 3 − secretion and epithelial repair.

Results
ERα stimulation of HCO 3 − secretion from the male duodenum and distal colon Since the lack of information on ER subtypes in intestinal HCO 3 − secretion, we performed Ussing chamber experiments with pH-stat to examine the roles of ER subtypes in HCO 3 − secretion from the duodenum and distal colon in male mice. First, we found that both estradiol-17β (E 2 , 100 nM) and ERα selective activator propyl pyrazole triol (PPT, 10 nM) stimulated rapid duodenal HCO 3 − secretion; however, ERβ selective activator diarylpropionitrile (DPN, 10 nM) did not (Fig. 1A). Second, we tested if there is any regional heterogeneity for the ER subtype-mediated HCO 3 − secretion in between the duodenum and distal colon. Like in the duodenum, both E 2 and PPT stimulated colonic HCO 3 − secretion but DPN alone did not (Fig. 1B). Moreover, compared to PPT alone, PPT plus DPN did not further stimulate additional colonic HCO 3 − secretion (Fig. 1B). Thus, ERα activation stimulated HCO 3 − secretion from the duodenum and distal colon, which has no regional heterogeneity in GI tract. In the subsequent experiments, distal colonic HCO 3 − secretion was focused because duodenal HCO 3 − secretion has been extensively studied (10,18,19).

Differing effects of ERα and ERβ receptors on Ca 2+ -mediated and cAMP-mediated colonic HCO 3 − secretion
It is well-known that Ca 2+ and cAMP signaling pathways play crucial roles in regulating intestinal anion secretion (11)(12)(13)(18)(19)(20)(21). Since carbachol (CCh, a cholinergic agonist) and forskolin (an adenylcyclase activator) stimulate colonic anion secretion by triggering intracellular Ca 2+ and cAMP signaling respectively (15), we examined if ER signaling may impact CCh (Ca 2+ signaling)-or forskolin (cAMP signaling)stimulated HCO 3 − secretion. As shown in Fig. 2, A  ERβ inhibits male colonic HCO 3 − secretion via protein kinase pathways After demonstrating ERβ inhibition of CCh-and forskolininduced colonic HCO 3 − secretion, we further elucidated the underlying mechanisms. Since protein kinase, such as tyrosine kinase (TK), protein kinase C (PKC), and phosphoinositide 3-kinase (PI3K) pathways play critical roles in intestinal HCO 3 − secretion (6,(22)(23)(24)(25), we next studied the intracellular Figure 1. Activation of ERα but not ERβ stimulates epithelial HCO 3 − secretion in both duodenum and distal colon. A, effects of estradiol-17β (E 2 , 100 nM, n = 6), propyl pyrazole triol (PPT, 10 nM, n = 6), and diarylpropionitrile (DPN, 10 nM, n = 5) on duodenal net peak HCO 3 − secretion. B, effects of E 2 (100 nM, n = 7), PPT (10 nM, n = 5), DPN (10 nM, n = 5) and PPT plus DPN (n = 5) on distal colonic net peak HCO 3 − secretion. E 2 , PPT, and DPN were added to both sides, but CCh and forskolin were add to serosal side. **p < 0.01. ns, no significant differences. Data are presented as mean ± SD. The statistical significance of differences in the means of experimental groups was determined using Student's t test or one-way ANOVA followed by post hoc test for multiple pairwise comparisons. CCh, carbachol; ER, estrogen receptor. PPT, DPN, and genistein were add to both sides, but CCh and forskolin were added to serosal side. **p < 0.01 and ***p < 0.001. n = 5 to 6 tissues for each signaling mechanisms behind ERβ inhibition of Ca 2+ -and cAMP-stimulated HCO 3 − secretion. First, as shown in ERα specifically inhibits CCh-and forskolin-induced female colonic short-circuit current After demonstrating the roles and the mechanisms of ER subtypes in the regulation of male colonic HCO 3 − secretion, we examined their roles in colonic Cl − secretion that is poorly understood although it is known about E 2 inhibition of Cl − secretion in rat distal colonic epithelium with a gender-specific mechanism (14,15,28). None of E 2 (1 μM), PPT (500 nM), and DPN (500 nM) at high concentrations affected basal colonic short-circuit current (I sc ) in both sexes (Fig. 4, A and B). Moreover, none of them affected CCh-and forskolin-induced male colonic I sc (Fig. 4, C and D), consistently with the previous reports that E 2 did not alter colonic epithelial Cl − secretion in male rats (15,28). We further examined the role of ER subtypes in female colonic Cl − secretion. As shown in Figure 4, E and F, ERα selective activator PPT (100 nM) significantly inhibited CChand forskolin-induced I sc of female mouse colon, which is consistent with the previous reports that E 2 inhibited colonic epithelial Cl − secretion through KCNQ1 channels in female rats (14,15,28). However, ERβ selective activator DPN at high concentration 500 nM did not affect CCh-and forskolininduced I sc of female mouse colon (Fig. 4, G and H). Taken together, ERα subtype is specifically responsible for the inhibition of female colonic I sc .
ERα induces Ca 2+ signaling via the store-operated Ca 2+ entry, but ERβ inhibits the store-operated Ca 2+  Although the initial Ca 2+ transient produced by CPA in Ca 2+ -free solutions is presumably Ca 2+ release from stores, it was reduced by pretreatment with SOCE blockers (Fig. 5, C, F, and I). We assume it is because the SOCE is important to refill the intracellular Ca 2+ store, and Ca 2+ release from stores would be reduced by pretreatment with SOCE blockers. To test this possibility, we added SOCE blockers after the store depletion but before extracellular Ca 2+ influx. Indeed, SKF96365 (50 μM) markedly reduced CPA-induced Ca 2+ influx ( Fig. 5, K-M). These data verify not only SOCE function per se but also its important role in refilling intracellular Ca 2+ store in HCoEpiC.

ERα promoted Ca 2+ -dependent proliferation and migration of HCoEpiC
Since proliferation and migration of epithelial cells are critical for the restitution of injured intestinal epithelium (33,34), we examined the roles of ER subtypes in epithelial regeneration. PPT at 5 to 50 nM promoted proliferation of HCoEpiC (  , forskolin (n = 4), DPN + forskolin (n = 4), and wortmannin + DPN + forskolin (n = 6). PPT, DPN, rottlerin, and wortmannin were added to both sides, but CCh and forskolin were added to serosal side. *p < 0.05 and **p < 0.01. ns, no significant differences. Data are presented as mean ± SD. The statistical significance of differences in the means of experimental groups was determined using Student's t test or one-way ANOVA followed by post hoc test for multiple pairwise comparisons. CCh, carbachol; DPN, diarylpropionitrile; ER, estrogen receptor; PPT, propyl pyrazole triol. Since cyclin D1, proliferating cell nuclear antigen (PCNA), and β-catenin play crucial roles in enterocyte proliferation (35,36), we examined their protein expression following ER subtype stimulation. As shown in Fig. 8I, the pretreatment with PPT (10 nM) for 48 h enhanced the expression of cyclin D1, PCNA, and β-catenin. However, DPN (10 nM) did not affect their expression (Fig. 8J). Therefore, ERα promotes Ca 2+dependent proliferation of HCoEpiC.
Next, we performed cell scratch assays to examine the roles of ER subtypes in HCoEpiC migration. As shown in Fig. 9, A and B, PPT at 10 to 50 nM promoted cell migration, but DPN (10-50 nM) did not affect it. Moreover, ERα selective inhibitor MPP (1 μM) and BAPTA-AM (1 μM) and shERα abolished PPT-induced cell migration (Fig. 9, C-E). Therefore, ERα also promotes Ca 2+ -dependent migration of HCoEpiC.

ER subtype expression in HCoEpiC and native mouse colonic epithelia
Since there is no information available in literature on ER subtype expression in HCoEpiC, we examined ER subtype expression in HCoEpiC and human umbilical vein endothelial cells (HUVECs) as a positive control (37,38). As shown in Figure 10, A and B, both ERα and ERβ mRNA expression were detected in HCoEpiC. Western blots analysis further confirmed ERα and ERβ protein expression in HCoEpiC like in HUVEC (Fig. 10, C and D). Moreover, we also performed immunofluorescence to examine the expression and localization of ER subtypes in HCoEpiC. As shown in Figure 10E, ERα proteins were predominately expressed in the cytoplasm and nucleus, but ERβ were mostly expressed in the cytoplasm compared to the nucleus. However, the immunofluorescence staining was not observed without the primary antibodies against ERα and ERβ in the negative control, indicating specific staining on these proteins in HCoEpiC.
ERα and ERβ mRNA expression was previously detected in human and mouse colonic epithelia (38)(39)(40); however, the protein expression of these receptors in colonic epithelia is unknown. So, we performed Western blots analysis to examine protein expression of ER subtypes in mouse colonic epithelia. As shown in Figure 10, F and G, both proteins were expressed in native colonic epithelia in male and female mice, further supporting both ERα and ERβ expression in mice (38,39).

Discussion
Although sex-based differences in E 2 regulation of intestinal Cl − and HCO 3 − secretion have previously been described  (9, 10, 14-16), it has been unknown what specific ER subtypes may be responsible for intestinal ion transport and epithelial restitution. In the present study, for the first time, we demonstrate ER subtypes in E 2 -mediated colonic ion transports and epithelial restitution in distinct ways (please see Fig. 11). Here we show that (1)     and cAMP pathways. We not only found that ERβ inhibited SOCE mechanism (49) but also revealed that inhibition of protein kinases reverses ERβ-inhibited colonic HCO 3 − secretion via cAMP pathway, suggesting that ERβ-activated protein kinases are involved in this process. In contrast, inhibition of protein kinases did not reverse ERα-inhibited HCO 3 − secretion via Ca 2+ -pathway, further supporting the notion that ERβ inhibits colonic HCO 3 − secretion by suppressing Ca 2+ signaling via SOCE and cAMP signaling via protein kinases.
We previously showed that activation of the Ca 2+ -sensing receptor in the duodenum resulted in [Ca 2+ ] cyt increase but cAMP decrease (19); the end result being increased duodenal HCO 3 − secretion without simultaneously altering I sc . Therefore, the data from Ca 2+ -sensing receptor and ER further support our notion that epithelial HCO 3 − and Cl − secretion could be triggered differentially by Ca 2+ and cAMP signaling (9,19). Determining how to specifically modulate Cl − versus HCO 3 − secretion is important as one develops drugs to improve acid-base balance and epithelial repair in GI diseases (e.g., duodenal ulcer disease, cystic fibrosis, ulcerative colitis [UC]) without triggering excessive Cl − secretion that might induce the unwanted diarrheal side effects. While epithelial restitution plays a critical role in the healing process of intestinal mucosa (17), little is known about the role of ER in colonic mucosal healing. Diseases like UC, a global intestinal autoimmune disease with no available cure, require an ongoing process of healing from injury. The foundational target of UC therapies is to suppress the immune system, thereby leading to less autoreactivity. However, large amounts and long-term immunosuppressive therapy increases the risk for infections and cancers (50). Therapies that may promote mucosal healing without immunosuppression are enticing as stand-alone or adjunctive therapies. In addition to modulating epithelial ion transport, ERα, but not ERβ, activation promotes Ca 2+ -dependent cell proliferation and migration, leading to colonic epithelial restitution; a process that may be beneficial to colonic disease like UC. Moreover, we found that activation of ERα but not ERβ increases protein levels of cyclin D1, PCNA, and β-catenin, indicating increased proliferation and migration.
In conclusion, we demonstrate for the first time that E 2 modulates colonic epithelial ion transports and epithelial restitution via ER subtype-dependent mechanisms. Since ERα and ERβ subtypes usually play opposite roles in regulating colonic epithelial function, E 2 may coordinate different ER subtypes to orchestrate functional homeostasis of colonic epithelial cells. Disruption of ERα-and ERβ-coordinated gut homeostasis may have underpinnings behind sex-based differences in GI disease, especially those involving epithelial injury and repair. While this requires further study, we have provided new insights into the cellular mechanisms of E 2mediated colonic epithelial ion transports and epithelial repair via different ER subtypes.

Animal studies and ethics
All animal studies were approved by the Ethics Committee of the Qingdao University Medical College. All animal care and experimental procedures complied with the "Guide for the Care and Use of Laboratory Animals" published by the National Institutes of Health. Animal studies are reported in compliance with the ARRIVE guidelines (51). The C57BL/6 mice (6-8 weeks old; 18-22 g) were purchased from HFK Bioscience Co., Ltd. Animals were assigned randomly to different experimental groups. Randomization and singleblinding were used for the measurement.

Measurement of intestinal I sc and HCO 3 − secretion in Ussing chamber experiments
Ussing chamber experiments were performed as previously described (52,53). C57BL/6J mice were anesthetized by halothane, and the abdomen was opened with a midline incision. Duodenal and colonic tissues were removed, stripped of seromuscular layers, divided, and mounted in Ussing chambers (aperture area, 0. luminal HCO 3 − secretion is expressed as micromoles per square centimeter per hour. The transepithelial I scs were measured via an automatic voltage clamp, in which μA was used for the original recordings, but μA⋅cm -2 was used for summary data. After a 30 min basal period, inhibitor or control vehicle was added for another 30 min, followed by addition of stimulus to both sides of the tissue. Electrophysiological parameters and HCO 3 − secretion were then recorded for 60 min. The maximal volumes of concentrated stock solutions of all compounds added to 3 ml Ussing chamber solutions were less than 30 μl. All added compounds did not alter pH values in chamber solutions, which is consistent with other report (25). The concentrations of all pharmacological inhibitors used in the present study were based on their IC 50 and the data from others' reports.

Measurement of [Ca 2+ ] cyt by single-cell imaging
[Ca 2+ ] cyt imaging experiments were performed as previously described (52). Briefly, cells were grown on glass coverslips for 24 h and incubated with 5 μM fura-2/AM (Invitrogen) for 1 h in physiological salt solution (PSS) at 37 C humidified atmosphere containing 5% CO 2 in the dark and then washed with PSS for 20 min. Then, cells on coverslips were mounted in a standard perfusion chamber on the stage of an inverted fluorescence microscope (Leica). Fluorescence signals were imaged using an intensified CCD camera (ICCD200) attached to an inverted fluorescence microscope (Leica) and recorded with MetaFluor software (Universal Imaging Corporation). Images were acquired every 3 s. The dual wavelength excitation method for the measurement of fura-2 fluorescence was used. The excitation wavelengths were 340 and 380 nm, and the emitted fluorescence was collected at 510 nm. [Ca 2+ ] cyt was presented as fluorescence ratios (F340/F380) after background subtraction. The PSS contained the following: 140 mM Na + , 5 mM K + , 2 mM Ca 2+ , 147 mM Cl − , 10 mM Hepes and 10 mM glucose (pH 7.4). The 0 Ca 2+ solution (0 Ca 2+ ) contained the following: 140 mM Na + , 5 mM K + , 145 mM Cl − , 0.5 mM EGTA, 10 mM Hepes, and 10 mM glucose (pH 7.4). The osmolality for the solution was 300 mosmol . kg −1 of H 2 O.

Western blotting
Western blotting was performed as previously described (52,54). Briefly, the cells and tissues were harvested and lysed using RIPA lysis buffer. The protein sample were separated using 4 to 20% SDS-PAGE and transferred to polyvinylidene difluoride membrane. Following blocking with 5% nonfat milk for 2 h at room temperature, the membranes were incubated with the   . ab29, Abcam). Then, the membranes were washed with Tris-buffered saline with 0.1% Tween 20 detergent three times and incubated with corresponding secondary antibodies for 2 h at room temperature. The signals were visualized using enhanced chemiluminescence (Millipore) in an ImageQuant LAS 400 digital biomolecular imaging system. Each experiment was repeated three times. The gray value of the bands was measured by ImageJ software for statistics.

Cell proliferation and scratch assays
Cell proliferation assay was performed as previously described (56). Briefly, cells were plated in 96-well plates. After 24 h, medium was replaced with medium containing different drugs. CCK-8 reagent (Cat. No. C0038, Beyotime Biotechnology) was added to each well at 0.5 to 2 h before the endpoint of incubation. A microplate reader (Thermo Fisher Scientific) was used to quantify viable cells by measuring the absorbance at 450 nm. Cell scratch assay was performed as previously described (57). After scratching, cell monolayers were gently washed to remove detached cells and replenished with serum free medium (with or without drugs) to inhibit cell proliferation. Images were obtained at 0 and 6 h postscratch. Experiments were repeated at least three times.
was dissolved in anhydrous alcohol. All salts were supplied by Sangon Biotech and dissolved in ultrapure water.

Data and statistical analysis
GraphPad Prism 7.0 (RRID: SCR_002798, USA) software was used for analysis and graph generation. All results shown are means ± SD. All experiments were repeated at least three times. The number of biological repeats (n) in the figures is the number of individual tissues or cells obtained from at least three mice or three independent experiments. The statistical significance of differences in the means of experimental groups was determined using Student's t test or one-way ANOVA followed by post hoc test for multiple pairwise comparisons. Significant differences (*p < 0.05) are expressed in the figures and figure legends.

Data availability
The data supporting the findings of this study are available within the article. Conflicts of interest-The authors declare no conflicts of interest with the content of this article.