Estrogen stimulates female biliary epithelial cell interleukin-6 expression in mice and humans


  • Kumiko Isse,

    1. Department of Pathology, Division of Transplantation, and Thomas E. Starzl Transplantation Institute, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • Susan M. Specht,

    1. Department of Pathology, Division of Transplantation, and Thomas E. Starzl Transplantation Institute, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • John G. Lunz III,

    1. Department of Pathology, Division of Transplantation, and Thomas E. Starzl Transplantation Institute, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • Liang-I Kang,

    1. Department of Pathology, Division of Transplantation, and Thomas E. Starzl Transplantation Institute, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • Yoshiaki Mizuguchi,

    1. Department of Pathology, Division of Transplantation, and Thomas E. Starzl Transplantation Institute, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • Anthony J. Demetris

    Corresponding author
    1. Department of Pathology, Division of Transplantation, and Thomas E. Starzl Transplantation Institute, University of Pittsburgh Medical Center, Pittsburgh, PA
    • University of Pittsburgh Medical Center, UPMC Montefiore, Room E741, 200 Lothrop Street, Pittsburgh, PA 15213
    Search for more papers by this author
    • fax: 412-647-2084

  • Potential conflict of interest: Nothing to report.


Females are more susceptible than males to several biliary tract diseases. Interleukin-6 (IL-6) is critical to triggering autoimmune reactions and contributes substantially to biliary epithelial cell (BEC) barrier function and wound repair, and estrogen differentially regulates IL-6 expression in various cell types. We hypothesized that estrogen might stimulate BEC IL-6 production. Exposure to physiologic levels of estradiol, in vitro, increased female mouse BEC (mBEC) IL-6 messenger RNA (mRNA) and protein expression, but either inhibited or had no effect on male mBECs. Female mBECs expressed higher concentrations of estrogen receptor-alpha (ERα) mRNA and protein and were also more dependent on estradiol for survival, in vitro. In vivo, elevated estrogen during estrous cycling in mice, and estrogen treatment of mice harboring an ERα+ human cholangiocarcinoma resulted in increased BEC IL-6 mRNA and tumor viability, respectively. Both responses could be blocked by an ERα antagonist. Human cholangiocarcinoma cell lines differentially expressing ERα were treated with specific ERα and ERβ agonists/antagonists to further test the relationship between estrogen stimulation, ERα expression, and IL-6 production. Results show that ERα, and not the underlying BEC sex, was responsible for estrogen-induced IL-6 production. Estrogen-induced proliferation of ERα-expressing cholangiocarcinoma was blocked by anti–IL-6 antibodies, indicating that at least some of the estrogen-trophic effects are mediated via IL-6. Finally, an association between ERα, IL-6, and phosphorylated signal transducer and activator of transcription 3 (pSTAT3) signaling was shown in female-predominant polycystic livers using immunohistochemical analyses, including multiplex quantum dot labeling. Conclusion: Estrogens stimulate IL-6 production in non-neoplastic female BECs and in neoplastic BECs expressing ERα. An association between these signaling pathways was demonstrated for female-predominant polycystic livers and might also influence autoimmune hepatitis, primary biliary cirrhosis, and cholangiocarcinogenesis. (HEPATOLOGY 2010.)

Estrogens promote female reproductive organ development and function, but estrogen receptors (ERs) are also found, at lower levels, in the skin,1 intestine,2 brain,3 and liver4 where they exert significant influence over diverse aspects of cellular physiology. For example, in the skin, the transition from cyclical estrogen fluctuations during reproductive life to an estrogen deficiency after menopause is associated with dryness, atrophy, fine wrinkling, and poor wound healing.5

Estrogens also intimately regulate interleukin-6 (IL-6) expression in various cell types.6 IL-6 is also critical to epithelial barrier function and wound healing in the skin7 and gastrointestinal8 and biliary tracts.9 For example, estrogens inhibit macrophage IL-6 production, which in turn, maintains serum IL-6 levels at relatively low levels during reproductive years.10 Menopausal loss of estrogens elevates serum IL-6 levels, which in turn, stimulates osteoclastic bone resorption.6, 11 IL-6−/− mice, however, are resistant to the osteopenic complications of estrogen deficiency.12 In contrast, estrogen stimulates IL-6 production in human ovarian epithelial cells and ovarian cancer cells.13 Promoter complexity controlling IL-6 gene expression14 and complexity of estrogen signaling15, 16 contribute to the tissue-specific regulation of IL-6 expression by estrogens.

Estrogens influence biliary tract pathophysiology.17 Females are significantly more susceptible than males to several chronic liver diseases that involve either the biliary tree and/or are influenced by IL-6 expression. Included in these chronic diseases are: (1) primary biliary cirrhosis (PBC),18 a disease associated with variations of IL-6 production19, 20; (2) debilitating/symptomatic adult polycystic liver disease (PCL) requiring liver transplantation,17 in which cyst fluid contains high levels of IL-621; and (3) autoimmune hepatitis, which requires IL-6 production to sustain T helper 17 (TH17)-type T lymphocytes that are critical to disease development.22

These observations raise the hypothesis that estrogens might influence biliary epithelial cell (BEC) IL-6 expression and thereby affect barrier epithelial function, wound repair, and peribiliary or portal tract immune responses. Using primary cultures of non-neoplastic mouse BECs (mBECs) and two human cholangiocarcinoma cell lines, we show that estrogens can stimulate BEC IL-6 production, but only in female or ERα-expressing neoplastic BECs. Estrogen-induced BEC IL-6 production, in turn, is related to ERα expression, which is higher in female than male BECs. Consequently, female BECs are more dependent on the trophic influences of estrogen for continued survival in vitro, and estrogen-induced stimulation of BEC growth can be inhibited by anti–IL-6 antibodies. Induction of ovulation in female mice showed that estrous, but not anestrous, female BECs produce IL-6 messenger RNA (mRNA) in vivo. Finally, clinical relevance was illustrated by showing a spatial-temporal relationship between ERα and IL-6/glycoprotein 130 (gp130) signaling in cystic BECs from adult polycystic liver disease.


BEC, biliary epithelial cell; C-DMEM, complete Dulbecco's modified Eagle's medium; C-SFM, complete serum-free medium; ELISA, enzyme-linked immunosorbent assay; ER, estrogen receptor; IL, interleukin; LPS, lipopolysaccharide; mRNA, messenger RNA; PBC, primary biliary cirrhosis; PCL, polycystic liver; PSLD, protected least significant difference test; pSTAT3, phosphorylated signal transducer and activator of transcription 3; RT-PCR, reverse transcription polymerase chain reaction; S-SFM, simple serum-free medium; TFF1, trefoil family factor 1.

Materials and Methods

Additional experimental procedures are described in the Supporting Materials.


Male and female IL-6−/− and corresponding wild-type littermates (8-12 weeks old) from C57BL/6 and a mixed predominant C57BL/6 strain23 were used for in vitro assays. Nonobese diabetic NOD.CB17-Prkdcscid/J (severe combined immunodeficient) mice (5-8 weeks old) were used for in vivo tumor studies. The mice were bred and maintained in the University of Pittsburgh animal facility, and all procedures were performed in compliance with Institutional Animal Care and Use Committee protocols #0701830-1 and #0803253A-1.

Primary Mouse BEC and Macrophage Cultures.

Primary mBEC cultures were prepared over a 3-week period as previously described.24 The media was changed to simple serum-free medium (S-SFM)24 for 24 hours, and cells were treated with 17β-estradiol (2-20,000 pg/mL) (Sigma-Aldrich, St. Louis, MO) or vehicle control in fresh S-SFM for 48 hours. The 200 pg/mL 17β-estradiol resulted in peak IL-6 mRNA production. Media containing forskolin (complete SFM [C-SFM])24 was used as a positive control for IL-6. BECs were then collected, seeded onto collagen-coated wells, and incubated for 24 hours in complete Dulbecco's modified Eagle medium (C-DMEM).24 Peritoneal macrophages were collected and seeded in Roswell Park Memorial Insitute 1640 medium (RPMI-1640; Sigma) with 2 mM L-glutamine, 5% fetal bovine serum, and gentamicin. Following macrophage attachment (30 minutes; 37°C), nonadherent cells were removed by washing. Macrophages were treated with lipopolysaccharide (LPS; 1, 10, 100 ng/mL; (Sigma) for 1 hour before adding estradiol (200 pg/mL) or vehicle.

Human Cell Lines.

Conditions for growth of cholangiocarcinoma cell lines SG231 and HuCCT-1 are described in the Supporting Materials. MCF7 breast carcinoma cells were the positive control for estrogen receptor expression.

Quantitative Real-Time Reverse Transcription Polymerase Chain Reaction.

Primers used for real-time reverse transcription polymerase chain reaction (RT-PCR) are shown in Table 1. See Supporting Materials for details.

Table 1. List of Primers Used in Real-Time PCR
Gene Name SequenceReal-Time PCR Product Size
  1. bp, base pair; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; h, human; m, mouse.


Protein Analysis.

Details for western blotting and enzyme-linked immunosorbent assay (ELISA) are in the Supporting Materials.

BEC IL-6 Expression in Mice In Vivo.

The estrous phase of cycling was induced by cohabitation of male and female mice for a period that incorporated two 12-hour nights and 2-3 hours into the third night. At this time, mice were sacrificed and tissues were harvested: (1) ovaries to confirm an estrogen surge and ovulation/follicle maturation (histological verification [data not shown]); (2) bile for IL-6 protein; and (3) the extrahepatic bile duct for BEC IL-6 mRNA. The extrahepatic bile duct was isolated and opened, BECs were scraped from the surface, and RNA was extracted immediately.

Cell Proliferation and Death.

The mBEC and SG231 cells were cultured for 2 days in growth media and changed to serum-free media (SFM) 24 hours prior to stimulation with estradiol in SFM. Cell counts/viability were measured using trypan blue exclusion. Goat anti-human IL-6 neutralizing antibody (1 μg/mL) or normal goat immunoglobulin G in 0.1% bovine serum albumin (R&D Systems, Minneapolis, MN) were incubated with SG231 cultures 1 hour after 20,000 pg/mL estradiol or vehicle stimulation.

Agonist and Antagonist Treatment.

The selective ERα agonist 4,4′,4″-(4-propyl-[1H]pyrazole-1,3,5-triyl)tris-phenol (PPT; 10 nM) and the selective ERβ agonist 2,3-bis(4-hydroxyphenyl)-propionitrile (DPN; 1 nM) (Tocris Bioscience, Ellisville, MO) were added to 24 hours serum-starved SG231 cells in parallel with 200 pg/mL estradiol or vehicle. Time points which gave maximum effect were 48 hours for estradiol and DPN, and 72 hours for PPT.

The estrogen antagonist fulvestrant (ICI 182,780; (Sigma) or vehicle was added to 24 hours serum-free SG231 cells (86 μM) 1 hour before stimulation with estradiol (200 pg/mL).


Formalin-fixed, paraffin-embedded sections (4 μm) were deparaffinized and underwent antigen unmasking, blocking of endogenous avidin/biotin, and incubation with primary antibodies (Table 2). Eight normal and 15 human adult PCLs were used for cytokine and growth factor staining. The patients with PCL included six premenopausal (five of six with known kidney involvement); six postmenopausal females (three of six with known kidney involvement); and three males (one of three with known kidney involvement). Tissues were selected from our files in accordance with Institutional Review Board protocols 9507150-980 and 0404010. For multispectral staining (Fig. 6A), biotinylated secondary antibodies were applied followed by streptavidin-conjugated quantum dots (Invitrogen, Carlsbad, CA).25 Images were taken using Nuance microscopy (CRi, Woburn, MA). The phosphorylated signal transducer and activator of transcription 3 (pSTAT3) specificity was confirmed by blocking with a recombinant peptide (Cell Signaling Technology, Danvers, MA) (Fig. 6B).

Table 2. List of Primary Antibodies Used
  1. ER, estrogen receptor; IGF, insulin-like growth factor; IL, interleukin; EGFR, endothelial growth factor receptor; ErbB2, erythroblastic leukemia viral oncogene homolog; pSTAT3, phosphorylated signal transducer and activator of transcription 3; VEGF, vascular endothelial growth factor.

ERαPolyclonal1:200Santa Cruz Biotechnology
ERβPolyclonal1:100Santa Cruz Biotechnology
EGFRPolyclonal1:250Santa Cruz Biotechnology
ErbB2Polyclonal1:50Cell Signaling Technology
Flk-1A-31:50Santa Cruz Biotechnology
IGF-1Polyclonal1:50Santa Cruz Biotechnology
pIGF-1RPolyclonal1:50Santa Cruz Biotechnology
IL-6Polyclonal1:100Santa Cruz Biotechnology
pSTAT3Polyclonal1:40Cell Signaling Technology
VEGFPolyclonal1:50R&D Systems

Mouse pSTAT3 staining was developed using a catalyzed signal amplification (CSA) system (DAKO, Carpinteria, CA).

In Vivo Tumor Studies.

Induction of SG231 tumors and animal treatments are described in the Supporting Materials.

Statistical Analysis.

All in vitro experiments were performed in triplicate and repeated ≥3 times. Values shown are the mean ± standard deviation of ≥3 experiments. In vivo analyses were done using ≥3 animals per group. A Student t test, Welch's t test, or Mann-Whitney U test was used to determine significance between two groups. A one-factor analysis of variance was used for intergroup comparisons and significance within individual study groups determined with a Fisher's protected least significant difference test (Fisher's PSLD). A forward stepwise regression test was used for correlations between pSTAT3 and growth factors/cytokines in stained tissues. For all, a P value of <0.05 was considered significant. Data was analyzed using Statcel (OMS Publishing Inc., Saitama, Japan).


Gender Differences in BECs, but not Macrophages, in Response to Estrogen Treatment.

Estrogen17 and IL-6 are known to influence BEC physiology,9 and estrogens regulate IL-6 expression in other cell types. Therefore, we tested whether increasing concentrations of estradiol affected BEC IL-6 mRNA and protein expression, and whether male and female BECs responded differently. The estrogen concentrations used correspond roughly to those in normal male or postmenopausal female (2 pg/mL), normal premenopausal female (200 pg/mL), and pregnant female (20,000 pg/mL). Treatment of male mBECs with increasing concentrations of 17β-estradiol for 48 hours either caused no significant change or inhibited IL-6 mRNA and protein expression compared to the vehicle control (Fig. 1). In contrast, the same treatment of female mBECs resulted in significantly increased IL-6 mRNA and protein expression with maximum induction at 200 pg/mL. Media containing forskolin (C-SFM), which stimulates BEC IL-6 production, was used as a positive control (Fig. 1B). Consistent with previous studies,10 estrogen treatment of male and female peritoneal macrophages inhibited LPS-induced IL-6 expression, as expected (Fig. 1C).

Figure 1.

Male and female mBECs (3 × 105 cells/24-well plate) were treated for 48 hours in S-SFM with estradiol at concentrations that correspond roughly to normal male or postmenopausal female (2 pg/mL), normal premenopausal female (200 pg/mL), and pregnant female (20,000 pg/mL). Vehicle control; 2 × 10−5 % ethanol (A) IL-6 mRNA measured by real-time PCR showed estradiol treatment either had no effect or significantly inhibited male but stimulated female mBEC IL-6 mRNA expression (n = 3). (B) IL-6 protein was measured by ELISA at 48 hours after treatment with 200 pg/mL estradiol. C-SFM containing forskolin was used as a positive control. Estradiol (200 pg/mL) stimulated female, but inhibited male mBEC IL-6. Female mBECs also released more IL-6 protein than did male mBECs after treatment with forskolin (n = 5). (C) Peritoneal macrophages (4 × 105 cells) were seeded in RPMI-based media and treated with LPS for 1 hour before stimulated with estradiol or vehicle control. Similar to previous studies,10 estradiol significantly inhibited LPS-induced IL-6 protein production by male and female macrophages as measured by ELISA (n = 4). (D) Treatment of IL-6–deficient mBECs with 200 pg/mL estradiol resulted in increased expression of the estrogen-responsive gene TFF1 mRNA by real-time PCR only in female mBECs (n = 3). Therefore, differential regulation of gene expression by estrogen in male and female mBECs is not restricted to IL-6 expression. Mean ± standard deviation; *P < 0.05; Fisher's PSLD test in (A), (B), and (C); Welch's t test in (D)

We next examined expression of another well-known estrogen-responsive gene, trefoil family factor 1 (TFF1), to determine whether the sex difference in mBEC estrogen responsiveness was specific for IL-6 production (Fig. 1D). Because TFF1 expression can also be induced by IL-6, we used IL-6−/− mBECs. The results show that estradiol increased TFF1 mRNA expression in female, but not in male, IL-6−/− mBECs. This shows that male mBECs also respond differently than female mBECs to estrogen stimulation when another estrogen-responsive gene is used as the read-out.

Estrogen Receptor Expression on mBECs In Vitro.

After ligand binding, ERα and ERβ form homodimers or heterodimers, interact with basal transcription factors and numerous coactivators, and bind to estrogen-responsive elements to influence transcription.16, 26 ERs can also modulate target gene expression by interacting indirectly with activator protein 1 (AP-1), Sp1, or cyclic AMP responsive element (CRE) to modulate expression in a tissue-specific manner. The IL-6 promoter has CRE and AP-1 binding sites.27 When expressed together, ERβ generally inhibits gene activation by ERα.16, 26 We hypothesized, therefore, that differential ERα and ERβ expression between male and female mBECs might account for the sex differential regulation of IL-6 expression by estrogens. To determine basal ER expression, mBECs were cultured for 3 weeks in media without exogenous estrogen (C-DMEM) and then acclimated to S-SFM for 24 hours. Results showed that female mBECs express significantly higher ERα mRNA and protein than do male mBECs, but there was no significant difference in ERβ mRNA or protein expression (Fig. 2A,B). ERα mRNA and protein expression were maintained even in the absence of exogenous estrogen.

Figure 2.

Basal ERα and ERβ mRNA (A) and protein (B) expression was determined for male and female mBECs cultured in non–estrogen containing media. Female mBECs expressed more ERα mRNA and protein than did male mBECs, but there were no differences in ERβ mRNA or protein. ERα and ERβ expression was then measured after short-term (48 hours) exposure to (C) 200 pg/mL estradiol or (D) continuous estradiol exposure throughout the 3-4 week culturing period (added every 48 hours). In male mBECs, ERα mRNA (C) increased after 48 hours; ERα protein expression also increased, but the difference was statistically insignificant (data not shown). However, ERα mRNA and protein significantly increased in male mBECs after chronic estrogen exposure (D). ERβ mRNA and protein did not change in male mBECs with either 48 hours or continuous exposure to estradiol. Neither ERα nor ERβ mRNA or protein expression levels changed significantly in female mBECs, regardless of the length of estradiol exposure (n ≥ 3). Mean ± standard deviation; *P < 0.05; Welch's t test in (A), (B), and (C); Fisher's PSLD test in (D).

We next determined whether short-term (48 hours) or chronic long-term (added every 48 hours over 4 weeks) estradiol exposure changed ERα/ERβ expression when compared to vehicle controls. Male mBECs showed increased ERα mRNA and protein after 48 hours exposure to estrogen, but the difference was statistically significant only for mRNA (Fig. 2C). Chronic estrogen exposure, however, significantly increased ERα mRNA and protein in male mBECs. Estradiol did not alter ERβ expression in male mBECs, regardless of the length of exposure (Fig. 2D). In contrast, female mBECs showed a tendency for decreased ERα/ERβ mRNA and protein expression after both short-term and chronic estradiol exposure, but the difference was statistically insignificant (Fig. 2C,D).

Female, but not Male, BECs are Dependent on Estradiol for Survival, In Vitro and In Vivo.

Because of the sex differences in BEC ERα expression, and the positive growth modulating influence of ERα,17 we hypothesized that survival of female mBECs would show more estrogen-dependence than male mBECs. Therefore, male and female mBECs were propagated in the absence of estrogen. Estradiol or vehicle was added for the final 48 hours of culture and the number of viable and nonviable mBECs were counted. As expected, female, but not male, BECs were dependent on environmental estrogens for a sustained level of viability compared to the vehicle controls (Fig. 3A–B).

Figure 3.

In vivo and in vitro dependence of ERα-expressing BECs on estrogen for viability. (A-B) Male and female mBECs were cultured without exposure to estrogen, then switched to S-SFM for 24 hours (0 hour) and estradiol or vehicle for the final 48 hours. Calculating percent live (A) or percent dead (B) cells by trypan blue exclusion shows that female, but not male, mBECs were significantly dependent on estradiol in the medium (n = 4). (C-G) In vivo effects of estrogen treatment on subcutaneous tumors from an ERα-expressing cholangiocarcinoma cell line (SG231). Mice bearing SG231 tumors were treated with 100 μL of 17β-estradiol (100 μg/kg), 17β-estradiol (100 μg/kg) + fulvestrant (10 mg/kg), or vehicle (10 μL dimethyl sulfoxide + 90 μL corn oil) for 3 weeks (daily subcutaneous injections). (C) No significant difference was seen in gross tumor size among the treatment groups. (D-F) Hematoxylin & eosin–stained tumor sections showed: (D) estrogen significantly reduced the percentage of apoptotic cells; (E) mitotic counts were highest in the controls, probably driven at least in part by the increased apoptosis and necrosis. (F) estrogen also significantly reduced the necrotic area of the tumor when compared to the control, and fulvestrant partially reversed the protective effect of estrogen. (G) Real-time PCR analysis shows increased IL-6 expression in the estrogen-treated tumors and this induction was blocked with fulvestrant. Percentage of cells were calculated from five high-power fields (HPF), 400×, and percent area was calculated using ImageJ software. Mean ± standard deviation; *P < 0.05; Fisher's PSLD test (A-G).

To verify the dependence of ERα-expressing BECs on estrogen, we used ERα-positive SG231 cells in a mouse tumor model. Treatment of mice harboring SG231 subcutaneous tumors showed that estrogen aided cell viability by yielding less apoptosis, less necrosis, and increased IL-6 expression in the tumors. The slightly larger, but not significant, tumor size and increased mitotic response of control tumors is likely a compensatory mechanism driven by increased necrosis in this group (Fig. 3C–G).

BEC IL-6 Expression in Male and Estrous and Anestrous Female Mice, In Vivo, and pSTAT3 Expression.

We next determined whether high estrogen levels in vivo during the estrous cycle stimulated BEC IL-6 expression compared to anestrous mice. Estrous cycling was induced and the mice were sacrificed for tissue analysis. Histologic examination of the ovaries and serum estradiol concentrations confirmed follicle maturation and elevated estrogen levels, respectively (data not shown). BECs gently scraped from the opened surface of the common bile duct (Fig. 4A) showed significantly higher IL-6 mRNA levels in estrous mice compared to male and anestrous mice (Fig. 4B). Verification that the RNA was obtained from the BECs was accomplished through histology (Fig. 4A) and real-time PCR for cytokeratin-19 (data not shown). IL-6 protein in the gallbladder bile was significantly higher in male and estrous mice than anestrous females (Fig. 4C), but neither males nor anestrous females showed BEC IL-6 mRNA expression. This suggests that bile IL-6 in males is derived from the liver and/or the peripheral circulation28 whereas the BECs make a larger contribution in estrous female mice.

Figure 4.

(A) Common bile ducts of male and estrous and anestrous female mice were opened showing an intact BEC lining (arrows) before (left frame), but not after scraping (right frame) (arrowheads) for RT-PCR. (B) Estrous female mBECs showed easily detectable IL-6 mRNA, which was absent in male and anestrous mBECs. (C) Gallbladder bile IL-6 protein was detected in all groups, but levels in male and estrous female mice were higher than bile from anestrous female mice. (D,E) IL-6 signaling stimulates nuclear translocation of pSTAT3. Immunohistochemical staining of intrahepatic bile ducts showed significantly increased nuclear pSTAT3 in BECs lining the intrahepatic bile ducts of estrous compared to anestrous female mice (n = 3; 5 bile ducts/n). Mean ± standard deviation; *P < 0.05; Fisher's PSLD test in (C); Welch's t test in (D)

Because pSTAT3 is a downstream signal of IL-6 stimulation in BECs, we compared intrahepatic BEC nuclear pSTAT3 expression by immunohistochemistry between estrous and anestrous female mice. Results showed that estrous female mBECs have increased pSTAT3 compared to anestrous mice (Fig. 4D–E).

Human BEC ERα and ERβ Expression and IL-6 Production in Response to Estrogen Stimulation.

Because ERα has been most closely linked with a positive modulatory effect on BEC physiology,17 we hypothesized that ERα expression, and not the underlying sex, was responsible for the differential BEC response to estrogen stimulation. Unable to sufficiently knock-out/knock-in protein expression in primary mBECs with transfection reagents, we decided to test this hypothesis using two male-derived cholangiocarcinoma cell lines that differed in ERα expression. SG231 cells strongly express ERα mRNA and protein, similar to female BECs and the positive control MCF7 cells. The HuCCT-1 cell line expresses ERα mRNA, but no ERα protein, making it an ideal model for testing the importance of ERα in estrogen-induced IL-6 signaling (Fig. 5A). ERβ mRNA and protein levels were similar between the two cell lines. Because HuCCT-1 is devoid of ERα protein, estradiol can only signal through ERβ. Figure 5B shows that ERα protein expression was tightly linked to the ability of estrogen to stimulate BEC IL-6 mRNA and protein. Estradiol treatment for 48 hours increased IL-6 mRNA production in SG231 cells, but either inhibited or had no effect on HuCCT-1 IL-6 production. The reduction of IL-6 in SG231 cells after high-dose estradiol (20,000 pg/mL) is likely due to IL-6 feedback inhibition through IL-6 or ERα expression pathways.

Figure 5.

(A) Two male human cholangiocarcinoma cell lines, SG231 and HuCCT-1, were cultured in serum-free media for 24 hours and harvested to examine ERα and ERβ mRNA (left) and protein expression (right) by RT-PCR and western blotting, respectively. MCF-7 was used as a positive control. SG231 showed strong ERα and ERβ protein expression, whereas HuCCT-1 expressed ERα mRNA, but was negative for ERα, protein. All cell lines expressed ERβ mRNA and protein. (B) Estradiol stimulated SG231 IL-6 mRNA production in a concentration-dependent fashion, but IL-6 mRNA expression in HuCCT-1 cells was either inhibited or unchanged. (C) Estradiol and the specific ERα agonist PPT increased whereas the ERβ agonist DPN inhibited BEC IL-6 mRNA expression in SG231 cells (n ≥ 3). (D) Estradiol, PPT, and DPN treatment of SG231 yielded the same results for IL-6 protein (n ≥ 3). (E) The ERα antagonist fulvestrant (86 μM), given 1 hour before estradiol stimulation for 48 hours, prevented estradiol-stimulated IL-6 mRNA expression by SG231. (F) Estradiol-stimulated SG231 proliferation and the increased proliferation was blocked by anti–IL-6 neutralizing antibody

If ERα protein expression determines whether BECs respond to estrogen with IL-6 production, then the selective ERα agonist PPT should also increase IL-6 mRNA and protein production. In contrast, the specific ERβ agonist DPN should have the opposite effect because ERβ activation generally inhibits gene activation by ERα.16, 26 Furthermore, fulvestrant, a specific ERα antagonist, which decreases ERα protein expression by accelerating proteosomal degradation,16 should prevent estrogen-induced BEC IL-6 expression in SG231 cells. The results were as expected (Fig. 5C–E).

Because estrogen and IL-6 promote the growth/survival of normal cholangiocytes17, 29 and some cholangiocarcinomas24, 30 and we have shown that estrogens stimulate BEC IL-6 production, we hypothesized that the trophic influence of estrogens on BECs might, at least in part, be mediated by IL-6. The estrogen-responsive BEC line SG231 was treated with estradiol in the presence and absence of anti–IL-6 blocking antibody. The results show that anti–IL-6 neutralizing antibodies significantly inhibit estradiol-induced BEC proliferation (Fig. 5F).

Immunohistochemical Localization of ERα Expression in Polycystic Livers and Relationship to IL-6 Signaling Pathways.

Consistent with previous studies, BECs in normal male and female human livers were negative for ERα by immunohistochemistry, whereas cystic BECs in polycystic livers showed focally strong nuclear ERα protein expression.31 Our goal was to extend these observations using multiplex quantum dot labeling to search for an association between ERα expression and IL-6 expression and evidence of gp130 downstream signaling in a clinically relevant situation in human BECs in vivo. Results showed nuclear ERα expression was associated with cytoplasmic IL-6 protein expression and evidence of downstream gp130 signaling, manifest as nuclear pSTAT3 (Fig. 6A). Cystic BEC pSTAT3 expression was entirely blocked with recombinant pSTAT3 peptide (Fig. 6B). BEC from the normal liver (Fig. 6A), in contrast, were negative for IL-6 expression and showed only rare pSTAT3-positive BECs; occasional portal-based periductal inflammatory cells expressed cytoplasmic IL-6 protein and pSTAT3 (Fig. 6).

Figure 6.

Expression of growth factors and cytokines in polycystic livers (PCL). (A) Multiplex Quantum dot staining of PCL and normal liver. Green: cytokeratin 19 (CK19); Red: IL-6; Pink: ERα; Cyan: pSTAT3; Blue: DAPI. In PCL, cystic BECs expressed nuclear ERα and pSTAT3 and cytoplasmic IL-6 protein and cytokeratin-19 (CK19). In normal liver, BECs are negative for ERα, IL-6, and pSTAT3, but some inflammatory cells near the bile ducts showed cytoplasmic IL-6/pSTAT3 (n = 3 normal livers; n = 3 PCL). (B) Excess recombinant pSTAT3 peptide (50 ng/mL) incubated with anti-pSTAT3 antibody before and at the time of staining confirmed the specificity of the pSTAT3 staining on serial sections: upper, pSTAT3 antibody only; lower, pSTAT3 antibody blocked with pSTAT3 peptide. (C) Immunohistochemical analysis of premenopausal (<45 years; range 36-44; average 40.8; n = 6), postmenopausal (>55 years; range 55-60; average 59.8; n = 6) and male (n = 3) PCL showed a significant correlation with menopausal status for ERα and pSTAT3. (D) All PCL (n = 15) were sorted on the basis of low and high BEC pSTAT3 expression (threshold at 25% positive pSTAT3/BEC) and analyzed by stepwise regression. Only IL-6 showed a significant correlation with pSTAT3 expression (mean ± standard deviation; R2 = 0.44; P = 0.025). IL-6, VEGF, Flk-1, IGF-1, EGFR, and HER2/ErbB-2 were scored as: 0 = negative; 1 = weakly and partially positive; 2 = moderately and mostly positive; or 3 = strongly and diffusely positive. ERα, pSTAT3, and pIGF-1R were scored as percent frequency of positive cells/total BECs. Five high-power fields (HPF; 400×) were examined for each case.

To determine the significance of estrogen signaling in PCL, we compared cyst BECs from three males and six premenopausal (<45 years) and six postmenopausal (>55 years) women for immunohistochemical expression of ERα, IL-6, pSTAT3, and a variety of other growth factor and receptor proteins (vascular endothelial growth factor [VEGF], Flk-1, insulin-like growth factor 1 [IGF-1], phosphorylated IGF-1 receptor [pIGF-1R], epidermal growth factor receptor [EGFR], erythroblastic leukemia viral oncogene homolog 2 [Her2/ErbB-2]) which are known to be expressed on PCL BECs or in the cyst fluid. ERα and pSTAT3 showed a significant correlation to menopausal state. Male pSTAT3 was comparable to that of premenopausal women and likely influenced by other environmental factors (androgens and testosterone) or IGF-1/pIGF-1R, whose expression was slightly higher in male PCL (data not shown) (Fig. 6C). Although it is not surprising that ERα was higher in the premenopausal group, the correlation between pSTAT3 and menopausal status implicates ERα signaling in disease progression. Because many of the factors tested might influence IL-6 or manifest downstream signaling as pSTAT3, we analyzed PCL on the basis of pSTAT3 high and low expression. The only significant relationship with pSTAT3 expression was IL-6 (Fig. 6D).


Our results on differential regulation of BEC IL-6 mRNA and protein expression by ERα according to sex are consistent with previous studies showing that: (1) ERα expression is complex and regulated at the level of transcription, translation, and protein degradation by the ubiquitin-proteasome pathway32; (2) ERβ generally blocks or significantly reduces gene activation mediated by ERα16, 26; and (3) ERα is most closely correlated with a positive modulatory effect on BEC physiology.17 The following observations support these conclusions. Non-neoplastic female BECs and the male BEC cell line SG231 express significantly more ERα than non-neoplastic male BECs and the HuCCT-1 cell line. Also, the higher level of ERα expression, not the sex, correlates with estrogen responsiveness. In addition, the selective ERα agonist PPT stimulates BEC IL-6 production, whereas the ERα antagonist fulvestrant16 and the selective ERβ agonist DPN inhibit the estrogen-induced BEC IL-6 production. Conversely, increasing ERα protein in non-neoplastic male BECs with chronic estrogen exposures enables estrogen-responsiveness in normally nonresponsive cells.

Persistent expression of ERα in female BECs, in the absence of an exogenous estrogen source, also renders them more vulnerable to estrogen deficiency in vitro. In vivo, estrogen enhanced tumor viability of ERα-positive cholangiocarcinoma cells by reducing apoptosis and necrosis and increasing IL-6 expression. Previous studies have shown similar results for non-neoplastic BECs.17

Regardless, we also showed that the trophic effect of estrogen on isolated BECs is mediated, at least in part, by stimulating IL-6 production. Alvaro et al. found a role for IGF-1 in estrogen-mediated BEC proliferation.31 The current study also showed IGF-1 and pIGF-1R expression in cystic BECs, which is not usually seen in normal BECs, but there was not a significant correlation with pSTAT3 signaling as seen with IL-6. Other factors, such as VEGF, Flk-1, EGFR, and HER2/erbB-2 were also investigated and some showed a trend, but the association was statistically significant only for pSTAT3 and IL-6. This is consistent with the observation that, in women, liver cysts frequently emerge at puberty and increase throughout the child-bearing years,33 suggesting that estrogen/IL-6/pSTAT3 signaling stimulates cyst growth.

Both in vitro and in vivo data showed that estrogen treatment reduced cell death and increased IL-6 expression in ERα-expressing BECs (Fig. 3). Because IL-6 is critical to BEC barrier function and wound healing,9 similar to the skin7 and gastrointestinal8 tracts, the estrogen deficiency of menopause and female-to-male liver transplants likely diminishes two critical trophic factors for BECs: direct estrogen stimulation17 and estrogen-induced IL-6 expression.9

Male BECs are able to adjust to an estrogen-rich environment, whereas female BECs still express higher ERα protein levels than male BECs in an estrogen-poor environment This result is consistent with the observation that ERα is normally expressed at low levels in some normal male human tissues.34 In addition, chronic estrogen therapy up-regulates ERα protein expression in tissues of human male-to-female transsexuals.35 This might help to explain the significantly lower survival of female-to-male liver allografts and why they experience a higher rate of biliary tract complications36 compared to other donor-recipient sex combinations.

A recent study in rodents showed that estrogens inhibit hepatocellular carcinogenesis by reducing Kupffer cell IL-6 expression,28 but opposite results were observed in humans, where higher serum IL-6 levels were an independent risk factor for hepatocellular carcinoma development in female but not male patients with chronic hepatitis C.37 We confirmed that estrogens inhibit macrophage IL-6 production in both sexes, but in BECs, estrogen-induced IL-6 expression is sex dependent. This might partially explain why the dramatic sex disparity for hepatocellular carcinoma does not exist for cholangiocarcinoma,38, 39 because STAT3 plays a critical role in tumor initiation and promotion.40 In fact, hepatocellular carcinoma is the most common primary liver neoplasm in males, whereas cholangiocarcinoma is the most common primary liver neoplasm in females.39 In addition, a majority of cholangiocarcinomas express ERα, regardless of sex, and preferential expression of ERα is associated with estradiol-induced cholangiocarcinoma growth.41 Our in vivo studies show that estrogen is involved in cell survival by inhibiting apoptosis and necrosis, which may have therapeutic implications for bile duct injury. In addition, fulvestrant significantly increased apoptosis and inhibited tumor growth, which might be a useful tool for cholangiocarcinomas and PCL disease.

Finally, this study provides some insights into BEC sex differences that could influence non-neoplastic disease pathogenesis. For example, liver cyst growth might be accelerated in females17 because of the estrogen-induced mitogenic influence of BEC IL-6 expression.29 We showed a temporal-spatial and statistical association among ERα, IL-6, and pSTAT3 signaling in cystic epithelium, consistent with previous studies showing increased IL-6 concentrations within cyst fluid.21 We also analyzed a variety of other factors associated with PCL. A significant relationship was found only between ERα, pSTAT3, and menopausal status and the strongest relationship with pSTAT3 levels was with IL-6. In women, liver cysts frequently emerge at puberty and continue to grow throughout the child-bearing years.33 This data suggests that a patient's menopausal status influences cyst enlargement via ERα/IL-6/pSTAT3 signaling, but we cannot exclude the contribution of other previously studied factors to cyst growth. We also showed that female mBEC IL-6 mRNA and bile IL-6 protein expression vary throughout the estrous cycle. IL-6 is required to sustain TH17-type T lymphocytes, but inhibits regulatory T cell production,22, 42 and is required for plasma cell differentiation.43 Therefore, it is not unreasonable to suggest that the differential hepatic IL-6 microenvironment that occurs as a consequence of BEC estrogen responsiveness might contribute to the pathogenesis of diseases such as PBC and autoimmune hepatitis, which are associated with TH17 autoimmunity44 and localized plasma cell differentiation.