11Beta-hydroxysteroid dehydrogenase-1 deficiency or inhibition enhances hepatic myofibroblast activation in murine liver fibrosis

MOUSE LIVER FIBROSIS MODELS

All experiments involving animals were approved by The University of Edinburgh Animal Welfare and Ethical Review Body and by the United Kingdom Home Office. Experiments used adult male (14 weeks of age) mice with global knockout (Hsd11b1^−/−; GKO).14 GKO mice have been backcrossed for over 10 generations on a C57Bl/6J genetic background, and C57Bl/6J mice were used as controls.15-18 GKO mice were maintained in parallel with the control C57Bl/6J mice; both lines were bred and maintained within our biomedical research facility, housed under standard conditions. To avoid interanimal variability that would be introduced due to differences in the stage of estrus in female mice, only male mice were used. Male mice (12 weeks of age) in which deletion was targeted to hepatocytes (liver-specific knockout [LKO]) were generated by crossing Albumin-Cre^Tg/+ mice (B6.Cg-Tg[Alb-cre]21Mgn/J; Jackson Laboratories) with mice homozygous for a “floxed” allele of Hsd11b1 (Hsd11b1^fl//fl) in which exon 3 is flanked by LoxP sites (generated by Artemis Pharmaceuticals directly onto a C57BL/6 background and designated Hsd11b1^tm1Arte, MGI 5784734). Cre-mediated excision of exon 3 generates a null allele.19 Cre-Hsd11b1^fl//fl littermates served as controls for LKO mice. To target deletion specifically at MFBs/HSCs (MFB/HSC-specific 11βHSD1 knockdown [MFKD]), Hsd11b1^fl//fl mice were crossed with Pdgfrb-Cre^Tg/+ mice.20 Cre-Hsd11b1^fl//fl littermates served as controls for MFKD mice.

CCl₄ MODEL

Mouse chronic liver fibrosis was induced by intraperitoneal injection of 25% CCl₄/g twice weekly for 12 weeks. Male GKO or LKO mice and their control littermates were culled at 24 hours (peak fibrosis), 72 hours, 1 week, and 1 month (scar resolution phases) after the last injection, as previously validated.3 MFKD male mice were culled 24 hours after the last CCl₄ injection to evaluate the role of 11βHSD1 deficiency at the peak fibrotic response. For acute injury, a single dose of 25% CCl₄/g intraperitoneally was given in GKO or control mice, and livers and plasma were collected after 24 hours.

In male C57Bl/6J mice, pharmacological 11βHSD1 inhibition was achieved by mixing a chow diet with 0.15% [4-(2-chlorophenyl-4-fluoro-1-piperidinyl][5-(1H-pyrazol-4-yl)-3-thienyl]-methanone (UE2316). Groups of C57Bl6/J mice were given either a chow diet or a diet containing UE2316 (UE group) ad libitum21 either throughout 12 weeks of CCl₄ administration and until sacrifice or only from 48 hours after the last CCl₄ injection until sacrifice, i.e., during resolution (UE-R group). Mice were sacrificed at 6 hours, 24 hours, 72 hours, and 8 days after the last CCl₄ injection.

ALTERNATIVE LIVER PATHOLOGY MODELS

An alternative model of liver fibrosis was induced in C57Bl/6J mice by administration of thioacetamide (TAA; 600 mg/L) in drinking water for 1 year. Livers were harvested 24 hours or 1 week after TAA termination. To induce steatosis and NASH in a model of NAFLD, GKO male mice and age-matched C57Bl/6J mice aged 6-10 months were fed commercially available diets (all from Dyets, Bethlehem, PA)—control diet (518574), choline-deficient diet (518753), or methionine/choline-deficient diet (MCDD, 518810)—for 2 weeks, as described.22

ISOLATION OF NONPARENCHYMAL LIVER CELLS AND FLOW CYTOMETRY

Fresh liver was perfused with 5 mL of saline through the portal vein at 100-300 mg, then digested with Roswell Park Memorial Institute 1640 medium with deoxyribonuclease I and collagenase IV at 37^oC. After digestion, hepatocytes were pelleted and discarded by centrifugation at 50g for 5 minutes. Nonparenchymal cells were washed with Roswell Park Memorial Institute 1640 medium and pelleted by centrifugation at 350g for 15 minutes, then washed and blocked with 10% mouse serum for 30 minutes. Antibodies against cluster of differentiation 11b (CD11b; clone M1/70; Ebioscience, Hatfield, UK), Ly-6C (clone HK1.4; Ebioscience), CD45.2 (clone 104; Ebioscience), F4/80 (dilution 1:50; clone BM8; Invitrogen), and Ly-6G (clone 1A8; Biolegend, San Diego, CA) were added for 30 minutes in a dilution of 1:100 unless otherwise specified. Cell viability was assessed with Fixable Viability Dye eFluor780 (Ebioscience) according to the manufacturer's protocols. Cells were formalin-fixed and analyzed by flow cytometry using a BD LSR Fortessa II (BD Bioscience, Oxford, UK).

IMMUNOFLUORESCENCE AND IMMUNOCHEMISTRY

For immunohistochemistry, livers were fixed with formalin for 16 hours directly after harvest. Sections, 4 μm, were dewaxed and rehydrated, followed by antigen retrieval in boiling sodium citrate. Avidin and biotin were blocked according to the manufacturer's protocol (Vector, Peterborough, UK). Protein block (Dako, Cambridge, UK) or serum was added, and sections were then incubated with alpha-smooth muscle actin (αSMA; Sigma-Aldrich, Dorset, UK), collagen I, or GR1 (Ly6G) (Cambridge Bioscience, Cambridge, UK), F4/80 (Abcam, Cambridge, UK), or reelin (Abcam) antibodies at 4^oC overnight and subsequent secondary antibody for 30 minutes. After ABC vector incubation, sections were incubated with 3,3'-diaminobenzidine for 10 minutes and counterstained with hematoxylin. Picrosirius red (PSR) staining was performed as described.3, 23 Thirty to 40 high-power fields (magnification ×80) per section were randomly selected for each slide by an assessor blind to genotype. Images were analyzed in Photoshop CS3 for positively stained pixels and normalized to the total number of pixels. In the period acid–Schiff stain, the necrotic cell area was quantified in a blinded manner using the Image J Trainable Weka Segmentation tool.

LIVER HISTOPATHOLOGY

Blinded hematoxylin and eosin–stained sections from the acute single-dose CCl₄ and the chronic CCl₄ injury models (at 24 hours peak fibrosis) were evaluated by a pathologist using two separate ordinal scales. Inflammation was scored using the inflammation component of the NAFLD activity scoring system for NASH24 (0, no inflammation; 1-<2, necroinflammatory foci/×20 field; 2-3, necroinflammatory foci/×20 field; 3->4, necroinflammatory foci/×20 field). Hepatocellular necrosis was scored using a scale based on the descriptors of severity used in the reporting of acute lobular hepatitis in human biopsies (0, none; 1, single-cell necrosis; 2, confluent necrosis; 3, zonal necrosis; 4, panacinar necrosis). The data are ordered categorically and, therefore, are presented as dot plots.

PRIMARY CULTURE OF MOUSE HSCs

Livers from GKO, C57BL/6J control, MFKD, and control littermate mice were perfused with 5 mL of Hank's balanced salt solution, excised, cut into 2 × 2 mm² cubes, and digested in Hank's balanced salt solution medium supplemented with deoxyribonuclease I, collagenase IV, and pronase. Released cells were purified through a 70-μm strainer and separated by density gradient centrifugation in sequential concentrations of OptiPrep (Sigma-Aldrich). Cells, 1 × 10⁶/well, were plated and cultured in Dulbecco's modified Eagle's medium with 16% fetal bovine serum and 1% penicillin/streptomycin. HSCs were left to spontaneously activate, then collected after 2, 5, 8, or 11 days in culture for mRNA and protein analysis. Pharmacological 11βHSD1 inhibition was performed in primary cultures of C57Bl/6J HSCs. HSCs were treated with vehicle, 500 nM corticosterone, 500 nM 11-dehydrocorticosterone, and/or 10 μM UE2316. HSC medium was replaced daily, and all treatments were added daily with the medium change, for the duration of the experiment. Following 8 days of treatment, HSCs were collected for mRNA and protein analysis.

OTHER STANDARD METHODS

Immunoblotting, RNA extraction, real-time PCR, 11βHSD1 activity assay, and in situ hybridization have been described25-27; and detailed methods can be found in the Supporting Information.

STATISTICAL ANALYSIS

All data are expressed as mean ± SEM. Statistical analysis was performed using GraphPad prism or Statistica 7. Two-way analysis of variance was used to test for the interaction of genotype/treatment analysis. Two-tailed unpaired Student t test or one-way analysis of variance was used for the comparison of two (i.e., control versus knockout mice) or three (i.e., control, chow diet, MCDD comparisons) groups, respectively.

11βHSD1 EXPRESSION IS REDUCED DURING HEPATIC INJURY

In the CCl₄ murine model of liver fibrosis, whole-liver 11βHSD1 mRNA (Fig. 1A) and protein (Fig. 1B) levels were reduced during injury. This injury-associated reduction in whole-liver 11βHSD1 mRNA was recapitulated in both the TAA (Fig. 1C) model of liver fibrosis and in the choline-deficient diet and MCDD models of steatosis and NASH (Fig. 1D). 11βHSD1 activity levels were not significantly altered during peak CCl₄-induced fibrosis (Supporting Fig. S1A). During the scar resolution phase following CCl₄ injury, 11βHSD1 mRNA and protein levels returned to levels seen in uninjured (vehicle-treated) mice (Fig. 1A,B).

11βHSD1 IS EXPRESSED IN HSCs, ATTENUATING THEIR ACTIVATION IN VITRO

Although 11βHSD1 is highly expressed in hepatocytes, the key cells orchestrating fibrogenesis in parenchymal liver injury models are the HSCs. HSCs are “spontaneously” activated in culture such that αSMA (a global marker of spontaneous HSC activation in vitro) protein levels are maximal at day 8 (Fig. 2A). Significant 11βHSD1 expression was detectable in primary murine HSCs, with a striking reduction observed in 11βHSD1 mRNA (Fig. 2B), protein (Fig. 2C), and activity (Fig. 2D) following spontaneous HSC activation (day 8) in vitro. Similarly, 11βHSD1 gene expression was present in the human LX-228 HSC cell line and was significantly reduced in response to activation with transforming growth factor beta, a classic profibrogenic stimulus (Supporting Fig. S1B).

To investigate the functional role of 11βHSD1 in HSC activation, primary HSCs were isolated from GKO (11βHSD1-deficient) mice and showed significantly higher Acta2 (Fig. 2E) and Col1a1 (Fig. 2F) mRNA levels at day 8 compared to control HSCs.

Consistent with this effect of 11βHSD1 being mediated by amplification of GCs, incubation of wild-type murine HSCs with either corticosterone or 11-dehydrocorticosterone, the inert substrate for 11βHSD1, reduced Acta2 (Fig. 3A) and Col1a1 (Fig. 3B) mRNA levels. The effects of 11-dehydrocorticosterone were abrogated by coadministration of the specific 11βHSD1 inhibitor UE2316 (Fig. 3A,B).

GLOBAL 11βHSD1-DEFICIENT MICE SHOW INCREASED HEPATIC MFB ACTIVATION FOLLOWING CCl₄ INJURY

To investigate whether reducing 11βHSD1 (in GKO mice) in vivo enhances MFB/HSC activation and delays the resolution of scarring, we used the reversible CCl₄ liver fibrosis model because the times of peak injury and maximal scar resolution have been well defined; time points were chosen to reflect the injury and resolution phases because collagen deposition is higher at 24 hours and resolves after 72 hours.3 After 12 weeks of CCl₄ treatment, GKO mice showed significantly higher PSR staining (Fig. 4A) and collagen I deposition (Supporting Fig. S2A) at peak (24 hours) fibrosis. Increased fibrogenesis in the absence of 11βHSD1 was supported by elevated liver mRNA levels of Col1a1 (Fig. 4B). In keeping with enhanced MFB activation, western blot analysis showed that αSMA protein levels were higher at 24 hours and stayed elevated at 8 days in GKO mice (Supporting Fig. S2B). Similarly, histological quantification of αSMA immunostaining showed that GKO mice retained higher αSMA during the resolution phase (72 hours; Fig. 4A), although no histological difference was detected at peak fibrosis. In addition, Acta2 (αSMA) gene expression remained higher in GKO mice during resolution (72 hours; Fig. 4C). To identify whether the observed alteration in MFB phenotype was specific to HSCs, we stained for reelin (Fig. 4D), a known marker of HSCs.29 We did not detect any significant difference in the number of reelin-positive cells at 72 hours, the critical point at which markers of fibrosis were elevated in GKO mice. Hence, no definitive conclusion could be drawn on the origins of MFBs in this model.

GKO MICE SHOW SIMILAR HEPATOCELLULAR DAMAGE AFTER CHRONIC OR ACUTE LIVER INJURY

To assess whether the exaggerated profibrotic response in GKO mice following chronic CCl₄ was due to more severe hepatocyte injury, we performed a detailed histopathological evaluation. This showed similar NAFLD activity scores for inflammation, hepatocellular necrosis, and total injury in GKO and control mice (Supporting Fig. S3A-C). In fact, plasma alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels were lower in GKO mice compared to control injured mice during peak fibrosis (Fig. 4E,F). Plasma albumin levels were comparable between genotypes at the peak injury (24 hours) time point (control versus GKO, 25.40 ± 2.12 U/L versus 22.80 ± 0.66 U/L) and remained at similar levels throughout the resolution phases (data not shown).

To further evaluate the effects of global 11βHSD1 deficiency on hepatocellular injury, a single dose of CCl₄ was administered to induce acute liver injury. As in the chronic CCl₄ model, the GKO mice showed significantly lower plasma ALT, AST, and alkaline phosphatase but similar albumin levels (Supporting Fig. S4A-D). The single dose confirmed similar histopathological scores between genotypes (Supporting Fig. S5A-C). Quantification of hepatocyte necrotic cell area using period acid–Schiff staining showed similar degrees of hepatocellular death in GKO and control mice (Supporting Fig. S5D).

GKO MICE SHOW SIMILAR HEPATIC MACROPHAGE OR NEUTROPHIL NUMBERS TO CONTROL MICE IN CCl₄ INJURY

Apart from HSCs, macrophages have been implicated in both promoting fibrosis and facilitating resolution; therefore, we further assessed if global 11βHSD1 deletion affects hepatic macrophages and neutrophil numbers during injury and resolution phases. Neutrophil numbers were similar between GKO and control livers at all time points (Fig. 5A,B). Macrophage numbers (Fig. 5A-C) and cytokine mRNA levels of Mcp1 and Il1 (Fig. 5D,E) were also similar in GKO and control mice. Taken together these data suggest that enhanced profibrotic response in GKO mice is not due to exacerbated inflammation or macrophage-induced fibrosis but is more likely to reflect a direct effect on HSC activation pathways.

MFB/HSC 11βHSD1 KNOCKDOWN ENHANCES FIBROSIS MARKERS IN RESPONSE TO CHRONIC CCl₄

To distinguish between possible contributions of 11βHSD1 deficiency in MFBs and hepatocytes to the profibrotic phenotype, we generated two independent mouse models targeting 11βHSD1 in either hepatocytes or MFBs/HSCs. Mice with hepatocyte-specific deficiency in 11βHSD1 (LKO) showed almost complete (94%-100% knockdown) loss of 11βHSD1 (Supporting Fig. S6A,B). At 24 hours after chronic CCl₄ the LKO mice showed identical hepatic profibrotic responses histologically (Fig. 6A), similar liver function test changes (Fig. 6B,C), and identical macrophage or neutrophil numbers (Fig. 6D,E) to control littermates. Thus, deletion of 11βHSD1 in hepatocytes does not mimic the changes in liver fibrosis in GKO mice.

To assess the specific role of 11βHSD1 in the regulation of MFBs/HSCs in the CCl₄ model, we used MFKD mice. HSCs isolated from MFKD mice demonstrate a 50%-60% reduction in 11βHSD1 expression in vitro (Supporting Fig. S6C,D) compared to littermate controls. Following chronic CCl₄ administration, MFKD mice showed increased COL1 and αSMA immunostaining (Fig. 7A) and significantly higher Acta2 mRNA (Fig. 7B) levels at 24 hours post–CCl₄ liver injury. No significant difference was detected in PSR staining (Fig. 7A) or plasma ALT/AST levels (Fig. 7C,D) between genotypes. Enhanced MFB activation in MFKD mice was also confirmed in ex vivo primary HSCs from MFKD mice that showed significantly higher Acta2 and Col1a1 levels at days 2, 5, and 8 in culture (Fig. 7E,F).

PHARMACOLOGICAL 11βHSD1 INHIBITION INCREASES MFB ACTIVATION AND EARLY FIBROSIS BUT ENHANCES SCAR RESOLUTION

To address the clinical relevance of our findings for 11βHSD1 inhibitor therapies and specifically test the potential intervention times (prophylactic/injury/repair), we used the 11βHSD1 inhibitor UE2316 in the CCl₄ liver fibrosis model (Fig. 8A). During CCl₄ administration, UE2316-treated animals had slightly lower body weights (Supporting Fig. S7A)21 but no difference in liver-to-body weight ratio (Supporting Fig. S7B). Overall, as observed in GKO mice, UE2316 treatment throughout the period of CCl₄ administration exacerbated hepatic fibrosis measured 24 hours following termination of CCl₄, as shown by PSR staining (Fig. 8B) and collagen 1 (Fig. 8C) immunohistochemistry. UE2316 treatment throughout CCl₄ injury also inhibited later scar resolution, as shown by elevated PSR and collagen 1 up to 8 days post-CCl₄. The UE2316-treated group had higher αSMA as early as 6 hours after the last CCl₄ injection, but this normalized during later resolution (Fig. 8D). Furthermore, as seen in GKO mice, plasma ALT and AST levels were significantly lower at both peak injury (24 hours) and late resolution (day 8) in the UE group compared to vehicle (Fig. 8E,F).

In order to investigate the therapeutic potential of 11βHSD1 inhibition and/or dissect the effects on injury/repair phases, we administered UE2316 only during scar resolution, i.e., commencing 48 hours following the final CCl₄ injection (Fig. 8A).There was no significant difference in total collagen or collagen 1 deposition in the UE-R group (Fig. 8B,C) compared to vehicle control, but administering UE2316 during the resolution phase accelerated the reduction in the number of αSMA-positive cells (Fig. 8D) and reduced plasma ALT levels (Fig. 8E) at day 8.