FXR-Mediated Cortical Cholesterol Accumulation Contributes to the Pathogenesis of Type A Hepatic Encephalopathy

Materials

All chemicals were purchased from MilliporeSigma (Burlington, MA) unless otherwise noted and were of the highest grade available. The Total Cholesterol and Cholesteryl Ester Colorimetric/Fluorometric Assay Kit was purchased from BioVision Inc (Milpitas, CA). Nile Red was purchased from Tokyo Chemical Industry (Tokyo, Japan). The Cyp46A1 antibody was purchased from GeneTex (Irvine, CA). The glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody was purchased from GeneTex (Irvine, CA). The primer for Cyp46A1 (catalog no: PPM03965A) and GAPDH (catalog no: PPM02946E) were purchased from Qiagen, SABiosciences (Frederick, MD). FXR vivo-morpholino sequences (FXR morpholino; 5′-CTGAAACTGCATCACCATCCTTAGC-3′), FXR mismatch vivo-morpholino sequences (FXR mismatch; 5′-CTCAAAGTGGATCACCATCGTTACC-3′), and Endo-Porter were purchased from Gene Tools (Philomath, OR). Guggulsterone was purchased from Tocris (Minneapolis, MN). Hematoxylin was purchased from Vector Laboratories (Burlingame, CA). Eosin was purchased from MilliporeSigma. Liver enzyme blood chemistry assays were purchased from IDEXX Laboratories, Inc (Westbrook, MA).

Mouse Model of Acute Liver Failure

In vivo experiments were performed using male C57Bl/6 mice (25–30 g; Charles River Laboratories, Wilmington, MA) or Cyp7A1^-/- mice bred on a C57Bl/6 background (kind gift from Dr Sandra Erickson, University of California, San Francisco, CA⁶). All animal experiments were approved by the Baylor Scott & White Research Institute Institutional Animal Care and Use Committee and were performed in accordance with the Animal Welfare Act and the Guide for the Care and Use of Laboratory Animals.

Acute liver failure and hepatic encephalopathy were induced using the hepatotoxin azoxymethane (AOM) as previously described.1, 7, 8, 9 Briefly, C57Bl/6 or Cyp7A1^-/- mice were injected with AOM (100 mg/kg intraperitoneally) and placed on heating pads set to 37°C to ensure normothermia. To reduce serum and cortical bile acid levels, C57Bl/6 mice were fed a diet supplemented with 2% cholestyramine (Dyets, Inc, Bethlehem, PA) or the control diet AIN-93G (Dyets, Inc) for 3 days before the injection of AOM. In parallel, C57Bl/6 mice underwent surgery to implant Alzet brain infusion cannulas coupled to subcutaneous implanted minipumps (Alzet, Cupertino, CA) to directly infuse FXR morpholino or FXR mismatch (1 mg/kg) made up in a 10 μmol/L Endo-Porter solution to knock down FXR expression in the brain, which effectively knocked down the expression of FXR in neurons located adjacent to the injection site shown and characterized in previous studies.¹ This same approach was used to infuse 2-hydroxypropyl-β-cyclodextrin (2-HβC, 6 mg/kg/day; MilliporeSigma) to inhibit cholesterol accumulation. The infusion cannulas were implanted using the co-ordinates anteroposterior -0.34, mediolateral -1.0, and dorsoventricular -2.0 from Bregma. This surgery was performed 3 days before the injection of AOM to allow the mice to recover before the onset of acute liver failure to minimize mortality.

Starting at 12 hours after injection, mice were monitored at least every 2 hours for body temperature, weight, and neurologic score as previously described.7, 9 The neurologic score was assessed by an investigator blind to the treatment groups and is a summation of the scores given for the following parameters: pinna reflex, corneal reflex, tail flexion, escape response, righting reflex, and gait. Each parameter was assigned a score between 2 (normal) and 0 (absence of reflexes and presence of severe ataxia), and summed to provide a neurologic score out of 12. At approximately 16 hours after AOM injection, the gait of these experimental mice was examined further using ventral plane imaging technology (DigiGait; Mouse Specifics, Inc, Framingham, MA), which images the underside of animals walking atop a motorized, transparent treadmill belt, thereby generating digital paw prints. These paw prints were analyzed using DigiGait Analysis and Imager software (Mouse Specifics, Inc) and measures of neuromuscular function were quantified, including the paw angle in forelimbs and hind limbs (to measure the degree of external rotation), gait symmetry (ratio of forelimb stepping frequency to hind limb stepping frequency), and the ataxia coefficient for fore and hind limbs (calculated as follows: [maximum stride length – minimum stride length]/average stride length).

Tissue was collected before the onset of neurologic symptoms (preneurologic), when minor ataxia and weakened reflexes were present (minor neurologic), when major ataxia and deficits in reflexes were evident (major neurologic), and at coma, as defined by a loss of righting and corneal reflexes. In experiments involving Cyp7A1^-/- mice, cholestyramine-supplemented mice, or 2-HβC-infused mice, only the coma time point was investigated.

Assessment of Liver Damage and Function

Liver damage was assessed by H&E staining according to previously published protocols.⁷ Paraffin-embedded livers were cut into 4-μm sections and mounted onto positively charged slides (VWR, Radnor, PA). Slides were deparaffinized and stained with Hematoxylin QS (Vector Laboratories) for 1 minute followed by staining for 1 minute with eosin Y (Amresco, Solon, OH) and rinsed in 95% ethanol. The slides then were dipped into 100% ethanol and subsequently through 2 xylene washes. Coverslips were mounted onto the slides using CytoSeal XYL mounting media (Thermo Fisher Scientific, Waltham, MA). The slides were viewed and imaged using an Olympus BX40 microscope with an Olympus DP25 imaging system (Center Valley, PA).

Liver function was assessed by measuring plasma alanine aminotransferase and aspartate aminotransferase using the IDEXX Catalyst One machine from IDEXX Laboratories, Inc.

In Vitro Primary Neuronal Culture

Primary cortical neurons were isolated and cultured using methodology previously described.⁸ Primary neurons were isolated from postnatal day 1 mouse pups. Mice were decapitated and whole brains were removed. The cortex was isolated and meninges and dura were removed. Cortical tissue was mechanically disrupted and filtered through a 100-μm filter. Neurons were pelleted by centrifugation at 1400 g. Neurons were suspended in media and plated on 12-well plates with 750,000 cells per well. After 24 hours, cells were washed and media was replaced and supplemented with 2% B27 growth supplement. Cells were cultured for 10–12 days and subsequently treated with deoxycholic acid (10 μmol/L) in the presence or absence of the FXR antagonist guggulsterone (10 μmol/L) for 24 hours. At this point, cells were lysed and used for immunoblot or reverse-transcription polymerase chain reaction (RT-PCR) analyses.

Assessment of Bile Acid Content in the Brain

The bile acid content of cortex tissue was determined using methodology previously used for other studies.¹ When AOM-treated mice progressed to coma, they were euthanized and transcardially perfused with ice-cold saline to remove the blood from the brain. Cortex homogenates were prepared by calculating the wet weight of brain tissue with subsequent homogenization in 100 mg/mL in ultrapure water using a Miltenyi Biotec gentleMACS Dissociator (San Diego, CA). Homogenates were spun down for 5 minutes at 16,100 g and supernatants were collected. Total bile acid content was assessed in homogenates of the frontal cortex following the manufacturer’s instructions (IDEXX, Westbrook, ME). Data are reported as the nanomole of bile acid per milligram of cortex tissue protein for each respective analysis.

Measurement of Brain Cholesterol

Total, free, and esterified cholesterol were assessed in brain homogenates using a BioVision Total Cholesterol and Cholesteryl Ester Assay Kit following the manufacturer's instructions with individual samples run in duplicate and averaged for final values. The subcellular location of cholesterol was assessed using Nile Red staining (for intracellular cholesterol¹⁰) and Filipin III staining (for membrane-bound cholesterol¹¹). Briefly, mice were perfused transcardially with ice-cold phosphate-buffered saline (PBS), followed by 4% paraformaldehyde (PFA) in PBS. Whole brains were rapidly removed and postfixed in PFA overnight at 4°C. Brains were cryoprotected in 30% sucrose (in PBS) and embedded in optimal cutting temperature compound. Thirty-micrometer sections were cut through the frontal cortex region; sections were rinsed in PBS and further fixed in 4% PFA for 10 minutes at room temperature. Free-floating sections were incubated with Filipin III (0.5 μg/mL in 10% bovine serum albumin/PBS) for 2 hours at room temperature, or with Nile Red (10 μg/mL in PBS) for 30 minutes at room temperature. Sections were washed in PBS, mounted onto microscope slides, and imaged using a Bio-Rad Zoe Fluorescent Cell Imager (Hercules, CA). Field images were quantified for the percentage positive area of Nile Red or Filipin III staining using ImageJ software (National Institutes of Health, Bethesda, MD). Each Nile Red or Filipin III analysis required 10 images taken per brain section with the percentage area quantified as the average of these images per mouse.

Expression of Cyp46A1

RNA was extracted from tissue or primary neurons using an RNeasy Mini Kit (Qiagen, Germantown, MD) according to the manufacturer’s instructions. Synthesis of complementary DNA was accomplished using a Bio-Rad iScript Complementary DNA Synthesis Kit. RT-PCR was performed as previously described¹² using commercially available primers designed against mouse Cyp46A1 and GAPDH (SABiosciences). A delta-delta threshold cycle analysis was performed using vehicle-treated tissue or untreated primary neurons as controls for subsequent experiments.13, 14 All individual RT-PCR samples were run in duplicates and averaged to determine final values.

Cortex tissue from all treatment groups was homogenized using a Miltenyi Biotec gentleMACS Dissociator and total protein was quantified using a ThermoFisher Pierce BCA Protein Assay kit. Sodium dodecyl sulfate–polyacrylamide gel electrophoresis gels (10% vol/vol) were loaded with 10–20 μg of protein diluted in Laemmli buffer per each tissue sample. Specific antibodies against Cyp46A1 and β-actin were used. All imaging was performed on an Odyssey 9120 Infrared Imaging System (LI-COR, Lincoln, NE). Data are expressed as the fold change in fluorescent band intensity of target antibody divided by β-actin or GAPDH, which are used as loading controls. The values of vehicle or control groups were used as a baseline and set to a relative protein expression value of 1. Band intensity quantifications were performed using ImageJ software.

Statistics

All statistical analyses were performed using GraphPad Prism software (GraphPad Software, La Jolla, CA). For data that passed normality tests, significance was established using the Student's t test when differences between 2 groups were analyzed, and analysis of variance when differences between 3 or more groups were compared, followed by the appropriate post hoc test. If tests for normality failed, 2 groups were compared with a Mann–Whitney U test or a Kruskal–Wallis ranked analysis when more than 2 groups were analyzed. Results were expressed as means ± SEM. Differences were considered significant for P < .05.

Cortical Cholesterol Concentrations Are Increased in AOM-Treated Mice

We previously showed that aberrant bile acid signaling may contribute to the neurologic complications of acute liver failure.¹ Little is known about the effects of bile acid signaling in the brain. However, the enzymatic machinery for bile acid synthesis has been shown to be present in the brain and is the predominant method by which the brain regulates cholesterol homeostasis.⁵ Therefore, we hypothesized that aberrant bile acid signaling during hepatic encephalopathy might disrupt cholesterol homeostasis in the brain. Indeed, total cholesterol (Figure 1A) and free cholesterol (Figure 1B), but not esterified cholesterol (Figure 1C), were increased significantly with the onset of major neurologic complications in AOM-treated mice. This increase in cholesterol content correlated with increases in both Nile Red (intracellular) and Filipin III (membrane-bound) staining in the brain (Figure 1D and E). Taken together, these data indicate that the total cholesterol content is increased in the brain during acute liver failure, which is distributed both intracellularly and in the membrane.

AOM-Treated Mice Have Reduced Cortical Cyp46A1 Activity

To determine if the cholesterol buildup may be owing to an impaired clearance pathway, the expression of Cyp46A1 was assessed in mice with acute liver failure. Expression of Cyp46A1 messenger RNA and protein was down-regulated in the cortex of AOM-treated mice showing neurologic symptoms of hepatic encephalopathy (Figure 2A), suggesting that the buildup in cholesterol in the brain might be attributable to an impaired clearance pathway. Interestingly, this down-regulation of Cyp46A1 was not evident in mouse models showing attenuated AOM-induced aberrant bile acid signaling in the brain.¹ Specifically, we previously have shown that increases in serum and cortical bile acid content are not evident in Cyp7A1^-/- mice after AOM injection, which correlated with a delay in the neurologic decline associated with acute liver failure.¹ Furthermore, we previously have shown that the effects of bile acid signaling in the development of hepatic encephalopathy are attributable, in part, to FXR signaling in neurons and specific knockdown of FXR activity in the brain using a cholestyramine-supplemented diet or direct infusion of FXR morpholino delayed the neurologic decline associated with AOM-induced acute liver failure.¹ In Cyp7A1^-/- mice, cholestyramine-fed mice, or mice infused with FXR morpholino, the down-regulation of Cyp46A1 expression in response to AOM injection was attenuated (Figure 2B–D). To determine if the suppression of Cyp46A1 was attributable to a direct action of FXR-mediated signaling and not a reflection of the degree of neurologic impairment in these mice, primary neurons were treated with deoxycholic acid, a known agonist of FXR,¹⁵ which decreased the expression of Cyp46A1. This could be prevented by pretreatment with the FXR antagonist guggulsterone (Figure 2E).

Reducing Cortical Bile Acid Levels Alleviates Cholesterol Accumulation in AOM-Treated Mice

The pathway mediated by Cyp46A1 is a primary method regulating cholesterol homeostasis in the brain.⁵ Therefore, we assessed whether preventing the suppression of Cyp46A1 expression in mouse models in which aberrant cortical bile acid signaling was attenuated could influence the degree of cholesterol buildup. Initially, Cyp7A1^-/- mice were assessed during AOM-induced hepatic encephalopathy for their relative increase of cortical bile acids compared with Cyp7A1^+/+ (wild-type [WT]) mice. Concentrations of cortical bile acids were increased significantly in WT AOM-treated mice and this effect was absent in Cyp7A1^-/- mice (Figure 3A). A similar trend was seen regarding cholesterol because an increase in total and free cholesterol content was observed in the WT mice after AOM injection that was not evident in Cyp7A1^-/- mice (Figure 3B and C). Furthermore, the increase in intracellular cholesterol (Nile Red staining) (Figure 3D) and membrane-bound cholesterol (Filipin III staining) (Figure 3E) observed in WT mice after AOM injection was absent in Cyp7A1^-/- mice.

To better validate that the effects observed in Cyp7A1^-/- mice were caused by bile acids alone and not a congenital or strain-specific effect, C57Bl/6 mice were fed a diet supplemented with 2% cholestyramine to reduce bile acid levels. Cholestyramine supplementation was found to significantly reduce the concentration of cortical bile acids in AOM-treated mice compared with the AOM-treated control diet-fed mice (Figure 4A). This effect correlated with a reduction in total and free cholesterol levels in the AOM-treated, cholestyramine-supplemented mice compared with AOM-treated controls (Figure 4B and C). In addition, the increase in Nile Red (Figure 4D) and Filipin III staining (Figure 4E) observed after AOM treatment was reduced by cholestyramine pretreatment, showing a reduction of intracellular and free cholesterol, respectively, after cholestyramine supplementation.

FXR Signaling Promotes Cholesterol Accumulation in the Cortex

If bile acids are the primary contributor to the accumulation of cortical cholesterol observed in AOM-treated mice, then this effect could be a result of bile acid-receptor–mediated signaling. Although bile acids can signal through both G-protein–coupled receptors and nuclear receptors, we previously have shown that FXR is present in neurons and FXR signaling contributes to neurologic decline in AOM-treated mice.¹ Therefore, the approach taken was to reduce neuronal FXR protein expression through FXR morpholino infusion in vehicle and AOM-treated mice and assess cholesterol concentrations. Similar to what was observed in Cyp7A1^-/- and cholestyramine-supplemented mice, total and free cholesterol (Figure 5A and B), as well as intracellular (Figure 5C) and membrane-bound (Figure 5D) cholesterol, were increased in AOM-treated mice infused with a control mismatched vivo morpholino sequence and this effect was attenuated in mice infused with FXR morpholino.

Infusion of 2-HβC Reduces Neurologic Deficits and Cortical Cholesterol Levels in AOM-Treated Mice

To determine whether cholesterol buildup in the brain plays a role in the neurologic deficits associated with acute liver failure, mice were infused with 2-HβC into the lateral ventricle. Treatment of mice with 2-HβC before AOM injection significantly delayed the neurologic decline (Figure 6A) and increased the time taken to reach hepatic coma (Figure 6B). Besides these neurologic measures, neuromuscular complications during hepatic encephalopathy also were assessed. Specifically, control mice walked with a paw angle of approximately 3.5° (forelimb) or 18° (hind limb) of external rotation, which is consistent with previously published studies¹⁶ (Figure 6C). In AOM-treated mice, there was an increased degree of external rotation of both the forelimbs and hind limbs, indicating a splaying of the paws often seen during ataxia.¹⁶ Treatment with 2-HβC reduced the paw angle of AOM-treated mice to values similar to controls (Figure 6C). Second, gait symmetry, defined as the ratio of forelimb stepping frequency to hind limb stepping frequency was effectively 1 in control mice as expected, but this was reduced significantly in mice injected with AOM (Figure 6D), indicating the mice were stepping more frequently on the hind limbs than forelimbs to compensate for the neuromuscular deficits in the hind limbs. Infusion with 2-HβC before AOM injection returned the gait symmetry to control levels (Figure 6D). Finally, we measured the ataxia co-efficient, which is an index of step-to-step variability for each limb. Control mice had a low ataxia co-efficient in both the forelimbs and hind limbs (Figure 6E), indicating a relatively consistent stride length in all limbs. However, mice with AOM-induced acute liver failure had a significantly higher ataxia co-efficient, indicating a greater variability in stride length, and infusion of 2-HβC returned the ataxia co-efficient back to control levels (Figure 6E).

The central infusion of 2-HβC was chosen over systemic administration to allow for the delineation of direct neuroprotective effects of 2-HβC on the development of hepatic encephalopathy vs the potential indirect hepatoprotective effects that then would impact the subsequent neurologic complications. Confirmatory experiments were performed to ensure that the local central infusion of 2-HβC had no effect on the underlying AOM-induced liver damage. The degree of liver damage significantly increased after AOM injection as shown by H&E staining (Figure 7A), serum alanine aminotransferase (Figure 7B), and aspartate aminotransferase (Figure 7C) to a similar degree as shown previously,1, 7, 8, 9 and these results were similar to mice infused with 2-HβC before AOM injection (Figure 7). These data indicate that central infusion of 2-HβC does not have any hepatoprotective effects against AOM-induced hepatotoxicity and that the protective effects of 2-HβC observed on the development of hepatic encephalopathy result from the direct actions in the brain.

Finally, 2-HβC treatment has been shown previously to prevent excessive cholesterol buildup in the brain in models of Niemann–Pick type C disease without depleting it completely.¹⁷ Similarly, 2-HβC infusion into the brain did not decrease the basal levels of cholesterol but did prevent the AOM-induced buildup in total (Figure 8A) and free (Figure 8B) cholesterol content in the cortex. Furthermore, 2-HβC treatment attenuated the AOM-induced increase in Nile Red (Figure 8C) and Filipin III (Figure 8D) staining. Taken together, these data suggest that neuroprotective actions of 2-HβC likely are owing to the prevention of cholesterol buildup in the brain and that the increase in cortical cholesterol observed after AOM likely is contributing to the neurologic complications of acute liver failure.