Development of mice with hepatocyte knockout of lanosterol 14α-demethylase (HCyp51−/−) from cholesterol synthesis is characterized by the progressive onset of liver injury with ductular reaction and fibrosis. These changes begin during puberty and are generally more aggravated in the knockout females. However, a subgroup of (pre)pubertal knockout mice (runts) exhibits a pronounced male prevalent liver dysfunction characterized by downregulated amino acid metabolism and elevated Casp12. RORC transcriptional activity is diminished in livers of all runt mice, in correlation with the depletion of potential RORC ligands subsequent to CYP51 disruption. Further evidence for this comes from the global analysis that identified a crucial overlap between hepatic Cyp51−/− and Rorc−/− expression profiles. Additionally, the reduction in RORA and RORC transcriptional activity was greater in adult HCyp51−/− females than males, which correlates well with their downregulated amino and fatty acid metabolism. Overall, we identify a global and sex-dependent transcriptional de-regulation due to the block in cholesterol synthesis during development of the Cyp51 knockout mice and provide in vivo evidence that sterol intermediates downstream of lanosterol may regulate the hepatic RORC activity.
The liver is to date the best example of a non-reproductive tissue that shows major differences in gene expression between males and females. The majority of liver-related sex differences emerge during pubertal development1 with adult mice exhibiting well over 1,000 genes differentially expressed (DE) between females and males2.
The development of liver begins at embryonic (E) day 8.5–9, with hepatocyte differentiation3 and cholesterol production4 starting around day E13.5. At birth, 55–60% of cholesterol in the liver is produced endogenously, while in adults hepatic cholesterol production provides 10% and 40% of total cholesterol in humans and mice, respectively5. Consequently, mouse liver knockouts of cholesterol synthesis genes survive embryonic development in contrast to the lethal phenotypes of the full knockout mice6. Blocking earlier steps of the pathway produces a stronger phenotype in both cases. The hepatocyte Hmgcr−/− mice survive for up to 5 weeks7, while mice lacking the rate limiting step of the post-lanosterol pathway, the hepatocyte Cyp51−/− mice, survive for over 19 weeks although with marked changes in liver structure and function8. Cholesterol intermediates after lanosterol have long been thought of as dedicated exclusively to cholesterol in contrast to earlier metabolites9. However, meiosis-activating sterols (MAS) have been identified for their importance in gametogenesis10 and intermediates further downstream most recently as potential ligands for RORC isoform 2 (RORγt) in Th17 cells11,12. RORC is an orphan nuclear receptor expressed in several tissues, including the liver13. The two isoforms have distinct functions and distribution, but with the same ligand- and DNA-binding domains14. RORγt is the primary isoform in thymocytes and is required for normal lymph node development15, whereas the more common isoform 1 (RORγ1 or RORC) modulates the circadian clock, as well as glucose and fat metabolism and in the liver peaks at circadian time ZT16–2016,17,18.
Herein we examine the development of the hepatocyte Cyp51−/− mice and show that normal liver function becomes impaired by the end of puberty, with earlier presentation and enhanced liver damage in females. A subgroup of male prevalent knockout mice exhibits augmented liver dysfunction combined with abnormal hepatic sterol profiles. Lastly, our work highlights the need for unperturbed cholesterol synthesis for normal RORC signalling in vivo.
Characterization of Cyp51 −/− mice during postnatal development
Hepatocyte Cyp51 knockout mice (HCyp51−/−) on the wild type (Cyp51+/+)8 or heterozygous (Cyp51+/−) background were born in accordance with the Mendelian ratio (Supplementary Table 1) and were euthanized at 0, 3, 6, 9 and 19 weeks of age (Supplementary Fig. 1).
From 750 mice, 704 were weaned and a further 46 (6%) exhibited severe developmental abnormalities (runt-HCyp51−/−; runts). Runts were exclusive to the HCyp51−/− genotypes. They died or had to be euthanized at 4–10 weeks with a male to female ratio of 2:1 (Supplementary Table 1). Runts are characterized by growth arrest after the weaning period (Fig. 1a), hunched posture, bristled hair and frequent icterus (Fig. 1b). Figures 1c,d show successful Cyp51 disruption with a progressively increasing gap in Cyp51 gene and protein levels between HCyp51−/− and control mice. Plasma parameters of these young mice showed mostly unperturbed cholesterol homeostasis, with total (Fig. 2a) and LDL (Fig. 2b) cholesterol remaining stable in HCyp51−/− mice of all examined ages, except for the runts that show a 40% decrease in total cholesterol and a 60% drop in HDL cholesterol (Fig. 2c). Increased ALT but not AST was found in 9-week HCyp51−/− mice and runts, which indicated hepatocellular damage, while elevated conjugated bilirubin in runts confirmed jaundice (Fig. 2d,e).
Haematoxylin and eosin (HE) and collagen staining (Sirius red) revealed initial signs of ductular reaction with mild inflammation (lymphocytes and granulocytes) surrounding the portal vein in all HCyp51−/− females (7/7) and about half of males (4/7) at 6 weeks. Enhanced ductular proliferation, inflammation and mild-to-moderate portal fibrosis were seen in all HCyp51−/− mice at 9 weeks (Fig. 3a–c) and were slightly more pronounced on the heterozygous background. Despite their youth, runts exhibited severe liver damage with widespread ductular reaction, immune cell infiltration and prominent fibrosis with portal bridging in the worst cases (Fig. 3b).
Gene expression profiling reveals progressive metabolic downregulation and a fall in RORC transcriptional potential resulting from Cyp51 disruption
To evaluate the impact of Cyp51 excision on liver development, we performed gene expression profiling using Affymetrix microarrays on Cyp51+/+ and HCyp51−/− mice, aged 3, 6 and 19 weeks, of both sexes, including runts. The number of DE genes between HCyp51−/− and control mice increased progressively from 3 weeks onward (Fig. 4a; Supplementary Table 2). Runts exhibited a global de-regulation with roughly a quarter of their hepatic transcriptome differentially expressed compared to 6-week Cyp51+/+ mice (Fig. 4b). Gene set enrichment analysis (PGSEA) based on pathways from Kyoto Encyclopedia of Genes and Genomes (KEGG) exhibited significant enrichment in runts (235/287 pathways), which are (pre)pubertal mice, and the established 19-week HCyp51−/− mice (202/287 pathways) (Table 1; Supplementary Table 3). Similarly to the 19-week mice (Lorbek et al.8 and this study), the runt phenotype also associated with downregulated metabolism of lipids, amino acids and bile acids as well as elevated inflammatory and cancer-related pathways, based on their transcriptional profiles (Supplementary Fig. 2). This points to a system-wide de-regulation in gene expression possibly stemming from altered transcriptional landscape. To evaluate this hypothesis, we employed transcription factor (TF) enrichment analysis, which indeed showed significant enrichment in the runt (176/250 TFs) and the adult 19-week HCyp51−/− mice (113/250 TFs) (Table 2; Supplementary Table 4). In both cases RORC and RORA transcriptional activities were among the most downregulated (Fig. 5a–c). Reduced HNF4A activity also indicated metabolic downregulation, while elevated NRF2 and NFY activities indicate that a block in cholesterol synthesis could lead to ROS production19 and protein misfolding20. The Interactome tool21 for visualizing interactions between selected enriched TFs in runts contained no known interactions of RORC with other TFs (Fig. 6), suggesting its diminished activity may stem from other sources, such as the decreased availability of endogenous sterol ligands.
Sterol imbalance as a major hallmark of Cyp51 ablation
To assess the availability of natural sterol RORC ligands we measured cholesterol and its biosynthesis intermediates by GC/MS. Total cholesterol was diminished in 6-week HCyp51−/− mice, but remained unchanged in runts. This was true also for esterified and free cholesterol (Fig. 7a). CYP51 substrate lanosterol (LAN) was increased 12- and 35-fold and 24,25-dihydrolanosterol (DHL) 800- and 6000-fold in 6-week HCyp51−/− mice and runts, respectively (Fig. 7b). Conversely, the direct product of CYP51 demethylation, follicular fluid meiosis-activating sterol (FF-MAS), was decreased 2.6- and 2.1-fold in 6-week HCyp51−/− mice and runts, respectively. Further downstream the pathway, zymosterol was decreased by 2.4- and 4.3-fold (Fig. 7b). Sterols between MAS and zymosterol (Fig. 7c) were recently proposed as strong RORC agonists11. Elevated levels of lathosterol and 7-dehydrocholesterol might support the existence of a shunt pathway, as previously proposed by us22,23 and others24.
An overlapping transcriptional network of the hepatocyte Cyp51 and Rorc knockout mice
We next set out to evaluate the contribution of diminished RORC activity on the phenotype of HCyp51−/− mice. We applied the hepatocyte Rorc (HRorc−/−) mice and compared the transcriptome data between the two models. We found 50 DE genes separating HRorc−/− and control mice (Supplementary Table 2), where 11 had the same response as in runts (Fig. 8a, Supplementary Table 5) and 5 as in 19-week HCyp51−/− mice. Due to the known circadian nature of RORC, hepatocytes from the HRorc−/− mice were isolated at ZT7 (daytime RORC nadir) and ZT19 (night-time RORC zenith). TF enrichment of the transcriptome data expectedly showed increased night-time RORC activity only in the control mice since its (circadian) expression is dampened in HRorc−/− livers. From 1264 diurnal DE genes in control mice, 791 lost their rhythm due to RORC abolishment in the HRorc−/− mice and may be considered as potential RORC targets. Moreover, 105 genes from this list were positive for RORC binding based on Chip-seq results by Takeda et al.17 and a further 49 differentially expressed (mostly downregulated) in runts (Fig. 8b, Supplementary Table 5). FIDEA25 showed that these genes are primarily involved amino acid metabolism (Supplementary Tables 6 and 7).
Age-dependent sexual dimorphism in Cyp51 −/− mice
The progressive increase in the number of DE genes between males and females (Supplementary Table 2) is in agreement with the current understanding that sex differences emerge from 4 weeks onward1. The number of DE genes in response to Cyp51 disruption was consistently higher in HCyp51−/− females compared to the males (61/1 at 6 weeks; 1821/941 at 19 weeks). The initial response in 6-week HCyp51−/− females favoured apoptosis and ErbB signalling, pointing to enhanced liver damage and regeneration. At 19 weeks, decreased fatty and amino acid metabolism and enhanced inflammatory and cancer-related pathways were also underlined (Supplementary Table 3). Male sex seems to confer a higher risk for development of the disease phenotype (M:F = 2:1) only in runts. Despite this, runt males and females differed in only 23 genes that showed little to no sex-variation at other ages. The elevation of Casp12 indicating ER stress and activated unfolded protein response (UPR) represents a hallmark of male runts. TF enrichment also showed elevated activity of NFYs (p = 0.06) and other damage response factors (e.g. JUN, FOS). KEGG pathway enrichment showed better preservation of metabolism and milder inflammation in runt females.
Expression of RORA and RORC is sexually dimorphic with RORC being generally higher in males and RORA in females26. The fall in transcriptional activity of both receptors was significantly higher in 19-week HCyp51−/− females, which correlates well with their reduced amino and fatty acid metabolism. These data further implicate ROR signalling as one of the sex-dependent metabolic regulators.
While a major part of cholesterol synthesis is carried out by the liver, other organs contribute to the whole-body cholesterol pool resulting in viable phenotypes of the liver conditional knockouts. Histological and transcriptional changes in HCyp51−/− mice became visible at 6 weeks of age (Supplementary Fig. 3) likely due to cholesterol compensation during the initial 3-week nursing period. The switch to standard laboratory chow lacking cholesterol27 raises the need for its endogenous synthesis. As this is disrupted in the knockout livers, the result is a fall in plasma and liver cholesterol at 6 weeks that is remedied at later ages, and alterations of sterol metabolites. Surprisingly, nursing was not always able to prevent liver failure, as indicated by the appearance of HCyp51−/− runts. Runts exhibited the strongest fall in Cyp51 levels. While the stability of Cyp51 mRNA and protein is estimated to be in line with other members of the cytochrome P450 family28,29, the elevated UPR in runts could lead to CYP51 protein misfolding and degradation. Additionally, increased removal of damaged hepatocytes and higher numbers of inflammatory cells that do not express Cyp51 could have contributed to these measurements. Liver damage in runts most likely started prior to weaning, as witnessed by one 3-week runt-like female. The normal hepatic cholesterol in runts could be explained by the steep fall in plasma HDL. Also based on the sterol profile of DHL we speculate that cholesterol synthesis insufficiency is most pronounced in runts.
From a structural standpoint, cholesterol is an important mediator of membrane fluidity and as a constituent of lipid rafts contributes to cellular signalling. Excess cellular cholesterol preferentially redistributes to intracellular compartments30, such as ER, where a raise in cholesterol can trigger UPR and apoptosis in macrophages31,32. Diminished cholesterol or its replacement with sterol intermediates may also interfere with membrane structure and function33,34,35, demonstrating the need for unperturbed cholesterol homeostasis. Due to methyl groups at carbon C4, cholesterol intermediates LAN and DHL cannot maintain the optimal membrane structure33,36. Livers of HCyp51−/− mice contained markedly elevated DHL that in runts even surpassed cholesterol. Accumulating DHL could interfere with membranes of mitochondria to produce ROS, with ER to hinder protein synthesis, and with peroxisomes to interrupt metabolic pathways. This is exemplified by activated inflammatory pathways, enrichment of several UPR genes (Casp12, Eif2ak3) and TFs (NFY), and decreased peroxisomal metabolism in 19-week HCyp51−/− and runt mice. The steep rise in DHL might therefore represent an initiator of hepatocellular stress and liver damage (Supplementary Fig. 4).
An interesting question is why is there such a large increase in DHL in runts as opposed to the other HCyp51−/− knockouts. In cell cultures, blocking CYP51 resulted in comprehensive LAN and DHL build-up with an almost complete replacement of cellular cholesterol37,38. Hepatic cholesterol intermediates (including LAN) generally correlate with their plasma concentration, but not in the case of DHL24. This suggested that DHL might be metabolized in a shunt pathway as most recently confirmed by us39, or might be exported to the bile, which still awaits experimental confirmation. The former option is enticing as it offers an explanation as to why we found no decrease in sterol intermediates immediately before their conversion to cholesterol in HCyp51−/− mice (e.g. desmosterol, 7-dehydrocholesterol). Conversion of DHL to one of the downstream intermediates from the Kandutsch-Russell branch of cholesterol synthesis could also explain the paradoxically increased lathosterol and 7-dehydrocholesterol in runts. Recently, it was also discovered that cholesterol biosynthesis intermediates act as RORγt ligands in Th17 cells11,12. Based on the current available evidence LAN and DHL could not serve as RORC agonists since a marked increase in RORC activity was observed upon Cyp51 overexpression that metabolizes both sterols11. In contrast, sterols downstream of LAN and DHL in the canonical pathway, as well as non-canonical sterols, such as metabolites of zymosterol, showed specific agonist activity for RORC.
Reduced availability of sterols after the CYP51 block in livers of 19-week HCyp51−/− and runt mice resulted in diminished RORC activity that could contribute to a fall in the overall metabolic capacity (Supplementary Fig. 4). However, the redundancy of binding to ROR response element (RORE) that enables RORA and REV-ERBs to regulate the overlapping set of target genes17,40, as well as the overlap in regulation potential of other TFs (PPARs, LXR etc.) make the prediction of RORC-specific effects difficult. By applying HRorc−/− mice and comparing genes that lost diurnality upon Rorc ablation with genes that bind RORC in their regulatory regions, we were able to overcome this hurdle and pinpointed to RORC transcriptional targets, half of which are de-regulated also in HCyp51−/− runts. These data show that RORC abolishment, after inactivation of Cyp51, exerts its effects either by acting directly on gene expression, or by disrupting the normal circadian rhythmicity of metabolic pathways.
With the exception of runts, female HCyp51−/− mice exhibited a more prominent liver phenotype compared to males throughout their development. This may attribute to the reduced production of oestrogens with antioxidant and anti-fibrotic properties41, as well as to the diminished amino acid and lipid metabolism, corresponding to a stronger fall in RORA and RORC activities in 19-week HCyp51−/− females. Conversely, higher prevalence of runt males with dampened metabolism and increased UPR points to pre-pubertal protection of females. Liver thus responds to cholesterol synthesis block in a sex- and age-dependent manner. This is exemplified also in the hepatocyte Hmgcr−/− mice7, that similarly to HCyp51−/− runts, died by 5 weeks, with minor changes in plasma and liver cholesterol and an activated ER stress response. While the hepatocyte Hmgcr−/− knockout was lethal in males, 30% of females survived for up to a year, reflecting a similar sexual dimorphism as observed in HCyp51−/− runts.
In summary, we provide evidence that cholesterol synthesis disruption reduces RORC activity in vivo, likely due to the lack of endogenous sterol ligands. We also show that during development towards adulthood female and male livers differ in their susceptibility to the cholesterol synthesis block, where dampened ROR signalling and decreased metabolism likely contribute to this progressive liver disease.
Mice experiments were performed in compliance with the current guidelines and regulations. In the case of Cyp51 mice, they were approved by the Veterinary Administration of the Republic of Slovenia (license numbers 34401-31/2011/4 and 34401-52/2012/3) and were conducted in agreement with the European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes (ETS 123) as well as with the guidelines for work with laboratory animals by the National Institute of Health. In the case of Rorc mice, experiments were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of the NYU School of Medicine.
Hepatocyte Cyp51 knockout mice were prepared by crossbreeding Cyp51flox/flox or Cyp51-/flox with Alb-Cre mice to generate hepatocyte specific Cyp51−/− mice on a wild type (Cyp51flox/flox Alb-Cre; designated as Cyp51+/+ HCyp51−/−) and heterozygous (Cyp51-/flox Alb-Cre; designated as Cyp51+/− HCyp51−/−) background8,42. Cyp51flox/flox (Cyp51+/+) and heterozygous Cyp51-/flox (Cyp51+/−) mice were used as controls27. Mice were euthanized at different time points during development – 0, 3, 6, 9 and 19 weeks. Blood and liver were collected for biochemical and histological analyses.
Hepatocyte Rorc knockout mice (HRorc−/−) were prepared by crossbreeding Rorcflox/flox and Alb-Cre mice. Hepatocytes were isolated from 12–13 week old mice at ZT7 and ZT19 using the two-step collagenase perfusion method.
Formalin-fixed and paraffin-embedded liver sections were cut at 5 μm and stained with hematoxylin and eosin for general histological evaluation or with Sirius red to assess collagen deposition.
Plasma parameters (cholesterol, ALT, AST, bilirubin) were measured using commercially available kits. Liver sterols were evaluated by coupled gas spectrometry/mass spectrometry as previously described43. Total proteins and RNA was isolated from frozen liver samples.
Microarray-based gene expression profiling
Affymetrix GeneChip® Mouse Gene 2.0 ST arrays were used to conduct gene expression profiling on Cyp51+/+ and Cyp51+/+ HCyp51−/− mice of both sexes, aged 3, 6, 19 weeks and runts (30 samples). Gene expression analysis was done in R using Bioconductor packages limma and PGSEA. Gene sets for enrichment analysis were taken from the KEGG and TRANSFAC databases.
Libraries were prepared using the Illumina TruSeq Stranded Total RNA library prep, with Ribozero Gold, starting from 500 ng of total RNA and loaded on high output HiSeq 2500 flow cells. RNA alignments were performed using STAR and reads mapping to exons counted using featureCounts. Counts were normalized in R using the DESeq2 package.
Detailed protocols can be found in the supplementary information file.
How to cite this article: Urlep, Z. et al. Disrupting Hepatocyte Cyp51 from Cholesterol Synthesis Leads to Progressive Liver Injury in the Developing Mouse and Decreases RORC Signalling. Sci. Rep. 7, 40775; doi: 10.1038/srep40775 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank K. Kodra, H. Klavžar, N. Pezdirc, M. Kužnik and S. Omahen for their help with experiments. We acknowledge S. Horvat and R. Keber for their previous contribution to development of the Cyp51 knockout mouse model, J. Aćimović and A. Lövgren-Sandblom for contributing to the gas chromatography/mass spectrometry analyses and A. Rives for help with statistical analysis of RNA-Seq data. This work was supported by the Slovenian Research Agency (ARRS) program grants P1-0104, P1-0390 and J1-4790 and by the Dale and Betty Frey Fellowship of the Damon Runyon Cancer Research Foundation 2105-12 (J.A.H). Z. Urlep and G. Lorbek were supported by the graduate fellowship of ARRS.