Liver-specific ablation of KRAB Associated Protein 1 in mice leads to male-predominant hepatosteatosis and development of liver adenoma

Abstract

The liver is characterized by sexually dimorphic gene expression translating into sex-specific differences in lipid, drug, steroid hormone and xenobiotic metabolism, with distinct responses of males and females to environmental challenges. Here, we investigated the role of the KRAB-associated protein 1 (KAP1) epigenetic regulator in this process. Liver-specific KAP1 knockout led to strikingly sexually dimorphic phenotypic disturbances, including male-predominant steatosis and hepatic tumors. This correlated with sex-specific transcriptional dysregulation of a wide range of metabolic genes, notably those involved in retinol and sex hormone processing as well as in detoxification. Furthermore, chromatin immunoprecipitation followed by deep sequencing indicated that a number of dysregulated genes are direct targets of the KRAB/KAP1 repression system. Those genes include sexually dimorphic Cyp2d9, Gstπ, Slp Cyp2a, Cyp2b and Cyp3a gene clusters. Additionally, we identified a male-restricted KAP1 binding site in the fsp27 (fat specific protein 27) gene, correlating with its male-predominant upregulation upon Kap1 deletion, suggesting that the latter might be an important trigger in the development of male-specific hepatosteatosis and secondary tumorigenesis.

Conclusion

This work reveals KRAB/KAP1-mediated transcriptional regulation as a central event in the metabolic control hormones, drugs and xenobiotics in the liver, and further links disturbances in these processes with hepatic carcinogenesis.

Metabolism is orchestrated by complex gene regulatory networks, and the liver is central to this process. This organ is characterized by sexually dimorphic gene expression, with numerous genes transcribed in a sex-dependent manner¹. In mice, examples of such genes include the male-predominant Cyp2d9, which encodes for a testosterone-16-α-hydroxylase inactivating the main male sex hormone and various drugs, and the female-predominant Cyp2a and Cyp2b genes involved in xenobiotic and drug metabolism². The sex-specific expression of these and other classes of enzymes results in the differential response of males and females to environmental challenges, with males displaying a higher predisposition than females for liver inflammation, cirrhosis and tumors following infections or non-alcoholic steato-hepatitis^{3, 4}.

Sexually dimorphic gene expression in the liver is established by sex-specific action of liver transcription factors¹. One of those, regulator of sex limitation (Rsl), belongs to the Krüppel-associated box zinc finger proteins (KRAB-ZFP) family of tetrapod-specific transcription repressors, which counts between 300 and 400 members in mouse and human⁵. KRAB-ZFPs are characterized by the presence of an N-terminal KRAB domain required for repression and a C-terminal array of DNA-binding ZFs. Few have been assigned a physiological role in vivo, yet enough data is available to indicate that these proteins display a wide range of biological activities, from the control of imprinting to regulation of metabolic processes in neurons^{6, 7}. Whereas the role and gene targets of the vast majority of KRAB-ZFPs remain ill defined, how they repress transcription is comparatively well understood. Many KRAB-ZFPs likely bind DNA in a sequence-specific manner via their ZFs and recruit KRAB-associated protein 1 (KAP1, also known as TIF1β, TRIM28 and KRIP-1), which serves as their universal cofactor. KAP1 in turn acts as a scaffold for a chromatin-remodeling complex comprising the histone methyltransferase SETDB1, which catalyses histone 3 lysine 9 tri-methylation (H3K9me3), histone deacetylases, nuclear remodeling factors as well as heterochromatin protein 1. The formation of heterochromatin ensues, leading to epigenetic silencing⁸.

In the mouse, constitutive Kap1 knockout is early embryonic lethal, correlating with an absence of gastrulation⁹. Deleting Kap1 in ESC leads to loss of pluripotentiality, imprinting defects, and failure to silence endogenous and some exogenous retroviruses^{6, 8, 10}. In adult tissues, KAP1 regulates spermatogenesis and impacts on the management of behavioral stress^{8, 11}. Furthermore, inbred mice heterozygous for a null mutation in Kap1 were found to exhibit greater variance in body weight than wild type littermates, and for some animals liver steatosis, adipocyte hypertrophy and impaired glucose tolerance¹².

Here, we explored the role of KRAB/KAP1 in liver function. The analysis of hepato-specific conditional Kap1 KO mouse revealed a sexually dimorphic metabolic syndrome, with male-preominant liver steatosis and tumorigenesis. This correlated with sex-specific transcriptional dysregulation of genes involved in retinol, xenobiotic and drug metabolism. The identification of KAP1-binding sites in the mouse liver further revealed that a number of the affected genes are direct KAP1 targets. This study thus unveils the important contribution of KRAB/KAP1-mediated regulation to liver function and metabolism.

Experimental Procedures

Mice generation and phenotyping

Generation and genotyping of mice with a floxable Kap1 allele (Kap1^fl/fl) and the Alb.Cre mouse strain have been previously described^{9, 13}. Animals were housed in standard conditions with water and chow or high fat diet provided ad libitum - for details see Supporting Experimental Procedures. All animal experiments were approved by the local veterinary office and carried out in accordance with the European Community Council Directive (86/609/EEC) for care and use of laboratory animals. Phenotyping was performed on at least 8 males and females. For details see Supplementary Methods. Levels of metabolic parameters in the plasma were determined with the Dimension Xpand automat (Siemens Healthcare Diagnostics, Deerfield, IL, USA). Liver levels of cytokines were determined as previously described¹⁴.

Histological tissue processing and analysis

Hematoxylin and erythroeosine and Oil Red O (ORO) stainings were performed according to standard procedures. KAP1 immunohistochemistry on paraffin-embedded liver sections was performed with the polyclonal anti-TRIM28 antibody (Proteintech, UK). Histopathology evaluation, ORO scoring and tumor occurrence analysis were done by a board certified pathologist (F.A.). For BrdU pulse-labeling, mice received 3 mg/ml of BrdU in drinking water for 6 days and the liver tissue was stained with anti-BrdU antibody (Oxford Biomedical Research). TUNEL staining was performed with the Apoptag kit (Millipore) according to manufacturer’s instructions.

Small scale DNA, RNA and protein analysis

Genotyping, RNA extraction and reverse-transcription as well as protein analysis were performed according to standard procedures as previously described, see Supplemental Methods¹¹. Primer specificity was confirmed by dissociation curve analysis. β-actin and RPS9 were used for normalization. Primers used in this work are listed in the Supporting Table 2.

MicroArray analysis

8-10 weeks old mice were food-deprived at 0700 hrs and liver was isolated at 1100-1400 hrs. Total RNA was extracted with the MirVana kit (Ambion) and treated with DNAse (Turbo DNA free kit, Ambion), following manufacturer instructions. Three RNA pools from 9 mutant and wild type male livers were prepared, while three individual female KO and wild type livers were used separately. Microarray analysis was performed as previously described¹¹. Significantly deregulated genes (p<0.05 by unpaired t-test and fold-change ≥ 2) were used to interrogate DAVID bioinformatic database¹⁵.

ChIP-sequencing and ChIP-PCR

Liver chromatin was performed as previously described with minor modifications¹⁶. For ChIP-seq, chromatin from two male and female Kap1^fl/fl and one male and female Alb.Cre Kap1^fl/fl control were prepared. Chromatin immunoprecipitation with an affinity-purified rabbit polyclonal antibody raised against KAP1 amino acids 20–418 (RBCC) kindly provided by Dr. Rauscher¹⁷ was performed as previously described¹⁸. 10 ng of IP and total input were used to prepare libraries for sequencing, following the ChIP-seq library preparation protocol (See Supplementary Methods). Sequencing was performed on an Illumina Genome Analyzer IIx. Reads with less than five matches were aligned to the Mus musculus genome (assembly NCBI37/mm9) using bowtie¹⁹. Peak calling was performed with the model-based analysis of ChIP-seq algorithm²⁰ for wild type samples, using as a background the KO control. The male and female peak lists were intersected and any two peaks distant by <1kb were called overlapping. Distribution of KAP1 peaks around TSS was done with Chip-Cor tool (http://ccg.vital-it.ch/chipseq/chip_cor.html) using M. musculus ENSEMBL59 TSS collection as the input feature. For intersection with transcriptome analysis, Illumina probes were mapped to their corresponding ENSEMBL genes and overlapping peaks annotated with the bioconductor package ChIPpeakAnno (PMID: 20459804) were identified. Comparison between the distance of deregulated genes and all the UCSC annotated genes (used as control) was done using a Wilcoxon rank sum test. Frequency of KAP1 peaks was calculated as a percentage of peaks found in a given region divided by the region length in kb. Correction for multiple testing issues was done using the Benjamini-Hochberg procedure²¹.

Statistical analysis

Body and organ weight measurements as well as plasma biochemistry were analyzed with the two-tailed Student’s t test or two-way ANOVA followed by the Bonferroni post-test. Contingency testing was done with Fisher’s exact test. For statistical analysis of high throughput data see related paragraphs.

Results

Generation of hepatocyte-specific Kap1 knockout mice

We crossed Kap1^fl/fl mice⁹ in a mixed C57/Bl6-Sv129 background with animals expressing the Cre recombinase under the control of an albumin promoter¹³. We confirmed the liver-specificity of Kap1-deletion by locus-specific PCR, and determined that, in whole liver from mutant mice, KAP1 mRNA was reduced by more than 75% and protein was undetectable by Western blot, while immunohistochemistry confirmed that its loss was restricted to hepatocytes (Supporting Fig. 1). Histology combined with Oil Red O (ORO) staining further revealed that 8-10 weeks-old male Alb.Cre Kap1^fl/fl mice, but not their female counterparts, suffered from mild liver steatosis (Supporting Fig. 1). These results suggested that loss of KAP1 perturbed metabolic homeostasis in the liver, and pointed to possible sex-specific differences in this process.

Sex-dependent phenotypic consequences of hepatocyte-specific Kap1 KO

Because of this early finding and of known sex-related differences in metabolism and liver function in rodents and other mammals including humans^{1, 22}, we followed a phenotyping protocol that separately examined male and female mice (see Experimental procedures). At 8-10 weeks of age, no differences in body weight, food intake and plasma markers of liver function and lipid metabolism between mutant and wild type mice were observed (not illustrated). However, histopathology revealed significant fat accumulation with small vacuoles in all of 12 KO male livers compared with only one third (4 out of 12) of controls (Fisher’s exact test, p= 0.0013; not shown), while females were exempt of this pathology. The steatosis noted in male KO mice was not due to Cre, because no difference was observed between Alb.Cre Kap1^wt/wt and Kap1^wt/wt animals (not illustrated).

We then asked whether a metabolic challenge would accentuate the phenotype of male Kap1 KO mice. For this, we fed 10 weeks-old male and female Alb.Cre Kap1^fl/fl and Kap1^fl/fl littermates with high fat diet (HFD; 60% calories from fat) or chow diet (CD) for 19 weeks. CD-fed Kap1 KO males displayed progressive hepatosteatosis characterized by increased liver weight and fatty change of mild to moderate degree at 6 months of age (Fig. 1B and 1C, Table 1). In contrast, control littermates exhibited minimal lipid accumulation in the liver, visible only through ORO staining (Fig. 1B and 1C). These changes were not accompanied by weight gain, increased food intake and only by a trend to upregulation of plasma markers of liver function (Fig. 1A and 1D; Table 1 and data not shown). Interestingly, Kap1-KO males displayed significantly lower plasma free fatty acid (FFA) concentration than littermate controls, suggesting an increased rate of hepatocytic FFA uptake following Kap1 deletion (Table 1).

Figure 1. Phenotypic analyses of Kap1 KO mice.

A. Body weight of male (n=9-12; top) and female (n=10; bottom) mice fed indicated diets. *: p<0.05 as assessed by two-way ANOVA followed by Bonferroni post-test. CD - chow diet, HFD - high fat diet. B. HE staining of liver sample from chow diet-fed 29-week-old indicated mice. Inlet: ORO staining. Scale bar: 50 μm. C. Semi-quantitate ORO staining score of male (top) and female (bottom) mice fed indicated diets ***: p<0.001 as assessed by two-way ANOVA followed by Bonferroni post-test. D. & E. Plasma levels of alanine transaminase (ALT) and lactate dehydrogenase (LDH) (n=7-9; D.) and liver protein levels of pro-inflammatory cytokines IL6 and TNFα (n=5; E.) in 29-week-old female (left) and male (right) mice. Black bars: CD, grey bars: HFD. *: p<0.05 as assessed by two-way ANOVA followed by Bonferroni post-test. Error bars represent S.E.M.

Upon HFD challenge both Alb.Cre Kap1^fl/fl and Kap1^fl/fl males developed severe hepatic steatosis (Fig. 1C), but the liver was heavier in Kap1-KO animals, which also presented with a bigger epididymal fat pad and a decreased testis weight (Table 1). Although this latter trait suggested an impaired sex hormone balance, a condition also observed in humans in cases of severe liver steatosis⁴, total plasma testosterone levels were not different between KO and WT littermates (Table 1).

In contrast to males, CD-fed females did not present significant liver lipid accumulation, although Kap1-KO animals showed plasma dyslipidemia and upregulation of liver enzymes (Fig. 1A-D and Table 1). When subjected to HFD, Kap1 mutant females displayed a higher body weight than littermate controls probably due to increased body fat content as assessed by echoMRI (Fig. 1A and Table 1). Furthermore, these mice were characterized by i) a moderate degree of liver steatosis (Fig. 1C); ii) mildly elevated plasma cholesterol and triglycerides; and iii) significant upregulation of ALT, AST as well as a clear trend in the upregulation of other plasma markers of hepatocytic dysfunction (Fig. 1D and Table 1). Nevertheless, Kap1-KO females were comparable to their littermates for food intake, physical activity and energy expenditure (not illustrated).

Development of hepatic adenomas in Kap1 KO males

Anatomo-pathological examination indicated that male Alb.Cre Kap1^fl/fl mice on HFD had a tendency to develop liver tumors that occasionally reached large size and markedly distorted the contour of the parenchyma (Fig. 2A and 2B). Hepatic adenomas were seen in 4/11 HFD-fed and 1/12 CD-fed males at 29 weeks, while no tumors were found in Kap1^fl/fl littermates (n=9 and n=10 on CD and HFD, respectively). Moreover, one of HFD-fed Kap1-KO male tumors displayed features of carcinomatous progression. Consistent with the previously reported protective effect of female hormones³, Kap1-KO female mice appeared largely to escape HFD-induced tumorigenesis as only 1/10 HFD-fed KO females had an adenoma and no tumor was observed in wild type (n=10 on CD and n=10 on HFD) nor in CD-fed female KO mice (n=10). To confirm these results, we analyzed the livers of CD-fed mice at age of 53-72 weeks (Table 2). Gross necropsy examination revealed that Kap1-KO mice presented with hepatomegaly (Table 2) and 9/15 (64%) male Alb.Cre Kap1^fl/fl mice developed tumors compared to 2/13 (15%; p<0.024) wild type controls, while females appeared partially protected against this process with a tumor in only 4/15 (27%) KO and 1/15 (7%) wild type controls (Table 2 and Fig. 2C). Histopathological examination performed on a subset of animals (13-15 in each subgroup) confirmed that all tumors observed in Kap1-deleted mice mice were adenomas, while focal cellular alterations were detected in both control and knockout mice. In addition, some infiltration by inflammatory cells was noted in about a quarter of the animals, but no significant association wit Kap1 status was found (not illustrated). We verified that similar results were obtained in a C57/Bl6 background (Supporting Fig. 2).

Figure 2. Liver tumorigenesis in KAP1 KO mice.

A. Left, liver adenoma with a well demarcated nodular proliferation and a clear expansile growth; right, hepatocellular carcinoma with poorly demarcated irregular borders suggestive of infiltrative growth (black solid line), increased anisocytosis and anisokaryosis within the tumor, when compared to the adjacent normal hepatocytes (right) in male mouse fed HFD for 19 weeks. Scale bar in A: 2 mm. Scale bar in B: 100 μm. B. Macroscopic view of rostral (left) and caudal (right) face of livers from indicated CD-fed, 58-week-old male mice. C. Gross necropsy analysis of tumor occurrence in livers of mice at age 53-70 weeks. *: p<0.024 by Fisher’s exact test. D. Western blot of indicated proteins in Kap1 KO tumor tissue and normal liver from WT and Kap1 KO mice (left) with quantification of band intensity normalized to actin and represented as fold-change over average WT liver (right). *** - p<0.001 by one-way ANOVA E. Phosphorylation status and total protein content of AKT and various MAPK in indicated samples determined by ELISA and represented as fold-change over average WT liver. *** - p<0.001 by one-way ANOVA F. Hepatocyte proliferation in livers of 35- and 68-week-old female (n=3-5; left) and male (n=4-5; right) mice. Data represent mean number of BrdU-positive nuclei per field of view. Error bars in D., E. and F. represent S.E.M.

Despite the lack of significant inflammatory infiltration in Kap1 KO livers, we observed higher intrahepatic protein levels of pro-inflammatory cytokines IL1β, IL6 and TNFα (Fig. 1E, Tables 1 and 2). IL6 and TNFα contribute to hepatoprotection and induce hepatocyte proliferation during liver injury²⁴ and have been implicated in the development of fatty liver and tumors in other mouse models of liver tumorigenesis³. To assess the contribution to the observed phenotype of hepatocyte proliferation and hepatic cell death, we performed BrdU pulse in 35- and 68-weeks-old mice and found that hepatocyte proliferation was comparable between KO and controls (Fig. 2D). Similarly, histological examination and TUNEL staining in KAP1-depleted livers did not reveal any significant signs of cell death in Kap1-KO livers (data not shown).

In summary these results indicate that loss of KAP1 in the liver leads to sexually dimorphic phenotypes, with males exhibiting progressive steatosis and age-related tumorigenesis, and females suffering from a less dramatic syndrome with mild metabolic defects, including obesity and steatosis revealed only upon exposure to an environmental stress.

Sex-specific gene dysregulation in Kap1 knockout liver

We performed gene expression profiling in liver from CD-fed 8-10-weeks-old wild type and mutant male and female mice, choosing this young age in order to minimize the potential impact of compensatory changes. Loss of KAP1 led to at least two-fold dysregulation of 170 and 149 genes in male and female mice, respectively. Interestingly, less than 50% of the deregulated genes were common to the two sexes (Fig. 3A), suggesting that the KRAB/KAP1 system targets different genes in males and females. Gene ontology analysis of 2-fold dysregulated transcripts (Supporting Table 1) indicated that loss of KAP1 impacted on genes encoding protein endowed with oxidoreductase activity involved in xenobiotic metabolism by cytochrome P450 (Cyp450) as well as the pathways of retinol, drug and arachidonic acid metabolism (Fig. 3B). Additionally, GO terms and Kegg pathway analyses singled out genes relevant for steroid hormone biosynthesis and metabolism, peptidase inhibitor activity and PPAR signaling in male KO mice, contrasting with genes linked to pheromone binding in their female counterparts. About 50% of genes with the highest fold-change differed between male and female mice. For example, the most highly upregulated genes in males belong to phenobarbital-inducible Cyp2b family, while in females those include Spink3, Krt23 and Cyp2d9 (Fig. 3C). We verified that the transcriptional changes noted upon Kap1 deletion were not due to the mixed genetic background of the mice nor to Cre expression (Supporting Fig. 2).

Figure 3. Hepato-specific KAP1 deletion leads to sex-specific gene dysregulation.

A. Venn diagram of genes ≥ 2-fold deregulated in 10-week-old female and male KO vs WT mice. Less than 50% of deregulated genes are shared between sexes. B. Functional annotation chart of significantly deregulated (fold change ≥ 2) in Kap1-mutant livers. C. Top 10 up- and down-regulated genes in male (left) and female (right) livers. D. Expression of Rsl-regulated sexually dimorphic genes in Kap1 KO and WT male and female livers (n≥5). Error bars represent SEM. E. Dysregulation of mRNA levels of selected direct KAP1-target genes; n=3-4. Error bars represent SEM.

KAP1 controls expression of Rsl-target genes

Female-restricted changes in Cyp2d9, sex-limited protein (Slp) and major urinary proteins (Mup), are hallmarks of the Rsl loss-of-function phenotype⁵. While we could document both Cyp2d9 and Slp derepression in female Kap1 KO livers (Fig. 3D), Slp appeared affected to a far lower extent than reported in rsl-deleted female mice, although a direct comparison is difficult owing to differences in the mouse strains used in the two studies. Furthermore, we observed a downregulation, rather than an upregulation of Mups in Kap1 mutant mice at both RNA and protein level (Fig. 3D and Supporting Fig. 3), suggesting that either Rsl2-mediated female-specific repression⁵ is not mediated by KAP1 or that Mups are indirect KAP1 or Rsl targets.

KAP1 binding sites and KAP1-related chromatin modifications in the mouse liver

To understand better the molecular mechanisms of KAP1-mediated gene expression control in the liver, we performed chromatin immunoprecipitation (ChIP) followed by deep sequencing on male and female wild type and KAP1-depleted nuclei, using an antibody directed against the RBCC domain of KAP1⁸. False-positive peaks originating from non-hepatocytic cells present in the samples were minimized by following a protocol optimized for hepatocyte-centered ChIP experiments¹⁶, and data obtained in Kap1-deleted hepatocytes served as a control sample for peak calling. We identified 7158 and 5223 KAP1 peaks in male and female livers, respectively (Fig. 4A), a number comparable to that extrapolated from genome-wide KAP1 binding studies previously performed in other tissues²⁵. Surprisingly few of the identified peaks (~10%) were common to males and females (Fig. 4A), consistent with the observed sex-specific gene dysregulations induced by the Kap1 knockout (Fig. 2). KAP1 peaks were significantly enriched in vicinity of transcriptional start sites (TSS; Fig. 4B).

Figure 4. KAP1 binding sites in mouse liver.

A. Number of KAP1 peaks on genome from male and female hepatocytes. B. Distribution of KAP1 putative binding sites relative to all annotated transcriptional start sites (TSS). For distance of -200, 0 and 200 bp relative to TSS KAP1 is significantly enriched (P<0.05) as assessed by the Wilcoxon sum rank test. C. KAP1-responsive genes are closer to the nearest KAP1 peak than random genes. Distance to the nearest peak was calculated for all KAP1 up- or down-regulated genes and compared to the average distance for all genes probed by the Illumina chip. D. Frequency of KAP1 peaks in different intervals from the TSS for genes downregulated (left) or upregulated (right) in Kap1 mutant livers. Frequency is defined at the % of peaks normalized for the interval size.

We then searched the nearest peak for each of the genes dysregulated at least 1.5-fold upon KAP1 removal (Supporting Table 1). We observed that relatively few of them had a KAP1-binding site in the gene body (less than 20%; not shown), however the average distance between both upregulated and downregulated genes and the nearest KAP1 peak was significantly shorter than for all protein-coding genes present on our Illumina chip (Fig. 4C). Furthermore, these peaks were usually found just upstream of the TSS of deregulated genes (Fig. 4D). This suggests that KAP1 might directly control many of the genes perturbed in its absence, but that it may not always be through repression.

Surprisingly, for several genes expressed and deregulated in a sex-discriminating fashion, KAP1-binding was found in both males and females (Fig. 3 and 5). These included peaks in the promoter of the male-predominant Cyp2d9 and Slp genes, both strongly upregulated in Kap1 KO females; and in the glutathione S-transferase π (Gstπ) cluster, a late-phase drug-responsive gene preferentially expressed in males, with strong downregulation upon KAP1 removal. Of note, the Gstπ cluster bears a male-specific KAP1 binding site downstream of the locus that could explain sex-biased dysregulation. Moreover we identified putative KAP1-binding sites in an intron of the gene encoding fat specific protein 27 (FSP27), which plays a role in lipid droplet formation²⁶ and was strongly upregulated in male KO livers and in the gene encoding for the cellular homologue of the viral Casitas B-lineage lymphoma proto-oncogene (c-Cbl), a regulator of various tyrosine kinase signaling pathways²⁷ upregulated in KO livers of both sexes (Fig. 3E, 5 and 6A). We also identified KAP1 binding sites in a cluster of phenobarbital-responsive, female predominant Cyp2b genes that were upregulated in males and downregulated in females upon loss of KAP1 and in clusters of other xenobiotics-processing Cyp450 genes, such as Cyp2a and Cyp3a (Fig. 3E, 5, 6A and Supporting Fig. 4).

Figure 5. Selected KAP1 putative binding sites in mouse liver.

Screen-shots from the UCSC Genome Browser illustrating the KAP1 ChIP-seq data from female and male Kap1 KO (n=1) and WT (n=2) hepatocytes for indicated loci. Direction of transcription is indicated by arrow. Regions targeted by primers used for the validation by qPCR are shown.

Figure 6. Chromatin studies on KAP1-regulated genes.

A. qPCR validation of KAP1 binding in female and male livers (n=4). Data are represented as an enrichment of wild type over KO sample relative to an average of KAP1 unbound control loci (Foxp3, RPS9 and β-actin) at sites depicted in A. Error bar is S.E.M. B. & C. H3K9me3 (B.) and H3 acetylation (C.) ChIP-qPCR in female and male nuclei (n=3) of wild type and KAP1-depleted liver. Data are represented as an enrichment of wild type over KO sample relative to an average of three control loci (GAPDH, RPS9 and β-actin). Error bar is S.E.M.

KAP1 recruits the SETDB1 histone methyltransferase, which induces deposition of the H3K9me3 repressive mark, and HDAC-containing complexes, which result in histone deacetylation¹⁷. We thus compared the presence of the relevant chromatin marks in wild type and Kap1 knockout liver at both nearest KAP-binding site and promoter of a few genes suspected to represent direct KAP1 targets. Loss of KAP1 correlated with a loss of H3K9me3 at the binding site for genes upregulated in Kap1-mutant livers, such as Cyp2d9 in females and fsp27, c-Cbl or the Cyp2b cluster in males. However, for male-downregulated Gstπ, Cyp2d9 or Slp no change in H3K9me3 at the KAP1-binding site was detected. Notably, gain of H3K9me3 at the male KAP1-binding site in the fsp27 locus in female liver suggests local chromatin changes that could be the cause of mild female-upregulation of the corresponding transcript. Finally, increase in H3Ac at the promoters of Cyp2d9 in females and Fsp27 in males coincided with gene dysregulation, however this was not the case for all tested KAP1-target loci (Fig. 6B-C).

Discussion

The present work reveals the prominent role of KRAB/KAP1-mediated regulation in the control of liver metabolism and endo/xenobiotics detoxifying genes. The liver-targeted knockout of KAP1, the universal cofactor of KRAB-ZFPs, resulted in a markedly sexually dimorphic phenotype in mice, which included male-predominant steatosis and hepatic tumors. This correlated with sex-specific transcriptional dysregulation of a wide range of metabolic genes, notably those involved in retinol and sex hormone processing as well as in detoxification. An examination of KAP1 genomic binding sites in the liver and of chromatin marks at the promoters of some of these genes further revealed that a number of detoxifying genes such as Cyp2d9, Gstπ, and the Cyp2a, Cyp2b and Cyp3a gene clusters are direct KAP1 targets. Moreover, loss of KAP1 binding at the fsp27 locus was associated with the male-predominant upregulation of its product.

Male liver-specific Kap1 knockout mice displayed early onset hepatosteatosis and age-related development of liver adenomas, while their female counterparts exhibited milder metabolic disturbances revealed only by a high fat diet challenge and had lower incidences of liver adenoma than their male counterparts. This correlates with the reported resistance of the female liver to environmental insults, which has been attributed to its greater ability to inactivate toxins and neutralize reactive oxygen species, and to the protective action of estrogens against inflammation and steatofibrosis, including during the course of chronic viral hepatitis^{22, 28}. Our results suggest that these mechanisms are influenced by KRAB/KAP1-mediated transcriptional control.

What specific mechanisms account for the accumulation of fat and tumor development in the Kap1-deleted male liver remains to be determined. We observed a male-predominant, Kap1 deletion-induced upregulation of the fat specific protein 27 (FSP27), correlating with a male-restricted strong KAP1 binding site in the body of this gene. FSP27 promotes fat droplet formation in adipocytes and its hepatocytic overexpression results in hepatosteatosis^{26, 29}. This phenotypic abnormality, which represents an early sign of hepatocyte suffering common to many pathological circumstances, has been linked to increased liver damage, inflammation and development of hepatocellular carcinoma in other mouse models³. Interestingly, KAP1 KO mice displayed an upregulation of liver levels of proinflammatory cytokines IL6 and TNFα, which is a sign of a response to hepatocyte damage. Aberrant activity of signaling pathways downstream of these cytokines have been linked to increased fat accumulation in hepatocytes and development of liver tumors³. The early development of hepatosteatosis due to de-repression of fsp27 in KAP1 KO males may be one important trigger in a cascade of events leading to development of liver steatosis and tumors in these mice.

Indirectly, our analyses suggest that xenobiotic stress, which if prolonged can lead to tumorigenesis³⁰, might contribute to the observed phenotype. In Kap1-deleted livers, we measured a dysregulation of drug and xenobiotic metabolizing phase I genes such as members of the Cyp2a, Cyp2b and Cyp3a families, of phase II and III genes such as Gstπ and θ, and of genes coding for various UDP-glucuronosyltransferases required for conjugation of molecules that need this modification for efficient excretion². We identified KAP1-binding sites in the vicinity of several of these genes, suggesting that they are direct targets of the transcriptional regulator.

The combined results of our transcriptome and chromatin analyses exclude a simple model for KAP1 action, whereby a given KRAB-ZFP would tether the cofactor to specific sequences, with secondary heterochromatin formation and gene silencing. Some sexually dimorphic genes, such as the male-predominant Cyp2d9, Gstπ and Slp, carry KAP1 signatures inconsistent with a simple repression model. The Cyp2d9 and Slp genes are Rsl1-targeted, male-predominant genes upregulated upon Kap1 KO only in female mice. However, KAP1 was detected at similar sites, for example the promoter of Cyp2d9, in liver chromatin from both sexes. As well, KAP1 binds upstream of the Gstπ locus in both male and female, whereas this gene is highly expressed in the former and strongly repressed in the latter. These observations might reflect the effect of sex-specific dominant factors counteracting the repressing influence of KAP1, or sex-differential post-translational modifications of the transcription regulator. KAP1 SUMOylation appears to be required for its co-repressor activity, while phosphorylation stimulates its chromatin opening properties⁸. In the case of Cyp2d9, it could be that phospho-KAP1 keeps the promoter accessible for transcription factors in males, whereas SUMO-KAP1 recruits SETDB1 and heterochromatin protein 1 to close it down in females.

It is generally acknowledged that KAP1 exerts its gene repressor activity via an interaction with DNA-binding KRAB-ZFP family members. It will be important to identify the mediators of its activity in the liver; however it remains that our data strongly suggest that KAP1 sometimes acts as a transcriptional activator rather than as a repressor. Future studies should explore this possibility, which challenges the current view of KAP1 function.

List of Abbreviations

KRAB

Krüppel associated box

KAP1

KRAB-associated protein 1

Cyp

cytochrome p450

GST

gluthatione S-transferase

Slp

sex limited protein

Rsl

regulator of sex limitation

ZFP

zinc finger protein

SETDB1

SET domain bifurcated 1

H3K9me3

histone 3 lysine 9 tri-methylation

ESC

embryonic stem cell

Alb

albumin

Cre

Cre recombinase

ORO

Oil Red O

HFD

high fat diet

CD

chow diet

FFA

free fatty acids

AST

aspartate transaminase

ALT

alanine transaminase

IL1β

interleukin 1 β

IL6

interleukin 6

TNFα

tumor necrosis factor α

BrdU

bromodeoxyuridine

TUNEL

terminal deoxynucleotidyl transferase dUTP nick end labeling

PPAR

peroxisome proliferator receptor

Mup

major urinary protein

FSP27

fat specific protein 27

c-Cbl

Casitas B-lineage lymphoma

SUMO

small ubiquitin-like modifier