Fxr/Shp double knockout (DKO) mouse model recapitulates sexdifference observed in HCC incidence.

(A-E) One-year-old DKO male mice developed hepatocellular carcinoma, which was not observed in age-matched wild type (WT) and DKO female mice. (B and E) Representative H&E stained liver sections from a (B) DKO male and (E) female. Inflammation and injury are evident at 1 year, and the dotted line (B) separates the HCC with large nuclei on lower right. Sirius red staining shows increased collagen in a perisinusoidal distribution, which is greater in the DKO males (C and F). The liver-to-body weight ratio was significantly higher in DKO males (G). Compared to WT animals, serum markers of liver injury, (H) AST and ALT were higher in DKO mice. (I-J) Analysis of five different HCC clinical cohorts (n=1000) patients reveals a reduction in Fxr and Shp transcript levels in patients with liver tumors. n=5-10 mice /group; mean ± SEM; *p<0.01, **p<0.001 compared to genotype or gender controls. One-way ANOVA with Bonferroni post hoc analysis was performed.

Transcriptome analysis reveals striking sex differences in hepatic metabolism.

Microarray was performed on liver tissue from WT and DKO mice of both sexes (n = 6/group). (A-B) GO categories were determined using genes with >1.3 fold change in expression between DKO males and females. Enrichment of overlapping GO categories between males and females was determined by comparing – log p-values for each term. (C) GO categories unique to the set of genes upregulated >1.3 fold in DKO males and (D) DKO females.

Gene signatures obtained from the DKO mouse model correlate well with the clinical outcomes of HCC patient.

The survival probability based on WT and DKO transcriptome changes was evaluated using five different HCC clinical cohorts. (A-C) Analysis of OS (Overall Survival) and RFS (Recurrence Free Survival) in patients using the gene signatures representative of either male WT or DKO mice (A), female WT or DKO mice (B), and unique changes observed in female DKO mice (C).

Estrogen signaling protects against liver tumorigenesis and can regulate BA synthesis in DKO female mice.

(A-B) Ovariectomized female DKO mice were aged to a year and examined for liver tumorigenesis, where a dotted line demarcates the tumor margin. (C) Serum total bile acid concentrations. (D) Experimental design of chow and 1% cholic acid (CA) diet for 1 week with or without (OVX). (E) Expression of hepatic Era was induced with CA diet in DKO female mice and reduced in both WT and DKO females following ovariectomy. (F) CA-mediated suppression of Cyp7a1 and (G) Cyp8b1 in WT and DKO females was lost in DKO females after OVX. (H) Sult2a1 has greater baseline expression in DKO mice, induced to a lesser extent upon CA challenge compared to WT animals (n = 4-5/group). (I) ChIP-PCR was performed in WT and DKO male and female livers to test ERα recruitment to BA synthesis and metabolism genes, Cyp7a1, Cyp8b1, and Sult2a1. Mean ± SEM; Two-way ANOVA with Bonferroni post hoc analysis was performed. #p<0.05, *p<0.01, **p<0.001 compared to controls.

BA composition and metabolism are differentially regulated between the sexes of DKO mice.

(A) Hepatic mRNA expression of classical BA synthetic enzymes was elevated in DKO compared to WT mice. While the alternative BA synthesis encoding gene, Cyp27a1, was increased in females only. (B) Expression of hepatic BA transporters and (C) BA sulfotransferase in WT and DKO mice. (D) Percentages of sulfated BAs in DKO male and female serum (one-tailed t-test). (E-F) BA composition is slightly varied in serum, whereas it remains unchanged in the liver between DKO males and females. (G-H) BA composition in the urine was variable between the sexes, and BAs constitute a higher proportion of fecal metabolites in the DKO females compared to males (n = 5-7/group). Mean ± SEM; #p<0.05, *p<0.01, **p<0.001 compared to genotype or gender controls. One-way ANOVA with Bonferroni post hoc analysis was performed.

Treating with BA-binding resin reduces the tumor burden in DKO male mice.

(A) DKO male mice were fed a 2% cholestyramine (CHR)-enriched diet for 3 months until one year of age. (B-C) Serum BA levels and composition upon feeding DKO male mice CHR-enriched diet. (D) CHR dramatically reduced the HCC burden in DKO males. (E) Histological analysis shows HCC, bland tumor cells, and enlarged nuclei with irregular membranes in DKO male mice. CHR treatment results in smaller and fewer nodules but increases steatosis. (n = 6-7/group). Mean ± SEM; *p<0.01; **p<0.001 compared to DKO controls.

DKO female mice exhibit reduced liver injury.

(A-B) Compared to DKO males, the female mice showed smaller liver size and liver-to-body weight ratio. (C) Sirius red staining revealed less collagen staining (red) in DKO females. (Six month old mice, n = 6/group). Mean ± SEM. *p<0.01, **p<0.001 compared to genotype or gender controls. Unpaired t-test was used to analyze the data.

Schematic of the pipeline used to obtain gene signatures to predict outcomes in HCC patient cohorts.

(A) Workflow schematic of the gene expression analysis. (B-C) Different gene signatures were extracted from the microarray data by comparing wild-type (WT) male and female gene expression patterns with that of DKO male and DKO female, respectively. These gene lists were used to predict overall survival and recurrence-free survival using five human clinical cohorts. (D) Gene changes in these defined sets DKO males (DKO_M), DKO females (DKO_F), or combined (DKO_ALL), or DKO_F vs M (DKO female gene signature that does not overlap with DKO males), DKO_Estrogen, DKO_BA, and DKO_Urea are shown. These genes were selected by t-test and log2 Fold Change (p<0.001 and log2 FC>1 or <-1).

Correlation of DKO gene signatures with various clinical stages of HCC.

Clinical relevance of combined gene changes in DKO male and female (DKO_Combined), or DKO male-specific, and the distinct gene set of DKO female vs. male (DKO_FvsM) were tested in five HCC cohorts. The trained BCCP (Bayesian compound covariate predictor) algorithm using the three DKO gene sets was tested across the HCC stages as determined by (A-C) CLIP (Cancer of the Liver Italian Program) or (D-F) TNM (Tumor, Node, and Metastasis). The probability of gene signatures was generated from 0 (not predictable) to 1 (predictable).

Expression of the urea cycle genes are reduced in human liver cancer.

RNA-seq from TCGA-LIHC (The Cancer Genome Atlas-Liver Hepatocellular Carcinoma Collection) database was analyzed. Several genes that encode enzymes involved in ureagenesis exhibited a reduction in their transcript levels in tumors compared to non-tumor (NT) tissue. (Males: n =29 NT, n=245 tumors; Females: n=20 NT and n=114 tumors). Mean ± SEM; *p<0.01, **p<0.001, ***p<0.0001 compared to their respective sex-specific controls. One-way ANOVA with Bonferroni post hoc analysis was performed.

Urea cycle genes elevated in DKO females correlate with better patient survival and reduced after ovariectomy.

(A) Schematic of the urea cycle, with a representative heat map of urea cycle genes from the microarray showing higher expression in DKO female livers (n = 6/group). (B) Survival curves from HCC patients who exhibit increased expression of urea cycle genes. (C) Expression of urea cycle genes was reduced in DKO female mice following OVX (n = 4-5/group). and (D) LC-MS quantification of urea cycle intermediates (n = 4-5/group). Mean ± SEM; *p<0.01, **p<0.001 compared to genotype or gender controls. Unpaired t-test was used for analysis.

Estrogen receptor signaling positively correlates with better survival in HCC clinical samples.

(A) RNA-seq from TCGA-LIHC (The Cancer Genome Atlas-Liver Hepatocellular Carcinoma Collection) data was analyzed. Erα transcript levels were reduced in tumors compared to non-tumor (NT) tissue. (Males: n =29 NT, n=245 tumors; Females: n=20 NT and n=114 tumors). Mean ± SEM; *p<0.01, **p<0.001, ***p<0.0001 compared to their respective sex-specific controls. One-way ANOVA with Bonferroni post hoc analysis was performed. (B) Kaplan Meier Survival graphs using Erα targets altered in DKO females (DKO_Estrogen) showed better clinical outcomes for HCC patients.

DKO females challenged with the DDC diet developed liver tumors.

(A-B) DKO females were fed a 0.1% DDC diet, as shown in the schematic, for 5 months, and this treatment led to visible tumors at one year. (C-D) The diet elevated liver-to-body weight ratio and serum BAs in DKO females. (E) Histology revealed cholangitis, ductular reaction, and tumor nodules (marked by a dotted line) in DKO female livers+ DDC treatment. (n = 4-7/group). Mean ± SEM; *p<0.01; **p<0.001 compared to chow-fed DKO controls. Unpaired t-test was used for analysis.

BA composition and expression of BA synthesis genes in DKO mice after different treatments.

(A) Correlation analysis of serum and hepatic composition of BAs post-DDC diet in DKO female mice (n = 5-8 mice/group) and post-CHR diet in DKO male mice (n = 4-5 mice/group) along with their sex-specific chow controls. (B-C) Relative expression of classical and alternative BA synthesis genes after CHR and DDC treatments. Mean ± SEM; *p<0.01; **p<0.001, *** p<0.0001 compared to their chow-fed DKO controls. Unpaired t-test was used for analysis.

Transcription Motifs Enriched in DKO male livers compared to DKO females.

Transcription Motifs Enriched in DKO female livers compared to DKO males.

Summary of HCC gene expression data sets

Serum BA composition in DKO mice.

Urine BA composition in DKO mice.

Primer sequences used