Estrogen Receptor-α Suppresses Liver Carcinogenesis and Establishes Sex-Specific Gene Expression

Abstract

Estrogen protects females from hepatocellular carcinoma (HCC). To determine whether this protection is mediated by classic estrogen receptors, we tested HCC susceptibility in estrogen receptor-deficient mice. In contrast to a previous study, we found that diethylnitrosamine induces hepatocarcinogenesis to a significantly greater extent when females lack Esr1, which encodes Estrogen Receptor-α. Relative to wild-type littermates, Esr1 knockout females developed 9-fold more tumors. Deficiency of Esr2, which encodes Estrogen Receptor-β, did not affect liver carcinogenesis in females. Using microarrays and QPCR to examine estrogen receptor effects on hepatic gene expression patterns, we found that germline Esr1 deficiency resulted in the masculinization of gene expression in the female liver. Six of the most dysregulated genes have previously been implicated in HCC. In contrast, Esr1 deletion specifically in hepatocytes of Esr1 conditional null female mice (in which Cre was expressed from the albumin promoter) resulted in the maintenance of female-specific liver gene expression. Wild-type adult females lacking ovarian estrogen due to ovariectomy, which is known to make females susceptible to HCC, also maintained female-specific expression in the liver of females. These studies indicate that Esr1 mediates liver cancer risk, and its control of sex-specific liver gene expression involves cells other than hepatocytes.

Keywords: estrogen receptor; gene expression; hepatocarcinogenesis; liver cancer; ovariectomy; sexual dimorphism.

Figure 1

Estrogen Receptor-α knockout females, but not Estrogen Receptor-β knockout females, show male-specific liver gene expression. This heat map shows Estrogen Receptor-α (Esr1) regulated transcripts in female mice that differ at least 5-fold compared to wild-type (WT) female littermates. Each lane represents microarray results from a pool of livers from individual mice sacrificed at 9–10 weeks of age. Duplicate arrays are shown for each group. WT females are represented by the first three lanes. Lane 1: C57BL/6J (B6) females (n = 18). Lane 2: WT littermates of Estrogen Receptor-α knockout (Esr1 KO) mice (n = 6). Lane 3: WT littermates of Estrogen Receptor-β knockout (Esr2 KO) mice (n = 6). Lane 4: Esr2 KO females (n = 8). Lane 5: Esr1 KO females (n = 6). WT males are represented by lanes 6–8. Lane 6: B6 males (n = 15). Lane 7: WT littermates of Esr1 KO mice (n = 6). Lane 8: WT littermates of Esr2 KO mice (n = 6). Lane 9: Esr1 KO males (n = 6).

Figure 2

Pairwise analysis shows liver transcripts affected by Esr1 knockout in females are largely sex-specific. Pairwise analysis showing log₂ fold-change in transcript expression from microarray analysis comparing Estrogen Receptor-α (Esr1) dependent transcripts in females with sex-specific transcripts. All transcripts that changed at least two-fold in either comparison were included (n = 988). Each set of transcripts represents microarray results from a pool of livers from individual mice sacrificed at 9–10 weeks of age. Wild-type (WT) females are represented by a pool of C57BL/6J (B6) females (n = 18) and WT female littermates of the Estrogen Receptor-α KO mice (n = 6) and Estrogen Receptor-β KO mice (n = 6). Estrogen Receptor-α KO females are represented by a pool of intact individuals (n = 6) and females that underwent a sham operation (n = 6).

Figure 3

QPCR analysis of Estrogen Receptor-α-sensitive transcripts identified by microarray. RNA from the livers of 9–10 week-old mice was reverse transcribed and quantified by real-time PCR. WT females (n = 5) were littermates of the Esr1 KO females. For Fmo3, Sult3a1, and Cyp4a12 the Esr1 KO female group was (n = 5) and for 3β-Hsd4/5 the Esr1 KO female group was (n = 4). The average fold change and the standard error of the mean are shown. The ΔΔCT method was used to calculate the fold change and standard error of the mean. The Wilcoxon rank sum test (two-sided) was used to test for changes in gene expression where each transcript was analyzed individually.

Figure 4

Estrogen Receptor-α-dependent gene expression is not altered in Estrogen Receptor heterozygotes, Estrogen Receptor-β knockout mice or in B6 females following alteration of ovarian hormones. Estrogen Receptor-α (Esr1) regulated transcripts in female mice showing at least 5-fold relative expression difference compared to wild-type (WT) female littermates are shown. Each lane represents microarray results from a pool of livers from individual mice sacrificed at 9–10 weeks of age. Duplicate arrays are shown for Lane 1: C57BL/6J (B6) sham operation placebo group (n = 7) and Lane 2: B6 ovariectomized animals with 17β-Estradiol add-back (n = 6) (Lane 2). A single array is shown for Lane 3: a pool of B6 females that underwent ovariectomy and implantation of a placebo pellet (n = 6), Lane 4: B6 animals with ovariectomy but no placebo implant (n = 5), Lane 5: Esr1 heterozygous females (n = 6), and Lane 6: Esr2 heterozygous females (n = 6). Duplicate arrays are shown for Lane 7: Esr2KO females (n = 8), and Lane 8: sham-operated Esr2KO females with placebo implants (n = 6).

Figure 5

Hepatocyte-specific loss of Estrogen Receptor-α has little effect on sex-specific liver gene expression. This heat map shows gene expression associated with hepatocyte-specific loss of Estrogen Receptor-α (LERKO). Transcripts differing ≥ 5-fold in the global Estrogen Receptor-α knockout female group compared to WT females are shown. Each lane represents microarray results from individual mice sacrificed at 9–10 weeks of age. A single array was run for each animal. Lane 1: wild-type (WT) female littermates (n = 3). Lane 2: Female LERKO mice (n = 3). Lane 3: WT male littermates (n = 3). Lane 4: Male LERKO mice (n = 3).