Specific genomic and transcriptomic aberrations in tumors induced by partial hepatectomy of a chronically inflamed murine liver

Abstract

Resection of hepatocellular carcinoma (HCC) tumors by partial hepatectomy (PHx) is associated with promoting hepatocarcinogenesis. We have previously reported that PHx promotes hepatocarcinogenesis in the Mdr2-knockout (Mdr2-KO) mouse, a model for inflammation-mediated HCC. Now, to explore the molecular mechanisms underlying the tumor-promoting effect of PHx, we compared genomic and transcriptomic profiles of HCC tumors developing in the Mdr2-KO mice either spontaneously or following PHx. PHx accelerated HCC development in these mice by four months. PHx-induced tumors had major chromosomal aberrations: all were amplifications affecting multiple chromosomes. Most of these amplifications were located near the acrocentric centromeres of murine chromosomes. Four different chromosomal regions were amplified each in at least three tumors. The human orthologs of these common amplified regions are known to be amplified in HCC. All tumors of untreated mice had chromosomal aberrations, including both deletions and amplifications. Amplifications in spontaneous tumors affected fewer chromosomes and were not located preferentially at the chromosomal edges. Comparison of gene expression profiles revealed a significantly enriched expression of oncogenes, chromosomal instability markers and E2F1 targets in the post-PHx compared to spontaneous tumors. Both tumor groups shared the same frequent amplification at chromosome 18. Here, we revealed that one of the regulatory genes encoded by this amplified region, Crem, was over-expressed in the nuclei of murine and human HCC cells in vivo, and that it stimulated proliferation of human HCC cells in vitro. Our results demonstrate that PHx of a chronically inflamed liver directed tumor development to a discrete pathway characterized by amplification of specific chromosomal regions and expression of specific tumor-promoting genes. Crem is a new candidate HCC oncogene frequently amplified in this model and frequently over-expressed in human HCC.

Figure 1. Accelerated HCC development in the Mdr2-KO/FVB mice following 70% PHx

Either PHx or sham surgery was performed at the age of either three (A-C) or six (D-F) months; all mice were sacrificed at the age of nine months. Tumor incidence in female (A) or male (B) mice, or females and males together (C) operated at the age of three months (females: PHx n=17, sham n=16; males: PHx n=18, sham n=22). Tumor incidence in female (D) or male (E) mice, or females and males together (F) operated at the age of six months (females: PHx n=25, sham n=10; males: PHx n=20, sham n=11).

Figure 2. Quantification of proliferation and DNA damage markers in the liver of hepatectomized and untreated Mdr2-KO/FVB mice

Comparative quantification of the appropriate markers in the distant non-tumor (N), surrounding non-tumor (S) and tumor (T) liver tissues from 9-month-old post-PHx (PHx) and 13-14-month-old untreated (Spon) Mdr2-KO mice. The representative IHC pictures are shown in Supplemental Figure 3. Frequency of hepatocytes positive for: (A) BrdU; (B) phospho-histone-3; (C) mitotic figures; (D) γH2AX. *, statistical significance between the surrounding and non-tumorous tissues of the same group (ttest p≤0.002); **, statistical significance between tumors and non-tumor/surrounding tissues (ttest p≤0.001). Six mice per each group. HPF: high power field.

Figure 3. Specific pattern of genomic amplifications observed in the post-PHx tumors of Mdr2-KO/FVB mice

(A, B) Typical profiles of chromosomal aberrations in either a post-PHx tumor (A), containing only amplifications (gray bars on the right of each chromosome), or a spontaneous tumor (B), containing both amplifications and deletions (black bars on the right of each chromosome); centromeres - at the top of each chromosome. (C) Four genomic regions that were frequently amplified in the post-PHx tumors. (D) In post-PHx tumors, the average amplification level positively correlated with tumor size. Axis “Y” - average amplification level (fold-change) of common amplified regions per tumor; axis “X” - tumor diameter, mm. Pearson correlation r=0.9234, p=0.0086.

Figure 4. Gene expression profiling of the post-PHx and spontaneous tumors from Mdr2-KO mice

(A, B) Venn diagram, comparing numbers of differentially expressed liver genes (tumor/non-tumor) in the 9-month-old post-PHx mice with those of untreated 16-month-old mice; six pairs of tumors and their matched non-tumor liver tissues from each group. (A) Genes down-regulated in at least four of six tumors below the threshold of −1.8-fold-change. (B) Genes up-regulated in at least three of six tumors above the threshold of 1.8-fold-change. (C-E) H19 expression (qRT-PCR) in the distant non-tumor (N) and tumor (T) liver tissues from 9-month-old post-PHx (PHx) and 13-14-month-old untreated (Spon) Mdr2-KO females (C) and males (D), or in 9-month-old controls (E): Hz, Mdr2-heterozygotes; F or M - sham-operated Mdr2-KO females or males. Four-to-five mice per group; statistics - using the F-test.

Figure 5. Post-PHx tumors and their matched non-tumor liver tissues of Mdr2-KO mice have prominent “poor prognosis in HCC” signatures

(A-B) Comparison of the “tumor/non-tumor tissue” gene expression signatures of Racgap1-interacting genes (axis “Y” – fold-change ratio, microarray data) between the six tumors from 9-month-old post-PHx Mdr2-KO mice (A) and the six tumors from untreated 16-month-old Mdr2-KO mice (B). (C-D) Comparison of “non-tumor poor prognosis in HCC signatures” between non-tumor livers from the following groups of 9-month-old mice: four Mdr2-KO post-PHx (PHx), four untreated Mdr2-KO (KO d0), and three untreated Mdr2-heterozygous controls (Ctrl d0). Axis “Y” - absolute gene expression levels, microarray data. (C) Genes of the “non-tumor poor prognosis in HCC signature”. (D) A set of 16 housekeeping genes used for normalization of two datasets.

Figure 6. Expression of the Crem protein isoforms in murine liver at tumor stage and in human and murine HCC cells

Immunoblotting of the Crem protein in murine HCC tumors (A) and in human and murine HCC cell lines (B); Crem isoforms are designated on the right. In (A): upper panel (T) – 12 murine tumors (6 post-PHx and 6 spontaneous) with the Crem amplification numbers above (tumor/non-tumor, based on aCGH); middle panel (N) – their matched non-tumor liver samples; bottom panel – beta-actin.

Figure 7. Increased Crem protein expression in the nuclei of HCC tumor cells

Representative images of IHC staining of the Crem protein in murine (A) and human (B) HCC (T) and adjacent non-tumor liver (N). Magnification x100.

Figure 8. Inhibitory effect of the Crem protein knockdown on proliferation of HCC cells

Reduced proliferation rate of two tested human HCC cell lines following CREM knockdown by siRNA: mock-treated (blue rectangles), or siLuc-treated (green squares), or siCREM-treated (red circles) Hep40 (A) and Huh7 (B) cells (ttest p=0.01 for both cell lines when compared siCREM to siLuc).