Systematic Identification of Characteristic Genes of Ovarian Clear Cell Carcinoma Compared with High-Grade Serous Carcinoma Based on RNA-Sequencing

Abstract

Objective: Ovarian cancer has the highest mortality among gynecological cancers. High-grade serous carcinoma (HGSC) is the most common histotype of ovarian cancer regardless of ethnicity, whereas clear cell carcinoma (CCC) is more common in East Asians than Caucasians. The elucidation of predominant signaling pathways in these cancers is the first step towards understanding their molecular mechanisms and developing their clinical management.

Methods: RNA sequencing was performed for 27 clinical ovarian specimens from Japanese women. Principal component analysis (PCA) was conducted on the sequence data mapped on RefSeq with normalized read counts, and functional annotation analysis was performed on genes with substantial weights in PCA. Knockdown experiments were conducted on the selected genes on the basis of PCA.

Results: Functional annotation analysis of PCA-defined genes showed predominant pathways, such as cell growth regulators and blood coagulators in CCC and transcription regulators in HGSC. Knockdown experiments showed that the inhibition of the calcium-dependent protein copine 8 (CPNE8) and the transcription factor basic helix-loop-helix family member e 41 (BHLHE41) repressed the proliferation of CCC- and HGSC-derived cells, respectively.

Conclusions: This study identified CPNE8 and BHLHE41 as characteristic genes for CCC and HGSC, respectively. The systemic identification of differentially expressed genes in CCC and HGSC will provide useful information to understand transcriptomic differences in these ovarian cancers and to further develop potential diagnostic and therapeutic options for advanced disease.

Keywords: RNA sequencing; clear cell carcinoma; gene expression; high-grade serous carcinoma; ovarian cancer.

Figure 1

Principal component analysis (PCA) using log₂ read per kilobase of transcript length per million mapped reads (RPKM) of RefSeq genes showed transcriptomic differences among pathological groups. (A) Schematic presentation of transcriptomic and functional analysis of ovarian cancer performed in this study. Briefly, RNAs were isolated from 27 clinical ovarian specimens and analyzed by RNA sequencing. Based on the log2RPKM values for each RefSeq gene, PCA using log2RPKM of Refseq genes was performed. Then, functional analysis using small interfering RNA (siRNA) was also performed. (B) Plot of the first two components from PCA of the log2RPKM of RefSeq genes. The prcomp function of R-package was used for the PCA of clear cell carcinoma (CCC) (n = 6), high-grade serous carcinoma (HGSC) (n = 15), and normal tissues (n = 6). PC1, 1st principal component; PC2, 2nd principal component. (C) Plot of the first two components of the CCC and normal tissue groups. (D) Plot of the first two components of the HGSC and normal tissue groups. (E) Plot of the first two components of the CCC and HGSC groups. CCC and HGSC are indicated by blue and red solid circles, respectively. Normal tissues are indicated by green circles, with the solid ones representing normal ovary tissues and the open ones representing normal oviduct tissue.

Figure 2

Expression levels of genes contributing to the first principal component of PCA in CCC and HGSC. (A) Expression levels (log2RPKM) of the top 12 genes contributing to PC1 negative in Figure 1E were collected from RNA-sequencing data. The box plot was created by the statistical analysis software JMP. Higher expression levels were shown in CCC compared with HGSC. (B) Expression levels (log2RPKM) of the top 12 genes contributing to PC1 positive in Figure 1E were collected from RNA-sequencing data. Higher expression levels were shown in HGSC compared with CCC. Results are shown as means ± SD. Statistical analysis was performed using Student’s t test followed by FDR (5%) analysis.

Figure 3

mRNA expression of CPNE8 and BHLHE41 in ovarian cancer cell lines. (A) The expression levels of the top 12 genes that separated CCC from HGSC (PC1 negativity in Figure 1E) were examined in HGSC-derived OVCAR3, CCC-derived RMG1, and adenocarcinoma-derived SKOV3 cells by qRT-PCR. (B) The expression levels of the top 12 genes that separated HGSC from CCC (PC1 positivity in Figure 1E) were examined in OVCAR3, RMG1, and SKOV3 cells by qRT-PCR. Relative mRNA levels are expressed as means ± SD (n = 3), by normalizing to GAPDH level. Row data are plotted with circles; *, p < 0.05; **, p < 0.01, using one-way ANOVA with Tukey’s HSD test; n.s., not significant.

Figure 4

Growth inhibitory effects of siRNAs targeting CPNE8 and BHLHE41 in ovarian cancer cells. (A) The knockdown efficiency of CPNE8-specific siRNAs (siCPNE8 #A and #B) was examined in OVCAR3, RMG1 and, SKOV3 cells (n = 3, left panel). A DNA assay was performed to assess cell growth 4 days after siCPNE8 #A, #B, or control siRNA (siControl) transfection (n = 5, right panel). (B) The knockdown efficiency of BHLHE41-specific siRNAs (siBHLHE41 #A and #B) (n = 3, left panel) and its effect on cell growth (n = 5, right panel) were examined as in (A). Row data are indicated by circles. Results are shown as means ± SD; *, p < 0.05; **, p < 0.01, using one-way ANOVA with Tukey’s HSD test; n.d., not detected.

Figure 5

Enriched pathways observed with siRNAs targeting CPNE8 and BHLHE41. (A) Gene Set Enrichment Analysis (GSEA) enrichment plot of microarray analysis of RMG1 cells in siControl versus CPNE8-specific siRNAs. (B) GSEA enrichment plot of microarray analysis of OVCAR3 cells in siControl versus BHLHE41-specific siRNAs. The X and Y axes represent “rank in ordered dataset” and “enrichment score”, respectively.

Figure 6

Schema of the present study. PCA based on mRNA levels analyzed by RNA sequencing was conducted in CCC and HGSC groups. The top three most highly expressed genes were chosen from the genes contributing to the first principal component of PCA in CCC and HGSC. Knockdown experiments for the selected genes showed that CPNE8 and BHLHE41 promoted the proliferation of CCC- and HGSC-derived cells, respectively. Functional enrichment and pathway analyses also showed that these two genes were related to cancer cell growth signaling.