亜森 マハムット

BACKGROUND: The intratumor heterogeneity (ITH) of cancer cells plays an important role in breast cancer resistance and recurrence. To develop better therapeutic strategies, it is necessary to understand the molecular mechanisms underlying ITH and their functional significance. Patient-derived organoids (PDOs) have recently been utilized in cancer research. They can also be used to study ITH as cancer cell diversity is thought to be maintained within the organoid line. However, no reports investigated intratumor transcriptomic heterogeneity in organoids derived from patients with breast cancer. This study aimed to investigate transcriptomic ITH in breast cancer PDOs. METHODS: We established PDO lines from ten patients with breast cancer and performed single-cell transcriptomic analysis. First, we clustered cancer cells for each PDO using the Seurat package. Then, we defined and compared the cluster-specific gene signature (ClustGS) corresponding to each cell cluster in each PDO. RESULTS: Cancer cells were clustered into 3-6 cell populations with distinct cellular states in each PDO line. We identified 38 clusters with ClustGS in 10 PDO lines and used Jaccard similarity index to compare the similarity of these signatures. We found that 29 signatures could be categorized into 7 shared meta-ClustGSs, such as those related to the cell cycle or epithelial-mesenchymal transition, and 9 signatures were unique to single PDO lines. These unique cell populations appeared to represent the characteristics of the original tumors derived from patients. CONCLUSIONS: We confirmed the existence of transcriptomic ITH in breast cancer PDOs. Some cellular states were commonly observed in multiple PDOs, whereas others were specific to single PDO lines. The combination of these shared and unique cellular states formed the ITH of each PDO.

PURPOSE: Recent genetic studies have suggested that tumor suppressor genes are often silenced during carcinogenesis via epigenetic modification caused by methylation of promoter CpG islands. Here, we characterized genes inactivated by DNA methylation in human hepatocellular carcinoma (HCC) to identify the genes and pathways involved in DNA methylation in hepatocellular carcinoma. METHODS: Eight HCC-derived cell lines were treated with a DNA demethylating agent, 5-aza-2'-deoxycytidine. Additionally, 100 pairs of primary HCC and adjacent non-cancerous tissues as well as 15 normal liver tissues were analyzed by comprehensive gene expression analysis using microarrays. Moreover, gene set enrichment analysis identified the major molecular pathways associated with DNA methylation. Validation of gene expression and DNA methylation status was performed by real-time PCR after bisulfite modification. RESULTS: We showed that CXCL2, an immune-related chemokine, expression was significantly downregulated in tumor tissues and also significantly upregulated by DAC treatment in cell lines. Furthermore, we observed a statistically significant difference in methylation status between normal liver tissues and tumor tissues (P < 0.05). In addition, tumors with higher CXCL2 expression included significantly more numbers of multiple tumors than the lower expression group. CONCLUSIONS: We identified CXCL2, an immune-related chemokine, decreased in hepatocellular carcinoma and the regulation mechanism may be controlled by methylation. Further studies should be warranted to examine if and to what extent the gene is actually suppressed by methylation and if there is a possibility that the CXCL2 axis plays a role for diagnosis and treatment of hepatocellular carcinoma.

Microarray analysis has been applied to comprehensively reveal the abnormalities of DNA copy number (CN) and gene expression in human cancer research during the last decade. These analyses have individually contributed to identify the genes associated with carcinogenesis, progression, metastasis of tumor cells and poor prognosis of cancer patients. However, it is known that the correlation between profiles of CN and gene expression does not highly correlate. Factors which determine the degree of correlation remain largely unexplained. To investigate one such factor, we performed trend analyses between the lengths of CN segments and corresponding gene expression profiles from microarray data in hepatocellular carcinoma (HCC) and colorectal carcinoma (CRC). Significant correlations were observed in CN gain of HCC and CRC (p<0.05). The trend of the CN loss showed a significant correlation in HCC although there was no correlation between the length of CN loss segments and gene expression in CRC. Our findings suggest that the influence of CN on gene expression highly depends on the length of CN region, especially in the case of CN gain. To the best of our knowledge, this is the first study describing the correlation between lengths of CNA segments and expression profiles of corresponding genes.