Chronic hypoxia is associated with transcriptomic reprogramming and increased genomic instability in cancer cells

Abstract

Hypoxia afflicts the microenvironment of solid tumors fueling malignancy. We investigated the impact of long hypoxia exposure on transcriptional remodeling, tumor mutational burden (TMB), and genomic instability of cancer cells that were grouped based on their inherent sensitivity or resistance to hypoxia. A hypoxia score was used as a metric to distinguish between the most hypoxia-sensitive (hypoxia high (HH)), and most resistant (hypoxia low (HL)) cancer cells. By applying whole exome sequencing and microarray analysis, we showed that the HH group was indeed more sensitive to hypoxia, having significantly higher TMB (p = 0.03) and copy number losses (p = 0.03), as well as a trend of higher transcriptional response. Globally cells adapted by decreasing expression of genes involved in metabolism, proliferation, and protein maturation, and increasing alternative splicing. They accumulated mutations, especially frameshift insertions, and harbored increased copy number alterations, indicating increased genomic instability. Cells showing highest TMB simultaneously experienced a significant downregulation of DNA replication and repair and chromosomal maintenance pathways. A sixteen-gene common response to chronic hypoxia was put forth, including genes regulating angiogenesis and proliferation. Our findings show that chronic hypoxia enables survival of tumor cells by metabolic reprogramming, modulating proliferation, and increasing genomic instability. They additionally highlight key adaptive pathways that can potentially be targeted to prevent cancer cells residing in chronically hypoxic tumor areas from thriving.

Keywords: chronic hypoxia; copy number variation; genomic instability; hypoxia; solid tumors; transcriptome remodeling; tumor mutational burden.

FIGURE 1

Hypoxia-scored cancer cell lines. (A) Heatmap of the fold change of the eight genes across the twelve cancer cell lines. (B) Relative hypoxia score of each cell line ordered from highest to lowest. Cell lines with relative hypoxia score greater than zero are considered as hypoxia high (HH) and those less than zero are considered as hypoxia low (HL).

FIGURE 2

Differentially expressed transcripts in hypoxia. (A) The total number of transcript clusters (TCs) and the downregulated and upregulated number of TCs in hypoxia (H) versus normoxia (N) in each cell line. (B) The mean number of differentially expressed TCs in hypoxia high (HH) versus hypoxia low (HL) groups. (C) Heatmaps of the HH specific differentially expressed genes (left panel) and HL specific differentially expressed genes (right panel). (D) Heatmap of the common differentially expressed genes in both HH and HL groups. Statistical analysis based on unpaired two-tailed t-test with p ≤ 0.05 considered statistically significant.

FIGURE 3

Pathways enriched following hypoxic stress. Cell signaling and response-related pathways (A), metabolism-related pathways (B), DNA, RNA, or protein processing-related pathways (C) and cell cycle-related pathways (D) significantly enriched in the hypoxic (NES>0) or the normoxic (NES<0) condition based on gene set enrichment analysis. Only pathways with FDR q-value ≤0.05 are included. NES: normalized enrichment score.

FIGURE 4

Mutational landscape of the highest and lowest scoring cancer cell lines. (A) Median tumor mutational burden (TMB) per megabase (Mb) of genome. Each symbol represents one sample. (B) Number of variants in the different variant classifications. (C) Number of single nucleotide variant (SNV) in each class. (D) Percentage of SNVs that are transitions (Trans) and Transversions (Transv) per cell line. Each symbol represents one sample.

FIGURE 5

Mutational landscape of the hypoxia high (HH) versus hypoxia low (HL) groups. (A) Mean tumor mutational burden (TMB) per megabase (Mb) of genome for each group. (B) Number of variants in the different variant classifications; number of single nucleotide variant (SNV) in the different classes and the percentage of SNVs that are transitions (Trans) and Transversions (Transv) (C) The number and type of variants in the top ten mutated genes with the percentage of samples harboring mutations in each gene.

FIGURE 6

Top ten mutated genes in each cell line and corresponding tumor TCGA dataset. (A, C) Number and type of variants in the top ten mutated genes with the percentage of samples harboring mutations in each gene. Color code reflects the variant class. (B, D) The frequency (in percent) of alteration events of the top ten mutated genes in hypoxia high (HH) versus hypoxia low (HL) patient groups with lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), breast invasive carcinoma (BRCA), colorectal adenocarcinoma (COAD), pancreatic ductal adenocarcinoma (PDAC) and ovarian serous cystadenocarcinoma (OV).

FIGURE 7

Copy number variation (CNV) profiles of cell lines in hypoxia. Number of windows with copy number variation (including both gains and losses), copy number gain (CNG) and copy number loss (CNL) per cell line (A) and per hypoxia group (B). Each window is 50 kilo base pair (kbp). HH: hypoxia high; HL: hypoxia low. Statistical analysis based on unpaired two-tailed t-test with p ≤ 0.05 considered statistically significant.

FIGURE 8

Downregulation of DNA replication and repair related pathways in hypoxia. (A) Significantly enriched pathways related to DNA replication or repair and chromosomal maintenance in the hypoxic (NES>0) or the normoxic (NES<0) condition based on gene set enrichment analysis. Only pathways with FDR q-value ≤0.05 are included. NES: normalized enrichment score. (B) Log2 fold change (FC) of genes related to the same pathways in H226, MCF-7 and Capan-1 in hypoxia versus normoxia. Statistical analysis based on ordinary one-way ANOVA with Sidak correction for multiple testing. p ≤ 0.05 considered statistically significant: *p = 0.032; **p = 0.0021; ***p = 0.0002; ****p < 0.0001.

FIGURE 9

Differential enrichment of immune-related pathways. (A) Significantly enriched pathways related to immune response enriched in the hypoxic (NES>0) or the normoxic (NES<0) condition based on gene set enrichment analysis. Only pathways with FDR q-value ≤0.05 are included. NES: normalized enrichment score. (B) Log2 fold change of genes related to the same pathways in H226, MCF-7 and MIA PaCa-2 in hypoxia versus normoxia. Statistical analysis based on ordinary one-way ANOVA with Sidak correction for multiple testing. p ≤ 0.05 considered statistically significant: *p = 0.032; **p = 0.0021; ***p = 0.0002; ****p < 0.0001.