Molecular Portrait of Hypoxia in Breast Cancer: A Prognostic Signature and Novel HIF-Regulated Genes

Abstract

Intratumoral hypoxia has been associated with invasion, metastasis, and treatment failure, prompting the need for a global characterization of the response to hypoxic conditions. The current study presents the results of a large-scale RNA sequencing (RNA-seq) effort, analyzing 31 breast cancer cell lines representative of breast cancer subtypes or normal mammary epithelial (NME) cells exposed to control tissue culture conditions (20% O₂) or hypoxic conditions (1% O₂). The results demonstrate that NME have a stronger response to hypoxia both in terms of number of genes induced by hypoxia as well as level of expression. A conserved 42-gene hypoxia signature shared across PAM50 subtypes and genes that are exclusively upregulated in Luminal A, Luminal B, and normal-like mammary epithelial cells is identified. The 42-gene expression signature is enriched in a subset of basal-like cell lines and tumors and differentiates survival among patients with basal-like tumors. Mechanistically, the hypoxia-inducible factors (HIF-1 and/or HIF-2) mediate the conserved hypoxic response. Also, four novel hypoxia-regulated and HIF-1-responsive genes were identified as part of the conserved signature. This dataset provides a novel resource to query transcriptional changes that occur in response to hypoxia and serves as a starting point for a clinical assay to aid in stratifying patients that would benefit from hypoxia-targeted therapies, some of which are currently in clinical trials. IMPLICATIONS: RNA-seq of 31 breast cancer cells exposed to control or hypoxic conditions reveals a conserved genomic signature that contains novel HIF-regulated genes and is prognostic for the survival of patients with triple-negative breast cancer.

Figure 1.

RNA sequencing of 31 cells line exposed to 20% O₂ and 1% O₂ reveals a conserved set of hypoxia-inducible genes. A-C, qPCR of RNA lysates from 34 breast cancer cells lines to determine the fold change in the induction of P4HA1 (A) BNIP3 (B) and VEGFA (C) upon exposure to 1% compared to 20% O₂ conditions (N = 3 biological × 3 technical repeats per cell line). Dash lines mark a fold change equal to 1 (left) or 2 (right). D, Differential expression statistics comparing samples exposed to 20% as compared with 1% O₂ were calculated using DESeq2. A volcano plot of the differential expression statistics show genes (in red) with a log-fold change (LFC) greater than 1 and a false-discovery rate (FDR) adjusted p-value less than 0.05. The 42-genes that meet this criteria are listed in Supplementary Table S2.

Figure 2.

Basal-like breast cancer cells express higher endogenous levels of the 42-gene hypoxia signature. A-B, Heatmap of gene expression values for the 42-gene hypoxia signature identified with CoGAPS analysis of gene expression data from 31 breast cancer cell lines cultured under standard tissue culture conditions (20% O₂) (A) or both 20% and 1% O₂ (B) Color-coding describes the subtype of the cancer cell line. htert-HME cells were not classified in the Marcotte et al. study. The z-score ranges from low (blue) to high (red).

Figure 3.

Breast cancer cells display a heterogeneous response to hypoxic conditions. A-B, The number of genes that are upregulated (A) or downregulated (B) in each cell line stratified by fold change after exposure to 1% compared to 20% O₂ conditions.

Figure 4.

Basal cell lines display the highest number of differentially expressed genes upon exposure to hypoxia. A-D, Differential expression statistics comparing samples exposed to 20% as compared with 1% O₂ were calculated using DESeq2 for Basal-like (A), Luminal A (B), Normal-like (C) and Luminal B (D) subtype cell lines. A volcano plot of the differential expression statistics show genes (in red) with a LFC greater than 1 and FDR adjusted p-value less than 0.05. E, VENN diagram displaying the number of genes that overlap or that are exclusive to each subtype.

Figure 5.

Patients with Basal breast cancer have a higher level of the 42-gene hypoxia signature. Supervised clustering analysis of all breast cancer patients in the TCGA database by the 42-gene hypoxia signature reveals that patients with basal breast cancer have a higher level of expression of hypoxia gene compared to other subtypes.

Figure 6.

Novel hypoxia regulated identified in the HIF-signature require HIF-1α expression. A, qPCR to determine CASP14, FUT11, DNAH11, and TCAF2 mRNA levels in 12 breast cancer cell lines exposed to 20% O₂ or 1% O₂ for 24 h, normalized to mean value for HCC1569 cells at 20% O₂ (for CASP14, FUT11, and TCAF2) or HME cells at 20% O₂ (for DNAH11) (mean ± SEM, n = 3). *, P < 0.05 versus 20% O₂ within the same cell line (Student’s t-test). B-C, CASP14 and FUT11 (B) or DNAH11 and TCAF2 (C) mRNA levels were analyzed by qPCR in MCF7 (B) or MDA-MB-231 (C) subclones, which were stably transfected with an non-target control (NTC) CRISPR vector or vectors encoding gDNA sequences that target either HIF-1α (1–1, 1–2) or HIF-2α (2–1, 2–2), and exposed to 20% or 1% O₂ for 24 h (mean ± SEM, n = 3–6); **, P < 0.01, ***, P < 0.001 versus NTC at 20% O₂ or ^###, P < .001 versus NTC at 1% O₂ (two-way ANOVA with Bonferroni posttest). D, Candidate HIF-binding sites (ovals) were identified in the 5’-untranslated region of the human CASP14 gene (exons 1 and 2 shown as a rectangles). E, BT474 cells were incubated at 20% or 1% O₂ for 4 hours and ChIP assays were performed using IgG or antibodies against HIF-1α, HIF-2α, or HIF-1β. Specific primers flanking candidate HIF-binding sites were used for qPCR (Supplementary Table S1), and values were normalized to IgG expression at 20% O₂ (mean ± SEM, n = 3). *, P < 0.05 vs. 20% O₂ within the same antibody (Student’s t-test).