Comprehensive characterization of claudin-low breast tumors reflects the impact of the cell-of-origin on cancer evolution

Abstract

Claudin-low breast cancers are aggressive tumors defined by the low expression of key components of cellular junctions, associated with mesenchymal and stemness features. Although they are generally considered as the most primitive breast malignancies, their histogenesis remains elusive. Here we show that this molecular subtype of breast cancers exhibits a significant diversity, comprising three main subgroups that emerge from unique evolutionary processes. Genetic, gene methylation and gene expression analyses reveal that two of the subgroups relate, respectively, to luminal breast cancers and basal-like breast cancers through the activation of an EMT process over the course of tumor progression. The third subgroup is closely related to normal human mammary stem cells. This unique subgroup of breast cancers shows a paucity of genomic aberrations and a low frequency of TP53 mutations, supporting the emerging notion that the intrinsic properties of the cell-of-origin constitute a major determinant of the genetic history of tumorigenesis.

Fig. 1. Claudin-low tumors are distributed across luminal-like IntClust3, basal-like IntClust10, and CNA-devoid IntClust4, which harbors rare focal genomic alterations.

a Repartition of breast tumors from the METABRIC cohort for each molecular subtype among integrative clusters. Number of tumors are represented on the y axis and the corresponding percentages are indicated above each bar. All tumors (n = 1266); claudin-low tumors (n = 45); basal tumors (n = 129); HER2 tumors (n = 102); luminal A tumors (n = 461); luminal B tumors (n = 429). b, c Identification of focal genomic alterations in a CNA-devoid IntClust4 claudin-low sample (MB-6052/FGA < 0.3%). b BAF and LRR plots of SNPs localized in genomic regions of copy number deletion (LRR < 0.4). c Mutation analysis from targeted sequencing data. BAF: B allele frequency; CNA: copy number alteration; FGA: fraction of genome altered; IntClust: integrative cluster; LRR: log R ratio; SNP: single-nucleotide polymorphism.

Fig. 2. Claudin-low tumors form a genomically and molecularly heterogeneous subgroup.

a Ploidy for each molecular subtype of the METABRIC cohort. Claudin-low tumors are mainly diploid. b FGA% for each molecular subtype. Claudin-low tumors display overall low levels of genomic alterations. c Gaussian mixture model application on FGA distribution across claudin-low tumors reveals three CNA-related subgroups of tumors. Each bar on the x axis corresponds to one claudin-low tumor. d FGA% in claudin-low subgroups compared with other molecular subtypes. e Integrative clusters and f breast cancer receptor status distribution in each claudin-low subgroup. The CNA-devoid CL1 subgroup is associated with IntClust4 and TN status, whereas CL2 and CL3 are respectively related to luminal (luminal clusters and hormone receptor expression) and basal-like (IntClust10 and TN tumor enrichment) subtypes. Respective colors of luminal-related clusters are shown in Fig. 1. Boxplot: center line, median; box limits, upper and lower quartiles; whiskers, minimum to maximum; all data points are shown. Claudin-low tumors (n = 45); basal tumors (n = 129); HER2 tumors (n = 102); luminal A tumors (n = 461); luminal B tumors (n = 429); CL1 tumors (n = 10); CL2 tumors (n = 17); CL3 tumors (n = 18).CL: claudin-low; TN: triple negative.

Fig. 3. Claudin-low subgroups show distinct gene expression signatures relative to their differentiation status.

a GSEA comparing global gene expression of each claudin-low subgroup from the METABRIC cohort to the two others. The three pathways with the highest and lowest enrichment scores are represented (>15,000 tested pathways—NES ranking). b, c Gene expression analysis for each molecular subtype of (b) MaSC1/2/3, LP, and mL1/2 transcriptomic signatures (generated from Morel et al. data) and (c) previously published MaSC, LP, and mL transcriptomic signatures (from Lim et al. publication). The CL1 subgroup displays strong stemness features, whereas CL2 and CL3 present luminal- and basal-like transcriptomic characteristics, respectively. Wilcoxon tests. Boxplot: center line, median; box limits, upper and lower quartiles; whiskers, minimum to maximum; all data points are shown. Basal tumors (n = 129); HER2 tumors (n = 102); luminal A tumors (n = 461); luminal B tumors (n = 429); CL1 tumors (n = 10); CL2 tumors (n = 17); CL3 tumors (n = 18). GSEA: gene-set enrichment analysis; LP: luminal progenitor; MaSC: mammary stem cell; mL: mature luminal; NES: normalized enrichment score.

Fig. 4. Claudin-low subgroups show distinct gene methylation profiles relative to their differentiation status.

Heatmaps and pathway-enrichment analysis of differentially methylated genes between a CL1 vs. CL2 and CL3 tumors from TCGA, b CL2 vs. CL1 and CL3 tumors from TCGA, and c CL3 vs. CL1 and CL2 tumors from TCGA. CL1 tumors mainly display hypomethylated genes that are related to a stemness phenotype, CL2 tumors are overall hypermethylated for genes associated with basal-like features, and CL3 tumors show more heterogeneous methylation profiles with both hypermethylated genes (linked to luminal phenotype) and hypomethylated genes (relative to basal-like characteristics) (see Supplementary Fig. 5 for methodology). Clustering method: Ward’s; distance: Spearman. FDR: false discovery rate.

Fig. 5. Claudin-low subgroups show distinct activated biological pathways.

Heatmaps of ssGSEA scores for each MSigDB hallmark gene-set signature (median by molecular subtype) for a METABRIC and b TCGA datasets. ASCAT median purity score of each subtype is indicated on the top heat map annotation. Claudin-low tumors cluster together in a specific branch of the dendrogram downstream distinguishing CL1 from CL2/CL3 tumors. Hallmark gene-set cluster in three main groups of pathways related to luminal, basal, and claudin-low tumors. ASCAT: allele-specific copy number analysis of tumors; MSigDB: molecular signature database; ssGSEA: single-sample GSEA. Clustering method: Complete; distance: Euclidean.

Fig. 6. Claudin-low subgroups jointly display high EMT features and MAPK pathway activation.

a, c, e ssGSEA score distribution for each molecular subtype, from METABRIC and TCGA cohorts, related to a EMT, c TP53, and e RAS/MAPK pathways. d TP53 somatic mutation distribution across breast molecular subtypes of METABRIC and TCGA cohorts. b, f Protein-based pathway scores (RPPA data) for each molecular subtype from TCGA breast tumors related to b EMT and f RAS/MAPK. g Principal component analysis (PCA) of TCGA breast tumors according to EMT, RAS/MAPK, cell cycle/proliferation, hormone-related and DDR pathways at RNA (ssGSEA scores), and protein (RPPA scores) levels. For each sample, molecular subtype is indicated in color and variables (pathways) are highlighted in black squares. h IC50 distribution according to molecular subtype for three different MEK inhibitors available in the GDSC database. Claudin-low cell lines commonly exhibit higher sensitivity to MAPK inhibitors than other breast molecular subgroups. Wilcoxon tests. Boxplot: center line, median; box limits, upper and lower quartiles; whiskers, minimum to maximum; all data points are shown. TP53 signaling pathway signature: GO positive regulation of signal transduction by p53 class mediator; MAPK signaling signature: KEGG MAPK signaling pathway. DDR: DNA damage response; EMT: epithelial-mesenchymal transition; RPPA: reverse-phase protein array.

Fig. 7. The diversity of claudin-low tumors reflects the impact of the cell-of-origin on the evolutionary process of breast tumorigenesis.

The malignant transformation of a normal MaSC leads to undifferentiated tumors with a low genomic instability and a low frequency of TP53 mutations. These features are characteristic of the CL1 subgroup. The activation of an EMT transdifferentiation process and the gain of mesenchymal features in a basal-like tumor promotes the development of a CL3 tumor with extensive genomic aberrations. EMT commitment in a luminal-like breast cancer leads gradually to a CL2 tumor with a moderate level of genomic aberrations.