The pretenders: Basal breast cancer cells mimic epidermal keratinocytes and rely on retinoic acid signaling for survival, invasion and metastasis.

Abstract

In breast cancer, the presence of cells expressing basal epithelial markers is associated with a high risk of tumor invasion, metastasis and therapy resistance. It has therefore been proposed that specifically killing basal breast cancer cells would be beneficial for patients. However, clinically feasible methods for specifically targeting this population of cells have not been reported. This, in part, can be attributed to the limited understanding of the proteins and signaling networks uniquely active in basal tumor cells. Basal cells are difficult to characterize due to their relative rarity in tumors and the transience of their existence. Rather than being a fixed cellular state, basal cells are typically formed through cellular plasticity, in which non-basal cells convert into basal cells, which can then revert to their original state. In this work, we enriched for basal cells using organotypic culture and then transcriptionally profiled each subpopulation of cells using single cell RNA-sequencing. We identified a unique transcriptional signature of invasive basal cells and discovered that it had significant overlap with the transcriptional signature of basal epidermal keratinocytes. Given that several transcription factors and pathways are known to regulate keratinocyte identity and function, we hypothesized that the same pathways are functional in basal tumor cells. To test our hypothesis, we first focused on retinoic acid signaling, which can be easily altered using pharmaceutical agents such as retinoids. As anticipated, activation of retinoic acid receptors suppressed transcription of several basal epidermal marker genes, but surprisingly lead to increased invasive behavior in 3D culture. Conversely, retinoic acid receptor antagonism prevented invasion and induced tumor cell death in 3D models of mouse and human breast cancers in which basal cells are present. We additionally tested the role of KLF4, a transcription factor required for proper epidermal barrier formation, in basal breast cancer cell invasion and viability. In a similar manner to retinoic acid addition, knock-down of KLF4 lead to decreased transcription of epidermal markers and increased rates of invasion in 3D culture. Overexpression of KLF4 induced cell death. Confirming the translational relevance of our findings, treatment with a retinoic acid receptor inhibitor in-vivo prevented tumor outgrowth and metastasis in-vivo. Vitamin supplements, including those with the retinoic acid precursor vitamin A, are used by up to 70 percent of cancer patients. Further, retinoic acid has shown some clinical promise as an approach for treating breast cancer patients, although basal-like breast cancers appear to be resistant. Given our results, we wanted to determine whether basal breast cancers are not only resistant to retinoic acid, but are in fact more metastatic in the presence of excess vitamin A. To do so, we performed post-hoc analysis of two nutritionally annotated clinical cohorts of women tracked for over a decade. In these studies, vitamin A supplementation was associated with poor clinical outcomes specifically in patients with estrogen receptor-negative tumors, including earlier recurrence and mortality. In total, we have reported a transcriptional program present in invasive basal cells and have identified factors that regulate this program. Further analysis of our transcriptional data may reveal novel markers and vulnerabilities of basal breast cancer cells. Additional work is warranted to clarify patient populations in which vitamin A intake may be hazardous rather than beneficial or harmless.