Genetic and epigenetic driven variation in regulatory regions activity contribute to adaptation and evolution under endocrine treatment

Abstract

Comprehensive profiling of hormone-dependent breast cancer (HDBC) has identified hundreds of protein-coding alterations contributing to cancer initiation, but only a handful have been linked to endocrine therapy resistance, potentially contributing to 40% of relapses. If other mechanisms underlie the evolution of HDBC under adjuvant therapy is currently unknown. In this work, we employ integrative functional genomics to dissect the contribution of cis-regulatory elements (CREs) to cancer evolution by focusing on 12 megabases of non-coding DNA, including clonal enhancers, gene promoters, and boundaries of topologically associating domains. Massive parallel perturbation in vitro reveals context-dependent roles for many of these CREs, with a specific impact on dormancy entrance and endocrine therapy resistance9. Profiling of CRE somatic alterations in a unique, longitudinal cohort of patients treated with endocrine therapies identifies non-coding changes involved in therapy resistance. Overall, our data uncover actionable transient transcriptional programs critical for dormant persister cells and unveil new regulatory nodes driving evolutionary trajectories towards disease progression.

Long-term multi-modal recording reveals unpredictable nongenetic adaptation routes in dormant breast cancer cells.

“…Therefore, we conclude that interim dormant frequencies are not associated with chances of awakening and that lineage-specific evolutionary trajectories become extremely divergent with time (compare the d winner in 60.1 and 60.2). These data agree with the result of a parallel CRISPR-Screen experiment showing that the relative frequency of non-targeting sgRNA randomly diverge between replicates during dormancy 30 . Overall, our data demonstrate that in vitro, ET can trigger dormancy in a large fraction of randomly selected cancer cells, but that awakening is largely unpredictable.…”

“…In agreement with phenotypic, genomic and barcode analyses (Figure 2c, d and 3ab), the bulk RNA-seq profile of TAM and AI awakening carbon copies significantly diverge in time, especially when compared to UT cells kept in culture for the same amount of time (Figure 5a-b). TEP carbon copies continue to evolve past the awakening bulk phenotype (Figure 5a-b) in agreement with our cis-regulatory screen, which shows that differently from MCF7, resistant cells have large margin of evolutionary space (see companion manuscript 30 ).…”

“…Our model fits with these observations, as additional awakening from the persister dormant pool will give rise to competing lineages which become unmanageable even with combinatorial therapy. On the other hand, our data strongly support an effort toward targeting dormant persister cells 30 . Current strategies have focused on combinatorial treatments with limited success 37 .…”

“…Longterm neutral drift did not significantly alter the RNA-seq profile of the UT arm despite the significant loss of lineages and the associated clonal expansion of four recurrent clones (Figure 3a and 5a). Taken together with our cis-regulatory element screen 30 (see companion manuscript), these data suggest that MCF7 cells have possibly reached an evolutionary apex under standard culture conditions and highlight how clonal heterogeneity has limited impact on the phenotype of HDBC cells. These results also challenge the idea of long-term passaging as a source of transcriptional variability in MCF7, at least for cells maintained in identical conditions 32 .…”

“…Current strategies have focused on combinatorial treatments with limited success 37 . We argue that this reflects the difficult task of drugging a moving target and advocates for sequential intervention designed to minimize transcriptional heterogeneity (via dormancy bottlenecking) followed by targeted approaches tailored to dormant persisters 30 .…”