Abstract
Systematically discerning functional drivers from inconsequential "passenger" mutations is a central challenge in cancer genomics. Using a somatic in vivo CRISPR/Cas9 mutagenesis screen, we investigated 215 recurrently mutated "long-tail" breast cancer genes. Our findings reveal that epigenetic regulation is a major tumor-suppressive mechanism. Specifically, loss of core components of the COMPASS histone methylase complex—including KMT2C, KDM6A, BAP1, and ASXL2 (collectively termed "EpiDrivers")—cooperates with the common PIK3CA H1047R mutation to accelerate tumorigenesis. This process is driven by "lactation mimicry," where cells lose their lineage integrity and reactivate a multipotent, alveolar-like program normally reserved for pregnancy. Given that these EpiDrivers are altered in 39% of human breast cancers, this mechanism represents a significant pathway in early disease progression.
Introduction: The "Long-Tail" Challenge
Breast cancer genomics is characterized by a few highly frequent mutations, such as TP53 and PIK3CA, followed by a "long tail" of hundreds of genes mutated in only 2–10% of patients. Determining which of these rare mutations are true drivers of cancer has historically been slow and resource-intensive. Furthermore, the functional consequences of these mutations on cellular identity—or lineage integrity—have remained largely unexplored. If a cell "forgets" its specific identity (e.g., whether it is a basal or luminal cell), it may become more plastic and susceptible to oncogenic transformation.
Systematic Discovery via In Vivo CRISPR Screening
To address this, we developed a high-throughput somatic in vivo CRISPR/Cas9 screen. Rather than relying on traditional in vitro models, we injected a library of 860 sgRNAs targeting 215 "long-tail" genes directly into the mammary glands of mice carrying a latent Pik3ca H1047R mutation. This approach allowed us to observe tumor development in a native physiological environment.
To analyze the resulting tumors, we employed a multi-omic approach:
- Single-cell Multi-omics: Combining scRNA-seq and snATAC-seq to map the transcriptomic and epigenetic landscape of early-stage transformation.
- Lineage Tracing: Utilizing viral vectors (Ad-K5-Cre and Ad-K8-Cre) to track whether tumors originated from basal or luminal cell populations.
- Human Validation: Validating findings using CRISPR knockouts in human MCF10A mammary cells and analyzing clinical data from the METABRIC and TCGA datasets.
The COMPASS Complex and "EpiDrivers"
The screen identified a cluster of genes that, when lost, significantly accelerated tumor growth. These genes—KMT2C, KDM6A, BAP1, and ASXL2—are all involved in the COMPASS histone methylase complex. This complex acts as a master epigenetic regulator, controlling the "openness" of chromatin and the accessibility of specific genetic programs.
We found that while these genes are individually rare, they are collectively altered in approximately 39% of human breast cancers. In mice, the combination of a Pik3ca mutation and an EpiDriver knockout reduced the median time to tumor formation from 154 days to just 74 days.
Lactation Mimicry: A Hijacked Developmental Program
The most striking mechanistic finding was the activation of "lactation mimicry." In the absence of these epigenetic regulators, mammary cells began expressing milk proteins (such as Caseins and Wap) and following a genetic program typically seen during pregnancy and lactation—even in virgin mice.
This is not merely a change in gene expression; it is a functional shift. The lactation program involves a massive proliferative burst intended to prepare the mammary gland for milk production. Cancer cells hijack this rapid growth phase to drive tumor progression. Epigenetic mapping (snATAC-seq) revealed that the loss of KDM6A led to increased chromatin accessibility at sites controlled by Elf5, the master transcription factor for alveolar cell fate.
Lineage Infidelity and the Basal Origin of Luminal Tumors
The study also highlights the role of "lineage infidelity." We observed that the loss of EpiDrivers caused basal cells to adopt luminal-like characteristics and eventually form luminal-like tumors. This suggests that a significant portion of human luminal breast cancers may actually originate from basal cells that have lost their lineage integrity due to epigenetic disruption. This plasticity allows cells to bypass normal developmental constraints, reactivating multipotency to fuel malignancy.
Clinical Significance and Future Directions
The clinical implications are profound. Patients harboring both an EpiDriver mutation and high PI3K signaling exhibit the poorest survival outcomes across all breast cancer subtypes. Furthermore, the discovery that even "shallow" deletions (haploinsufficiency) of genes like KDM6A can drive poor outcomes suggests that these epigenetic regulators are highly dosage-sensitive.
Future research must determine if these epigenetic states are reversible. If we can develop therapies—such as BET inhibitors or other epigenetic modifiers—that restore the COMPASS complex's function or suppress the hijacked Elf5 program, we may be able to "re-program" these cells back to their original identities and halt tumor progression. Understanding the metabolic requirements of cells in this pseudo-lactating state may also reveal new therapeutic vulnerabilities.
Summary updated