Recurrent ASXL1 Mutations Promote Myeloid Malignancies through Unleashed Condensation

ASXL1 mutations are very frequent in all forms of myeloid malignancies, always associated with adverse prognosis, relapse, and therapy resistance (Asada et al. Cell Mol Life Sci 2019, Panuzzo et al. J Clin Med 2020). ASXL1 is an obligate co-factor for BAP1 in erasing H2AK119 ubiquitination, an epigenetic modification associated with gene repression. ASXL1 mutations are mostly truncating mutations and concentrated in a central intrinsically disordered region (IDR) of ASXL1. ASXL1 truncations are gain-of-function mutations that promote myeloid malignancies, but the underlying molecular mechanisms remain poorly understood.

Our lab previously showed that the truncating mutation of another epigenetic protein, UTX, loses the central IDR-mediated phase separation in cancer (Shi et al. Nature 2021). As the frequent mutations in ASXL1 also lose the central IDR, we hypothesized that ASXL1 forms condensates to suppress cancer and the truncations lose this property to promote cancer. We were struck to find that ASXL1 truncations form strong phase-separated condensates whereas the wild type (WT) does not. The most disordered region, which is frequently deleted in cancer, does NOT form condensates, opposite to the common belief.

We then found that the frequently truncated mutants of ASXL1 possess an intrinsic property of phase separation, and this property is normally suppressed by the cancer-deleted regions of the WT ASXL1. An endogenous ASXL1 mutant in leukemia cells forms dynamic nuclear co-condensates with other endogenous factors important for gene activation. Through studies including in vitro biochemical, cell-based, and animal tumor assays, we show that condensation of the truncations is crucial to promote myeloid malignancies through activating endogenous retroviruses and promoting expression of inflammation and myeloid tumorigenesis genes, at least in part through concentrating and enhancing H2A deubiquitination activity of BAP1 (BAP1 hyperactivity is known to be critical in ASXL1-associated myeloid malignancies).

How is ASXL1 condensation normally controlled by the cancer-deleted region? We found that this is through the evolutionally conserved negative charges in the cancer-deleted region. This highly disordered region does not form condensates at the physiological pH level, but forms condensates at lower pH levels, which restores the charge neutrality. Mutations that restore charge neutrality of this IDR also restore condensation of the full-length ASXL1, and are sufficient to turn ASXL1 (which was identified as a tumor-suppressor) into a leukemogenic protein in mice. We show that, for the full-length ASXL1, the INTRA-molecular interactions between the negatively charged IDR and the positively charged N-terminus block multivalent INTER-molecular interactions necessary for phase separation. Cancer-deletion of the charged region (and the charge-neutralizing mutations) release the N-terminal region, unleashing the phase separation property to drive the formation of nuclear hubs with aberrant activity in modifying chromatin and activating leukemic transcription.

The ASXL1 patient mutation frequency profile peaks at residue 646 in the IDR and gradually reduces on both sides. We show a striking correlation between the mutation frequencies, condensation abilities, and the leukemia-promoting activities of a series of patient-derived ASXL1 mutations, all peaking at truncation 646 and gradually dropping on both sides. It strongly suggests that dysregulation of the phase separation property is a central mechanism for ASXL1 tumorigenesis.

A fundamental theme in cancer etiology is that cancer mutations commonly highjack developmental regulators by escaping the biological regulation. ASXL1 is an important developmental regulator. By deleting the regulatory region, cancer mutations unleash the intrinsic condensation property and render ASXL1 constitutively and overly active in the condensates. This greatly expands our understanding of the fundamental theme in cancer etiology from a very different and understudied angle of spatiotemporal organization of the regulator. Dysregulation of phase separation might be a crucial pathogenic mechanism for the hotspot mutations in IDRs, which are pervasive but poorly understood.

Participation in the conference was partially funded by the Minister of Science under the „Regional Initiative of Excellence Program”.