# A druggable dependency in low-MITF/high-AXL melanoma: preclinical efficacy and mechanism of action in a key treatment-resistant subclass.

> **NIH NIH R01** · MASSACHUSETTS GENERAL HOSPITAL · 2021 · $373,547

## Abstract

PROJECT SUMMARY
Despite major advances in targeted and immune therapies for melanoma, most patients experience relapses
or do not respond adequately. Acquired and intrinsic resistance to these therapies in both patient biopsies and
cell lines have been associated with a unique undifferentiated cell state characterized by low levels of MITF,
the master regulator of melanocyte development, and high levels of the AXL tyrosine kinase. This cell state is
reported in a high fraction of melanomas exhibiting de novo resistance to BRAF inhibition and in 50-100% with
acquired resistance. It is also a prominent feature of anti-PD1 resistant melanomas. Single-cell RNA-seq
demonstrated that all melanomas, even high-MITF ones, contain low-MITF subpopulations that are positively
selected by targeted therapy. To eradicate this population, AXL itself is not a therapeutic target, based on
published tests of multiple inhibitors and shRNAs. We therefore interrogated genetic dependencies using
Project Achilles—an unbiased genome-scale loss-of-function screen—and found that low-MITF/high-AXL
melanomas exhibit a striking dependency, not on AXL, but on LSD1, a lysine-specific histone demethylase that
has been implicated in oncogenic processes. We validated LSD1 dependence in multiple melanoma lines
using both genetic and pharmacologic inhibition. Analysis of melanomas in the TCGA and CCLE databases for
candidate LSD1 target genes mediating its dependency revealed three candidates that LSD1 selectively
modulates in low-MITF melanomas. The first of these, NDRG1, was shown to be required for LSD1-targeted
lethality. Concordance of these LSD1-target expression patterns is seen across both cell lines and tumors
(TCGA). We also show that SP2509, an LSD1 inhibitor related to a compound in clinical development, is highly
lethal to multiple low-MITF melanomas, an effect that is dependent upon (and modulated by) the low-MITF
state. Deeper analysis of the Achilles database identified multiple subunits of CoREST (known LSD1 co-
repressive complex) as strong dependency factors, phenocopying selective lethality of low-MITF/high-AXL
melanomas and suggesting a mechanistic connection to LSD1-dependency. In Aim 1, we will test the
hypothesis that therapeutic resistance of low-MITF melanomas can be mitigated by combining targeted or
immune therapies with LSD inhibition in early passage melanoma cell lines and PDX. Our preliminary results
demonstrate profound cooperative lethality by combining genetic or pharmacologic LSD1 inhibition with BRAF
inhibitor in vitro and in xenografts. We will test the ability of LSD1 inhibition to antagonize the emergence of
resistance in high-MITF melanoma, as well as to cooperate with agents that actively suppress MITF
expression. In Aim 2, we will mechanistically examine functional roles of four candidate mediators of LSD1
dependency and use genomic approaches to systematically scrutinize the low-MITF state, identifying potential
pharmacodynamic markers and...

## Key facts

- **NIH application ID:** 10074542
- **Project number:** 5R01CA222871-04
- **Recipient organization:** MASSACHUSETTS GENERAL HOSPITAL
- **Principal Investigator:** DAVID E FISHER
- **Activity code:** R01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2021
- **Award amount:** $373,547
- **Award type:** 5
- **Project period:** 2018-01-01 → 2022-12-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10074542

## Citation

> US National Institutes of Health, RePORTER application 10074542, A druggable dependency in low-MITF/high-AXL melanoma: preclinical efficacy and mechanism of action in a key treatment-resistant subclass. (5R01CA222871-04). Retrieved via AI Analytics 2026-05-21 from https://api.ai-analytics.org/grant/nih/10074542. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*