9247 Identification of Novel Therapies for Monogenic Diabetes Using Connectivity Map (CMap)

Abstract Disclosure: R.J. Kreienkamp: None. F.R. Dos Santos: None. C. Bryan: None. J.C. Florez: None. M. Claussnitzer: None. M.S. Udler: None. Monogenic diabetes is an underdiagnosed form of diabetes that is caused by DNA variation of strong effect in a single gene. While it is an exciting area for targeted drug development, as some forms of monogenic diabetes can be treated with gene-specific therapies, there are currently few targeted therapies available. Additionally, existing treatments are not able to cure or modify the underlying disease. Therefore, to identify novel disease-modifying therapies, we leveraged CMap, a data resource developed at the Broad Institute of MIT and Harvard that contains over one million gene expression signatures from treatment of cell types with perturbagens including drugs, bioactive small molecules, gene overexpression, and gene knockdown reagents. We analyzed the connectivity, i.e. the similarity, between two gene expression signatures, to search for perturbagen gene expression signatures opposing gene expression changes associated with monogenic diabetes. We were able to identify 15 classes of medications with significantly enriched connectivity with HNF1A-knockdown. Included in these classes were sulfonylureas, which are known to have preferential therapeutic benefit for this specific condition, supporting this method for drug discovery. Other medications included HDAC inhibitors, tyrosine kinase inhibitors, and riboflavin. We also surveyed other genes causing monogenic syndromes that include diabetes, such as PPARG, where CMap indicated high connectivity with thiazolidinediones, a type 2 diabetes drug class with known mechanism targeting PPARγ, opposing PPARγ-knockdown and beneficial for treating patients with PPARγ-driven familial partial lipodystrophy. We then validated CMap results in a high-fidelity cellular model system - primary human adipocytes constitutively expressing Cas9 - using a high-dimensional metabolic disease oriented read-out LipocyteProfiler for systematic image-based profiling of cellular phenotypes in metabolic cell types. To establish causality between the causal monogenic target PPARG and its agonizing compound effect, we performed a targeted CRISPR-Cas9 based knockout of PPARG in human adipocytes, and linked changes at the cellular level to changes related to thiazolidinediones. The data revealed a striking association between the PPARγ perturbation and the TZD effect, along with an increasing CMap absolute value and LipocyteProfiler connectivity (p=9.8 E-6). Our findings support the use of expression signature mapping as a novel approach for identifying therapies for monogenic diabetes. Further exploration of databases like CMap and validation of identified medication classes could lead to new pharmacological treatments for diabetes. Presentation: 6/2/2024