High-Content Deep-Phenotypic Screening of Existing Drugs for the Treatment of HHT



Chadwick Davis, PhD, is a Post Doctoral Fellow at Recursion Pharmaceuticals working under the direction of Dr. Dean Li, Professor of Medicine, at the University of Utah.  In 2015, Dr. Davis was awarded a $30,000 Young Investigator Grant by Cure HHT.



Proposed Research Summary

Recursion Pharmaceuticals models loss of function genetic diseases, such as HHT, utilizing high-content cellular imaging and advanced machine learning software that can differentiate between normal and diseased cells. A high-throughput drug screening method identifies repurposed drugs that revert the disease model back to a healthy state. Dr. Davis will utilize this system to discover a known drug with the potential to be repositioned to successfully treat HHT.

  1. Aim 1 – Characterize ENG, ACVRL1, and SMAD4 deficient cell lines using RNAi
  2. Aim 2 – Identify known drugs that fully or partially ameliorate the HHT phenoprint

Dr. Davis will (a) screen at least 2,727 known drugs and bioactive compounds against the phenoprints identified in Aim 1 and (b) evaluate and prioritize all hits for scientific and medical potential. Dr. Davis expects to find a minimum of one known bioactive compound that fully or partially reverses the phenoprint and can be repurposed for the treatment of HHT.


Research Study Update

Development of a cellular model of HHT – Dr. Davis established cellular phenotypic profiles associated with decreased ACVRL1 and SMAD4 in three separate human cell lines using high-content imaging. First, Dr. Davis generated and characterized cellular HHT models via siRNA knockdown of ACVRL1, SMAD4, and ENG, using six independent siRNA per gene. Using CellProfiler, he identified each cell in every image (millions of total cells) and quantified ~1000 features for each cell, including shape features, texture features, pixel intensity features, and abstract features, depositing the measurements to a Structured Query Language database.

Identification of hits – To identify highly translatable treatments for HHT, we screened an internal compound library of over 2,800 FDA-approved drugs and bioactive compounds against our ACVRL1/HUVEC model. For drug selection, high-dimensional cellular phenotypes representing treated, untreated, and control (“healthy”) conditions were projected onto a 2-dimensional plot, as in Figure 7. Drugs that shifted the phenotype away from the disease controls (decrease in disease score) with minimal off-target effects (decrease in side-effect score) were prioritized as potential candidates in this assay. We performed our drug screen in 3 phases.

Validation of findings – To confirm that compounds identified in the preliminary phenotypic screening campaign are active in disease specific signaling pathways, we Dr. Davis performed a series of in vitro assays – western blots, toxicity assays, evaluations of barrier function and migration rate, and limited RT-qPCR of control (ACVRL1 siRNA knockdown in HUVEC) and drug-treated HHT-modeling cells to determine whether our prioritized compounds rescued cellular phenotypes of HHT orthogonal to our screen (data not shown). Based on these results, Dr. Davis prioritized 5 compounds for additional study.

Summary – Dr. Davis established a robust phenotypic profile for HHT based on ACVRL1 knockdown in human endothelial cells. He screened a library of highly translatable compounds to identify compounds that rescued cellular HHT phenotypes, and identified several efficacious compounds. Orthogonal in vitro assays demonstrated the potential of these compounds to modify disease-relevant cellular biology. Based on these results, Dr. Davis has prioritized REC-0001529 or a similar VEGF/R inhibitor for advancement to investigational new drug.