Reversal of Infected Host Gene Expression Identifies Repurposed Drug Candidates for COVID-19

Jing Xing; Rama Shankar; Aleksandra Drelich; Shreya Paithankar; Eugene Chekalin; Thomas Dexheimer; Surender Rajasekaran; Chien-Te Kent Tseng; Bin Chen

In total, 430 samples infected by either MERS-or SARS-CoV under different models (e.g., cell line, mouse models) across multiple time points from public repositories were used for the identification of disease signatures (Table S1, 12 studies in total). Their expression profiles were created using either microarray or RNA-Sequencing. Depending on the profiling platform, data processing and signature creation methods For each comparison, we computed a disease signature to characterize the infection status, followed by the prediction of which drugs may have activity. As we could not directly evaluate the quality and pathologic relevance of each disease signature, we turned to validating their action using those drugs that are positive in vitro MERS/SARS testing (41 positive drugs in total, 30 with EC50 values, Table S2 ). Among 215 MERS-CoV or SARS-CoV infection signatures, 13 signatures led to the drug list, where positive drugs are highly enriched on the top (Extended Data 2). Moreover, EC50 of these drugs significantly correlated with sRGES ( Figure 2A and Figure S4 ). In contrast, for the H1N1 infection signatures, we didn't observe any significant enrichment of anti-coronavirus positive drugs (Extended Data 2). The recent clinical data suggested that the combination of two drugs lopinavir/ritonavir present clinical benefits to patients with coronavirus 18, 19 . Both drugs were profiled in LINCS 20 individually; however, none of them could induce substantial gene expression change (absolute z score < 2) under 10 µM concentration in cancer cell lines; therefore, neither was predicted as a hit. We next found another dataset where HepaRG cells were treated by ritonavir under 10 different concentrations (from 9 nM to 300 µM). We observed that ritonavir could reverse nearly all the 13 disease signatures under high concentration (i.e., 10 µM to 200 µM) ( Figure S5 ).