Viruses interact with host targets in many, often unpredictable ways. While viral replication itself may lead to clinical symptoms and dictate disease outcomes, often it is the host’s reaction to the virus that causes problems. Such is the case with a subset of coronavirus disease-2019 (COVID-19) cases. COVID-19, caused by the novel severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) presents with a broad range of symptoms and disease outcomes. While most patients infected with SARS-CoV-2 experience mild symptoms reminiscent in many ways of other respiratory infections, 15-20% of patients experience moderate to severe symptoms leading to hospitalization and in many cases, death. Examination of lung tissues from patients with severe COVID-19 cases led to the identification of a severe immune reaction to the viral infection leading to over-activation of inflammatory cytokines, the so-called cytokine storm, and resulting tissue damage. Indeed, several early therapies tested clinically involved blocking or at least reducing inflammation with agents such as the steroid dexamethasone. While partly successful, anti-inflammatory therapies run the risk of suppressing the immune system’s ability to eliminate the virus. A major challenge of developing therapies to treat the symptoms of COVID-19 has been to dampen the toxic effects of excessive immune activation without immunosuppressing the patient. A recent publication from the laboratory of Dr. Taran Gujral in the Human Biology Division in collaboration with the labs of M. Juliana McElrath (Vaccine and Infectious Disease Division) and Eric Holland (Human Biology) described the identification of a small molecule therapeutic previously approved by the Food and Drug Administration for the treatment of Chronic Myelogenous Leukemia as a potential inhibitor of the tissue damaging cytokine storm. Their report was published in Molecular Systems Biology.
Conventional thinking about interrogating a biological process with small molecule compounds dictates using highly specific agents to modulate the action of a single target selectively and specifically. In that way, the results are unambiguous. The Gujral group has developed an alternative strategy that turns conventional wisdom on its head. What if, instead of a single selective and specific agent, a collection of agents with similar, overlapping but non-identical selectivities were used to generate a complex phenotypic profile, one that could be disentangled using advanced computational tools. The real power of such an approach lies in the ability to determine the functional roles of not just single targets, but in a way that is more relevant to the complex biology, collections of targets that participate or cooperate in a biological process. The Gujral, McElrath, and Holland groups applied this strategy to the challenge of preventing cytokine storm in response to challenge with SARS-CoV-2 Spike protein.
Spike is presented on the surface of the SARS-CoV-2 virion and is known to interact with the human receptor angiotensin-converting enzyme 2 (ACE-2). The Spike/ACE2 interaction is necessary for the internalization of SARS-CoV-2. Other functional domains of Spike, however, are responsible for activation of macrophages and massive cytokine release. Using recombinant Spike expressed in mammalian cells, the authors showed that the N-terminal domain of Spike was responsible for activation of transcription factors regulating cytokine gene expression. With this phenotype as a readout, the Gujral group used a carefully selected collection of 35 kinase inhibitors to interrogate the cytokine release pathway. Critical to the success of this experiment was that none of the 35 compounds was specific for any of the over 500 kinases found in the human genome. The process was used to identify 30 most informative kinases out of the >300 in the computational model. Examining the Spike-induced expression of seven cytokines, the authors identified kinases known to regulate cytokine expression such as JAK1, but also several kinases not associated with cytokine expression. Using siRNA knockdown of 13 candidate kinases, 11 were shown to have a statistically significant effect on cytokine expression compared to controls. Since more than a single kinase was shown to play a role in cytokine release, the researchers used the computational model to predict the activity of 427 available kinase inhibitors as single agents as well as a 427x427 matrix of two inhibitor combinations. Somewhat surprisingly, a drug approved by the FDA for the treatment of chronic myelogenous leukemia (CML), ponatinib, robustly inhibited Spike-induced cytokine expression in macrophages as well as donor-derived peripheral blood mononuclear cells (PBMC). Ponatinib, developed to inhibit the BCR-ABL tyrosine kinase driver of CML, also showed activity against several of the kinases involved in cytokine expression. Interestingly, ABL kinase, the wild-type version of BCR-ABL was not among the cytokine-linked kinases. Treatment with ponatinib outperformed treatment with baricitinb, a JAK1 inhibitor approved by the FDA for prevention of SARS-CoV-2-related cytokine storm.
Although prevention of excessive cytokine release is critical to ameliorating severe COVID-19 symptoms, it also runs the risk of suppressing normal response of the immune system to the viral infection. To test the specificity of the inhibition of SARS-CoV-2-induced cytokine release, the authors stimulated PBMCs with a pool of peptides derived from cytomegalovirus in the presence of ponatinib and found the drug to have little or no effect, unlike baricitinib or dexamethasone, which inhibited the response. These results confirm that ponatinib specifically blocks cytokine release triggered by exposure to the N-terminal domain of SARS-CoV-2. Since ponatinib is already approved by the FDA for human use, clinical trials in COVID-19 patients experiencing cytokine storm could happen soon. The results reported in the current publication highlight the power of the polypharmacology approach to drug discovery and target validation.
Chan, M., Vijay, S., McNevin, J., McElrath, M.J., Holland, E.C. and Gujral, T.S. Machine Learning Identifies Molecular Regulators and Therapeutics for Targeting SARS-CoV-2-induced Cytokine Storm, Mol. Systems Biol. 17, e10426|2021
This research was supported by grants from Fred Hutch COVID19 Pilot Fund and computational resource grant from ORIP.
Fred Hutch/UW Cancer Consortium members M. Juliana McElrath, Eric C. Holland and Taranjit S. Gujral contributed to this study.