COVID-19 disproportionately sickens and kills older patients, but the precise reasons why remain a mystery. A common hypothesis is that as we age, our immune system weakens, making our cells more susceptible to the virus. Prof. Caroline Uhler at MIT, in collaboration with Prof. G.V. Shivashankar at ETH Zurich, proposes an alternative explanation: as our lung tissue grows stiffer with age, the coronavirus may take advantage of its altered mechanical state.
Uhler and Shivashankar have incorporated the signatures of an aging cell into a computational search for drugs. Uhler will speak about “Causal Inference and Autoencoders for Drug Repurposing,” at the AI Cures Drug Discovery Conference on October 30th.
Previous research, including that of Uhler and Shivashankar, shows that the cytoskeleton, a complex network of protein fibers in a cell, gives shape not only to the cell, but also to its nucleus. As the cytoskeleton keeps nearly six feet of genetic material tightly coiled in the nucleus, it also plays a role in locking and unlocking that genetic information, affecting how DNA is expressed.
Because stiffness of the lung tissue affects the cytoskeleton, stiffness also influences gene regulation. Here’s where the coronavirus comes in: once inside the cell, SARS-CoV-2 hijacks specific signaling pathways, including a pathway named NF-κB, to start replicating its own genome. Interestingly, the NF-κB pathway is also modulated when cells become stiffer. Uhler and her colleagues believe this means the virus can harness this signaling pathway more effectively in aged cells, aiding itself in replication. The group plans to test their hypothesis by analyzing lung tissue samples from patients who died of COVID-19.
The idea has guided Uhler and her group in a computational search for drugs to repurpose against COVID-19. Large datasets, which contain profiles of many drugs and their interactions with different cell types, already exist. Many of these cell types, including those in a large dataset called The Connectivity Map (CMap), are related to cancer, but can be adapted to the coronavirus by including SARS-CoV-2 gene expression data.
“In particular cell types, we know the effect of these drugs,” Uhler says. “Based on this data we want to predict the effects of these drugs on other cell types,” namely those found in the respiratory system infected with SARS-CoV-2.
“If we don’t take age into account, we’ve got this huge list of drugs, which was all over the place,” she says. “But once we took age into account, all of the drugs were from the same class. Age acted as an important filter in our analysis.”
Their analysis pointed to serine/threonine and tyrosine kinases as potential targets that are important to both SARS-CoV-2 replication and signal pathways in aging cells. Drugs that affect these targets already exist on the market. In this case, Uhler says, it’s a good sign that the computational analysis came out with a specific class of drugs without having any prior expectations.
With a list of drugs in hand, Uhler and the group’s next step is to test the therapeutic effects of particular drugs in her lab, a hopeful step toward identifying drugs that can disrupt the replication of the virus.