NEW YORK – Comparative analyses of viral-human protein-protein interactions and viral protein localization for SARS-CoV-2, SARS-CoV-1, and MERS-CoV have revealed common cellular pathways that the viruses use to proliferate in their hosts, as well as possible targets for therapeutic inhibition.
In a study published on Thursday in Science, a team of almost 200 researchers from the University of California, San Francisco; the Gladstone Institutes; EMBL's European Bioinformatics Institute; the Icahn School of Medicine at Mount Sinai; Institut Pasteur; and many others, as well as the companies Aetion and Synthego, described their use of proteomic analyses and functional genetic screening, combined with reviews of COVID-19 patient genetic data and medical billing records, to identify molecular mechanisms and potential drug treatments that would merit further study.
The paper builds on two studies published earlier this year by some of the same authors that characterized the protein landscape of SARS-CoV-2. In Nature in April, the researchers described a protein interaction map for the virus that revealed targets for drug repurposing, and in Cell in August, they published a phosphoproteomic analysis of SARS-CoV-2-infected cells that identified small molecules that target dysregulated pathways and elicit potent antiviral efficacy.
For the Science study, the researchers added to the protein interaction map of SARS-CoV-2 by identifying the host factors that physically interact with each SARS-CoV-1 and MERS-CoV viral protein. They observed 389 high-confidence interactors for SARS-CoV-2, 366 for SARS-CoV-1, and 296 for MERS-CoV.
In a presentation and press conference on the paper, Nevan Krogan, director of the Quantitative Biosciences Institute at the UCSF School of Pharmacy and the study's lead investigator, said that studying all three viruses at once has the potential for leading to pan-viral coronavirus therapies in the future. The research looked to the past with SARS-CoV-1, he added, and the knowledge gained can hopefully shed some light in the future on a potential SARS-CoV-3.
"We'll be in a much better position for the next pandemic, because we know it's coming," Krogan said.
The researchers also assessed the cellular localization of individually expressed coronavirus proteins in order to learn about their functions in the cell. Their analyses showed that the vast majority of shared SARS-CoV-2, SARS-CoV-1, and MERS-CoV protein homologs had similar patterns of localization, supporting the hypothesis that conserved proteins share functional similarities.
In experiments to study how host proteins interact with each virus, the researchers identified cellular processes that were significantly enriched in the interactomes of all three viruses and ranked them by the degree of overlapping proteins. They identified proteins related to the nuclear envelope, proteasomal catabolism, cellular response to heat, and regulation of intracellular protein transport as biological functions that are hijacked by these viruses through different human proteins.
The protein-protein interaction analyses found that nearly 21 percent of all the human proteins overlapped between the three viruses, and that 23 percent of the proteins overlapped between SARS-CoV-1 and SARS-CoV-2, Krogan added.
Another important aspect of the study was the identification of host factors that are critical for infection and could serve as targets for host-directed therapies. The researchers performed genetic perturbations of 332 human proteins to observed their effect on infectivity, using an siRNA screen to knock down the targets and a CRISPR-based screen to knock them out. The host dependency factors for SARS-CoV-2 infection, which represented potential targets for drug development and repurposing, were of particular interest. For example, non-opioid receptor sigma 1 (sigma-1, encoded by SIGMAR1) was identified as a functional host-dependency factor. The researchers also generated a network that integrated the hits from both screens and the protein-protein interactions with SARS-CoV-2, SARS-CoV-1, and MERS-CoV proteins and observed an enrichment of genetic hits that encoded proteins interacting with viral Nsp7, which has a high degree of interactions shared across all the three viruses.
Krogan noted that the researchers were able to map host factors to 69 different existing drugs and compounds, including 13 that are currently being tested in various clinical trials as possible treatment for COVID-19. Of particular interest to the investigators were sigma receptor 1 modulators, which are used in a variety of treatments for neuropathological conditions. But anticancer agents such as the investigational compound plitidepsin (PharmaMar's Aplidin) and the neoplastic activity inhibitor zotatifin are also in trials, Krogan said.
The siRNA and CRISPR screens revealed that 73 genes in total seemed to affect infectivity, he noted, adding that "many of these are druggable targets."
Significantly, they found that Orf9b, an alternative open reading frame within the nucleocapsid (N) gene of both SARS-CoV-1 and SARS-CoV-2, interacts with mitochondrial chaperone protein Tom70. That protein recognizes and mediates the translocation of mitochondrial pre-proteins from the cytosol into the mitochondria in a chaperone-dependent manner, and is involved in the activation of the mitochondrial antiviral signaling protein, which leads to apoptosis upon viral infection, the researchers said. In the case of SARS-CoV-2, it may act as a host dependency factor.
According to Krogan, the team is still studying why Orf9b attaches itself to Tom70. UCSF researcher Kliment Verba noted that Tom70 is involved in establishing the innate immune response, and that the virus interferes with that process, which is also reflected in clinical data from patients infected with SARS-CoV-2.
Most pertinent to the ongoing pandemic, the researchers used real-world data from COVID-19 patients to provide a possible rationale for repurposing existing drugs as treatments for SARS-CoV-2 infection. They identified 738,933 patients in the US with documented SARS-CoV-2 infection and probed the use of drugs against targets they had identified through the genetic perturbation screens. In particular, they analyzed outcomes for an inhibitor of PGES-2 and for potential ligands of sigma-1 and asked whether these patients fared better than matched patients treated with clinically similar drugs that do not act on coronavirus host factors.
PGES-2, an interactor of Nsp7 from all three viruses, is a dependency factor for SARS-CoV-2. It is inhibited by indomethacin, a US Food and Drug Administration-approved prescription nonsteroidal anti-inflammatory drug. Indomethacin did not inhibit SARS-CoV-2 in vitro at reasonable antiviral concentrations, the researchers said, but a previous study found that the drug showed efficacy against canine coronavirus in vivo. They analyzed outcomes in a cohort of outpatients with confirmed SARS-CoV-2 infection who had initiated a course of indomethacin and compared them to patients who had been treated with the prescription NSAID celecoxib, which lacks anti-PGES-2 activity. The analysis showed that new users of indomethacin in the outpatient setting were less likely than matched new users of celecoxib to require hospitalization or inpatient services.
The researchers next grouped drugs that shared activity against sigma receptors. They had previously identified sigma-1 and sigma-2 as drug targets in their SARS-CoV-2-human protein-protein interaction map, and knockout/knockdown screens of SIGMAR1, but not SIGMAR2, led to robust decreases in SARS-CoV-2 replication, suggesting that sigma-1 may be a key therapeutic target. There were several clinical drug classes that could have worked as inhibitors, including typical antipsychotics and antihistamines, but over-the-counter antihistamines are not well represented in medical billing data. However, users of typical antipsychotics were easily identified in the real-world data.
The researchers constructed a cohort for retrospective analysis of new, inpatient users of antipsychotics and compared the effectiveness of typical versus atypical antipsychotics, which are not predicted to bind sigma receptors and do not have antiviral activity, for treatment of COVID-19. In the primary analysis, half as many new users of the typical antipsychotics compared to new users of atypical antipsychotics progressed to the point of requiring mechanical ventilation.
It's not certain that sigma receptor interaction is the mechanism underpinning this effect, as typical antipsychotics are known to bind to a multitude of cellular targets, the researchers cautioned. Replication in other patient cohorts and further work will be needed to see if there is therapeutic value in these connections. But the study did demonstrate a strategy of protein network analyses combined with real-world, clinical information.
"Certainly, these results suggest doing more research, and whether that's a randomized trial or more research with real-world data, I think that's an open question," said Aetion Cofounder and CSO Jeremy Rassen. "I think we can determine a causal effect between a drug and an endpoint in real-world data. We can certainly do that in a randomized trial. And in terms of whether that takes place in an inpatient or outpatient setting, I think it's really about the drug and about how the drug is used and what the endpoint that the drug would particularly prevent would be."
Institut Pasteur investigator Marco Vignuzzi added that the first goal would be to repurpose existing drugs in order to find treatments for COVID-19 as quickly as possible. However, "what this can lead to is that ultimately, medicinal chemists could work on those drugs to find derivatives or versions of it that are even better for coronavirus down the road," he said. "Of course, this would require more in vivo work to prove the derivatives also function. But in this case, these kinds of studies have an immediate possible potential and more downstream potential."
The researchers also debated the merits of targeting host proteins versus viral proteins as treatment strategies. Although antiviral targeting can be "a magic bullet," Vignuzzi noted, viruses mutate very quickly, and they code for only a dozen or so proteins at most.
"We have hundreds of host proteins that interact and therefore many more targets that we could use," he said. "Also, while viruses are fast and violent intruders, the host is strong and resilient. Viruses have just a few hours in which to replicate and do what they need to do. So, for a lot of these host-targeting therapies, we can hit a host protein, inhibit a key host protein for a certain amount of time, and not affect the host cell. The host cell will be able to recover while the virus is missing key essential resources."