Drug Repurposing for COVID19: What has been done, and how we're tackling it.

Originally published on LinkedIn: https://bit.ly/3EqDDeN

Over the past few weeks, we’ve strategized potential repurposing options for the treatment of Coronavirus (2019-nCov). Before diving into more details on our approach and the proposed molecules, it’s important to catch everyone up on what is happening, where we stand, and the heroic contributions by the scientific community to date. We’ll then present how we thought a bit differently about what we could do to leverage our computational drug discovery platform to present alternative treatment options.

Coronaviruses are a large family of viruses found in both animals and humans. They are responsible for 15% of all common cold cases, while some family members are responsible for the more severe Severe Acute Respiratory Syndrome (SARS) and Middle East Respiratory Syndrome (MERS). While genetically related to the above mentioned viruses, significant differences exist; consequently, no vaccine or antiviral treatment is approved to address 2019-nCoV. In addition to government efforts, a number of companies have suggested potential repurposed treatments for 2019-nCoV. The benefit of going down the repurposing path is that the drug is often readily available, and has been approved by regulatory agencies. Gilead made arguably the most noteworthy and impactful contribution to date by suggesting remdesivir, a drug-like molecule they attempted to introduce for Ebola previously. After showing positive data from the first 2019-nCoV patient in the US, Gilead and the Chinese government are working through IP issues to push the molecule forward. Recently AbbVie jumped in and offered $1.5M of their HIV drug Kaletra, with the hypothesis that the drug’s two antiretroviral components, lopinavir and ritonavir, which are protease inhibitors designed to block HIV viral replication, can do the same thing for 2019-nCoV.

In addition to pharma’s repurposing efforts, Artificial Intelligence (AI) approaches are being deployed in multiple ways to expedite the recommendation of potential therapeutic solutions. Using their AI-based platform, BenevolentAI identified Baricitinib, a drug commonly used to treat rheumatoid arthritis, as a repurposing option, with a host-based target. As of this writing, it has not been confirmed whether Barcitinib will work and BenevolentAI has suggested that further tests are required to support their suggestion. Around the same time, Insilico Medicine used their system to identify thousands of new molecules that could be turned into medicines that attack the virus directly. The company says that they are moving forward with synthesizing the most promising molecules. While one or more of these new molecules could become effective treatments, the reality is that synthesizing molecules can be time-consuming, establishing appropriate biological assays for novel viral targets is challenging, and given the molecules novelty, the path to get to the clinic can take a number of years after pre-clinical testing is complete. As such, the repurposing path offers more immediate benefits, which is important for this current outbreak.

With this in mind, we sought to evaluate the therapeutic potential of not only FDA-approved compounds but also those deemed safe to enter the clinic. Toward this end we used our deep learning proteome screening engine, MatchMaker, which powers our Ligand Design and Ligand Express drug discovery platform, to screen a library of >6,700 approved drugs and drug candidates with at least Phase 1 clinical data against the entire human proteome. The resulting database, dubbed PolypharmDB, is a unique collection of clinically assessed molecules and their polypharmacology profiles that can be searched rapidly to identify possible drug candidates for selected protein targets. We then probed PolypharmDB to identify potential binders of human targets linked to coronavirus pathogenesis. We credit the scientific community for enabling our target selection who, following the SARS and MERS outbreaks, used a variety of approaches to identify potential human and non-human therapeutic targets. The results of these studies are several promising human target candidates, including the TCDD-inducible poly(ADP-ribose) polymerase (TiPARP). Recently, it was shown that increased TiPARP activity was required for coronavirus to achieve maximal viral replication. Therefore, it is suggested that suppression of TiPARP may be a novel, previously unappreciated, a mechanism for treating viral infection. Overall, by leveraging PolypharmDB, we identified molecules that are predicted to interact with several human targets of possible therapeutic relevance.

While targeting human proteins is a promising strategy towards treating viral infection, there are many reasons to target the virus outright (e.g. less chance of side effects, direct binding leads to increased chance of viral inhibition, etc.). Harnessing the flexibility and benefits that PolypharmDB affords, we turned our attention to assessing proteins sourced from the 2019-nCoV virus. We began by modelling two viral 2019-nCoV proteins with established therapeutic relevance based on existing coronavirus structures and recent genomic studies. The first being an enzyme called 3C protease (3Cprot), which is essential for viral replication, and the second a protein called spike glycoprotein (S) which is responsible for binding to human host cells and inducing viral uptake. We then supplemented the results within PolypharmDB to include the likelihood of all clinical molecules interacting with the novel viral targets by assessing them with MatchMaker. By searching the expanded PolypharmDB, we identified molecules predicted to interact with the designated viral targets. Using this method, we identified 10 molecules that ranked highly for binding to the 3Cprot relative to our collection of the human proteins (~9,000 proteins), that were then reviewed by our internal applied science team which comprises of medicinal chemists and biochemists.

To be clear, we are not suggesting that these molecules will work. We have shown that computationally MatchMaker has the ability to generalize over human protein structures, and we have prospectively validated it in other therapeutic areas, but we have not yet done this on non-human structures. In addition, while we approached this situation delicately by reviewing the literature and evaluating the overall scientific fidelity of our approach, we have not experimentally validated these molecules for their intended target - we simply do not have access to the biological assays required for validation. With that said, given the urgency of the situation, we believe that the scientific community should consider the merits of the molecules we are proposing. If there is interest to gain access to the shortlisted molecules from PolypharmDB, please reach out to us directly. We believe it is our moral obligation to go about this the right way by explaining our methodology thoroughly and by setting expectations appropriately. Following a discussion, we can make the molecules available for testing to the right partner at no cost (we are not looking to profit off this at all).

Naheed Kurji, Chief Executive Officer

Naheed Kurji, Chief Executive Officer

Naheed Kurji is the Co-Founder, President and CEO of Cyclica. Naheed is passionate about building AI-augmented technologies that enable researchers to make more strategic and informed decisions in Healthcare and the life sciences. He spends the majority of his time obsessing over Cyclica’s culture, defining its strategy to best effect change in the pharma industry to achieve the company’s vision, and exploring opportunities for continued innovation.

Related Posts

What differentiation really means (and what it does not)

It’s only a minor exaggeration to say that I get asked about differentiation on a daily basis....


Part I: Leadership Lessons: Leading During Uncertainty (a COVID19 motivated article)

Originally published on LinkedIn: https://bit.ly/3BnnNzB

Reflecting on the past four months, many...


Neither AI nor HI are the silver bullet, but combined, they are

Recently, the Reproducibility Project: Cancer Biology published work that suggests that “more than...