A significant proportion of patients who develop severe disease after being infected with SARS-COV-2 appear to experience cytokine storm syndrome (CSS). CSS is characterized by excessive production of inflammatory cytokines which causes over-activation of various types of immune cells, infiltration of and damage to patient’s tissues. It is associated with a high mortality rate. There are several types of cytokine storms caused by various factors such as genetic mutations, response to immunotherapy and bacterial and viral infections, including SARS-CoV-2. The mechanism of the cytokine storm is not well understood, and specific cytokines involved differ between different types of cytokine storms. There are several drugs developed to treat the condition and some of these are already being trialed in COVID-19 patients. As with all drug development, success is not guaranteed. In addition, a potential drawback of these treatments is that they may suppress the overall immune response and interfere with virus clearance.
We are applying computational methods including data-driven network modelling in conjunction with molecular modelling to create new possibilities for drug repositioning to treat cytokine storm in COVID-19 patients. A detailed mathematical model based on patient and in vitro data will allow us to simulate CSS, understand its hidden dynamics and predict best protein targets for intervention, the same way we can understand and manipulate complex man-made systems such as airplanes and computers. The chosen drug targets should suppress the cytokine storm without compromising the immune response to the virus. Of course, not all protein targets are druggable, so to increase our probability of success we will apply simulation to mimic protein dynamics and potentially reveal hidden or ‘cryptic’ allosteric binding pockets. All identified druggable pockets will be subsequently screened virtually with market drugs or those compounds that are in clinical trials with the ultimate goal of repurposing a drug for COVID-19.
Progress during the Hackathon
During the Hackathon, we have created the first version of the model of interactions between cytokines elevated in COVID-19 patients. As a first approximation we have used the information gathered through intensive literature research to infer interactions between cytokines. We calibrated the model using published levels of cytokines in COVID-19 patients and estimated decay rates of the cytokines and have encouragingly achieved a good fit to the data. The model does not yet have full predictive ability but has suggested speculative targets and the directions for further investigation. We have taken one of these targets (IP-10/CXCL10 in its oligomeric form) as revealed by our network analysis and subjected it to analysis for druggable binding pockets (including cryptic allosteric sites). We have identified a potentially druggable site that will next be probed by virtual screening in a drug repurposing campaign.
The solution’s impact on the crisis
As of now, the extent of the COVID-19 crisis and its duration are unknown, hence, an understanding of CSS is critical in order to assist in the treatment of patients with severe symptoms. To our knowledge, there are no studies as of yet complete that have determined the number of patients suffering from CSS in COVID-19, however, it appears to be a significant proportion. We believe that our platform approach to modelling CSS within the context of COVID-19 will help to identify potential drugs for repurposing for CSS treatment and have the potential to have a direct impact on the healthcare system and patients’ well-being.
What we need to continue
In order to continue the model development we will need to establish collaborations with hospitals and research groups studying cytokines. It will allow us to obtain dynamic data of cytokine levels in patients and more precise data on cytokine-cell interactions such as phosphoproteomic and transcriptomic responses which will be used to dramatically improve model’s predictive capability and expand the range of potential targets. This will require external funding to cover the costs of the experimental data generation and target validation. We will also need to establish links with pharmaceutical companies that could carry out clinical trials with identified compounds.
The value our solution after the crisis
CSS is an immune system reaction not exclusive to COVID-19, hence, a developed model will be extremely useful for developing drugs against other types of CSS and may even aid in the preparation against upcoming potential viruses that may lead to CSS. The model could be potentially further extended to find or reposition drugs against challenging immune diseases such rheumatoid arthritis and systemic lupus.