INTRODUCTION

One of the crucial steps in the virus life cycle is the synthesis of the virus particles within the host cell. This includes the synthesis of viral structural proteins and new genomic material. For these processes, all viruses are fully dependent on their host for the required energy (Mahmoudabadi 2017) and building blocks (Berzin 1974; Waldbauer 2019). This dependency is evidenced by experimental findings showing significant metabolic flux alterations in host cells upon infection (Maynard 2010a; Yu 2011). Systems level metabolic studies have particularly highlighted changes in glucose uptake and glycolysis (El-Bacha 2004; Munger 2006), which might be related to an increased demand for biosynthetic precursors as viral production becomes the dominant process within infected cells (Berzin 1974).
The observation of virus synthesis dominating the metabolism and physiology of infected cells suggests that it might be possible to manipulate the latter to control the former (Maynard 2010a; Ikeda 2007). Indeed, several of the existing antivirals such as Ribavirin, Remdesivir, and Gemcitabine are nucleoside analogs that target metabolic enzymes in the nucleotide biosynthesis pathways and are thought to function through their impact on free nucleotide pools in the cell (Wang 2011; Leyssen 2008). An even more specific metabolic approach to inhibit virus production was demonstrated in the case of human cytomegalovirus. For this virus, metabolic analyses highlighted a shifting of metabolic fluxes within central carbon metabolism and fatty acid biosynthesis pathways during infection (Munger 2008). It was predicted that these flux changes could be blocked by perturbation of specific enzymes, which were then targeted with available inhibitors and resulted in reduced virus production (Munger 2008). Systematic analysis of gene knockout effects on infection of bacteria with phage also identified many metabolic genes associated with central carbon metabolism and substrate transport (Maynard 2010b), leading to the proposition of utilising host metabolic engineering to modulate viral production (Maynard 2010a). Such metabolic control has been explored in virus-based bioproduction using insect cells, where alterations in the culture media allowed alteration of metabolic fluxes and production levels (Carinhas 2010).
These experimental findings show that viral biomass synthesis causes significant metabolic flux changes in host cells and that metabolic perturbations can directly alter virus reproduction. Thus, system-level metabolic models could be utilised to predict what types of metabolic alterations can cause what kinds of impact on virus reproduction. While modelling of virus reproduction in host cells has mostly taken a kinetic approach, focusing on translation and transcription processes (Endy 1997; You 2002, Yin 2018), it has been possible to combine such kinetic models with genome-scale metabolic models to account for both host and virus biomass (Jain 2009). This allowed predicting the effects of metabolites available in the culture media on the dynamics of the infection process (Birch 2012). In a human cell context, genome-scale metabolic models were utilised to analyse the metabolic impact of infection of macrophages with bacteria or viruses (Bordbar 2010; Aller 2018). One of these studies incorporated viral production into the macrophage metabolic model and predicted specific reaction perturbations that can cause a reduction in viral reproduction (Aller 2018). These predictions correctly identified enzyme targets of the aforementioned antiviral drugs in nucleotide pathways and highlighted new target enzymes (Aller 2018). Such findings from the virus-host metabolic modelling aligns with the observations that genome-scale metabolic models can provide a comprehensive stoichiometric catalogue of possible biochemical conversions in a cell (Thiele 2013; Swainston 2016) and that can generate useful qualitative predictions on the impact of environmental or genetic alterations on the cellular metabolic flux distributions (Edwards 2000; Segré 2002; Papp 2004).
Motivated by the qualitative predictive power of stoichiometric metabolic models and flux analysis, we apply it here to simulate the production of SARS-CoV-2 virus particles as part of the host metabolism and predict metabolic inhibitions against this virus. We extend the available human genome-scale metabolic model with a viral biomass reaction, estimated using structural information available from SARS-CoV-2 and related viruses. By optimising flux distributions in this model for viral reproduction, we were able to predict reactions that can inhibit this process. We explored the possible impact of these predicted inhibitions on the host metabolism itself, using an estimation of host cell maintenance in human lung cells based on available protein expression data. In addition, we have explored experimental feasibility of implementing the predicted metabolic perturbations using available drug and inhibitor information on metabolic enzymes. Our results indicate that individual and double perturbation of several metabolic reactions from central metabolic pathways can inhibit SARS-CoV-2 reproduction production in general and some of these can do so selectively without affecting normal metabolic functions of the host. We highlight these reactions as experimentally testable drug targets for inhibiting SARS-CoV-2 reproduction in human lung cells and provide details of the implemented computational approach for further development.