Therapeutic approaches for COVID-19 based on the dynamics of interferon- mediated immune responses

1 Pharmaceutical Sciences Research Center, Shiraz University of Medical Sciences, Shiraz, Iran 2 Department of Pharmaceutical Biotechnology, School of Pharmacy, Shiraz University of Medical Sciences, Shiraz, Iran 3 Student Research Committee, Shiraz University of Medical Sciences, Shiraz, Iran 4 Cellular and Molecular Medicine Student Research Group, School of Medicine, Shiraz University of Medical Science, Shiraz, Iran 5 Non‐Communicable Diseases Research Center, Shiraz University of Medical Sciences, Shiraz, Iran Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 12 March 2020 doi:10.20944/preprints202003.0206.v1


Introduction
Coronaviruses (CoVs) are a group of RNA viruses that have the largest RNA genome among all the viruses known to date (1,2). Their genome is surrounded by a bilayer lipid envelope containing the spike and membrane proteins (3). CoVs replicate by the attachment of their spike protein to the host cell receptors resulting in release of the viral genome into the cell (4). They have several hosts including animals and human (5). They mainly cause respiratory disease and common cold (6), but can also cause central nervous system (CNS) infection (7).
The recent outbreak of COVID-19, the new disease caused by a novel coronavirus species, named SARS-CoV-2, has alerted many researchers around the world to find treatment for this condition.
Severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East respiratory syndrome coronavirus (MERS-CoV), two other known viruses from the same genera, were identified in 2002 and 2012, respectively, causing serious respiratory ailments (8).
COVID-19, though a new virus, seems to have a similar pattern to SARS and MERS (9). Despite the differences in the mortality and epidemiological rates of these three diseases, the pattern of age-specific mortality is similar; and the mortality rates get higher as the age increases with in the highest mortality rates among the elderly (9) ( Table 1). The zero fatality in children under nine years old (Table 1) seems contradictory to the fact that the immune system gets stronger when a child grows up (11). However, there are differences at the timing of the initiation of immune responses in children versus adults (11).
It has been shown that the potential first lines of defense against SARS are mediated through mannose-binding lectin as a pattern recognition molecule (PRM) of innate immunity (12).
Additionally, interleukin (IL)-12 seems to play a vital role in SARS (13). IL-12 activation would lead to the induction of interferons (IFNs) (14). IFN-γ is a key moderator in linking the innate immunity to adaptive immune responses (13).
IFNs are a group of cytokines, which communicate between cells against pathogens and have a critical role in the immune system, such as activating natural killer (NK) cells and macrophages, in addition to the flu-like symptoms of various diseases. There are three classes of IFNs: I (such as IFN-α and -β), II (IFN-γ), and III, all of which play roles against viral infections (15).
In SARS-CoV and MERS-CoV, the reaction to viral infections by type I IFNs is suppressed. Both CoVs use variant strategies to decrease type I IFN production. This dampening approach is highly associated with the disease severity and increased mortality (16).
On the other hand, in the lethal cases of SARS-CoV or MERS-CoV infection, the increased influx of inflammatory cells is always observed. In a mouse model of SARS-CoV infection, imbalanced type I IFN and inflammatory cells were shown as the main causes of fatal pneumonia (17).
Understanding the pattern of the immune system induction in adults and children in the CoVassociated respiratory syndromes could help to find treatment strategies for these fatal diseases.
Considering the lack of available data on COVID-19, SARS can be a helpful model in this regard.
Because SARS-CoV-2 has the highest similarity in structure and nucleotide sequence to SARS-CoV among other viruses of this family, showing 96% and 89.6% sequence identity in the proteins of their envelope and nucleocapsid, respectively (18).
In this study, to figure out the underlying mechanism for the differences in the age-specific mortality of SARS, the most important signaling pathways activated by the virus will be studied using bioinformatics tools and a systems biology approach (19). The obtained results will be validated through a literature review on SARS, MERS, and COVID-19 to indicate how the dynamics of IFN-mediated antiviral response in adults, elderly, and children could determine the severity of the disease and treatment outcomes.

Identification of the deferentially expressed genes using high-throughput RNA-Seq datasets
At first, "coronavirus" was searched through iLINCS at http://ilincs.org (20). iLINCS is an integrative online platform that brings together different levels of physiological data and integrates them with a bioinformatics analysis engine aiming to analyze and interpret omics data. GREIN is a web application with comprehensive analytical toolbox, which provides manipulation and analysis of RNA-seq data. The obtained results were then exported to iLINCS (20) for further analysis.

Gene set enrichment analysis
Gene set enrichment analysis is a method to interpret deferentially expressed genes )DEGs( in terms of the affected biological pathways and obtain information regarding signature. Gene ontology (GO) enrichment analysis on DEGs was performed by Enrichr (22,23)  pathway databases were used for pathway enrichment analysis to assess the potential association of the signature with pathways.

Protein-protein interactions network reconstruction
In order to find the essential proteins and pathways in the gene set, the protein-protein interaction (PPI) network of the signature was extracted from the International Molecular Exchange Consortium (IMEx) protein interactions database (27) through NetworkAnalyst (https://www.networkanalyst.ca) (28).
NetworkAnalyst is a powerful and user-friendly analytics platform, which assists biologists in the interpretation of systems-level data. This tool was implemented to visualize and analyze the PPInetwork of top 100 DEGs. Noteworthy, in order to control the network size, the minimum network tool was selected amongst network tools, which keeps seed proteins and non-seed proteins that are crucial for network connections.

Identification of DEGs between control and experimental groups and enrichment analysis
The currently available wealth of omics data prompt the researchers to develop tools and algorithms to fully exploit the information contained within these data. The list of the DEGs obtained through iLINCS is represented in the supplementary, Tables S1 (top 100 selected genes) and S2 (complete signature). As seen on Fig

The PPI network of top100 DEGs
The protein-protein interaction network for the signature was constructed using IMEx protein interactions database through NetworkAnalyst (Fig. 2), representing the crucial proteins in the network, which are called hub proteins. It is well established that the virus-host interaction has a crucial role in the disease outcome, and infecting the host system is mainly mediated through affecting host's important proteins. Though studying the related PPI network can clarify some routes that virus uses in this regard.
The list of all proteins in the network and node centralities (degree and betweenness) is available in Table S3. Noteworthy, the degree of a node is defined as the number of connections that a node has to other nodes and betweenness centrality is the number of the shortest paths passing through the node in a graph (28).
As shown in the PPI network ( Fig. 2

Dynamics of antiviral response determines the severity of the disease
The dynamics of IFN-related antiviral responses may be the lost circle in understanding the virulence of CoVs. There are some observations and facts supporting this notion.
The zero fatality in children under nine years old (Table 1) seems contradictory to the fact that the immune system gets stronger when a child grows up.
Considering the immunologic differences between adults and children, the IFN-γ induction by NK cells are higher in adults but has a lower threshold in children (11,37). It seems that children respond faster to the virus in the incubation period (38), so that their immune system inhibits the virus replication and prevents high virus titers. On the other hand, in adults, the immunologic response is postponed as the virus impairs the innate immune response by shutting down the signaling pathways.
In a study (39)  To validate this assumption, the dataset of the above-mentioned study (41), GDS1028, was searched and reanalyzed with the help of iLINCS (42). Noteworthy, this dataset contains the expression profiling of peripheral blood mononuclear cells (PBMC) from 10 adult hospitalized patients with SARS and four healthy controls.
The pathway (24,25,43,44) and diseases (45,46) enrichment analysis of the top 100 DEGs was also performed via Enrichr (22,23). Interestingly, disease enrichment analysis demonstrated that the signature was highly associated with some diseases that are the result of the immune system malfunction, including septic shock, obstructive pulmonary bronchitis, allergic diseases, and autoimmune diseases.
Besides, the pathway enrichment analysis revealed that the signature is highly associated with the apoptosis pathway. This result is consistent with the recent study, which represented that the number of T cells in patients with COVID-19 were reduced and functionally exhausted, especially among elderly patients (≥60) and in patients requiring Intensive Care Unit (ICU) (47).
The enrichment analysis results are represented in Fig. S2 and Tables S3. Moreover, the list of the DEGs obtained through iLINCS is represented in the supplementary, Tables S4 (top 100 selected genes) and S5 (complete signature).
The importance of T cell-mediated immune response in respiratory CoV is well established (48).
Type I IFN response is shown crucial in T cell survival (49,50). Moreover, the phosphorylation of STAT1 and STAT5 was increased in the activated naïve CD4+ T cells taken from young adults in order to lower their response threshold to type I IFN stimulation. However, this mechanism was subdued in the elderly naïve CD4+ T cells (51). Likewise, the impaired innate IFN secretion in the elderly is well documented in several studies (46,48).
As mentioned above, SARS-CoV ORF6 protein antagonizes the function of STAT1. Therefore, STAT1 inactivation by ORF6 protein might be the cause of reduced and functionally-exhausted T cells in patients with COVID-19, especially in the elderly patients (33).
Channappanavar et al. (29) have also demonstrated that suboptimal T cell responses occurred in SARS-CoV-infected BALB/c mice. Although the authors concluded that IFN-I-mediated inflammatory responses caused impaired T cell function, it seems that antagonizing the function of STAT1 by SARS-CoV can play a crucial role in this impairment. Considering the fact that T cells reduce cytokine storm by modulating the innate immune response (52), it seems that the higher response threshold in the elderly, which is aggravated by the antagonizing effects of ORF6 on STAT1 (33), leads to their poor clinical outcome.
Altogether, it might be concluded that the patients with SARS-CoV at the late stages of the disease suffer from many abnormalities, which are the result of immune system imbalance and malfunction and lack of effective IFN-specific immune responses that can lead to proinflammatory reactions and immunopathological conditions, presented by lethal inflammations in the lungs and vascular leakage (29). which finally led to a more robust antiviral symphony against virus replication (59).

Approaches to control COVID19: a systems biology perspective
Altogether, it seems that combinational IFN therapy could significantly inhibit virus replication and overcome the increased response threshold of IFN induction that has been resulted by STAT1 inhibition in the immune cells by CoVs, especially in the elderlies Additionally, another study uncovered that the timing of IFN therapy would be of great importance. In an in vivo study, mice were protected against the virus when IFN-I was given before the maximum rise of the virus, during one day after the infection, though the expression of the ISGs and inflammatory cytokine genes was reduced. On the other hand, treatment failure was seen in case of later injection of IFN-β, because the virus titer went up, and monocytes, macrophages, and neutrophils were accumulated and activated in the lungs, and proinflammatory cytokines were induced, which finally led to severe lethal pneumonia (60).
In this regard, two distinct studies (65,66)  Interestingly, chloroquine, a recently proposed medication for COVID-19 (69), interacts with poly (I:C) as an endosomal acidification inhibitor, which inhibits poly (I:C)-mediated IFN-β expression (70). Therefore, it can be concluded that TLR3 agonists can be a proper option for employment in development of vaccines against COVID-19.
Nevertheless, some of them may strongly stimulate the immune system and lead to unwanted reactions or toxicity. For example, PHA is a natural compound found in high concentrations in red kidney beans and with lower concentrations in other beans (75). The oral consumption of uncooked red kidney bean has been announced to induce gastrointestinal toxicity and mitogenicity (75) because of high levels of PHA. However, whether a low concentration of PHA could be beneficial at the early stages of the disease or incubation period to stimulate IFN production can be a subject for further research.
Although IFNs are available as medicinal products, some adverse effects such as bone marrow suppression should be considered for their direct indication (76). Moreover, the protocol for their indication including proper timing and dosing should be confirmed.
All in all, IFN induction in the incubation period and at the very early stages of the disease could be the key to prevent CoV-associated mortalities, yet the proper dosing needs further investigations. Research and clinical trials for finding the right timing for such interventions as well as introducing the proper dose-adjustment protocols are urgently needed.
On the other hand, at the later stages of the disease, the balance of the immune system becomes impaired, hence probable inflammatory over-reactions, cytokine storm, and possible autoimmune responses should be considered. In such circumstances, therapeutic approaches to reduce possible lung inflammations may be needed. The key for success in reducing the disease fatality might be the stimulation of the innate immune responses to trigger IFN production at the very early stages of the disease, which might be done through administration of agents that are able to augment IFNs production such as poly ICLC.

Conclusion
Despite the evidences for the efficacy of IFNs in treating CoV-induced infections, the proper dosing and ideal timing for such interventions needs to be verified in clinical trials. Moreover, adding IFN-γ to an IFN-I as a combination therapy is strongly suggested.
At the later stages of the disease, the balance of the immune reactions would be disrupted and the responses would shift toward immnopathogenic over-reactions and probably cytokine storm presented by severe respiratory syndrome, which might indicate a need for tempering the immune system activity, although this suggestion might need more clinical evidences.

Conflict of interest
Authors declare no conflict of interests. Figure S1. Gene set enrichment analysis of top DEGs of GDS1028 dataset. Table S1. Top 100 selected DEGs (deferentially expressed genes) of some subgroups from GreinGSE52405 dataset. Table S2. The complete signature of the MA15 group data from GreinGSE52405 dataset. Table S3. The list of all proteins in the network and node centralities (degree and betweenness) of the tested subgroups of GreinGSE52405 dataset. Table S4. Top 100 selected DEGs (deferentially expressed genes) of GDS1028 dataset. Table S5. The complete signature of GDS1028 dataset.  Table S1. Top 100 selected DEGs (deferentially expressed genes) of some subgroups from GreinGSE52405 dataset. The data is obtained through iLINCS and analyzed by Grein. The tested subgroups included the C57BL/6J mice infected with MA15 (mouse-adopted severe acute respiratory syndrome coronavirus) at four days post-infection. The differences in the gene expression level (signature) between three experimental groups and two control groups were analyzed in this study. Table S2. The complete signature of the MA15 group data from GreinGSE52405 dataset obtained through iLINCS and analyzed by Grein. The tested subgroups included the C57BL/6J mice infected with MA15 (mouse-adopted severe acute respiratory syndrome coronavirus) at four days post-infection. The differences in the gene expression level (signature) between three experimental groups and two control groups were analyzed in this study. Table S3. The list of all proteins in the network and node centralities (degree and betweenness) of the tested subgroups of GreinGSE52405 dataset (the C57BL/6J mice at four days post-infection with MA15, as three experimental groups, and two control groups) obtained through analysis of the protein-protein interaction network. Table S4. Top 100 selected DEGs (deferentially expressed genes) of GDS1028 dataset, obtained through iLINCS. This dataset contains the expression profiling of peripheral blood mononuclear cells (PBMC) from 10 adult hospitalized patients with SARS and four healthy controls (41). Table S5. The complete signature of GDS1028 dataset, obtained through iLINCS and analyzed by Grein. This dataset contains the expression profiling of peripheral blood mononuclear cells (PBMC) from 10 adult hospitalized patients with SARS and four healthy controls (41).