A parsimonious approach for recognizing SARS-CoV-2 and host interactions.

Effective countermeasures against the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) demand a better understanding of the pathogen-host interactions. However, such information about the targets, responses, and effects in the host due to the virus is limited, especially so in the case of newly emerged pathogens. The peptide domains that form the interfaces of host and pathogen interacting proteins being evolutionarily conserved, it may be hypothesized that such interactions can be inferred from the similarities in the nucleotide sequences between the host and the pathogen. This communication reports the results of a study based on a parsimonious approach for the identification of the host-virus interactions, where sequence complementarity between the human and SARS-Cov-2 genomes was used to predict several interactions between the host and SARS-CoV-2 at different levels of biological organization. In particular, the findings are suggestive of a direct effect of SARS-CoV-2 on cardiac health. The existing literature on host responses to SARS-CoV-2 and other viruses attest to many of these predicted interactions, supporting the utility of the proposed approach for the identification of host interactions with other novel pathogens.

INTRODUCTION

Following the first report from the city of Wuhan in the Hubei province of China in December 2019, a novel coronavirus‐induced disease, coronavirus disease 2019 (COVID‐19), has spread rapidly, triggering a global pandemic. COVID‐19 is caused by a hitherto unknown beta‐coronavirus which has been named severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) due to its high sequence similarity with SARS‐CoV. Besides the pneumonia‐like manifestations, such as cough, fever, and acute respiratory failure, evidence of the attack to multiple organs, such as digestive, cardiovascular, urinary, and reproductive systems have been reported. , , , Still, information on most aspects of this virus, including its interactions with the host, is limited.

Knowledge of the complete repertoire of host cell molecules that a pathogen can interact with can be extremely helpful in understanding the pathobiology of the disease but is distinctly lacking for newly emerged pathogens. The peptide domains forming the interfaces of host‐pathogen interactions are evolutionarily conserved and parsimonious comparison of host and pathogen genomes can uncover hitherto unknown interaction networks. Beyond the protein‐protein interactions, complementary RNA–RNA interactions between the host and the pathogen that may result in altered expression of certain host genes can also be identified, but not distinguished from the former, by parsimonious associations. This communication reports the results of a study based on a parsimonious, sequence complementarity‐based approach for the identification of the human and SARS‐CoV‐2 interactions.

METHODS

All complete genome sequences of SARS‐CoV‐2 available as of March 25, 2020 at the Virus Pathogen Database and Analysis Resource (VIPR) were added to the workbench and aligned using MUSCLE with Unclust function at default settings. After completion of the multiple sequence alignment (MSA), the consensus sequence was generated using VIPR workbench analysis tools and gaps were removed. BioEdit was used for visualization of the MSA and calculation of its entropy.

The consensus genome sequence of SARS‐CoV‐2 was used to query the human RefSeq Gene, genomic plus transcript (G + T), and PDB nucleotide databases, respectively, by blastn at default settings. Irrespective of their expect values, all nonredundant hits from the blastn outputs were selected for further analysis; the hits from the human RefSeq Gene and transcript datasets were used to query STRING database for network reconstruction and enrichment.

RESULTS AND DISCUSSION

It was hypothesized that if a viral gene abc, coding for a protein ABC, was similar in sequence to a host gene xyz, coding for protein XYZ, then the viral protein ABC could possibly interact with some of the proteins in the host that the protein XYZ interacted with and thereby, compromise the functions and pathways served by XYZ. Otherwise, the sequence similarity between abc and xyz could also possibly result in a gene silencing event in the host. Therefore, a consensus sequence of SARS‐CoV‐2 was derived and its local similarity to the human genome was investigated.

As of 25 March 2020, 110 complete genomes of SARS‐CoV‐2 were available at VIPR. The entropy of the MSA (Supplementary File 1) of these 110 genomes is shown in Figure 1A. Overall low entropy for the MSA implies that the findings based on the consensus sequence can be generalized for all other sequences. The blastn search against the human RefSeq Gene database yielded 69 hits, and that against the G + T database yielded 93 hits, respectively. In all, 73 nonredundant hits were identified from the human RefSeq Gene and transcript databases (Supplementary File 2).

Figure 1

(A) Entropy of multiple sequence alignment (MSA) of severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) genomes. One hundred and ten complete genomes of SARS‐CoV‐2 available at VIPR were selected and aligned with MUSCLE. The MSA was visualized in BioEdit and an entropy (Hx) plot was generated. (B) Interaction network of the host molecules showing parsimonious association with the consensus SARS‐CoV‐2 genome. STRING was used for network reconstruction and gene enrichment of the 73 nonredundant blastn hits. Thirty of the 73 hits did not interact with each other. (C) Gene ontologies (GO) of molecular functions in the host associated with SARS‐CoV‐2 infection. (D) GO of biological processes in the host associated with SARS‐CoV‐2 infection. The numbers of genes assigned to a particular molecular function by STRING, that is, observed gene counts have been depicted; for details of GO identifiers, please refer to Table 1. VIPR, Virus Pathogen Database and Analysis Resource

Table 1

Gene ontologies and molecular pathways enriched in the host associated with SARS‐CoV‐2 infection

Term ID	Term description	False discovery rate	Matching proteins in network
			Molecular function
GO:0005515	protein binding	0.02	ADD1, BLM, BRE, CACNA1D, CDH13, CDH2, CECR1, DMD, DOCK4, DST, EGR2, ENSG00000196689, FOXO1, GHR, GLRA1, GNB1, GRIK2, GRM7, HCN1, HDAC9, HERC1, HPSE2, IKZF5, JUP, KCNQ1, KRT17, MCF2, MYO18A, NTNG1, OCRL, PARK2, PICALM, PPP4R2, PSMA6, PTMA, RALGAPA1, SIN3A, SLC12A6, TLN2, TNNI3K, TYR
GO:0005200	structural constituent of cytoskeleton	0.0481	DMD, KRT14, KRT17, TLN2
GO:0005488	binding	0.0481	ADD1, ARSB, BAZ1A, BAZ2B, BLM, BRE, CACNA1D, CDH13, CDH2, CECR1, DLD, DMD, DOCK4, DSPP, DST, EGR2, ENSG00000196689, FOXO1, GHR, GLRA1, GNB1, GRIK2, GRM7, HCN1, HDAC9, HERC1, HPSE2, IKZF5, JUP, KCNA4, KCNQ1, KRT14, KRT17, MCF2, MRC1, MYLK4, MYO18A, NLRP12, NTNG1, NUBPL, OCRL, PARK2, PAX5, PICALM, PPIP5K2, PPP4R2, PSMA6, PTMA, RALGAPA1, SIN3A, SLC12A6, SULF1, TLN2, TNNI3K, TOX, TTBK1, TYR, ZNF654
GO:0015079	potassium ion transmembrane transporter activity	0.0481	GRIK2, HCN1, KCNA4, KCNQ1, SLC12A6
GO:0022839	ion gated channel activity	0.0481	CACNA1D, ENSG00000196689, GLRA1, GRIK2, HCN1, KCNA4, KCNQ1
GO:0030165	PDZ domain binding	0.0481	DOCK4, GRIK2, GRM7, PARK2
GO:0004065	arylsulfatase activity	0.0482	ARSB, SULF1
GO:0005216	ion channel activity	0.0482	CACNA1D, ENSG00000196689, GLRA1, GRIK2, HCN1, KCNA4, KCNQ1
GO:0005231	excitatory extracellular ligand‐gated ion channel activity	0.0482	ENSG00000196689, GLRA1, GRIK2
GO:0005261	cation channel activity	0.0482	CACNA1D, ENSG00000196689, GRIK2, HCN1, KCNA4, KCNQ1
GO:0005267	potassium channel activity	0.0482	GRIK2, HCN1, KCNA4, KCNQ1
GO:0015276	ligand‐gated ion channel activity	0.0482	ENSG00000196689, GLRA1, GRIK2, HCN1
GO:0019899	enzyme binding	0.0482	CDH2, DMD, DOCK4, EGR2, FOXO1, GHR, GNB1, GRIK2, HDAC9, HERC1, JUP, KCNQ1, MCF2, OCRL, PARK2, PICALM, SLC12A6
GO:0019901	protein kinase binding	0.0482	CDH2, DOCK4, GHR, HDAC9, JUP, KCNQ1, PARK2, SLC12A6
GO:0019903	protein phosphatase binding	0.0482	CDH2, FOXO1, JUP, KCNQ1
GO:0022843	voltage‐gated cation channel activity	0.0482	CACNA1D, HCN1, KCNA4, KCNQ1
GO:0033613	activating transcription factor binding	0.0482	EGR2, PTMA, SIN3A
GO:0043167	ion binding	0.0482	ARSB, BAZ1A, BAZ2B, BLM, CACNA1D, CDH13, CDH2, CECR1, DLD, DMD, DSPP, DST, EGR2, ENSG00000196689, GLRA1, GRM7, HCN1, HDAC9, HPSE2, IKZF5, KCNA4, KCNQ1, MYLK4, MYO18A, NLRP12, NUBPL, PARK2, PICALM, PPIP5K2, PSMA6, SULF1, TNNI3K, TTBK1, TYR, ZNF654
GO:0045294	alpha‐catenin binding	0.0482	CDH2, JUP
GO:0046873	metal ion transmembrane transporter activity	0.0482	CACNA1D, ENSG00000196689, GRIK2, HCN1, KCNA4, KCNQ1, SLC12A6
GO:0098632	cell‐cell adhesion mediator activity	0.0482	JUP, NTNG1

Abbreviation: SARS‐CoV‐2, severe acute respiratory syndrome coronavirus 2.

Network reconstruction (Figure 1B) and enrichment (Figure1C,D; Table 1) of the 73 hits, performed with STRING, identified the prominent interactions of the pathogen with the host at the bioprocess, molecular function, and pathway levels. Thirty of the 73 hits were found not to interact with each other. The proteins (ADA2, ADD1, HDAC9, JUP, MRC1, PPP4R2, and SIN3A), their paralogs (GNB1, KCNA4, KRT14, KRT17, MYO18A, NLRP12, PSMA6, SLC25A43, TLN2, and TTBK2), or regulatory subunits thereof (CACNA1A), coded by 18 of these 73 genes have been shown to be expressed differentially during SARS‐CoV‐2 infection. Intriguingly, sequence complementarity of any segment of the consensus viral genome was not seen withace2 or tmprss2. A blastn query of the RefSeq genes of the family Coronaviridae with human ace2 reference sequence at default settings showed sequence complementarity within orf6 of SARS‐related bat Coronaviruses but neither with any sequence of SARS‐CoV‐2 nor with any spike protein of the viral family (data not shown). However, at the molecular function level, an interaction between the host and the virus, involving PDZ domain binding (GO:0030165), was inferred; previously, the PDZ‐binding motif of the SARS envelope protein has been established as a determinant of viral pathogenesis.

Sequence similarity with dentin sialophosphoprotein (dspp) gene as well as the involvement of enamel mineralization (GO:0070166) could be also inferred, hinting at the possibility of developmental defects in dentition and tooth decay in COVID‐19 patients. The canine distemper virus, a member of Paramyxoviridae, is also known to interfere with enamel mineralization in its host, resulting in poor dentition in animals that suffer from the disease while their adult teeth are forming.

The involvement of two pathways viz. arrhythmogenic right ventricular cardiomyopathy (ARVC) (hsa05412) and glutamatergic synapse (hsa04724) was also identified (Table 1). The enrichment of ARVC at the pathway level and of regulation of heart rate (GO:0002027), regulation of cardiac muscle contraction (GO:0055117), regulation of heart rate by cardiac conduction (GO:0086091), regulation of atrial cardiac muscle cell membrane repolarization (GO:0060372), and cardiac muscle cell action potential (GO:0086001) at the bioprocess level is highly suggestive of a direct cardiomyopathic effect of SARS‐CoV‐2; myocardial injury associated with in‐hospital mortality has already been reported in confirmed and suspected COVID‐19 patients. Xiong et al. have also reported the enrichment of ARVC‐related transcriptional responses in COVID19 patients. It is important to note that cardiac comorbidities have been a major risk factor for COVID19‐related deaths and that the penetrance of ARVC‐related mutations is very high in the populations of the Mediterranean basin, especially Italy, where the highest death rates due to this disease have been experienced.

The involvement of glutamatergic synapse (hsa04724) at the pathway level was predicted with much lower confidence than the ARVC pathway. However, some other viruses, including the neurotropic and neuroinvasive human coronaviruses, the HCoV strain OC43, for example, are already known to elicit glutamatergic excitotoxicity. Based on this finding, N‐methyl‐ ‐aspartate receptor antagonists such as memantine may be tested in the management of COVID‐19 patients should they show specific signs of excitotoxicity. Interestingly, our analysis was also able to identify the sequence similarity between viral NSPs 14‐15 and hostntng1, previously reported by Lehrer and Rheinstein. Given the interactions of hostntng1 with cntn5 and cdh13 (Figure 1B), the competitions arising from this sequence similarity between the virus and ntng1 may account for the sensory disturbances, such as hyposmia/anosmia and dysgeusia observed in COVID‐19 patients. Clustering analysis of SARS‐CoV‐2, SARS, and MERS virus genes with human genes based on the codon usage and molecular features also associated the human genes with diseases of the nervous and cardiovascular systems.

Finally, the blastn search against the PDB nucleotide database yielded 6 redundant hits (Supplementary File 2), all showing complementarity over a 16 base‐long stretch, in the immediate proximity of an RGD domain‐coding sequence and within the receptor binding domain‐coding sequence, of the viral spike glycoprotein gene and the human 18S rRNA. The occurrence of an 18S rRNA‐complementary sequence juxtaposed with an integrin‐binding domain‐coding sequence within the receptor‐binding domain of SARS‐CoV‐2 may have strategic implications in the successful arrest of host translational machinery and virulence that is worthy of further investigations.

In conclusion, a computational, sequence complementarity‐based parsimonious approach was used to identify different types of host‐virus interactions. Existing literature on SARS‐CoV‐2 and related viruses is in support of many of these predicted interactions whereas further studies are warranted for attesting some of the other predicted interactions. It is believed that such studies will help in the development of new antiviral and disease management strategies.

CONFLICT OF INTERESTS

The author declares that there are no conflict of interests.

Supporting information.

Click here for additional data file.

Supporting information.

Click here for additional data file.