SARS CoV-2 spike protein variants exploit DC-SIGN/DC-SIGNR receptor for evolution and severity: an in-silico insight

Abstract

The Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) is related with the COVID-19 pandemic. Recent spike protein variations have had an effect on the transmission of the virus. In addition to ACE-2, spike proteins can employ DC-SIGN and its analogous receptor, DC-SIGNR, for host evasion. Spike variations in the DC-SIGN interaction region and role of DC-SIGN in immune evasion have not been well defined. To understand the spike protein variations and their binding mode, phylogenetic analysis of the complete GISAID (Global Initiative for Sharing Avian Influenza Data) data of the SARS-CoV-2 spike protein was considered. In addition, an in silico knockout network evaluation of the SARS-CoV-2 single-cell transcriptome was conducted to determine the key role of DC-SIGN/R in immunological dysregulation. Within the DC-SIGN-interacting region of the SARS-CoV spike protein, the spike protein of SARS-CoV-2 displayed remarkable similarity to the SARS-CoV spike protein. Surprisingly, the phylogenetic analysis revealed that the SARS-CoV-2’s spike exhibited significantly diverse variants in the DC-SIGN interaction domain, which altered the frequency of these variants. The variation within the DC-SIGN-interacting domain of spike proteins affected the binding of a limited number of variants with DC-SIGN and DC-SIGNR and affected their evolution. MMGBSA binding free energies evaluation differed for variants from those of the wild type, suggesting the influence of substitution mutations on the interaction pattern. In silico knockout network analysis of the single-cell transcriptome of Bronchoalveolar Lavage and peripheral blood mononuclear cells revealed that SARS-CoV-2 altered DC-SIGN/R signaling. Early surveillance of diverse SARS-CoV-2 strains could preclude a worsening of the pandemic and facilitate the development of an optimum vaccine against variations. The spike Receptor Binding Domain genetic variants are thought to boost SARS CoV-2 immune evasion, resulting in its higher longevity.

Background

A novel contiguous virus was found in Wuhan city of China, in late December 2019. It belongs to the coronavirus and is classified as SARS-CoV-2 [1]. It has infected billions of people and caused millions of deaths across the globe and poses a risk to human life. a pandemic situation was declared on March 11, 2020 (WHO) [2, 3]. A pandemic infection on March 11th, 2020 [2, 4]. SARS-CoV-2 is the seventh virus known to infect humans date [3, 5], and it is a member of the beta coronavirus family [3], which possesses single-stranded RNA such as SARS CoV and MERS. It reduces lung function [6] by causing vascular thrombosis, endothelial damage, and microangiopathy. The fusing of cellular membranes is a crucial stage in viral infection [6]. The fusing of cellular membranes is a crucial stage in viral infection. The spike (S) protein, which consists of 1273 amino acids and is a transmembrane fusion glycoprotein [7], is essential for membrane fusion. It has two domains named S1 and S2 [4, 6, 8], and a region between amino acids 682 and 685 where furin can disassemble it. Angiotensin-converting enzyme 2 (ACE2) is the major receptor for SARS-CoV spike [9] and SARS-CoV-2 [10], which regulates how the virus travels from one cell to the next [11]. In addition, the receptor binding domain (RBD) amino acid alteration improved spike protein stability and transmission antigenicity [12]. Nevertheless, the low expression of ACE-2 on a subset of lung epithelial and endothelial cells demonstrates that additional receptors are involved in infection and entrance [11, 13,14,15,16]. Additionally, alterations in the RBD of spike protein impact the propagation of the virus [17]. Spike N501Y (U.K. variant B.1.1.7) [18] and D614G have recently found, with the potential to increase ACE2 recognition and effect the considerable dissemination of these variants across different populations [19]. The D614G variant had a high transmission rate, indicating significant viremia in COVID patients but it had no effect on disease severity [19].

DC-SIGN and DC-SIGNR are member of c-type lectins protein family [20] which recognizes carbohydrates. They function as adhesion and pathogen pattern-recognition receptors (PRR) that identify high mannose and complex glycan substrates and influence viral spread [20]. These interactions disrupt the signaling of other PRRs, such as TLRs, and induce immunological regulation. The infection of SARS CoV-2 begins by recognizing ACE2 [8, 15]. Utilizing auxiliary receptors restricts virus transmission, though. CD209L (DC-SIGNR) and CD209 (DC-SIGN) are known as alternative receptors for SARS CoV, SARS CoV-2, and HIV-1 to thrive on the host [13]. DC-SIGN/CD209 is found on PBMCs [19] such as dendritic cells and tissue-specific alveolar and dermal macrophages [20]. Similarly, DC-SIGNR is strongly expressed on type II alveolar lung epithelium and on the endothelium surface of the liver, lung, and lymph node [20]. Despite 77% similarities, their glycan recognition profiles are distinct. [21]. Moreover, DC-SIGN/R participation in infection of SARS-CoV-2 was previously demonstrated [13]. The correlation between DC-SIGN expression and disease severity suggests a potential target for SARS-CoV-2. Therefore, analyzing spike variations in regions where DC-SIGN/DC-SIGNR interact, may be essential for understanding how infections and evasions arise.

Our study describes the region of the SARS CoV-2 spike RBD where DC-SIGN and DC-SIGNR interacted and their variants. Based on the amino acid sequence, the spike RBD domain interacts variably with DC-SIGN/DC-SIGNR in GISAID sequences from 2020 to 2021. Also, the link between amino acid substitution in spike RBD, evolution, and binding was examined. Spike protein variations had unique molecular interactions with DC-SIGN/R, but not ACE2. In addition, we found variation in their binding free energy and per residue contribution utilizing MMGBSA of HawkDock [22]. By analyzing the single-cell transcriptomes of BALF and PBMC from SARS-CoV-2 patients, we further showed that DC-SIGN/R influenced the propagation of network signals. The diverse variations may boost fitness by making infections in several organs more likely to cause their failure. The development of new treatments to limit the spread of covid infection may depend on the identification of spike RBD variants and the manner in which they evolve.

Methods

Retrieval of DC-SIGN structure and active site identification

The 3D structures of DC-SIGN, DC-SIGNR (CD209/L) and ACE2 (PDB ID: 2XR5 1XPH and 6M17) were retrieved from the Protein Data Bank. The models were refined and minimized using the protein preparation wizard tool of Maestro in Schrodinger’s software suite. The interacting residues of DC-SIGN, DC-SIGNR and ACE2 were obtained from literature [23]. Specifically, residues N365, D366, N367, K368, F313, F374, D320, L321, Q323, G325, T326, and W327 comprise the active site of DC-SIGN and DC-SIGNR, and amino-acid stretches 30–41, 82–84 and 353–357 form the ACE2 active site.

Sequence alignment and modeling of spike protein

Primary amino acid sequence of the spike glycoprotein of SARS CoV (Accession no. AAR86788.1) and SARS CoV-2 (YP_009724390.1) was retrieved from the NCBI database. A model of SARS-CoV-1 and SARS-CoV-2 was generated using the SWISS-MODEL server. The SARS-CoV (PDB ID: 2GHV) was used as a template because it shares 99.89% sequence similarity with SARS-CoV-2. The reliable models were recognized based on GMQE and QMEAN [24]. GMQE represents the alignment and coverage of a template to query sequence, and QMEAN indicates the similarity of a model to the experimentally solved protein structures. The stereochemical property of models were evaluated by the Ramachandran plot generated using PROCHECK [25]. Moreover, quality of the generated models was evaluated using the Mol Probity scores. Mol Probity score provides an estimate of the computed sum of clashes, unfavored residues in the Ramachandran plot (%), and bad rotamers (%). A low numerical score signifies considerable quality of a particular model. It exhibits the resolution of structure at which the corresponding values might be applicable. Prior to studying protein–protein interaction, 3D models were further refined and energy minimized using Schrodinger’s Maestro protein preparation wizard [26]. The sequence of spike glycoprotein of SARS CoV and SARS CoV-2 was aligned by using Clustal W [27] and the percentage identity between the two was determined. Next, Multiple sequence alignment (MSA) was employed to evaluate the potential DC-SIGN binding residues in SARS-CoV-2 Spike RBD using SARS CoV and DC-SIGN interacting residues as a template.

Mutational analysis of DC-SIGN/R binding region of spike protein

Mutations in SARS CoV-2 DC-SIGN/R binding region (337–399 amino acid) were analyzed for sequences available in GISAID from Oct 2020 to dec 2021. Mutations were analyzed using the GISAID CoV Server (www.gisaid.org/epiflu-applications/covsurver-mutations-app) with ‘hCoV-19/Wuhan/WIV04/2019’ (Accession number NC-045512.2). The information about the collected dataset is presented as Table S1. Noisy sequences were excluded from the analysis. The frequency of identified mutations was followed for a year (from oct 2020 to Oct 2021).

The 3D modeling of the mutated spike protein

FASTA file with altered amino acid sequences was submitted to the SWISS MODEL server to build mutated protein variants of the wild type spike protein (Wuhan SARS CoV-2). The stereochemical quality of the modeled mutants was verified using Procheck module of PDBsum tool [12, 28]. The original and mutant protein structures were viewed and studied using PyMOL visualization program [16].

Molecular docking of spike protein with DC-SIGN/R

Solving the crystal structures of interacting proteins is an inconvenient and laborious task. Several methods have been devised to study protein–protein modeling. The High Ambiguity Driven protein–protein Docking (HADDOCK) modeling approach is employed because it uses biochemical and biophysical data and provides the lowest energy structure [15, 16]. The protein–protein modeling was employed to study the intermolecular interaction of spike glycoprotein with DC-SIGN using HADDOCK 2.4 server [29]. For initiating docking, actively participating residues were provided to the HADDOCK 2.4 server [30]. The active residues of DC-SIGN were retrieved from literature [23] which encompasses calcium coordinated primary binding sites and secondary binding sites. The active site residues from 324 to 386 of spike SARS CoV were utilized. For SARS CoV-2 spike protein, the active site residues were predicted via sequence alignment of spike proteins of SARS CoV and SARS COV-2, that displays high sequence similarity [31] and the corresponding region was considered as the active site for the intermolecular interaction. The passive residues were generated automatically to a distance of 6.5A^° around the protein.

The structural models were ranked based on their Van-der wall and electrostatic interactions, desolvation, Ambiguous interaction restraints (AIRs), and energies of the buried surface energies. The most favored model as indicated by the lowest HADDOCK score was further considered for analyzing details of the intermolecular interactions between SARS CoV-2 and DC-SIGN/R. Structural models with lowest energy score were analyzed and visualized using [32] PyMOL and Mapiya [33]. The intermolecular interacting residues across the proteins involved in interaction were recognized by PDB sum (http://www.ebi.ac.uk/pdbsum) and visualized as a pictorial interaction map [34]Mapiya visualizes the interaction using a contact map (CM) and a distance map (DM). Mapiya showed a unique feature of coloring and filtering contacts based on the physiochemical properties of interactions Mapiya uses the cutoff distance in angstroms between two heavy molecules of different residues to make the CM interaction map. The pair of residues is colored when the distance between the residues is shorter than a cutoff value. We used a filter for electrostatic and hydrophobic interactions to analyze their respective interactions. Complex 3D structures were displaced using Mapiya's built-in Molstar viewer [33].

Estimation of binding free energy of spike variant-DC-SIGN/R using MMGBSA/MMPBSA method

The free binding energy of intermolecular interaction of Spike protein and DC-SIGN/R was evaluated by utilizing MMGBSA (Molecular Mechanics Generalized Born Surface Area)/ /MMPBSA (Molecular mechanics Poisson Boltzmann surface area) which exhibits more crucial than scoring functions, through Hawk Dock web server [22]. Moreover, it provides the contribution of each residue in binding free energy of PP interaction and depicts the key residues which indicate the near native interaction and their binding affinity. MM/GBSA was employed to determine the interface residues of Spike protein and DC-SIGN/R using Hawk dock scoring used in Hawk Dock webserver.

Influence of point mutation on dynamic of wild type and Spike variants

The effect of mutation on the flexibility, structural conformation and molecular stability of the SARS-CoV-2 protein was investigated using the DynaMut software [35].DynaMut is a software server that uses well established normal mode techniques to analyze and visualize protein dynamics. By uploading the modelled spike structure to the DynaMut software, the influence of point mutations on several protein structure stability characteristics such as deformation energies, atomic fluctuations and vibrational entropy was studied. DynaMut calculates the vibrational entropy changes to determine the impact of mutations on protein stability and dynamics.

DC-SIGN regulatory network construction

The DC-SIGN regulatory network was constructed using the STRING v.10.5 database [36] and Cytoscape [36, 37]. In the constructed network, proteins with highest degree of binding in the network explains a key regulatory role as a hub. The topological properties of the network were calculated using Network Analyzer [38], a plugin of Cytoscape v3.7.2. The network was functionally annotated using Metascape.

Topological properties of PPI network

Topological analysis facilitates the comprehension of a network's structure, hence facilitating the comprehension of its hidden mechanisms. Betweenness (CB) and closeness characterize the centrality assessment of the PPI network of differentially regulated proteins (DEGs) (Cc). Using Network Analyzer, measurements of centrality were determined [38]. These topological characteristics were utilized to track the topological alterations brought about by network disruption. To explore the network's fundamental characteristics, the following network properties were considered.

Degree

In PPI network analysis, the degree k indicates the total number of links established by a network node. This is also used to determine the local significance of a node in the process of network control. G = (N, E) represents the graph, where N represents the nodes and E represents the edges. Betweenness and proximity centrality are the fundamental centrality metrics and the parameters for estimating a node's global functional importance in a network regulation [39].

Betweenness centrality

Betweenness centrality quantifies a node's participation in the quickest communication between other pairs of nodes in the network. This nodal metric, c B (u), represents the number of shortest routes that pass through a node u. (Eq. 1).

$${c}_{B}\left(u\right)={\sum }_{v\in w\ne u}\frac{{\sigma }_{vw}\left(u\right)}{{\sigma }_{vw}}$$

(1)

Betweenness centrality is the degree to which a node shares the shortest-path traffic from all viable routes through other nodes. Consequently, it is the parameter of a node's ability to benefit from the flow of information throughout the network [40] and to exert control over the signal processing of other nodes in the network [41].

Closeness centrality

Closeness centrality is a measure of a node's proximity (via short pathways) to other nodes in a network. This nodal measure C_C (u) is the inverse of the sum of the distances between the node of interest u and every other node v $\in$ V\u (Eq. 2).

$${c}_{C}\left(u\right)=\frac{1}{{\sum }_{v\ne u}d\left(u,v\right)}$$

(2)

The closeness centrality of a network indicates how quickly information spreads from one node to another related node [38]. The function of a node is reliant on the centralities of its neighbors and changes across networks of highly connected nodes [42]. It is regarded as a significant measure of a node's information transmission capability inside a network since the likelihood of node isolation is reduced within a zone of tightly connected nodes.

Integration of single cell sequence data from COVID-19 patients

The expression analysis of DC-SIGN and associated proteins were validated through a database portal (http://covid19.cancer-pku.cn/). This database is repository of single cell transcriptome dataset of BALF and PBMC from mild and severely infected patients. We retrieved the gene expression data of different cell type from http://covid19.cancer-pku.cn/. We visualized the expression of DC-SIGN and associated proteins as violin plot and heat map by using ggplot2.

Results

DC-SIGN/R interacts with a unique area of spike protein's receptor binding domain (RBD)

Analysis of the spike proteins of SARS-CoV and SARS-CoV-2 revealed that their amino acid sequences were 98.89% similar [18] and that eight conserved and semi-conserved areas interacted [3,

Data availability

Data is with the authors and will be provided on request through corresponding author.