SARS-CoV-2 Prion-Like Domains in Spike Proteins Enable Higher Affinity to ACE2

Like other β-CoVs, the genome of the novel SARS-CoV-2 virus encodes structural proteins required for the efficient formation of infectious virions; these include the spike (S), envelope (E), membrane (M), and nucleocapsid (N) proteins (7).

The key determinant of the host specificity of a β-CoVs is the surface-located S protein, which plays critical roles in infection by mediating viral attachment to host cell surface receptors and facilitating viral entry (8). The S protein consists of two large regions: N-terminal S1 and C-terminal S2 (9). S1 is responsible for recognizing host-cell receptors, including the receptor-binding domain (RBD), and has higher sequence variability than S2 (S1 shares around 70% identity with that of other human β-CoVs). Moreover, the membrane-embedded S2 region responsible for fusion is more highly conserved than that of S1 (8,9) In SARS-CoV-2, the RBD in S1 allows the

virus to bind directly to the peptidase domain of the host angiotensin-converting enzyme 2 (ACE2)

complex, mediating virus entry into sensitive cells (10). Notably, compared to SARS-CoV, SARS- CoV-2 has a higher binding affinity to ACE2 (which is the common receptor for both SARS-CoV- 2 and SARS-CoV), with a broader interaction with ACE2 expressed not only in the lungs but also in kidney, testis, and heart (10,11).

Recently we have conducted an analysis and identified for the first time viral prion-like domains (PrDs), which we suggest are novel regulators of virion assembly with a role in virus-host cell interactions (12,13). These studies were in alignment with previous studies, showing that in addition to the pathological role of prions that they play in humans being implicated in Alzheimer’s and Parkinson’s diseases, diabetes, and many other human pathologies, protein misfolding plays important physiological roles in eukaryotes and prokaryotes (14-17).

Though the detailed molecular mechanisms underlying prion formation remain elusive, asparagine (Q)- and glutamine (N)-rich regions characterized by altered hydrophobicity and net sequence charge are known to drive prion formation. This is the basis for a number of algorithms for identifying candidate prionogenic domains (18,19). One such algorithm is prion-like amino acid composition (PLAAC) analysis, which allows the evaluation of prion-like domains based on the hidden Markov model (HMM) (20).

In this study, for the first time, we performed a detailed study of the prion-like domains in spike

protein of SARS-CoV-2 and a comparison of SARS-CoV-2 to other human-pathogenic β-CoVs. Our findings can contribute to a better understanding of the pathogenicity of SARS-CoV-2 and will help to uncover new targets for the development of drugs and vaccines based on the prionogenic properties of particular viral protein regions.

Results

Using the prion-prediction PLAAC algorithm, we analyzed structural proteins derived from UniProtKB and NCBI databases and identified PrDs in the S proteins of all β-CoVs analyzed in this study (Supplementary figure S1). The LLR scores of PrDs of the S proteins were practically identical within the studied β-CoVs, ranging from 4.431 to 4.991 (Supplementary Figure S2). Notably, with more precise mapping of PrDs within these proteins, we found a striking difference in their localization with SARS-CoV-2 being the only virus with PrDs identified within the RBD of the S protein (Table 1).

Considering that although SARS-CoV-2 and SARS-CoV (which are the closest related human β- CoVs pathogens) share the same host-cell receptor ACE2, SARS-CoV-2 binds tighter to it; therefore, we hypothesized that the presence of PrDs in the RBD of the SARS-CoV-2 might

explain this phenomenon (10). Consistent with this hypothesis, we found that SARS-CoV-2 along with other residue substitutions has five substituted amino acids in the RBD compared to SARS- CoV; the following are the substitutions in the RBD: S460→Q474, T488→N481, N480→Q493, Y485 → Q498 and T488 → N501, therefore forming a hydrophobic Q/N rich region that enables the prionogenity of the SARS-CoV-2 RBD (Figure 1).

We next analyzed the presence of prion-like domains in ACE2 protein and found PrDs within the α1 helix of ACE2 (aa 40-65 and 93-106) (Supplementary figure S2). Based on previous analysis by Yan et al. we modulated the interface between the SARS-CoV-2 RBD and ACE2 identified in this study, by aligning their sequences (Figure 2A, B) (21). Interestingly, we identified a pattern in which five of seven amino acids that interact between the SARS-CoV-2 RBD and host cell ACE2 are localized within the PrDs of SARS-CoV-2 RBD, ACE2 or both of them (Figure 2A). Thus, Q498 and T500 from the PrD of the SARS-CoV2 RBD interact with Y41 and Q42 within the PrD of ACE2; while Q474, F486 and N501 from the PrD of the SARS-CoV2 RBD bind to Q24, M82 and K343 of a non-PrD of ACE2. Notably, only K417 and Y453 were the only residues of the SARS-CoV-2 RBD that were outside the viral PrD and bound to a non-PrD of ACE2 (Figure 2B).

This study is the most complete evaluation of PrDs in the S protein of SARS-CoV-2. The results

highlight some previously unknown, unique characteristics of SARS-CoV-2 that may play

important roles in the pathogenesis and inform the development of new therapeutic strategies.

In this study, we used a high threshold of the PLAAC score for protein identification: only proteins with a high probability of prionogenic properties were included in the analysis. We found that all members of β-CoVs members contain PrDs in the S proteins. However, SARS-CoV-2 is the only member of β-CoVs that has a PrD in the RBD of the S protein that binds to the ACE2 receptor employed for host cell entry. Furthermore, we discovered specific amino acids (Q474, N481, Q493, Q498 and N501) that enable the prionogenity of the SARS-CoV-2 RBD that are not found in the RBD of SARS-CoV, of which Q474, Q498 and N501 directly contact within ACE2.

From these analyses, we conclude that the presence of these intrinsically disordered regions in the SARS-CoV-2 RBD, might be the reason for its optimized binding to the human ACE2 receptor in comparison to the RBD of SARS-CoV, since the distinguishing characteristic of PrDs is their ability to rapidly shift between multiple conformations due to residue hydrophobicity and net sequence charge (18, 22).

Notably, since five of seven amino-acid interactions that occur between the RBD of SARS-CoV- 2 and ACE2 are within their PrDs, it is also interesting to consider whether the prion-prion interaction between the virus and human receptor participates in COVID-19 and does it add a special value for the higher affinity to their binding. Since other β-CoVs were shown to lack the PrDs in the RBD, this means that the presence of PrDs is beneficial, but not necessary, for

receptor-mediated virion attachment to the host cell. One of the critical goals of our previous

studies was to show that PrDs identified in viruses may have important functional roles in virulence and are particularly associated with viral adhesion and entry.

This study provides a proof of this concept, showing that the presence of PrDs in the RBD of SARS-CoV-2 enhances viral binding to its host receptor compared to that of SARS-CoV, which lacks PrDs in its RBD structure. Further analyses of these PrD-containing proteins in SARS-CoV- 2 may improve our understanding of COVID-19 infection and provide new insights into its pathophysiology novel targets for developing therapies.

Materials and Methods

Protein Sequences

To identify the PrDs present in viral proteomes, protein sequences were obtained from the UniProt Knowledge Base and National Center for Biotechnology Information (NCBI) database (23, http://www.ncbi.nlm.nih.gov/). Protein functions were manually curated using information from UniProt and NCBI databases. The structure of the RBD-ACE2 complex was established based on the data from PDB ID: 6VW1 and visualized using the YASARA software

(http://www.yasara.com) (24, 25).

Identification of PrDs in viral proteomes

The presence of PrDs in β-CoV proteomes found using the PLAAC algorithm and the output probabilities for the PrDs were constructed based on amino-acid frequencies and similarities with PrDs in Saccharomyces cerevisiae. We used a cutoff of 3.0 log-likelihood ratio (LLR) and alpha = 1.0, representing S. cerevisiae background scanning, to identify the PrDs. Prion-like

domain amino acid positions were determined based on the PLAAC algorithm program analysis or manually.

Statistical analysis

All statistical analyses were conducted using the Statistical package for Windows (version 5.0) (StatSoft, Inc.). Data were compared between viruses using a χ2 test or Fisher’s exact test. To detect differences in multiple comparisons, one-way analysis of variance (ANOVA) was fitted with the standard confidence interval of 95%. P values < 0.05 were considered statistically significant.

Tables

Table 1. Comparison of the distribution of PrDs within the S protein among different β-CoV human pathogens

Figures

Figure 1. Analysis and comparison of mutations in the RBD of SARS-CoV-2 and SARS- CoV.

Supplementary figure S1. Graphical representation of the LLR score in the PrDs of the S protein from different β-CoVs.

The LLR value of the S protein from (A) SARS-CoV-2, (B) SARS-CoV, (C) MERS-CoV and (D) HCoV-OC43.

Supplementary figure S2. LLR score showing the predicted putative PrDs in S proteins of β-CoVs.

Heatmap of PrD distribution in S proteins in β-CoVs. Cells indexed by rows and columns are marked using a color gradient, ranging from white (LLR < 3.0) to saturated red (LLR = 5). The results were analyzed using one-way ANOVA.

Supplementary figure S3. Graphical representation of the LLR score of PrDs in the ACE2 protein.

The authors declare no competing interests.