The team’s approach rests on a protein side chain-based network method, a technique in which each amino acid residue is represented as a node, and edges are drawn between residues whose side chains come into close non-covalent contact. Such contacts include the hydrogen bonds, hydrophobic interactions, and electrostatic attractions that hold proteins together without forming chemical bonds. Rather than analyzing a single static structure, the researchers averaged their network construction over many snapshots drawn from molecular dynamics simulations produced by D. E. Shaw Research, an unusually long and detailed set of simulations that capture the complex in motion at atomic resolution. Averaging across these frames filters out fleeting, transient contacts and reveals the persistent connectivity patterns that define the complex’s functional architecture.

Once the network is built, the analysis proceeds at multiple hierarchical levels. At the most local scale, the researchers examined cliques—fully connected subgroups of residues in which every member interacts with every other member. Cliques at the viral–human interface act as tightly knit hubs of interaction, and their abundance and composition reveal how densely the two proteins interlock. At the global scale, the team analyzed clusters of connected residues and performed percolation analysis, a concept borrowed from statistical physics that asks how a network holds together as connections are progressively removed. In a percolation framework, the strongest substructures are identified by stripping away weaker links and observing which clusters survive.

The central finding of the study is unambiguous: the SARS-CoV-2 spike protein forms a more dominant and robust connection with the human ACE2 receptor than SARS-CoV-1 does. At the global level, this superiority shows up clearly in the percolation results. When the researchers restricted the networks to only the strongest interfacial bonds, the interface cluster of the SARS-CoV-2–ACE2 complex persisted, remaining a single connected unit, whereas the corresponding cluster in the SARS-CoV-1–ACE2 complex fragmented. In other words, the stronger non-covalent interactions between the newer virus and its receptor form a self-reinforcing web that cannot be dismantled by discarding weak contacts—a hallmark of structural robustness.

The atomic-scale analysis tells a complementary story. In the SARS-CoV-2 complex, network cliques at the interface form in a manner that reflects denser and better-organized inter-protein communication. Cliques are demanding structures: because every residue in a clique must contact every other, their emergence signals a level of mutual reinforcement that looser collections of contacts lack. The fact that these clique structures at the SARS-CoV-2 interface consistently encode stronger connectivity provides a mechanistic, residue-level explanation for the virus’s enhanced receptor engagement, one that emerges naturally from the geometry of interactions rather than from any single pinned atom or bond.

Beyond confirming what virologists had suspected from binding-affinity measurements, the network framework delivers something more practical: an objective method to map how atomic-level non-covalent interactions give rise to global connectivity between any two amino acids in the complex. Bridging these scales has long been a challenge in molecular biophysics. A single hydrogen bond at an interface may seem inconsequential in isolation, yet through the network it can contribute to pathways of communication that span the entire protein assembly. The side chain-network approach makes this propagation explicit, allowing researchers to trace exactly which local interactions feed into which large-scale structural couplings, and to rank residues by their contribution to overall interface integrity.

Using this ranking ability, the team pinpointed key functional residues in SARS-CoV-2 that play important roles in establishing the higher connectivity with ACE2. These residues, identified through their network centrality and their participation in robust interfacial clusters and cliques, constitute the structural backbone of the viral grip on human cells. The authors also examined specific individual snapshots from the molecular dynamics trajectories rather than only time-averaged networks, and this analysis highlighted prominent variations in network topology across different conformations. The spike–ACE2 complex does not sit in a single rigid state; it explores a diverse conformational landscape, and the connectivity architecture shifts meaningfully from one moment to another. Capturing these fluctuations matters because a receptor-binding interface is a dynamic entity, and transient configurations may expose vulnerabilities or reveal alternative interaction modes.

Perhaps the most striking connection to real-world virology emerges when the network map is overlaid with the evolutionary record of the pandemic. The researchers found that most mutations occurring in the SARS-CoV-2 spike protein in variants of concern and variants of interest—including the currently active JN.1 variant and its subvariants—have been observed at the interface with the ACE2 receptor. This is precisely the region that the network analysis identifies as the seat of the virus’s enhanced connectivity. The convergence is telling: as the virus evolves to evade immunity generated by prior infection or vaccination, natural selection keeps returning to the same interfacial neighborhood that network science flags as functionally dominant. Mutations there can simultaneously alter antibody recognition and tune receptor binding, which explains why this region is both the engine of immune escape and the object of intense surveillance by public health laboratories worldwide.

This convergence between evolutionary observation and network topology gives the study practical significance for therapeutic design. If a defined set of residues sustains the robust connectivity between the viral spike and ACE2, then molecules that mimic the relevant receptor surface could, in principle, compete with ACE2 for spike binding and neutralize the virus. The authors note that their analyses highlight the importance of interface interactions and provide a rationale for designing receptor-like peptides and proteins to combat immunity-escaping variants. Because such decoys target the physical grip on ACE2 rather than the antibody-recognized epitopes that mutate most rapidly under immune pressure, they may retain effectiveness even as the virus drifts genetically—a property that conventional antibody drugs increasingly struggle to maintain against JN.1-descended lineages.

More broadly, the work demonstrates the value of complex-systems thinking in molecular biology. A virus–receptor complex is not merely a collection of atoms or a static ribbon diagram; it is a network whose resilience, communication pathways, and failure points determine biological function. By providing an objective, quantitative bridge from atomic non-covalent interactions to global connectivity between any two amino acids, the framework developed by Subramanyan and colleagues can in principle be applied to any protein complex, well beyond coronaviruses. For SARS-CoV-2 specifically, it delivers a durable conceptual map of why this virus binds human cells so effectively, why evolution keeps redrawing the same interfacial territory, and where next-generation antiviral designs might most productively aim as the virus continues to diversify.

Subject of Research: Network analysis of the SARS-CoV-2 spike–human ACE2 receptor complex connectivity

Article Title: From atomic to global connectivity in the structure of the SARS-CoV2-human ACE2 receptor complex

Article References: Subramanyan, V., Anto, A., Bhattacharyya, M., Vishveshwara, S., & Vishveshwara, S. (2026). From atomic to global connectivity in the structure of the SARS-CoV2-human ACE2 receptor complex. PLOS Complex Systems, 3(4), e0000096. https://doi.org/10.1371/journal.pcsy.0000096

Image Credits: AI Generated

DOI: 10.1371/journal.pcsy.0000096

Keywords: SARS-CoV-2, ACE2 receptor, spike protein, SARS-CoV-1, molecular dynamics, protein structure networks, percolation analysis, network cliques, variants of concern, JN.1 variant, immune escape, peptide design

The stakes of this kind of work are hard to overstate. In the past two decades, three zoonotic coronaviruses—SARS-CoV-1, MERS-CoV, and SARS-CoV-2—have caused major outbreaks, and each emergence began with a virus crossing a species barrier. For sarbecoviruses, that barrier is defined largely by compatibility between the viral receptor-binding domain (RBD) on the spike protein and the host’s angiotensin-converting enzyme 2 (ACE2) receptor. If a virus’s RBD can engage a species’ ACE2 efficiently, that species becomes a plausible host; if it cannot, the chain of transmission breaks at the very first step. Understanding which receptors are vulnerable to which viruses is therefore central to anticipating where the next spillover might come from.

To build the atlas, the researchers first established an evolutionary framework. Phylogenetic analysis of the RBD sequences divided the 46 viruses—31 from bats, 10 from humans, three from civets, and two from pangolins—into three major clades, with clade 1 further splitting into subclades 1a and 1b. Clade 1a contains SARS-CoV-1 and related bat and civet strains; clade 1b contains SARS-CoV-2 and its closest relatives, including the bat virus RaTG13 and pangolin coronaviruses. Clade 2, which comprises most known sarbecoviruses, carries a distinctive double-segment deletion in the receptor-binding motif, while clade 3 viruses from African and European bats carry a single-segment deletion in the same region. These structural differences in the RBM, the loop that physically contacts ACE2, turned out to predict much of the functional variation the team later observed.

The functional screen itself relied on pseudotyped reporter viruses: vesicular stomatitis virus particles decorated with each sarbecovirus spike protein, whose entry into cells expressing a given ACE2 ortholog could be quantified by luminescence. The team confirmed comparable spike expression and incorporation across pseudovirus preparations by western blot and normalized viral input by quantitative RT-PCR, allowing a fair comparison across the roughly 3,000 possible virus-receptor pairings. The result is a heatmap of cross-species infectivity that functions as a kind of vulnerability index for each species’ receptor.

Clade 1 viruses emerged as the clear standouts in receptor breadth. Within clade 1b, the pangolin viruses PCoV-GD and PCoV-GX, which share 89.6 and 92.3 percent spike identity with SARS-CoV-2 respectively, showed broad ACE2 compatibility, with PCoV-GD efficiently infecting nearly all tested non-bat mammalian ACE2s and most bat ACE2s. Intriguingly, RaTG13—despite sharing 97.4 percent identity with SARS-CoV-2 across the full spike protein—displayed a more restricted host range, implying that subtle differences within the receptor-binding motif underlie the expanded tropism of SARS-CoV-2. Within clade 1a, human SARS-CoV-1 isolates used ACE2 orthologs from a wide range of bat families, whereas most bat- and civet-derived relatives showed narrower usage, with the bat virus WIV1 bridging human and bat lineages through its ability to engage ACE2 from multiple Rhinolophus and Myotis species.

Clade 2 and clade 3 viruses told very different stories. None of the clade 2 pseudoviruses, which represent the majority of known sarbecoviruses, mediated entry through any of the 66 ACE2 orthologs, consistent with their double RBM deletion and previously reported ACE2-independent entry mechanisms. Clade 3 viruses showed intermediate phenotypes: BtKY72 used ACE2 from only a few Rhinolophus species and did not engage human ACE2, while Khosta-2 showed broader tropism, infecting cells expressing rabbit ACE2 more than ten-fold more efficiently than those expressing the human receptor—a striking example of lineage-specific adaptation.

The study also delivered a decisive verdict on several proposed alternative receptors. Host factors including NRP1, CD147, AXL, and TMEM106B have each been suggested to facilitate SARS-CoV-2 entry in various experimental contexts. Yet across the entire 46-virus panel, none of these factors supported detectable infection when overexpressed in the assay cells, indicating that none functions as an independent entry receptor for any sarbecovirus tested. The authors are careful to note that context-dependent accessory roles in specific SARS-CoV-2 settings cannot be excluded, but the message is clear: ACE2 remains the only demonstrated primary entry receptor across this viral diversity.

Among the most consequential findings is the identification of two bat species whose receptors behave remarkably like the human version. Only the ACE2 orthologs of Rhinolophus affinis and Miniopterus schreibersii correlated strongly with human ACE2 usage across the viral panel, with correlation coefficients of 0.8314 and 0.8777 respectively. Surface plasmon resonance confirmed that M. schreibersii ACE2 binds the SARS-CoV-1 and SARS-CoV-2 RBDs with high affinity—dissociation constants of 8.14 × 10⁻⁸ M and 8.06 × 10⁻⁷ M, lower than human ACE2’s nanomolar binding but fully consistent with efficient entry. Because both species range across Southeast Asia, Europe, and North Africa in regions overlapping human habitats, the authors argue they should be priorities for ongoing surveillance. The data also flagged rabbits, cattle, sheep, raccoon dogs, badgers, Siberian chipmunks, and pigs as mammalian species whose ACE2 usage correlates strongly with human ACE2, marking them as candidate intermediate hosts worth investigating.

Perhaps the most mechanistically revealing part of the study concerns two closely related Miniopterus bats whose receptors sit at opposite ends of the permissiveness spectrum. M. schreibersii ACE2 supported entry by nearly all ACE2-dependent pseudoviruses, while the ACE2 of M. natalensis was entirely non-permissive—despite the two proteins differing by only seven amino acids. Three of those substitutions, at positions 27, 31, and 42, sit directly at the predicted RBD-binding interface. Molecular dynamics simulations using the molecular mechanics generalized Born surface area method showed that mutating M. schreibersii ACE2 residue I27 to lysine raised the binding free energy by roughly 5 kcal/mol, while the K42E substitution raised it by nearly 7 kcal/mol, disrupting a salt-bridge network involving residue D38 and the SARS-CoV-1 RBD residue Y442. Entry assays confirmed the simulations: the I27K mutation selectively impaired civet-derived SARS-CoV-1 entry, N31G had minimal impact, and K42E broadly abolished usage by nearly all pseudotypes. Reciprocally, introducing the corresponding M. schreibersii residues into M. natalensis ACE2 restored entry competence for multiple SARS-CoV-1 pseudotypes. Residues 27 and 42, in short, are the principal molecular switches governing Miniopterus receptor recognition.

The authors are candid about the limits of their approach. Pseudoviruses built from available sequences may not capture the full diversity of natural sarbecoviruses, and receptor compatibility is only one barrier to cross-species transmission; protease availability, replication competence, and immune evasion all matter for productive infection in a living animal. Efficient ACE2 engagement alone does not guarantee disease. Still, the atlas provides exactly the kind of mechanistic framework that field surveillance has lacked: a way to prioritize which of the world’s more than 1,400 bat species, and which of the hundreds of circulating sarbecovirus sequences, warrant the closest monitoring. As the authors put it, identifying species with ACE2 structures resembling the human receptor may help focus receptor-based risk assessment where it matters most—before, rather than after, the next spillover.

Subject of Research: Cross-species ACE2 receptor usage and host-range determinants of sarbecoviruses

Article Title: An atlas of sarbecovirus infectivity across diverse ACE2 orthologs

Article References: Sun, Y., Cheng, Z., Wu, X., Liu, K., Wang, L., Yang, D., Huang, W., & Nie, J. (2026). An atlas of sarbecovirus infectivity across diverse ACE2 orthologs. iScience, 29(10), Article 117769. https://doi.org/10.1016/j.isci.2026.117769

Image Credits: AI Generated

DOI: 10.1016/j.isci.2026.117769

Keywords: sarbecovirus, ACE2, SARS-CoV-2, SARS-CoV-1, bats, spillover, pseudovirus, receptor-binding domain, host range, surveillance, Miniopterus, Rhinolophus