🧬 genetics

Combinatorial mosaics dissect the residue level basis of ACE2 binding divergence between SARS-CoV-2 and RaTG13

By constructing a combinatorial library of RBD mosaics between SARS-CoV-2 and RaTG13, this study identifies eight key positions driving ACE2 binding divergence and reveals that the SARS-CoV-2 residues Q493 and Q498, which individually reduce binding affinity, are functionally compensated by specific residues at positions 483, 484, and 494 through sequence context dependence.

Original authors: Giovanetti, S. M., Sadhu, M. J.

Published 2026-10-20

📖 4 min read☕ Coffee break read

CC0 1.0

Original authors: Giovanetti, S. M., Sadhu, M. J.

Original paper dedicated to the public domain under CC0 1.0 (https://creativecommons.org/publicdomain/zero/1.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Viruses are masters of adaptation, constantly reshaping their outer shells to slip past our immune systems or to latch onto new hosts. At the heart of this ability lies a specific interaction: a protein on the virus's surface must fit perfectly into a receptor on a human cell, much like a key finding the right lock. For the virus that causes COVID-19, this key is a part of its spike protein called the receptor-binding domain, and the lock is a human protein called ACE2. The virus's ability to bind to this receptor determines how easily it can infect us. Scientists have long known that the virus responsible for the pandemic, SARS-CoV-2, binds to human cells with remarkable strength. However, a closely related virus found in bats, known as RaTG13, binds to the same human receptor very poorly. The two viruses are nearly identical in their genetic code, yet they differ significantly in the specific region responsible for this binding. Understanding exactly which tiny changes in their genetic code cause such a dramatic difference in behavior is crucial for predicting how viruses might evolve in the future.

To solve this puzzle, researchers Simone Giovanetti and Meru Sadhu decided to look at the problem not by studying the viruses as they exist in nature, but by building a vast library of artificial hybrids. They focused on eighteen specific spots where the SARS-CoV-2 and RaTG13 viruses differ. In nature, these differences travel together in fixed blocks, making it hard to tell which single change is responsible for the strong binding of SARS-CoV-2 and the weak binding of RaTG13. The scientists created thousands of unique mosaic versions of the virus's binding region. Each mosaic was a patchwork, carrying a random mix of the eighteen differences, with some spots taken from the human-infecting virus and others from the bat virus. They then displayed these mosaics on the surface of yeast cells and measured how strongly each one grabbed onto the human ACE2 receptor.

The results revealed that the difference in binding strength was not spread evenly across the entire region. Instead, the power to bind tightly was concentrated in just eight specific locations. Most of these spots were right on the surface where the virus touches the human cell. When the researchers swapped the bat virus's amino acids at these spots for the human virus's version, the binding generally got stronger. However, the story was not simple. Two specific spots, numbered 493 and 498, behaved in a surprising way. In the bat virus, these spots held a rigid, ring-shaped molecule that helped it bind. In the human virus, these spots held a flexible, polar molecule that, on its own, actually made binding weaker. If the human virus had simply swapped these two spots for the bat version, it would have bound even better. But it didn't.

The reason for this counterintuitive result lies in the company these spots keep. The researchers found that the effect of the molecules at spots 493 and 498 depended entirely on the neighbors surrounding them. In the bat virus, the surrounding environment was set up to support the rigid ring-shaped molecules. In the human virus, the surrounding environment had changed at three other spots. These changes created a new chemical landscape that made the flexible molecules at 493 and 498 work much better than they would have otherwise. It was as if the human virus had built a custom support structure around these two spots, allowing them to function effectively despite being individually weaker. Without this specific background, the human virus's version of these spots would have been a liability.

This study demonstrates that you cannot understand a virus's behavior by looking at its parts in isolation. The researchers showed that the impact of a single genetic change can flip from helpful to harmful depending on the other changes present in the same protein. By breaking apart the natural combinations of mutations and testing them in every possible mix, they were able to map out exactly which differences matter most and how they interact. They found that while a small number of spots drive the major differences in binding, the full picture requires understanding how those spots talk to their neighbors. This approach provides a clear, residue-level account of why SARS-CoV-2 binds so strongly to humans while its close relative does not, highlighting that the path to a new host is paved not just by individual mutations, but by the complex context in which they appear.

Problem
SARS-CoV-2 and the closely related bat coronavirus RaTG13 exhibit a significant divergence in their ability to bind human ACE2, with SARS-CoV-2 binding approximately 70-fold more strongly. While the two viruses share 96% amino acid identity across their genomes, their Receptor Binding Domains (RBDs) are only 89% identical. Previous studies have identified specific residues that influence this binding, but a comprehensive understanding of how these naturally occurring differences interact across diverse sequence backgrounds remains elusive. Traditional approaches, such as single-site mutagenesis in a single background or chimeric swaps of contiguous regions, fail to distinguish individual residue contributions from linked sequence effects or to fully capture how the impact of a specific mutation depends on the surrounding genetic context (epistasis).

Methodology
The authors constructed a high-throughput combinatorial library of "mosaic" RBDs to dissect the residue-level basis of ACE2 binding divergence.

Library Design: They focused on 18 amino acid positions that differ between SARS-CoV-2 and RaTG13, including 17 within the Receptor Binding Motif (RBM) and one (residue 403) immediately outside it. They generated a library of mosaic constructs where each position independently carried either the SARS-CoV-2 or RaTG13 residue, theoretically creating $2^{18}$ (262,144) combinations.
Synthesis: Using oligonucleotide synthesis with IUPAC degenerate bases, they encoded 32 oligonucleotides to cover all combinations of the 17 RBM sites. Residue 403 was varied by cloning these pools into parallel plasmid backbones.
Screening: The library was expressed on the surface of Saccharomyces cerevisiae. Yeast cells displaying the mosaic RBDs were incubated with 100 nM human ACE2 and sorted via Fluorescence-Activated Cell Sorting (FACS) into bins based on binding intensity.
Sequencing and Analysis: Deep sequencing (Illumina) of the sorted bins allowed the authors to calculate a "mean bin score" for each genotype, serving as a proxy for binding affinity. They analyzed the data using linear models to estimate the main effects of each position and to detect sequence context dependence (epistasis) by examining how the effect of specific residues changed across different combinations of other residues.

Key Results

Phenotypic Diversity: The combinatorial mosaics revealed a broad spectrum of ACE2 binding phenotypes. While 59% of mosaics showed intermediate binding, 26% exceeded the binding affinity of wild-type SARS-CoV-2, and 14% were weaker than RaTG13.
Concentration of Effects: The variation in ACE2 binding was concentrated at eight specific positions: 403, 439, 449, 486, 493, 498, 501, and 505. These eight positions explained nearly all (49%) of the variance in binding observed across the library.
- SARS-CoV-2 residues at positions 403, 449, 486, 501, and 505 generally enhanced binding relative to RaTG13.
- Conversely, SARS-CoV-2 residues at positions 439, 493, and 498 were associated with reduced binding on average compared to their RaTG13 counterparts.
Sequence Context Dependence: The study identified significant context dependence, particularly for residues 493 and 498. Although the SARS-CoV-2 residues (Q493 and Q498) generally reduced binding, their deleterious effects were attenuated in the presence of specific SARS-CoV-2 residues at positions 484 (E484) and 494 (S494). In a RaTG13-like background (T484/R494), Q493 and Q498 severely impaired binding, whereas in a SARS-CoV-2-like background (E484/S494), the negative impact was mild. This suggests that the two RBDs favor different local configurations of the receptor-binding interface.
Sampling Efficiency: Downsampling analysis demonstrated that exhaustive sampling of the combinatorial space is not required to identify major determinants. Libraries as small as 300–400 genotypes were sufficient to recover the correct direction of effect for the top eight positions, though detecting context-dependent interactions required larger libraries (approx. 3,000–10,000 genotypes).

Significance
The paper demonstrates that combinatorial mosaic libraries provide a robust approach for resolving the contributions of sequence differences between homologous proteins at the residue level. By breaking the linkage between naturally occurring differences, this method allows researchers to:

Identify which specific residues drive functional divergence (in this case, eight key positions).
Reveal how the functional impact of a residue depends on the sequence context, a factor often missed by single-site mutagenesis or traditional chimeras.
Achieve these insights without requiring exhaustive sampling of the entire genotype space, making the approach scalable to proteins with more differences.

The work provides a detailed residue-level account of why SARS-CoV-2 binds human ACE2 more strongly than RaTG13, highlighting that the divergence is driven by a small set of residues whose effects are modulated by the surrounding sequence environment.

More like this