Deep mutational scanning of recent SARS-CoV-2 variants highlights changing amino acid preferences within epistatic hotspot residues

This study utilizes deep mutational scanning of Omicron KP.3.1.1 and LP.8.1 RBDs to demonstrate that key evolutionary hotspots (residues 455, 456, and 493) exhibit shifting amino acid preferences due to ongoing epistatic reconfiguration, while also identifying mutations like H505W that may drive future viral evolution by stabilizing the closed spike conformation.

The Gist

Imagine the SARS-CoV-2 virus as a master thief trying to break into a house (our cells). The thief wears a special key (the Spike protein) to unlock the front door (the ACE2 receptor). However, the house has a security system (our immune system/antibodies) that tries to grab the thief's hand and stop them from turning the key.

For the virus to survive, it needs to keep its key working perfectly to open the door, but it also needs to change the shape of its hand so the security guards can't grab it. This is a delicate balancing act.

This paper is like a massive, high-tech "what-if" simulation lab run by scientists at the University of Utah. They wanted to see how the virus's key has changed in its latest versions (called KP.3.1.1 and LP.8.1) and, more importantly, how those changes affect the virus's ability to adapt in the future.

Here is the breakdown of their findings using simple analogies:

1. The "Giant Lego" Experiment

Instead of waiting for the virus to mutate naturally in the wild (which is slow), the scientists built a library of every possible single change the virus could make to its key.

• The Setup: They took the virus's key (the Receptor Binding Domain, or RBD) and created millions of versions of it. In some versions, they swapped one Lego brick for a different color; in others, they removed a brick entirely.
• The Test: They put all these mutant keys on the surface of yeast cells (tiny biological test tubes) and threw them into a pool of human door locks (ACE2 receptors).
• The Result: They watched which keys still fit the lock and which ones fell apart. This gave them a "heat map" showing exactly which changes help the virus and which ones break it.

2. The "Moving Target" Problem (Epistasis)

The most important discovery in this paper is about Epistasis. Think of this as a game of musical chairs where the rules change every time a new person sits down.

• The Analogy: Imagine you are trying to fix a car engine. In an old car (the original virus), swapping a specific part might make the engine run faster. But in a newer model (the Omicron variants), that same swap might cause the engine to explode.
• The Finding: The scientists found that the "rules" for the virus have changed. A mutation that was helpful in the past might now be harmful, and vice versa. The virus is constantly rewriting its own instruction manual based on what mutations it already has.

3. The "Hotspots" That Won't Settle Down

The study focused on three specific spots on the key (positions 455, 456, and 493).

• The Situation: These spots have been changing over and over again as new variants emerged.
• The Surprise: You might think the virus would eventually find the "perfect" shape for these spots and stop changing. But the data shows the opposite. These spots are still in a state of flux. The virus is still experimenting, trying out different shapes, and the "best" shape keeps changing depending on what other mutations are present. It's like a chef who keeps tasting the soup and adding different spices, never quite deciding on the final recipe.

4. The "Double Agent" Strategy

The virus has two ways to hide from the immune system:

1. Changing the Key's Shape: Making the key look different so antibodies don't recognize it.
2. Hiding the Key: Keeping the key folded away (the "down" position) so antibodies can't see it at all, only opening it when it's ready to attack the cell.

The scientists compared their "isolated key" tests with tests done on the whole virus. They found a sneaky mutation called H505W.

• The Trick: On its own, this mutation doesn't seem to change how well the key fits the lock. But when it's part of the whole virus, it acts like a locksmith's trick, forcing the key to stay folded away (hidden) more often. This helps the virus hide from antibodies while still being ready to strike when the time is right.

Why Does This Matter?

This research is like a weather forecast for the virus.

• Predicting the Future: Because the scientists know which "rules" are changing at the hotspots (455, 456, 493), they can better predict which new variants might emerge next.
• Better Vaccines: By understanding that the virus is still experimenting with these specific spots, vaccine designers can create shots that target these unstable areas, making it harder for the virus to escape.
• The Big Picture: The virus hasn't "finished" evolving. It is still actively reshaping its strategy to balance between breaking into our cells and hiding from our immune system.

In short: The virus is a master of disguise that keeps changing its costume. This paper gives us a detailed map of its current costume and predicts that it will keep trying on new outfits, especially at three specific spots, making it a moving target for our immune systems and vaccines.

Technical Summary

1. Problem Statement

SARS-CoV-2 evolution is driven by the dual pressures of evading host immunity (neutralizing antibodies) and maintaining infectivity (ACE2 receptor binding). A critical challenge in forecasting viral evolution is epistasis: the phenomenon where the fitness effect of a mutation depends on the genetic background (other existing mutations).

• As the virus evolves (e.g., from Omicron BA.2.86 to KP.3.1.1 and LP.8.1), the "rules" of mutational tolerance change. A mutation that was deleterious in an ancestral strain may become beneficial in a new variant, and vice versa.
• Existing Deep Mutational Scanning (DMS) datasets often rely on older variant backgrounds (e.g., Wuhan-Hu-1 or early Omicron), which may no longer accurately predict the evolutionary trajectories of current and future variants.
• There is a need to update DMS maps for recent dominant variants to understand how epistatic hotspots (sites like 455, 456, and 493) continue to reconfigure, and to distinguish between mutations that affect direct receptor affinity versus those that alter spike protein conformational dynamics (RBD "up" vs. "down" states).

2. Methodology

The authors utilized a yeast surface display Deep Mutational Scanning (DMS) platform to comprehensively map the mutational landscape of the Receptor-Binding Domain (RBD) for two recent Omicron sub-variants: KP.3.1.1 and LP.8.1.

• Library Construction:

  • Site-saturation mutagenesis libraries were created for the RBDs of KP.3.1.1 and LP.8.1.
  • Libraries included all 19 possible single amino acid substitutions and single-codon deletions at each of the ~200 RBD residues.
  • Variants were linked to unique 16-nucleotide barcodes (N16) to allow for multiplexed tracking.
  • Libraries were validated using long-read PacBio sequencing to ensure library integrity and even coverage.

• Phenotypic Assays (FACS-seq):

  • ACE2 Binding: Yeast displaying mutant RBDs were incubated with a concentration gradient of monomeric human ACE2. Fluorescence-Activated Cell Sorting (FACS) separated cells based on binding affinity.
  • RBD Expression: A parallel assay measured surface expression levels (folding stability) using an anti-Myc antibody tag.
  • Quantification: High-throughput sequencing of barcodes in sorted bins allowed the calculation of binding affinity ($K_D$) and expression levels for every single mutant.

• Comparative Analysis:

  • Epistasis Quantification: The authors calculated an "epistatic shift" metric (Jensen-Shannon divergence) to compare the mutational effect profiles of KP.3.1.1/LP.8.1 against their ancestor (BA.2.86) and earlier variants (Wuhan-Hu-1, Alpha, etc.).
  • Conformational Dynamics Disentanglement: The authors compared their isolated RBD DMS data (direct affinity only) with recently published pseudovirus whole-spike DMS data (which integrates both direct affinity and RBD up/down conformational dynamics). This allowed them to identify mutations that alter binding via conformational shifts rather than direct interface contact.

3. Key Contributions

1. Comprehensive DMS Maps for KP.3.1.1 and LP.8.1: The study provides the first high-resolution maps of mutational effects for these specific variants, covering virtually all single amino acid changes and deletions.
2. Identification of Ongoing Epistatic Hotspots: The paper identifies residues 455, 456, and 493 as persistent sites of evolutionary flux. Unlike other sites that have "entrenched" (stabilized), these hotspots continue to undergo sign-epistasis (where the effect of a mutation flips from beneficial to deleterious or vice versa) as the viral background changes.
3. Mechanistic Disentanglement: By comparing isolated RBD data with whole-spike data, the study successfully separates direct binding affinity effects from indirect conformational effects.
4. Discovery of H505W: The study highlights the H505W mutation as a specific outlier that reduces ACE2 binding in the whole-spike context (likely by stabilizing the "down"/closed conformation) despite having minimal effect on direct RBD:ACE2 affinity in isolation.

4. Key Results

• Epistatic Shifts at Hotspots (455, 456, 493):
  • These sites show significant divergence in amino acid preferences between BA.2.86 and KP.3.1.1/LP.8.1.
  • Example: The Q493E mutation is deleterious in BA.2.86 but becomes accessible/beneficial in KP.3.1.1/LP.8.1 due to the presence of L455S and F456L. Conversely, mutations like V493 are beneficial in BA.2.86 but deleterious in the newer variants.
  • This indicates that these sites have not converged on a fixed solution; rather, they remain in a state of "epistatic reconfiguration," allowing the virus to explore new evolutionary paths.
• Conformational Dynamics (The H505W Case):
  • Most mutations near the ACE2 interface affect binding primarily through direct affinity.
  • However, H505W showed a strong reduction in ACE2 binding in whole-spike assays but no significant change in isolated RBD assays. This suggests H505W biases the spike toward the RBD-down (closed) conformation, a strategy to hide epitopes from antibodies while potentially sacrificing some receptor engagement efficiency.
  • Mutations at F374 were also identified as potentially biasing the RBD toward the "up" conformation.
• Stability vs. Plasticity:
  • Sites like 498 and 501 (the R498R/Y501Y synergistic pair) showed strong epistatic shifts early in Omicron evolution but have since stabilized ("entrenched").
  • In contrast, sites 455, 456, and 493 continue to exhibit fluctuating preferences, making future evolution at these positions harder to predict but highly relevant for antibody escape (e.g., escape from monoclonal antibodies SA55 and VIR-7229).
• Constraint Patterns:
  • Disulfide bridge residues (e.g., C480, C488) and sterically constrained ACE2 contact residues (e.g., G502) remain highly constrained.
  • Residues 475, 478, and 487 show high tolerance for mutation, supporting the emergence of variants like XFG (N487D) and NB.1.8.1 (K478I).

5. Significance

• Viral Surveillance and Forecasting: The study demonstrates that relying on historical DMS data (from older variants) is insufficient for predicting the evolution of current SARS-CoV-2 strains. Continuous updating of mutational maps is essential to identify which mutations are now "accessible" and likely to emerge.
• Therapeutic Design: Understanding that sites 455, 456, and 493 are in a state of flux helps in designing monoclonal antibodies and vaccines that target more conserved regions or account for these specific epistatic dependencies.
• Evolutionary Mechanisms: The paper provides a concrete example of how epistasis drives viral evolution, showing that the virus is not just accumulating mutations but actively reconfiguring its fitness landscape. The identification of H505W as a conformational switch highlights a sophisticated viral strategy: trading receptor binding efficiency for immune evasion by modulating spike dynamics.
• Methodological Integration: The work validates the utility of combining isolated RBD DMS with whole-spike pseudovirus DMS to deconvolve the complex interplay between direct binding affinity and conformational dynamics.

In summary, this paper provides a critical, up-to-date resource for understanding the evolving constraints of SARS-CoV-2, revealing that the virus is still actively exploring new mutational combinations at key epistatic hotspots, with significant implications for future pandemic preparedness.