TL;DR
This paper introduces a novel structural profiling method to analyze antibody repertoires, revealing a surprising degree of shared structural "blueprints" (Public Baseline structures) across individuals, challenging previous sequence-based assumptions of repertoire diversity. This structural commonality extends to immune responses, identifying "Public Response" structures after vaccination, and suggests a powerful geometric basis for designing therapeutic antibodies with reduced immunogenicity.
What Problem Does This Paper Address?
The adaptive immune system's ability to recognize a vast array of pathogens is attributed to the immense diversity of antibody repertoires. However, this diversity has historically made it difficult to reconcile with the observation that individuals can mount similar immune responses to the same pathogens. Traditional methods like "clonotyping," which rely on sequence similarity (especially in the CDRH3 region), have only identified a tiny fraction of shared antibody clonotypes, suggesting that functional commonality might be much lower than expected. This disconnect leaves a gap in understanding how the immune system achieves broad yet coordinated responses, and how to effectively leverage this commonality for therapeutic antibody development.
Methods
The core of this work is an updated Repertoire Structural Profiling (RSP) pipeline. This method moves beyond sequence identity to analyze the three-dimensional structure of antibody binding sites. Key steps include:
- CDR Modellability Analysis: Assessing which complementarity-determining regions (CDRs) of antibody sequences can be reliably modeled.
- Sequence Clustering: Grouping similar VH and VL chains to reduce computational complexity.
- VH-VL Pairing and Orientation Prediction: Identifying likely pairings of heavy and light chains and predicting their relative orientation based on key interface residues.
- Structural Clustering: Grouping predicted Fv (variable fragment) structures based on their geometric similarity, specifically focusing on CDR loop structures and their orientations. This is the novel aspect, moving away from strict sequence identity.
- Antibody Model Library (AML) Construction: Building libraries of 3D antibody models representing these structurally defined clusters.
The authors also generated 'Random Repertoires' to statistically assess whether the observed structural sharing was greater than expected by chance.
Datasets
The study utilized two primary datasets:
- Baseline Repertoires: Ig-seq data from 41 unrelated individuals (Gidoni et al. [31]), focusing on those with sufficient sequencing depth.
- Vaccine Response Repertoires: Longitudinal Ig-seq data from three unrelated individuals before and after flu vaccination (Gupta et al. [5]).
These datasets were downloaded from the Observed Antibody Space (OAS) database [9]. The authors did not introduce a new benchmark dataset but rather applied their novel structural profiling method to existing, publicly available data.
Key Insights and Ideas
- Structural Commonality Exists: Contrary to sequence-based clonotyping, structural analysis reveals a significant degree of shared antibody binding site geometries ("Public Baseline" structures) across unrelated individuals, suggesting a higher functional commonality than previously appreciated.
- Beyond Sequence Identity: The paper strongly advocates for structural similarity as a more accurate proxy for functional equivalence than sequence identity, especially for CDRH3, which can be highly variable yet still achieve similar binding site shapes.
- Convergent Immune Responses: The method can identify "Public Response" structures that emerge in multiple individuals after vaccination, indicating a shared structural adaptation to common antigens.
- Therapeutic Potential: "Public Baseline" structures are hypothesized to be less immunogenic and more versatile, making them excellent starting points for designing therapeutic antibodies. The AMLs built from these structures show geometric similarity to existing clinical-stage therapeutics.
Key Results
- Public Baseline Structures: Approximately 3% of distinct antibody structures were found to be common across the ten most diverse individual baseline repertoires. This is over 100 times higher than the proportion of public clonotypes found by previous sequence-based studies.
- Focused Structural Space: Antibody repertoires explore a highly focused region of structural space (~10% of what would be expected by random sampling), indicating a degree of selection or constraint.
- Vaccine Response: After flu vaccination, structural sharing increased, with 17.78% of structures being "Public Before Vaccination" and 19.23% being "Public After Vaccination".
- Public Response Structures: A subset of 6.18% of structures were identified as "Public Response" structures, emerging specifically after vaccination and distinct from baseline structures.
- Therapeutic Similarity: Antibody models derived from "Public Baseline" structures showed high geometric similarity (within 1.00Å Fv RMSD) to 37 out of 66 clinical-stage therapeutics.
Limitations and Future Work
- Modellability Bias: The analysis is limited to antibody chains that are predicted to be modellable. Regions of natural structural space that are currently difficult to model remain uninvestigated.
- CDR Length Variation: The current implementation may not fully capture antibodies that use loops of different lengths or alternative binding modes to achieve similar binding.
- Combinatorial Pairing: The study relies on pairing VH and VL sequences, which are often analyzed separately. While this is a common approach, it's an artificial pairing. The advent of single-cell sequencing that preserves VH-VL pairing could address this.
- Future Work: The authors suggest that their approach can be extended to design in silico/in vitro screening libraries and assist in antibody functional annotation. They also plan to continue building and sharing AMLs. Integrating their structural profiling with methods that predict paratope similarity across all six CDRs could further enhance the identification of functionally equivalent antibodies. The "Public Baseline" and "Public Response" AMLs are made available for further research.
Summary updated