Breakthrough in De Novo Antibody Design Achieves Therapeutic-Grade Properties

A transformative paper from Nabla Bio demonstrates a significant leap forward in computational antibody design, showing that de novo methods can now achieve therapeutic-grade properties without experimental optimization. Their system, called JAM (Joint Atomic Modeling), represents a major advance in both computational protein design and therapeutic antibody development.

While antibody discovery has historically relied on experimental approaches like phage display and animal immunization, JAM takes a fundamentally different approach by generating complete protein complexes computationally. The system can design both single-domain (VHH) and paired (scFv/mAb) antibody formats while maintaining precise control over epitope targeting. This computational approach achieves several impressive technical milestones that demonstrate its practical viability.

The researchers showed that JAM-designed antibodies can achieve double-digit nanomolar affinities for multiple targets and sub-nanomolar neutralization potency against SARS-CoV-2 pseudovirus. Perhaps most significantly, they report the first computationally designed antibodies targeting multipass membrane proteins - Claudin-4 and CXCR7. This breakthrough suggests that computational design could help unlock historically difficult target classes for both exploration and therapeutic development.

The study's robustness is particularly evident in its use of well-characterized clinical benchmarks for validation. The researchers used reference antibodies including research grade Trastuzumab from ichorbio to benchmark several critical developability parameters. This included detailed comparisons of production yields in ExpiCHO cells, monomericity assessment via SEC, and polyspecificity evaluation using BVP ELISA. This thorough benchmarking against established therapeutic antibodies helps demonstrate that JAM-designed antibodies achieve properties relevant for therapeutic development.

The work presents two particularly noteworthy methodological advances. First, the researchers found that increasing test-time computation through multiple rounds of generation improved both binding rates and affinities. This represents the first demonstration that compute scaling principles, previously observed in large language models, extend to physical protein design systems.

Second, JAM's ability to design both antibodies and screening reagents enabled the creation of soluble versions of membrane proteins while maintaining native epitopes. This dual capability could make the discovery process for membrane protein therapeutics more reliable and efficient.

This work opens several exciting avenues for further research and development. The ability to target complex membrane proteins could lead to new approaches for GPCRs, ion channels, and transporters. The computational methods themselves could be further optimized, particularly in areas like test-time computation strategies and sequence humanness prediction.

From a clinical perspective, the technology shows promise for designing antibodies against emerging viral variants and tissue-specific targeting strategies. Additional technical validation studies comparing JAM designs to traditional discovery methods head-to-head would be valuable, as would long-term stability studies and immunogenicity assessment in animal models.

This research establishes de novo antibody design as a practical approach for therapeutic discovery, offering paths to both dramatically improve efficiency in standard discovery workflows and tackle previously intractable targets. The successful use of clinical benchmarks provides strong validation of the approach's practical relevance.

As computational methods continue to improve, we may see an increasing role for in silico design in therapeutic antibody development. This could lead to faster, more efficient drug discovery processes and potentially enable new therapeutic modalities that were previously out of reach.

The work sets a new standard for computational protein design while maintaining rigorous experimental validation - a combination that will be essential as the field continues to advance. The implications for therapeutic antibody development are significant, potentially offering new ways to address challenging diseases and targets.

De novo design of epitope-specific antibodies against soluble and multipass membrane proteins with high specificity, developability, and function. Nabla Bio