Combining structural biology with AI-guided protein design to develop new antibody-degrading enzymes.

The development of new enzymes can be greatly accelerated by the use of new computational methods. Our approach is inspired by the naturally occurring enzymes secreted by pathogens that specifically degrade human antibodies in order to “hide” from the host immune response. Such specific antibody degradation can be useful in both biotechnological and therapeutic settings, and so there is considerable interest in repurposing and developing more of these immune evasion enzymes. A notable example is an enzyme (named IdeS) from the bacterium Streptococcus pyogenes, which degrades human IgG antibodies and has been repurposed for clinical use in removing unwanted IgG-mediated immune responses during kidney transplantation. Proteases that target other human antibody subclasses, such as IgM, IgA and IgE, have also been discovered, and provide further potential opportunity to reprogram the immune system. The specificity of such proteases allows for their use in disease treatment, whilst avoiding significant off-target protein degradation, and enables their use as laboratory reagents for antibody manipulation and analysis. Currently the clinical use of these pathogen-derived enzymes is hindered by the prevalence of immunity arising from previous pathogen infections, which can limit their application to a single dose. Engineering novel enzymes would provide a greater toolkit of specific antibody-degrading enzymes, which would enable their more widespread use. The revolution in deep learning computational methodology has unlocked a new paradigm for protein engineering. Technologies such as AlphaFold and Protein MPNN unlock the potential for enzymatic redesign to overcome the limitations of immunoglobulin degrading enzymes. Protein redesign can additionally improve properties of native enzymes, such as stability at varying temperature and pH, immunogenicity and antigenicity (i.e. how the protein is recognised by the immune system). We propose to utilise an interdisciplinary synthetic biology approach to redesign existing natural biological systems for uses across human health. We aim to employ generalisable computational methodologies to engineer novel proteases with specificities for each human antibody subclass, whilst also addressing some current drawbacks of existing immunoglobulin-degrading enzymes. This proposal will also facilitate the knowledge transfer of these pioneering computational technologies, developed at the Institute for Protein Design (IPD; Seattle, Washington), to enhance the UK's capability for computational protein design. We will firstly use established structural biology pipelines to generate high-resolution structural information about each of these antibody proteases, and subsequently use this information to inform computational enzyme redesign. We will develop novel variants of the IgG-degrading enzyme IdeS for multi-use therapies in organ transplantation, IgA-specific proteases for use in IgA nephropathy, IgE proteases for allergy desensitization and IgM proteases for potential B-cell lymphoma therapeutics. The primary goal of this proposal is the demonstration of the broad applicability of these methodologies, with the potential to develop novel biotherapeutics. The proposal will additionally provide tools and resources of potential application to broad communities in the biosciences.