IDERT: Intelligent Deimmunization for Enzyme Replacement Therapies
Lead Research Organisation:
University of Edinburgh
Department Name: Sch of Biological Sciences
Abstract
Lysosomal storage diseases (LSDs) are rare inherited diseases caused by the deficiency of lysosomal enzymes, which cause the accumulation of substrates in the lysosome and ultimately lead to organ damage and premature death.
Since LSDs are caused by inherited mutations, they cannot be cured but can be treated using Enzyme Replacement Therapies (ERTs), which consist in restoring physiological enzymatic levels through the intravenous infusion of a recombinant version of the defective enzyme. Response to ERT is variable, because recombinant enzymes are less active than the human wildtype, are unstable in blood and are not properly absorbed by the cells. Moreover, repeated infusions usually trigger an immune response and the formation of anti-drug antibodies (ADAs), which limit the efficacy of ERTs and preclude long-term treatment for many patients.
Fabry Disease (FD) is the most common LSD. It is characterised by X-linked mutations in the gene encoding the alpha galactosidase (AGAL) enzyme, which cause irreversible damage to heart, vasculature, and kidney. Immune response is common, especially in males, who usually experience the most severe form since they have no enzyme activity. With the increasing number of FD patients due to better diagnosis, there is the pressing clinical need to improve ERT efficacy and reduce their immunogenicity.
Since the formation of anti-drug antibodies is initiated by T-cell recognition of peptides of antigen presenting cells, we propose to lower the immunogenicity of a recombinant enzyme by "recoding" known epitopes in the enzyme sequence, while preserving its catalytic function.
Here we will use Artificial Intelligence (AI) to design enzymes with a desired function but lacking known epitopes, a process called deimmunization, to deliver better therapies for FD patients. We will manufacture these enzymes (up to 200) using a mammalian cell line expression system, to ensure our ERTs have human-like biochemical properties, and then we will measure their catalytic activity in vitro and the immune reaction using patients' sera. Importantly, since AI has the potential to impact the life of millions of patients and their families, we will engage with patients through open meetings to show and explain the potential of this new technology in addressing an unmet clinical need.
Our project builds on the team's unique expertise in AI, biologics production and medicine and represents one of the largest ERT deimmunization study to date, with the potential to provide effective treatment options for FD and other LSD patients in the future.
Since LSDs are caused by inherited mutations, they cannot be cured but can be treated using Enzyme Replacement Therapies (ERTs), which consist in restoring physiological enzymatic levels through the intravenous infusion of a recombinant version of the defective enzyme. Response to ERT is variable, because recombinant enzymes are less active than the human wildtype, are unstable in blood and are not properly absorbed by the cells. Moreover, repeated infusions usually trigger an immune response and the formation of anti-drug antibodies (ADAs), which limit the efficacy of ERTs and preclude long-term treatment for many patients.
Fabry Disease (FD) is the most common LSD. It is characterised by X-linked mutations in the gene encoding the alpha galactosidase (AGAL) enzyme, which cause irreversible damage to heart, vasculature, and kidney. Immune response is common, especially in males, who usually experience the most severe form since they have no enzyme activity. With the increasing number of FD patients due to better diagnosis, there is the pressing clinical need to improve ERT efficacy and reduce their immunogenicity.
Since the formation of anti-drug antibodies is initiated by T-cell recognition of peptides of antigen presenting cells, we propose to lower the immunogenicity of a recombinant enzyme by "recoding" known epitopes in the enzyme sequence, while preserving its catalytic function.
Here we will use Artificial Intelligence (AI) to design enzymes with a desired function but lacking known epitopes, a process called deimmunization, to deliver better therapies for FD patients. We will manufacture these enzymes (up to 200) using a mammalian cell line expression system, to ensure our ERTs have human-like biochemical properties, and then we will measure their catalytic activity in vitro and the immune reaction using patients' sera. Importantly, since AI has the potential to impact the life of millions of patients and their families, we will engage with patients through open meetings to show and explain the potential of this new technology in addressing an unmet clinical need.
Our project builds on the team's unique expertise in AI, biologics production and medicine and represents one of the largest ERT deimmunization study to date, with the potential to provide effective treatment options for FD and other LSD patients in the future.
Publications
Kasprzyk M
(2024)
APEX: Automated Protein EXpression in Escherichia coli
Lobzaev E
(2024)
Dirichlet latent modelling enables effective learning and sampling of the functional protein design space.
in Nature communications
Lobzaev E
(2024)
Protein engineering using variational free energy approximation
in Nature Communications
| Description | The IDERT project has currently reached two important objectives: 1. Developing a new generative deep learning model that enables the conservative recoding of T-cell and B-cell epitopes to reduce their immunogenicity. 2. Establishing a CHO cell line for robust expression of human alpha galactosidase to be used as replacement therapy. |
| Exploitation Route | Methods developed as part of the project could be used in the biologics manufacturing sector for lowering the immunogenicity of protein-based therapies. |
| Sectors | Healthcare Manufacturing including Industrial Biotechology Pharmaceuticals and Medical Biotechnology |
| Description | As part of the IDERT project, we have engaged with families of Fabry patients through lab visit events. The scope has been to explain how AI and engineering biology can be exploited for good use in the field of drug discovery, especially in speeding up the identification of new therapeutic agents for rare diseases. The impact of these activities has generally been a shift in view in patients, who initially saw AI as a dangerous tool to use in healthcare, towards accepting AI as a powerful tool to support clinical personnel. |
| First Year Of Impact | 2023 |
| Sector | Healthcare,Pharmaceuticals and Medical Biotechnology |
| Description | AI-Bioscience Collaborative Summit |
| Geographic Reach | Multiple continents/international |
| Policy Influence Type | Participation in a guidance/advisory committee |
| Description | Cabinet Office & Alan Turing Institute Roundtable on "Narrow AI Models and Biological Research" |
| Geographic Reach | National |
| Policy Influence Type | Participation in a guidance/advisory committee |
| Description | Engineered Genetic Control Systems for Advanced Therapeutics |
| Amount | £12,367,418 (GBP) |
| Funding ID | BB/Y008545/1 |
| Organisation | Biotechnology and Biological Sciences Research Council (BBSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 02/2024 |
| End | 02/2029 |
| Description | Scottish Enterprise High growth spin out - Company Creation Scheme |
| Amount | £199,250 (GBP) |
| Organisation | Scottish Enterprise |
| Sector | Public |
| Country | United Kingdom |
| Start | 01/2025 |
| End | 06/2026 |
| Title | Dirichlet latent modelling enables effective learning and sampling of the functional protein design space |
| Description | Data generated by TDVAE and TGVAE models and used in manuscript "Dirichlet latent modelling enables effective learning and sampling of the functional protein design space". Contains raw input data, R scripts to process data and output figures used in the main text and supplementary materials of the manuscripts, as well as two supplementary tables in xlsx format. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2024 |
| Provided To Others? | Yes |
| Impact | Training and analysis data of the benchmarking of our TDVAE model. |
| URL | https://zenodo.org/doi/10.5281/zenodo.13269310 |
| Title | Protein engineering using variational free energy approximation |
| Description | Data generated by PREVENT model and used in manuscript "Protein engineering using variational free energy approximation". Contains raw input data, R scripts and Jupyter Notebooks to process data and output figures used in the main text and supplementary materials of the manuscripts. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2024 |
| Provided To Others? | Yes |
| Impact | Validation of a new protein engineering method, including experimental validation. |
| URL | https://zenodo.org/doi/10.5281/zenodo.13763481 |
| Description | Edinburgh Kidney research initiative |
| Organisation | University of Edinburgh |
| Department | Renal Medicine Edinburgh |
| Country | United Kingdom |
| Sector | Academic/University |
| PI Contribution | I have been invited to joined a network of researchers and clinicians at University of Edinburgh, who work on renal diseases which include also Fabry disease. My role here is to promote the use of AI and engineering biology in drug discovery and to engage with patients to make them aware on the progress enabled by these technologies. The collaboration is very productive and we are working on joint UKRI proposal. |
| Collaborator Contribution | Partner provide expertise into the clinical implications of my fellowship work, and has allowed me to bridge my research with patients in the clinic. |
| Impact | The collaboration is interdisciplinary since it involves work with clinicians. |
| Start Year | 2023 |
| Title | PREVENT: PRotein Engineering by Variational frEe eNergy approximaTion |
| Description | A generative model to learn the sequence and thermodynamic landscape of a target protein and generate more stable variants. Full Changelog: https://github.com/stracquadaniolab/prevent-nf/commits/v1.0.0 |
| Type Of Technology | Software |
| Year Produced | 2024 |
| Open Source License? | Yes |
| Impact | New method for protein engineering using biophysical information. |
| URL | https://zenodo.org/doi/10.5281/zenodo.13880396 |
| Title | PREVENT: PRotein Engineering by Variational frEe eNergy approximaTion |
| Description | A generative model to learn the sequence and thermodynamic landscape of a target protein and generate more stable variants. Full Changelog: https://github.com/stracquadaniolab/prevent-nf/commits/v1.0.0 |
| Type Of Technology | Software |
| Year Produced | 2024 |
| Impact | The software implements a generative AI method combining sequence and biophysical information for the design of highly stable proteins. |
| URL | https://zenodo.org/doi/10.5281/zenodo.13880397 |
| Title | PROTON: PROtein engineering by TempOral convolutional Networks |
| Description | The PROtein engineering by TempOral convolutional Networks (PROTON) is a deep learning software to design protein libraries using sequence information of protein families. PROTON it implements a generative model, called Temporal Dirichlet Variational Auto Encoder (TDVAE), which maps a protein family design space into a discrete mathematical space and uses temporal convolution to output new, unseen protein sequences. The software offers to design options: prior sampling design, which generates sequences using information learned by the entire protein family, or posterior sampling design, which generates variants of a user- defined protein. PROTON can performs biochemical characterisation of the designed sequences, and can rank and prioritise sequences for downstream experimental testing using two new analyses, namely coverage and confidence analysis: the former estimates the amount of data supporting the predicted amino acid, the latter estimates how confident the model is about its prediction. PROTON can also optimise the training process by performing sequence clustering, and similarly create highly diverse protein libraries by using sequence clustering methods like MMseq2, as already shown in our preprint: this step is completely optional or can be replaced by any other clustering software. PROTON is designed to work in high-performance computing environments and exploits parallelism to minimise the computational burden. PROTON is licenses through TTO at University of Edinburgh under the new technology disclosure "TEC1104509 - PROTON: PROtein engineering by TempOral convolutional Networks". |
| Type Of Technology | Software |
| Year Produced | 2023 |
| Open Source License? | Yes |
| Impact | PROTON is enabling the |
| URL | https://github.com/stracquadaniolab/prevent-nf |
| Description | Patients' Engagement, QMRI, University of Edinburgh. |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Patients, carers and/or patient groups |
| Results and Impact | 40 people attended the Edinburgh Kidney PPI event at QMRI, which focused on showing the new Mission Hub activities and the body work regarding rare diseases. The activity contributed to create a better understanding around gene and cell therapies.. |
| Year(s) Of Engagement Activity | 2024 |
| Description | Patients' Engagement, QMRI, University of Edinburgh. |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Patients, carers and/or patient groups |
| Results and Impact | 40 people attended the Edinburgh Kidney PPI event at QMRI, which focused on showing the using of AI in drug discovery for rare diseases. The activity contributed to shift the widespread negative opinion the patients had about AI, by explaining that AI is an assistive tool to help scientists to rapidly identify new potential treatments. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://edinburghkidney.co.uk |
| Description | Patients' Engagement, School of Biological Sciences, University of Edinburgh. |
| Form Of Engagement Activity | Participation in an open day or visit at my research institution |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Public/other audiences |
| Results and Impact | I organised a lab visit for a family of Fabry patients, who got in touch with me to know more about my research work. The visit was organised as follows: 1. I delivered a talk describing the work of my research group and what are our long term goals, and how we use AI and engineering biology to achieve them. 2. Q&A sessions to gather patients' feedback and views. 3. Visit of my laboratory and the Edinburgh Genome Foundry (EGF). |
| Year(s) Of Engagement Activity | 2024 |
