Presentation Information
[2ASPR-21]Accelerating PHA Material Design Through Predictive Modelling and Biofoundry-enabled Strain Engineering
○Bilge Elitok1, Kaisa Peltonen1, Emily Bennett1, Laura Salusjärvi1, Mervi Toivari1, Anna Ylinen1, Jukka Vaari1, Olli Pakarinen1, Antti Paajanen1, Sandra Castillo1, Tuula Tenkanen1, Antti Puisto1, Merja Penttilä1, Anssi Laukkanen1, Yvonne Nygård1 (1. VTT Technical Research Centre of Finland (Finland))
Keywords:
polyhydroxyalkanoates,synthetic biology,biomanufacturing,computational biomaterial design,biofoundry
Polyhydroxyalkanoates (PHAs) are bio-based and biodegradable polyesters produced by microbes, providing a sustainable alternative to current petrochemical-based plastics. PHAs can be synthesized as homopolymers, random or block copolymers, unlocking a vast design space of possible biomaterials with tunable properties. Because this design space is so large, it is difficult to identify optimal materials based on intuition or experimental screening.
The synthetic biology-based material accelerator platform (SynBioMAP) at VTT offers a targeted exploration of this vast design space by combining material modelling with iterative design-build-test-learn cycles consisting of biofoundry-based strain construction and screening, automated bioprocessing, and Bayesian active learning.
SynBioMAP accelerates the development of new PHAs with targeted performance properties by leveraging machine learning (ML) based predictive modelling that forecasts material behavior. We are working on predictive and generative modelling of barrier properties which are critical for the functional performance of PHAs in applied contexts such as bio-based coatings or packaging. For instance, we developed a method which predicts the glass transition temperature (Tg), a key parameter influencing polymer properties, blend formulation, and the material’s end-use behavior. Molecular dynamics simulations of amorphous homopolymer polyhydroxybutyrate (PHB) and its random copolymers successfully predicted the Tg as a function of fraction copolymer composition. These models can be used to identify PHA compositions with desired properties before building production strains.
To translate these material targets into biological production, we are developing Biofoundry pipelines for tailored PHA production in Escherichia coli. For quantitative measurements of produced PHAs, we are comparing several high-throughput compatible screening assays that report both total production and compositional information. We have also used AI to design novel PHA synthases that can produce PHAs with desired compositions. By generating PHA production strains whose monomer pathways can be tuned via an orthogonal set of small-molecule inducers, we can set the ratio of monomers to be incorporated into PHA polymers. Strain construction is supported by metabolic modelling and Bayesian learning-based design of strains for consecutive design-build-test-learn cycles in the Biofoundry. Our approach supports informed decision making, minimizes experimental cycle time, and thus enables efficient production of new biomaterials.
The synthetic biology-based material accelerator platform (SynBioMAP) at VTT offers a targeted exploration of this vast design space by combining material modelling with iterative design-build-test-learn cycles consisting of biofoundry-based strain construction and screening, automated bioprocessing, and Bayesian active learning.
SynBioMAP accelerates the development of new PHAs with targeted performance properties by leveraging machine learning (ML) based predictive modelling that forecasts material behavior. We are working on predictive and generative modelling of barrier properties which are critical for the functional performance of PHAs in applied contexts such as bio-based coatings or packaging. For instance, we developed a method which predicts the glass transition temperature (Tg), a key parameter influencing polymer properties, blend formulation, and the material’s end-use behavior. Molecular dynamics simulations of amorphous homopolymer polyhydroxybutyrate (PHB) and its random copolymers successfully predicted the Tg as a function of fraction copolymer composition. These models can be used to identify PHA compositions with desired properties before building production strains.
To translate these material targets into biological production, we are developing Biofoundry pipelines for tailored PHA production in Escherichia coli. For quantitative measurements of produced PHAs, we are comparing several high-throughput compatible screening assays that report both total production and compositional information. We have also used AI to design novel PHA synthases that can produce PHAs with desired compositions. By generating PHA production strains whose monomer pathways can be tuned via an orthogonal set of small-molecule inducers, we can set the ratio of monomers to be incorporated into PHA polymers. Strain construction is supported by metabolic modelling and Bayesian learning-based design of strains for consecutive design-build-test-learn cycles in the Biofoundry. Our approach supports informed decision making, minimizes experimental cycle time, and thus enables efficient production of new biomaterials.
Comment
To browse or post comments, you must log in.Log in