Use of Computational Modelling to Derive Structural Insights of SFB Protein P3340

Segmented Filamentous Bacteria (SFB) are gram-positive gut bacteria distinguished for their roles in developing Th17 and Th22 immune cells. These immune cells are well known for their robust response against pathogens and infections and the role of SFB in immune response is well studied. SFB_NYU_P3340 (P3340) is a surface protein of SFB possessing high expression in the human gut, where SFB binds with intestinal epithelial cells. While the amino-acid sequence of P3340 is well known, the exact structure remains elusive. AlphaFold 3, released in May 2024, can be used to predict the structure of P3340 using artificial intelligence and deep learning, but the accuracy of this prediction requires further testing and optimization. To bridge this gap, the protein simulation package OpenMM can be employed to run simulations of P3340 and computationally assess results, including the stability of the protein structure under standard force fields. In this study, OpenMM was used to find a stable structure of P3340 and simulate P3340 under variable solvents conditions. In-silico analysis was conducted on the P3340 simulations, finding important data such as conformational energy minimization, radius of gyration, density, contact distance to assess the stability of P3340 at various instances of simulating the protein. P3340 has been produced in our lab from synthetic DNA that was codon-optimized for E. coli expression, but with low yields. Our experiments have shown that the use of small-molecule additives is promising in increasing the solubility of P3340, which is characteristically insoluble, thus increasing the yield of P3340 during purification. OpenMM was also used to simulate P3340 in the presence of different additives and the solubility of P3340 was computationally measured using various solubility parameters. In summary, we have used machine learning enabled protein structure determination and molecular simulations to well characterize the three-dimensional structure of P3340 and explore how we can improve yields in our processes. In the future, we aim to use our models to explore mechanistic questions about this particular protein.