Design

Contribution Header

Design

The design of our project began with a simple but important question: Why does the immune system sometimes fail to recognize cancer cells, even though it has the capacity to detect abnormal proteins? Our answer was that tumor-specific antigens are often poorly presented by MHC Class I molecules, leading to weak or undetectable T-cell activation (1,2). With this in mind, our design strategy focused on enhancing the function of MHC Class I itself by combining computational protein modeling with wet-lab testing.

Diagram of our project work-flow

Figure 1. Diagram of our project work-flow

Computational Modeling

The first stage of our design was to explore the structural basis of antigen presentation. MHC Class I molecules contain a peptide-binding groove that accommodates short peptides, typically 8--11 amino acids in length. The strength and stability of peptide binding determine whether the antigen will be effectively presented to T-cells (3,4). Using AI-assisted docking platforms such as DiffDock, we modeled interactions between tumor-derived peptides and the wild-type HLA-A*02:01 molecule, one of the most studied human MHC Class I alleles.

Through these simulations, we examined the predicted binding energy and stability of the peptide--MHC complexes. When the interaction was weak, we hypothesized that introducing point mutations in the MHC binding groove could create a more favorable environment for peptide anchoring. This step was crucial because it allowed us to identify specific candidate mutations without relying solely on trial-and-error experiments, thereby saving time and resources (5). Importantly, computational modeling also gave us a visual and quantitative framework to compare different mutations and rank them according to their potential benefit.

Protein Production

Once the computational predictions were in place, we needed to test them in the laboratory. Producing human proteins in bacteria such as E. coli is a well-known challenge, as these proteins often misfold and form insoluble aggregates called inclusion bodies (6). To overcome this, we codon-optimized the gene sequence of the MHC Class I heavy chain for bacterial expression and cloned it into an expression vector. We also included a biotinylation tag at the C-terminal end of the heavy chain, which would later enable tetramer formation and detection assays.

At the same time, we prepared the complementary β2-microglobulin (B2M) protein, which is required for proper folding and stability of MHC Class I. Both the heavy chain and B2M were expressed separately in E. coli, purified under denaturing conditions, and later refolded together in vitro with the tumor-specific peptides. This three-component assembly --- heavy chain, B2M, and peptide --- was necessary for functional MHC formation (7,8).

Iteration in Protein Expression

One of the most important parts of our design process was improving protein expression. In our first attempts, we noticed that colonies transformed with the MHC construct grew poorly, and protein expression levels were very low. This suggested that expression of the human MHC heavy chain might have placed a burden on the bacterial host, possibly affecting cell survival.

To address this, we used an iterative strategy: instead of relying on a single culture, we divided the colonies into multiple flasks and monitored their growth separately. Some cultures continued to show poor growth and were discarded, while others achieved significantly better growth rates. By selecting colonies that tolerated expression more effectively, we were able to obtain improved yields of our target protein.

This simple yet effective adjustment represented the "learn and refine" step of our design cycle. While we did not make changes to the DNA sequence at this stage, the process demonstrated how careful observation and iterative problem-solving can overcome practical challenges in recombinant protein expression.

Binding Assays

With refolded proteins in hand, the next design step was to experimentally test whether our engineered MHC molecules indeed bound tumor-derived peptides more effectively than the wild type. For this, we used ELISA-based assays, where peptide-loaded MHC tetramers were immobilized on plates and their binding interactions were quantified through absorbance measurements. This setup allowed us to directly compare the relative affinity of wild-type and engineered variants (9).

By connecting computational predictions to experimental validation, we established a workflow that not only tested our ideas but also generated insights into the strengths and weaknesses of our models.

Differentiating Aspects of Our Design

Unlike many immunotherapy strategies that focus primarily on modifying T-cells or designing new peptide vaccines, our project is unique in its focus on the MHC molecule itself. By directly engineering the protein responsible for presenting antigens, we address a fundamental bottleneck in immune recognition. This design philosophy sets us apart from existing approaches and has the potential to contribute not only to cancer therapy but also to basic immunology research.

References (Design Section)

  1. Schumacher, T. N., & Schreiber, R. D. (2015). Neoantigens in cancer immunotherapy. Science, 348(6230), 69--74.
  2. Sahin, U., & Türeci, Ö. (2018). Personalized vaccines for cancer immunotherapy. Science, 359(6382), 1355--1360.
  3. Neefjes, J., et al. (2011). Towards a systems understanding of MHC class I and class II antigen presentation. Nat Rev Immunol, 11(12), 823--836.
  4. Rock, K. L., Reits, E., & Neefjes, J. (2016). Present yourself! By MHC class I and MHC class II molecules. Trends Immunol, 37(11), 724--737.
  5. Jumper, J., et al. (2021). Highly accurate protein structure prediction with AlphaFold. Nature, 596(7873), 583--589.
  6. Rosano, G. L., & Ceccarelli, E. A. (2014). Recombinant protein expression in Escherichia coli: advances and challenges. Front Microbiol, 5, 172.
  7. Garboczi, D. N., et al. (1992). Assembly, specific binding, and crystallization of a human TCR--MHC complex. Proc Natl Acad Sci USA, 89(8), 3429--3433.
  8. O'Callaghan, C. A., & Bell, J. I. (1998). Structure and function of HLA-A*0201. Immunol Rev, 163, 59--72.
  9. Crowther, J. R. (2000). The ELISA Guidebook. Humana Press.