Immersive insights from virtual reality: Protein engineering at our fingertips

Zahra Ouaray, PhD , Principal Computational Structural Biologist at Etcembly, explains how we’re using our expertise in structural modelling to streamline TCR discovery and optimisation, and how new VR tools are helping us with this task.

At Etcembly, we are focusing on developing novel biotherapeutics based on T cell receptors (TCRs). TCRs bind to intracellular peptides presented on the cell surface by the human leukocyte antigen (HLA) system, enabling T cells to identify and remove unhealthy or infected cells.?

This interface has been fine-tuned over millions of years of evolution, resulting in astonishing TCR diversity. We are now tapping into this diversity by discovering and optimising TCRs that bind to specific targets relevant to disease and turning them into next-generation immunotherapies for cancer, auto-immune conditions and more.

Taking a closer look at TCR structures

Proteins have highly complex three-dimensional structures, which define how they interact with and bind to other molecules. Understanding and manipulating these intricate structures therefore lies at the heart of modern drug discovery, particularly in the fast-growing field of biotherapeutics.?

The interaction between a TCR and its target is one of the most complicated in biology, involving multiple layers of intricacy. Each chain of the heterodimeric TCR consists of complementarity-determining region (CDR) loops, which interact with specific regions within the peptide:HLA complex to distinguish healthy cells from aberrant ones (Figure 1).

Therefore, an in-depth understanding of the structural interface between a TCR and its target is essential to design and optimise potent TCR-based therapeutics with high specificity.

Using virtual reality to enhance protein structure analysis

Most conventional tools for structural analysis of proteins, such as PYMOL, VMD, and RASMOL, display 3D models of proteins on 2D computer screens. While these models can be clicked, rotated, moved about, and zoomed in - they are still mere projections of the protein's 3D structure, with an inherent loss of information.

However, new virtual reality (VR) tools are enabling us to overcome the limits of 2D visualisation and fully immerse ourselves and interact with proteins in all their three-dimensional glory. VR creates an interactive, intuitive and collaborative environment so we can fully explore and manipulate these complex structures to accelerate TCR drug discovery and optimisation.?

At Etcembly, we’re experimenting with VR molecular visualisation software from Nanome to compare crystal structures of real TCRs with in silico models generated by our EMLy? discovery engine, which combines generative AI and structural modelling to discover and design novel TCRs with picomolar affinity.?

VR allows for detailed exploration of the intricate structural components of TCRs and their targets. Variable regions and CDRs can be highlighted and analysed in detail, either on their own or bound to a target peptide:HLA complex. This allows us to see and explore this interaction from all angles and highlight specific protein regions responsible for TCR binding and activation (Figure 1).?

We can use this tool to overlay the two structures and not only visualise differences, but also quantify them as well. We can calculate the root-mean-square deviation of atomic positions (RMSD) to assess the level of similarity between our AI-generated model and the crystal structure to see how our modelling is performing.

We then use this information to further train our system to generate TCRs with higher affinity and more specific binding.

TCR engineering at our fingertips

Another strength of VR protein visualisation tools is the ability to incorporate information from molecular dynamics analysis. This allows us to examine how molecules move in solution and how the designs generated by AI affect and enhance this movement.

We can also use these tools to virtually manipulate and modify proteins as well. Structural adjustments, followed by quick analysis and immediate feedback within VR, streamlines the protein structure refinement process so we can engineer best-in-class TCRs with high potency and specificity.?

Importantly, this process can also be done in a collaborative fashion. Instead of crowding around a 2-D model on a single computer screen, scientists can get together in VR and explore the TCR structure together.?

Furthermore, manipulating the structure directly with virtual hands makes this process incredibly intuitive. Team members without deep domain expertise in structural biology can quickly get to grips with our TCRs and see what needs fixing, fostering better collaboration across our multi-disciplinary team.?

What will the future of TCR design look like?

VR is becoming an integral part of our TCR design and optimisation workflow at Etcembly, and we’re looking to expand how we use it to explore and validate novel targets as well as design, model and optimise TCR therapeutics.?

It’s exciting to think about how virtual reality might expand our capabilities in structural modelling.

At the moment we are exploring protein structures using VR headsets. But maybe one day our team will be gathered around a holotable like the crew on a sci-fi spaceship, visualising and manipulating molecules projected into the space in front of us.?

We’re excited to see how further advances will transform how we understand, analyse, and modify complex protein structures to create the next generation of life-changing TCR therapeutics.?