Imagine if we could design life’s most essential tools from scratch. That’s exactly what’s happening in the world of protein design, and it’s a game-changer. A groundbreaking model has just produced enzymes—nature’s workhorses—that rival the efficiency of those found in the wild. But here’s where it gets controversial: can we truly replicate the complexity of nature’s designs, or are we just scratching the surface? Let’s dive in.
In a remarkable leap forward, the latest iteration of the protein design model RoseTTAFold Diffusion, developed in the lab of 2024 Nobel Prize-winner David Baker, has successfully created blueprints for functional enzymes entirely from scratch. These enzymes, synthesized and tested in the lab, exhibit activities nearly as effective as those found in nature. This achievement, led by postdoctoral fellow Rohith Krishna at the University of Washington, addresses two major hurdles in enzyme design. First, enzymes often interact with small molecules rather than proteins, a challenge previous models struggled with. Second, the precise positioning of protein side chains—especially their atoms in the catalytic site—is critical for function. Krishna explains, ‘For enzyme design, exactly where the side chains are in the active site is super important.’
To tackle this, the team expanded the model, now called RF Diffusion 2, to include side chain atoms, allowing each atom to interact dynamically within the network. They also relaxed constraints on sequence order, enabling the model to explore a wider range of designs. Krishna notes, ‘There’s so many different solutions, and [the model] gets to explore all of them.’ This flexibility led to a significant increase in design diversity.
The model was put to the test by designing zinc-based enzymes capable of breaking ester bonds. Using quantum chemistry calculations from natural zinc metallohydrolases, the team trained RF Diffusion 2 to position atoms precisely in the active site. The resulting designs, when synthesized, achieved enzymatic activities comparable to those in nature. ‘They’re not the best enzymes ever, but they’re in that range of natural activity,’ Krishna remarks.
What’s even more striking is that these new enzyme sequences bear little resemblance to known proteins, confirming the model’s ability to generate truly novel designs. Steffen Lindner-Mehlich, a biochemist at Charité University Hospital in Berlin, calls this development ‘very exciting,’ highlighting its potential to create tailored synthetic pathways for diverse applications.
But here’s the part most people miss: despite this progress, the success rate for highly efficient enzymes remains around 1%. Carlos Acevedo-Rocha, a senior researcher at the Technical University of Denmark, points out that this requires synthesizing and screening numerous designs, a costly endeavor for many labs. Additionally, factors like stability in industrial environments, such as bioreactors, still need thorough assessment.
However, the pace of innovation is encouraging. The latest version, RF Diffusion 3, is now freely available and boasts significant improvements. According to Krishna, it handles more non-protein molecules, places atoms with greater precision, and operates 10 times faster. And this is where it gets controversial: as we refine these models, are we truly mastering nature’s complexity, or are we merely mimicking it? What do you think? Let’s spark a discussion in the comments—do these advancements bring us closer to fully replicating nature, or are there limits we’ll never surpass?
