We are on the threshold of a new dawn in the pharmaceutical industry, a revolution driven by decades of publicly funded research and today's most advanced technologies. Artificial intelligence (AI) offers hope for developing drugs against some of the most devastating diseases affecting humanity, from cancer and diabetes to Alzheimer's disease. At the heart of these diseases are often proteins, key molecular players whose malfunction can have catastrophic consequences for human health.

If we imagine our body as a perfectly coordinated orchestra, proteins are its conductors. In the form of hormones, they manage fundamental processes like growth, metabolism, and reproduction. As enzymes, they dictate the tempo of chemical reactions necessary for digestion or DNA replication. In the role of antibodies, they lead our immune system in the fight against pathogens. In their many other forms, they decide the life and death of every cell. But when these masterful molecular machines fail, either in their structure or function, the harmony is disrupted and disease arises.

The Potential of Turning Proteins into Drugs

The idea of using proteins as drugs is not new. We use them when the body does not produce enough of a certain protein or when that protein does not function correctly. The most obvious example is insulin, a protein hormone that regulates sugar metabolism. In people with type 1 diabetes, the body does not produce enough insulin, so they must take it externally to control their blood glucose levels. Insulin is one of the first triumphs of bioengineering, but today's array of protein-based drugs is far broader and more sophisticated.

Besides insulin, other protein-based drugs are used in modern medicine. Popular weight-loss drugs, such as Ozempic and Wegovy, are based on GLP-1 receptor agonists, which are also protein-based. In oncology, antibody treatments, such as Herceptin for certain types of breast cancer, target specific proteins on the surface of tumor cells, marking them for destruction by the immune system. The potential is vast, but traditional methods of discovering and modifying existing proteins have their limitations. This is where artificial intelligence comes into play.

How Does Artificial Intelligence Create the Proteins of the Future?

The real breakthrough occurs when we start designing proteins completely from scratch, freed from the constraints of what already exists in nature. "If you design proteins completely from scratch, you are no longer limited to proteins that already exist. We can build proteins with completely new properties, and that could be incredibly powerful in addressing the challenges we face in medicine," explains Dr. Tanja Kortemme, a professor of bioengineering at the University of California, San Francisco (UCSF) and associate dean for research at the UCSF School of Pharmacy. Her lab recently created the world's first synthetic proteins that can change shape, mimicking the dynamic nature of natural proteins.

The process can be compared to the workings of popular AI models like ChatGPT, but instead of text, these models "think" in three dimensions. The AI is trained on vast databases containing the precise three-dimensional structures of hundreds of thousands of known proteins. Once the AI learns the "language" of protein structures – where each atom is positioned and how the protein folds in space – scientists can give it a task. For example, they can request it to generate a completely new protein that will perfectly bind to a protein responsible for the spread of cancer and block its function. The possibilities, as Dr. Kortemme says, are almost limitless.

The Foundation of the Revolution: The Power of Data and Decades of Research

The success of these advanced AI models would not be possible without one key resource: the global, publicly available Protein Data Bank (PDB). This incredible repository of knowledge has been built over decades through the efforts of the global scientific community. Scientists from all over the world have deposited their findings on protein structures into this open database to advance science together. It is this monumental data warehouse, containing more than 200,000 detailed molecular structures, that has enabled the development of AI tools that now promise a revolution.

This collaborative venture was made possible by continuous funding from public, federal sources, primarily from the National Institutes of Health (NIH) and the National Science Foundation (NSF) in the US. Dr. Kortemme's lab uses this large-scale data to produce new proteins with entirely new functions, using both existing generative AI models and developing their own.

From Digital Code to a Real Drug: The Journey of a Protein

After an AI model generates a promising digital protein structure, the crucial step of translating that design into a real, physical molecule follows. This is achieved through DNA synthesis and recombinant DNA technology. Scientists can synthesize a DNA segment that contains the "recipe" for the desired protein, and then insert that genetic code into a living organism, most commonly a bacterium or yeast. These cells become miniature factories that produce the new, artificially designed protein in large quantities.

Interestingly, UCSF was one of the pioneers in the development of recombinant DNA technology, an innovation that laid the foundation for modern biotechnology and is now key to bringing AI-created designs to life.

UCSF as an Epicenter of Innovation and Multidisciplinarity

For more than twenty years, the University of California, San Francisco (UCSF) has been at the forefront of protein engineering technologies, modifying existing proteins to make better drugs. Today, research groups at UCSF, led by exceptionally innovative graduate students and postdoctoral researchers, have developed advanced computational methods, including AI, to design proteins from scratch. Students with backgrounds in computer science, engineering, or mathematics come to UCSF, attracted not only by the scientific challenges but also by the unique environment the city offers. Although finding accommodation in San Francisco can be challenging, the academic opportunities offered far outweigh this.

The secret to this success lies in interdisciplinary science and collaboration. UCSF provides a unique environment where different fields converge – computer science, engineering, fundamental biology, biomedical sciences, and pharmaceutical development. This synergy creates fertile ground for innovation. Students with strong quantitative skills are fascinated by the biological and biomedical problems that drive research at UCSF, and they are the key drivers of progress. The academic reputation and research opportunities make UCSF a magnet for talent from all over the world, who often also seek accommodation in San Francisco to be part of this dynamic ecosystem.

When Can We Expect the First AI-Designed Drugs?

Although many companies are already using AI methods to aid in the discovery and optimization of potential drugs, we still do not have a drug on the market that has been completely, from start to finish, designed exclusively by artificial intelligence. However, we are witnessing a true explosion of efforts in the biotechnology industry aimed at developing AI-generated proteins with therapeutic potential.

Experts are optimistic. "I expect that we will see a large number of these designed proteins entering preclinical development in the next five years, and then, hopefully, clinical trials to really help people," predicts Dr. Kortemme. The path from a computer model to a patient is long and demanding, but the speed at which AI is accelerating the initial stages of drug discovery gives great reason for optimism. The revolution has already begun, and its fruits could forever change the way we treat the most serious diseases.

Source: University of California