Proteins that should not exist according to the rules of nature, yet it worked.
At the boundary between computation and biology, researchers have engineered proteins that glow in wavelengths nature never produced — opening a window through human tissue that has, until now, remained closed. Oliver Bruns and an international team including Nobel laureate David Baker have created the first fluorescent proteins designed to emit light in the near-infrared and short-wave infrared ranges, allowing surgeons to see cancer cells hiding at tumor margins and lymph nodes in real time. This is not merely a technical refinement; it is evidence that artificial intelligence can invent biological functions that evolution never arrived at — and that medicine may be entering an era where the tools of life itself are written, not found.
- The fundamental obstacle — that visible light scatters and vanishes within millimeters of human tissue — has long left surgeons operating at the edge of what they can see.
- Two entirely new proteins, paired with custom dyes, now emit light in infrared ranges that no natural protein has ever reached, producing sharp images where conventional imaging produces only noise.
- In cell cultures and animal experiments, the proteins revealed biological structures with a clarity that exposes individual cancer cells at tumor margins — the precise detail surgeons need to remove disease completely.
- The technology positions real-time surgical imaging as a near-term clinical possibility, with the potential to distinguish metastatic cells in lymph nodes from healthy tissue while the patient is still on the table.
- Beyond medicine, the work signals that AI-driven protein design has crossed from theory into practice — raising the question of what other biological functions might now be written from scratch.
Oliver Bruns has long pursued a single, exacting vision: the ability to see cancer cells hiding at tumor edges and in lymph nodes while a surgeon's hands are still at work. The obstacle has always been light itself — visible wavelengths scatter and fade within millimeters of human tissue. Infrared light penetrates deeper and cleaner, but no natural protein has ever been able to harness it.
Working with an international team that includes David Baker, the 2024 Nobel laureate in chemistry, Bruns and colleagues at the National Center for Tumor Diseases in Dresden have now engineered the first proteins ever designed to fluoresce in the near-infrared and short-wave infrared ranges. Published in the Journal of the American Chemical Society, the work pairs computationally designed proteins with custom-made dyes to produce light at wavelengths that evolution never reached. As one researcher on the team put it: they had created something that should not exist according to the rules of nature — yet it worked.
Two proteins emerged. One glows intensely at the far-red edge of the visible spectrum; the second pushes deeper into the short-wave infrared, where tissue becomes nearly transparent to light. In laboratory and animal experiments, both revealed biological structures with a sharpness and low background noise that conventional imaging cannot match. For surgeons, the implications are immediate: tumor margins become visible, blood vessels light up, and metastatic cells in lymph nodes can be distinguished from healthy tissue in real time.
Bruns has spent years building the hardware and methodology to exploit this infrared window. His department, a collaboration anchored at the German Cancer Research Center and University Hospital Dresden, has become a center of gravity for the field. But the new proteins represent something beyond better tools — they demonstrate that AI and computational design can invent biological functions that nature never produced.
The question the work leaves open is as large as the achievement itself: if proteins can be redesigned to fluoresce in the infrared, what else might be created from scratch? The age of de novo protein design, the study suggests, is no longer a promise. It has arrived.
Oliver Bruns stands at the intersection of two worlds: the surgeon's operating room and the computational laboratory. For years, he has pursued a particular vision of precision medicine—the ability to see cancer cells hiding at the edges of tumors, nestled in lymph nodes, while a surgeon's hands are still steady and the patient is still open. The obstacle has always been the same: light does not travel far through human tissue. Visible wavelengths scatter and fade within millimeters. But infrared light—the kind we cannot see—penetrates deeper, cleaner, with less interference from the body's own background noise.
Now, working with an international team that includes David Baker, the 2024 Nobel laureate in chemistry, Bruns and his colleagues at the National Center for Tumor Diseases in Dresden have crossed a threshold. They have engineered the first proteins ever designed to glow in the near-infrared and short-wave infrared ranges—wavelengths that exist nowhere in nature. The work, published recently in the Journal of the American Chemical Society, represents something more than an incremental improvement. It is a proof that artificial intelligence and computational protein design can create biological functions that evolution never produced.
The challenge was formidable. Fluorescent proteins have long been workhorses in biological research, allowing scientists to tag and track living processes inside cells. But the proteins that nature provides—the green fluorescent protein from jellyfish, the red variants engineered from coral—emit light in the visible spectrum. To push into the infrared, researchers needed to redesign proteins from the ground up, pairing them with custom-made dyes that could absorb and emit light at wavelengths far beyond what any natural protein could manage. Dr. Bernardo Arús, a research associate in Bruns' group, describes the moment of realization: the team had created something that should not exist according to the rules of nature, yet it worked.
Two proteins emerged from the effort. One produces a particularly intense glow in the far-red spectrum, the visible edge of the infrared world. The second ventures deeper still, its fluorescence extending into the short-wave infrared range where tissue becomes nearly transparent to light. In cell cultures and animal experiments, these proteins revealed biological structures with a clarity that conventional imaging cannot match. The sensitivity is high. The background noise is low. The potential is obvious to anyone who understands surgery.
Bruns has spent years developing the hardware and methodology to exploit this window into the body. Short-wave infrared imaging, combined with specialized cameras and fluorescent dyes, allows surgeons to see what was previously invisible during the operation itself. A tumor's edge becomes visible. Blood vessels light up. Metastatic cells in nearby lymph nodes can be distinguished from healthy tissue. The goal is not merely to see better—it is to act better, to remove cancer more completely while preserving healthy tissue, all in real time.
The recognition has followed. In 2024, Bruns received the Helmholtz High Impact Award for his work on SWIR imaging technology. His department at the NCT/UCC, a collaboration between the German Cancer Research Center, the University Hospital Dresden, and other institutions, has become a hub for this emerging field. Yet the new proteins represent something different: they show that the toolkit itself can be expanded, that artificial intelligence can be used not merely to optimize existing biology but to invent new biology altogether.
What Bruns and his team have demonstrated is that the future of medical imaging may not depend on finding better dyes or better cameras alone. It may depend on designing better proteins—proteins that do not exist in nature, that could not exist without human intervention and computational power. The implications ripple outward. If proteins can be redesigned to fluoresce in the infrared, what else might be redesigned? What other biological functions might be created from scratch? The study, with its shared authorship across continents and its reliance on Nobel Prize-winning methods, suggests that the age of de novo protein design is no longer theoretical. It is here, and it is beginning to reshape medicine.
Citas Notables
Thanks to computational protein design and custom-made dyes, we were able, for the first time, to develop proteins that exhibit fluorescence in the NIR and SWIR ranges. This represents a major breakthrough in the field of de novo protein design, as these properties have not been observed in nature to date.— Dr. Bernardo Arús, research associate at NCT/UCC Dresden
These newly developed proteins expand the toolkit for imaging in living organisms and could help us study disease mechanisms, biological processes, or therapeutic effects even in deeper tissue layers in the future.— Oliver Bruns, head of Department of Functional Imaging in Surgical Oncology
La Conversación del Hearth Otra perspectiva de la historia
Why does it matter that these proteins glow in infrared rather than visible light? What's the practical difference?
Visible light scatters in tissue almost immediately—you can only see a few millimeters deep. Infrared light travels much farther, and it creates less background noise from the tissue itself. So a surgeon can see deeper structures, and see them more clearly.
And these proteins didn't exist before. They're entirely artificial?
Yes. Nature never produced proteins that fluoresce in these wavelengths. The team had to design them computationally, then pair them with custom dyes. It's not just engineering an existing protein—it's inventing one from scratch.
How does that change what a surgeon can actually do in the operating room?
Right now, surgeons remove tumors based on what they can see and feel. But cancer cells at the edges, or in nearby lymph nodes, are invisible. With these proteins, you could mark those cells beforehand, and then during surgery, they would glow. You'd know exactly where to cut.
So this is about precision—removing all the cancer without taking too much healthy tissue?
Exactly. And doing it while the patient is still on the table, not waiting for pathology reports days later. That's the real power.
What does it say that a Nobel laureate was involved in this?
It says this isn't a small incremental step. Computational protein design is cutting-edge work. The fact that Baker's methods are now being used to create entirely new biological functions—functions that don't exist in nature—that's significant. It opens doors.