Teaching the body’s own cells to fight cancer

Cancer immunotherapy has been hailed as a revolution, but the reality is more complicated. Today, advanced treatments such as chimeric antigen receptor T-cell (CAR-T) therapy – in which doctors reprogramme a person’s own immune cells to attack cancer – have saved some people with blood cancer. But so far, these therapies have struggled in solid tumours, with limited progress in early trials.

Many promising immunotherapies have failed because the engineered cells attacked healthy tissue. Patients even died in one infamous trial when killer cells, meant to target tumours, also recognised a protein in the heart.

“The biggest unsolved problem is specificity,” explains Timothy Jenkins, Associate Professor at the Department of Biotechnology and Biomedicine, Technical University of Denmark, Kongens Lyngby.

“It is not enough to turn the immune system on. You have to make sure it only hits the tumour – and nothing else.”

That was the roadblock faced by Kristoffer Haurum Johansen, a postdoctoral researcher who works with T-cell biology every day. The turning point came when he and Jenkins realised they might combine forces.

“Kristoffer had the clinical problem, and I had the design tools,” Jenkins recalls. “Suddenly it felt like maybe we could build exactly the kind of targeting molecule his field had been missing.”

The first results were striking.

“When we saw our designed proteins guiding human killer cells to recognise and kill cancer in the laboratory, we knew that the concept was real,” Jenkins says. “And the fact that we can now do this in weeks instead of years changes everything. However, computer predictions are not perfect, and more work in animals and eventually humans is needed to prove safety.”

How a casual chat became a breakthrough

Big scientific projects often grow from years of planning and funding. But sometimes, a breakthrough begins with something as simple as a conversation between colleagues. At the Technical University of Denmark, Timothy Jenkins was teaching students how proteins can be designed to act like precision tools, while Kristoffer Haurum Johansen was covering how the body’s own killer cells, T cells, can be harnessed to fight cancer.

“It was just a casual chat after teaching,” Jenkins recalls. “We were answering questions from students, and suddenly we realised that our two worlds fit together in a way that neither of us had thought about before.”

That spark quickly led them to Johansen’s group leader, Professor Sine Reker Hadrup. Together, the three began exploring whether protein design could solve some of the toughest challenges in immunotherapy.

 “We immediately saw extreme synergy,” Jenkins says. “If we could bring the computational tools from my laboratory together with their deep biology expertise, we could create something that is not just a cool academic project but something with real clinical potential.”

What made the collaboration powerful was that Jenkins’ group sits exactly between two worlds: computer science and laboratory biology. “There are brilliant machine-learning groups and brilliant biologists – but very few speak both languages,” he explains. “That is where we sit, and that is why we could see the potential straight away.”

Within months, they had secured funding, recruited a PhD student and laid the groundwork for what would become a Science article – an unusually rapid journey from classroom discussion to cutting-edge cancer research.

Why current cancer therapies are not precise enough

Turning the body’s own killer cells against cancer is one of the most exciting ideas in medicine. But there is a catch: the immune system has to tell friend from foe with absolute precision.

“That is much harder than it sounds. If a therapy misses its target even slightly, the consequences can be deadly – as seen in trials in which killer cells, meant to attack tumours, also attacked healthy tissue.”

This constant risk has been the single biggest barrier for immunotherapy, holding back treatments that otherwise look promising. For Kristoffer Haurum Johansen, who works daily with T-cell biology, this was the roadblock. How do you accurately guide killer cells to their target rather than taking a shotgun approach?

Designing cancer-fighting tools on a computer

A new generation of digital design tools made the project possible. Instead of screening millions of natural antibodies in the hope of finding one that fits a tumour target, Jenkins’ laboratory could create tiny proteins on a computer and test their potential with algorithms.

“These mini-binders had to be made as small as possible so that cells could produce them easily – but not so small that they lost stability or accuracy. Finding that balance was part of the design challenge.

Jenkins says that the researchers do not want to keep this knowledge to themselves.

“These tools have so many uses that sharing them openly helps everyone move faster.”

This marks a sharp break from current practice, in which many of the antibody fragments used in immunotherapies are far from ideal.

“These short antibody fragments – the broken pieces of natural antibodies used in some therapies – are notorious,” Jenkins explains. “They are unstable, they clump together, and when you need ultraspecific binders, it is like searching for a needle in a haystack. Designing the needle directly is much faster and far more reliable.”

From digital design to real experiments

To their surprise, the virtual designs often behaved just like the real ones.

“In some tests, about 8 of 10 predictions matched the real results. Even if this is not perfect, this can save months of wasted work and reduce risks early.”

That means that they could spot promising designs before ever entering the laboratory. Instead of wasting time and resources building proteins that would fail, the researchers could focus only on the most promising candidates. In cancer research, in which every experiment can be costly and slow, that kind of filtering is a huge advantage.

“Once the best candidates were chosen, they were built into human killer cells and tested in culture dishes. We could go from identifying a target to having a validated design in just weeks,” Jenkins explains. “That is lightning fast compared with traditional drug discovery.”

This was the first time the team saw their computer models translate into living cells – a critical proof that the designs were more than theory. This might be exactly the kind of shortcut immunotherapy has been waiting for.

Killer cells show that they can wipe out cancer

The first tests showed that the designed proteins really could guide killer cells with remarkable accuracy.

“We targeted a classic cancer marker called NY-ESO-1 – a protein present on many tumour cells but not healthy cells – and the computer-designed mini-binders locked onto it exactly as predicted.”

Advanced imaging confirmed the fit, providing solid proof that the approach worked. When these binders were engineered into human killer cells, something even more striking happened: the cells began to recognise tumour cells and destroy them.

The researchers built the designs into CAR-T – a laboratory strain of human T cells engineered to mimic patient-derived ones – and showed that they could successfully wipe out tumour cells in the laboratory.

“We proved that our designed proteins can guide killer cells to recognise and kill cancer,” Jenkins says. “Seeing that work in human cells was the moment we knew the concept was real.”

A first step toward personalised therapy

Encouraged, the group pushed further. They also tested the method on a marker unique to one patient’s tumour, designed entirely from computer models.

“This showed that the approach could be extended to personalised cancer therapy,” Jenkins adds.

The fact that the approach worked here too hinted at something once thought impossible: tailor-made therapy designed for an individual patient within weeks.

“Normally, if a patient is terminal, there is zero chance of developing a therapy in time,” Jenkins explains. “With this, you actually have a shot in a month.”

That kind of turnaround time is unheard of in immunotherapy, in which developing and testing antibodies can take years. For a field that has long struggled with both speed and specificity, these results were not just incremental progress – they pointed toward a fundamentally different way of thinking about cancer treatment.

Cancer treatments tailored to each patient

The promise now lies in what could come next.

For decades, immunotherapy has been held back by slow development and safety concerns – but Jenkins and colleagues believe that their computer-designed proteins could change that.

“This is not just about a faster way to make drugs,” Jenkins says. “It is about opening the door to personalised therapies that could be ready in weeks instead of years.”

Nevertheless, he cautions that specificity remains the biggest hurdle. “We are closer than before, but it is not fully solved,” he notes.

The team is now working to sharpen the specificity even further, testing whether their designed proteins can avoid any unintended targets in the body. The next steps include studies in mice and collaborations with hospitals to understand how such therapies could be designed under the strict regulations that apply in medicine.

But Jenkins also stresses that success requires more than science. “At some point, the technology will be ready, but without a commercial pipeline it will not reach patients,” he explains. “We want to avoid being caught flat-footed. That means thinking about regulation, good manufacturing practice production and funding now – not later.”

Good manufacturing practice is the strict quality standard for making medicines, so there is a long road ahead, but the principle is clear: cancer treatments could one day be designed and not discovered – and delivered on a timeline that truly matters to patients.

Hope for patients, caution for scientists

The article in Science quickly drew international attention, not only from researchers but also from patients and families desperate for new treatments. For Jenkins, the messages were both inspiring and sobering.

“We had cancer patients write to us after the article came out,” he recalls. “It is motivating to see how much hope this research creates, but it is also heartbreaking to tell them that it is still years away.”

These conversations underscored why the team feels urgency. Although the technology is still at an early stage, the demand for better options is painfully real. In addition, clinicians have shown unusual interest in working with the group. “Doctors want to help their patients – that is why they are in medicine,” Jenkins says. “When they see tools with real potential, clinicians are often quick to engage. That collaboration is critical, because we as scientists cannot do this alone.”

For many physicians, the motivation is simple: some types of cancer still have no good treatment options. “There are cases in which a diagnosis is essentially a death sentence,” Jenkins notes. “That pushes doctors to engage with us, because they meet that desperation every day.”

Behind the science lies a human truth: for many patients, current therapies are inadequate. The hope is that by reprogramming the body’s own killer cells, a new chapter in cancer treatment may be within reach.