A conversation over lunch at Queen’s has led to a breakthrough collaboration spanning neuroscience, applied mathematics, and international research. What began as an informal exchange in the Senior Common Room between neuroscientist Dr David Menassa and applied mathematician Prof José Carrillo quickly revealed an unexpected overlap in their work on microglial development, the brain’s immune cells. Read their paper online.
Their serendipitous meeting set in motion a partnership with Prof Amanda Sierra’s team in Spain, ultimately uncovering a fundamental “switch” in early brain development shared between mice and humans. This cross-disciplinary effort, rooted in the everyday collegiate life of Queen’s, shows how ideas sparked over lunch can grow into research with implications for understanding neurodevelopmental and neurodegenerative disease. We asked the trio to tell us more about their work.
How did your chance meeting in the Senior Common Room at Queen’s lead to a research collaboration and published paper? How did you realise there was common ground in your research?
David: I met José over SCR lunch. We got chatting and slowly realised that the work I had done on human microglial development overlapped with his interests on applying mathematical models to predict changes of this cell’s phenotype during development. José had been collaborating closely with Prof Amanda Sierra from the Achucarro Basque Centre for Neuroscience in Spain and they were planning to get in touch with me regarding the human data I generated in a previous paper! It was a beautiful coincidence. We had our first meeting about the current work two years ago and I went on to validate some of the murine microglial findings in the human fetal hippocampus in samples from the Oxford Brain Bank.
Could you explain, in simple terms, what your collaborative study reveals about microglial development?
David: The brain has its own immune cells, called microglia, that help clean up waste and shape brain development. This study shows that early in life, microglia go through a key switch that is conserved in mice and human developments: first they divide a lot, then they slow down and become better cleaners/phagocytes. If this early growth phase is disturbed, the cells don’t develop properly and can’t clean the brain as well. This research identifies that this step-by-step maturation of microglia is controlled by changes in how DNA is packed and regulated inside the cells. These findings suggest there is an early ‘critical window’ where problems with microglial development could increase the risk of brain disorders of development and advanced ageing.
What does mathematical modelling bring to neuroscience that traditional lab work alone can’t? How did this model help reveal the developmental ‘switch’ in microglia?
José: Amanda and I began discussing her project online in early 2022 during the pandemic, through contacts at the Basque Center for Applied Mathematics. She approached me with the idea that, given the amount of data her group had collected, mathematical modelling could help advance their research. Her team had been measuring microglial cell counts in various regions of the developing mouse brain, focusing on a time window that began five days after birth. They noticed that microglial density initially increased, then plateaued, and eventually decreased as development progressed.
Through a series of conversations involving members of both our labs and by asking targeted questions about microglial proliferation and maturation, we converged on the idea that a model with two cell populations, proliferative and quiescent, might explain the observed density dynamics although we did not discard a one population model. Amanda’s team then realised they needed experimental data specifically on proliferative microglia.
While they worked on collecting that data, we developed three differential-equation–based models: two assuming a single microglial population, and one incorporating separate proliferative and quiescent compartments along with brain-volume growth and other effects. Our parameter-estimation analyses showed that, of the three, the two-population model best captured the experimental data. Remarkably, it predicted a switch from proliferative to quiescent behaviour between postnatal days three and five, much earlier than the time points Amanda’s lab had been examining, which began at day five.
In response, Amanda’s lab collected new data starting as early as postnatal day two. The results confirmed that the predicted transition indeed occurred between the third and the fourth day postnatal, precisely as the model suggested. That confirmation was the first “light-bulb moment” in our collaboration, nearly two years ago and opened lots of questions and experimental verifications of this fact as next steps.
Mathematical modelling, numerical analysis, and statistical parameter estimation are powerful tools that, when combined with precise data from experimentalists, can create a productive feedback loop, advancing both mathematics and the biological sciences. In our case, this interplay prompted experiments that might not otherwise have been pursued, and it also pushed us to incorporate time-dependent brain growth into our cell-population models. This synergy ultimately allowed us to develop a mathematical model with genuine predictive power, the “holy grail” of mathematical modelling.
It soon became evident that we should test the model on human data, if available, to see whether the same switching behaviour appeared. That’s when David entered the picture. Interestingly, the first time I heard of him was through a paper Amanda had sent us; he had led this work during his postdoctoral Fellowship in Southampton. When I looked him up, I discovered that he had just joined Oxford and, remarkably, he was at Queen’s. What a coincidence! We met at lunch, talked science, and the rest is history.
What makes this finding (the early proliferative-to-quiescent switch) so important for understanding brain development?
Amanda: The moment we realised the biological implications of the predictions of the two-population mathematical model was really a light-bulb moment, as José says. It is actually a really basic biological behaviour that was beautifully identified through mathematics, something we called “terminal differentiation”: cells divide, then synchronously stop dividing (what we call the switch). We can now ask the question of what drives this synchronisation and try to understand what is specifically happening in the brain and in microglia at this precise time point, and what happens when we disturb it.
In the paper we tested the prediction that impairing proliferation led to dysfunctional phagocytosis. You may think of microglia as a three-dimensional network of soldiers, perfectly positioned – not unlike the testudo formation of the roman army. To protect the brain, microglia need to be distributed throughout the brain. If you alter proliferation and have fewer microglia, they are not going to be effective defending the brain from damage and will not become efficient phagocytes.
Another aspect that we are exploring now is how to exploit what we have learnt from the synchronised switch to generate better alternative models to study human brain diseases. We are particularly interested in the so-called “minibrains”: human brain organoids that we culture in plates in the lab. As the real brain, they also need their army of defenders and we believe we need to control microglial proliferation within the organoids to have them mature as phagocytes.
Why is it significant that these findings are conserved between mice and humans? How might this work inform our understanding of neurodegenerative diseases or neurodevelopmental disorders? What are the implications of identifying an early “window of vulnerability” in microglial development?
David: It’s significant because it means this isn’t just a mouse quirk and it is a fundamental brain process shared across mammals, including us. Because mice and humans use the same “grow first, clean later” program for microglia, mouse experiments become much more relevant for understanding human brain development and disease. We can test drugs or interventions in mice with more confidence that they might work similarly in people. It suggests there’s a real early-life window in humans where disrupted microglia development could raise the risk of later brain disorders – and that window might be targeted for prevention.
How did you bridge the gap between applied mathematics and neuroscience, two very different scientific disciplines?
José: Talking about science openly, while recognising our own biases about other fields, and understanding that true excellence comes from different disciplines working together toward a shared goal is key to success. Social skills matter too; being professional yet friendly, communicative, and joyful helps create a positive work atmosphere. And, of course, good food and wine, whether in Bilbao or back at the College, helped spark creativity. 😊
True excellence comes from different disciplines working together toward a shared goal.
Amanda: I think we were extremely lucky to find José and his team, particularly Duncan Martinson and Carles Falco. The three of them have a unique combination of deep mathematical knowledge, curiosity for biology, and the rare ability to be able to explain what they were doing. On my side, I was very fortunate to have a talented, creative, and hard-working PhD student, Marta Pereira-Iglesias, who was able to manage a huge amount of experimental work. And then meeting David was the cherry on the pie. This was the most exhilarating experience in my scientific career, one that I really valued and enjoyed.
Were there any surprising insights or moments of misunderstanding that led to breakthroughs?
Amanda: This is an interesting story because we started off on the wrong foot. José’s team had been working on simple models but they did not fit the data. Then, they got captivated by one of our microscopy images where one could see microglia very close to the protein scaffolding that neurons use to migrate through the brain, and created the two-population model with the idea that there might be “free” and “bound” microglia, so they allowed two types of microglia with different behaviours. This two-population model fit perfectly will all our datasets, but by analysing the model parameters we learnt that the difference between the two populations was not that they were bound or free. They were proliferative and not proliferative, and this is how we discovered the switch. So, by being wrong but open to exploration we ended up discovering a completely unexpected biological behaviour that now makes a lot of sense.
By being wrong but open to exploration we ended up discovering a completely unexpected biological behaviour that now makes a lot of sense.
What have you learned from working outside your immediate discipline?
David: That although we may speak different technical languages, we get closer to the truth by teamwork and by interacting with disciplines when it does not seem immediately obvious to do so. This is scientific research at its best.
Where might this research go next?
David: The next steps will be to further investigate the mechanisms of this switch in more thoroughly collected human data and closer-to-human models, the minibrains we mentioned previously, test the mathematical model on these data, and to investigate the relevance of this switch in sections from specific neuro-developmental and neurodegenerative disorders in humans.
What advice would you give to other researchers at Queen’s who might want to collaborate across fields?
David: Collaborate as much as possible. Science is about teamwork and cross-over between disciplines is the way forward to getting closer to the truth in science. And surround yourself with a team of people you can have fun with!
What’s the best thing about doing this kind of cross-disciplinary science at Oxford?
David: The best thing is how easy it is to discuss with people who think in completely different ways but care about the same big questions. In one building you can go from talking to a clinician who sees patients with Parkinson’s, to an immunologist, to a mathematician who wants to model your data. Oxford is small enough that these conversations happen over coffee rather than through months of emails, but big enough that you have world-class expertise and excellent international collaboration in almost any method you might need. That mix makes it very natural to turn a biological observation into a quantitative model and then back into a hypothesis you can test in the lab or even in patients.
Oxford is small enough that these conversations happen over coffee rather than through months of emails, but big enough that you have world-class expertise and excellent international collaboration in almost any method you might need.
