Current Position: Lecturer, interdisciplinary Biology Laboratory (iBLab), Division of Natural Science, Graduate School of Science, Nagoya University, Japan
Research Interest: Why the same process, running in different people, ends up somewhere different?
Favorite Cell: Hematopoietic stem cell
What inspired you to study mathematical immunology? How did you discover mathematical immunology?
I was interested in biology long before I went to university, and what interested me was not any particular organism or molecule. It was the fact that living things are made of many parts, each of which is simple enough to describe on its own, and that together those parts produce behavior that none of them has by itself.
I came into this field from biology rather than from mathematics. As an undergraduate I was not good at experimental work, and I could not reliably run a gel. What appealed to me about mathematical work was that a method written entirely as equations and code is written down completely enough for anyone else to repeat it.
In my final undergraduate year, I joined the Mathematical Biology Laboratory in Kyushu University, where Dr. Shingo Iwami and Dr. Hiroshi Haeno, who were then an associate professor and an assistant professor, gave me a concrete problem to work on. The problem was to model how a single hematopoietic stem cell differentiates after transplantation. I had assumed that mathematics would be used to describe biology after the experiments were finished. What I found instead was that it decided what the data could and could not tell us. I have worked on that problem ever since, and we only finished the paper last year.
My interest settled on immunity because it is the clearest example I know of a system whose task is to absorb disturbances from outside and keep working. It detects, responds, resolves and remembers, and it usually manages all of that without destroying the body it defends. What holds my attention is not any single mechanism inside it, but the question of how the whole system stays functional at all.
That question takes me back to the problem I was first given. An immune system is not a fixed set of cells. It is a population that is being replaced continuously, and every lymphocyte and myeloid cell in it comes from a hematopoietic stem cell that had to decide what to become. The question I was handed as an undergraduate, which is how a single stem cell divides its output among lineages, is therefore also a question about how immunity is maintained, and not only about how it responds. That is still the part I find more interesting.
I also think this is the right moment to ask such questions quantitatively, because we are only now able to observe living systems at the resolution that these questions require. That is where mathematical modelling and machine learning are useful.
What are your research interests?
I want to understand why the same process, running in different people, ends up somewhere different. I work on this in some settings, for instance, infection in patients and blood cell production, and infection is where most of my own work has been. The same virus, and often the same drug, produces very different courses in different people, and one person clears an infection within a few days while another is still shedding virus weeks later. At the moment we cannot say in advance which person will be which, or what it is about them that makes the difference. The question matters in practice as well, because the answer is what decides when treatment should start, who actually benefits from a drug, and how a trial should be designed.
Where I would like this to end up is with the individual patient. At the moment, differences between people enter the models we build as a distribution, so we can say how much people differ but not who differs or why. I would like the parameters of a model to be tied closely enough to what we already know about a person that the model can tell us what is going to happen to that person. It should also tell us which measurements would have let us see it coming sooner.
The same question is what draws me to cell differentiation. There the outcome is which lineages a stem cell produces rather than whether an infection is cleared, but the problem has the same shape, because the decision itself is never observed and has to be reconstructed from what it left behind. That is why I also model hematopoietic clonal dynamics from single-cell lineage data.
Favorite immune cell to model, and why?
My answer is the hematopoietic stem cell, which is a slightly odd choice, because it is not an immune cell but the cell that keeps producing them. What interests me about this cell is that it is a unique stem cell that passes through various states on its way to producing a wide variety of lineages. the fate decision of a single cell cannot be seen at the moment it is made, and only becomes visible much later, in the sizes of the clones that the cell has left behind. You are always reconstructing a decision from its consequences, which is the same kind of problem I deal with when I work on viruses.
If you could unravel any mystery of the immune system, what would it be?
The thing I would most like to understand is how the immune system stays stable when the cells it is made of are being replaced all the time and the challenges it meets are never the same twice. In most people it keeps working for decades. That kind of stability is not contained in any single cell. It appears out of the interaction between many of them, and I would like to know what actually produces it.
The way to get there, I think, is to describe what happens inside the body in quantitative rather than qualitative terms. If we can write down how quickly cells are produced and lost, how fast a response builds up and resolves, and how the output of individual stem cell clones changes over a lifetime, then the stability we observe stops being a general statement and becomes something we can put numbers on. The same applies to the ways it fails. A chronic infection, a response that does not stop in time, and the slow expansion of a mutant blood cell clone are all cases where a balance that normally holds has been lost, and each of them looks different once it is written in numbers.
What I find attractive about this is that the answer would not stay inside the model. Once we can say which quantities keep the system stable, and which of them change first when that stability is lost, we also know what to measure and when to act. That is the information we need in order to step in while the balance can still be recovered rather than afterwards. A quantitative account of how immunity maintains itself and how it fails is therefore not only an explanation. It is also the basis for choosing treatments and designing studies that work for an individual person rather than for the average one.
I do not expect there to be a single answer, because the stability has more than one source. Part of it is dynamics, and a model can capture that. Part of it is immunology that we have not written down yet. And part of it may not be the kind of thing that a model like mine can represent at all. Being able to say which is which would already be a large step, because it tells us where to look next instead of absorbing everything into a random effect without comment.
What advice would you give to your younger self or young researchers?
When I was assigned to a laboratory in April of my fourth undergraduate year, I told the professor who led the group at the time, Prof. Iwasa, that I wanted to go on to graduate school. He asked me what I wanted to do there, and I had nothing to say. He told me that there is not much point in going if there is nothing you want to do. I do not think that was unkind. Graduate school is a place for people who have something they want to do, and asking that question early is part of taking responsibility for someone else’s life.
What I would say to my younger self is that the thing you need at the start is not a fixed research subject. What I work on has widened a good deal since then and it is still changing. What matters is having one thing you want to find out, or something you believe about how the work should be done, that runs through everything and stays there. People who have that seem to grow on their own, because it tells them which problems are worth their time and which ones are not. I did not have it at that point, and it took me a while to find it.
The need for it does not go away once you have a position, and it becomes more practical rather than less. Even in a small group, deciding what to apply for, what to write up and what to turn down are all decisions you have to measure against something. Without that core I find that I hesitate and stop making progress. So the advice is to be able to say what you would like to find out, even if you say it badly, and to say it out loud to someone who will argue with you about it.
When I could not answer that question myself, two people in the same group gave me a problem to work on rather than a lecture. I am still working on it.
What is your favorite part about your job?
The part I enjoy most is the moment when something nobody could observe directly turns into a number. A stem cell makes its decision without being watched, and a person is infected before anyone starts measuring them, so in both cases the event itself is over by the time the data arrive. Getting from what was left behind to a quantity that can be compared between individuals, and then seeing that number hold up when a colleague tests it with a new experiment, is the part of the work I would not want to give up.
Another thing I enjoy is that the same question can turn up again in a completely different system. The mathematics I use for virus dynamics is close to what I need for blood cell differentiation, so when I move between them I am not starting from nothing, although the biology has to be learned again each time and I learn it from the people who do the measurements. Working alongside them is a large part of what makes the job interesting.
I also enjoy that this work takes me to other countries. I like how many kinds of beer there are and how much they change from one region to the next. Ordering whatever is made locally turns out to be an easy way into the history of the place, because the beer a region ended up making usually has a reason behind it, and asking about that reason is a good way to end a day of talks.