Interviewers: Takashi Mikuriya, professor

Well, I was born in 1953, the year Watson and Crick published their famous paper on the double helix. When I was studying biology in high school, I became very interested in DNA. As a student in the school of medicine at university, I did research on the genetic causes of disease. After graduation I searched around for a research topic and got interested in atherosclerosis, so I began to research on cholesterol.

Researchers begin with a theory and then proceed from there with their research. But nobody knows better than a researcher how unreliable theories actually are. In 1990, when I was in the United States, I succeeded in cloning a gene that codes for storing cholesterol in fat tissue, and I published an article about it in the journal Nature. Some people misinterpreted my findings as a discovery of the gene that causes atherosclerosis. But based on my experiments, I knew that the identified gene was not the direct cause of atherosclerosis. I was surprised at how my research got oversimplified.

After returning to Japan, I did some more experiments using mice. Atherosclerosis generally doesn't occur in mice, so I crossed a mouse lacking the gene I discovered and two other types of mice with genetic irregularities that make them more susceptible to atherosclerosis. As a result, on one hand, the mice that were prone to atherosclerosis improved-again I published my findings in Nature. On the other hand, atherosclerosis worsened in the mice lacking the gene that codes for cholesterol storing. Generally this way of thinking does not work; what would happen if you take gene A, which makes atherosclerosis worse, and combine it with gene B, which makes atherosclerosis better? It's easier to publish an article in a world-renowned journal like Nature if your work is easy to understand and confirms preexisting theories. It's harder to publish complicated findings about a phenomenon that flies in the face of popular opinion. This sort of "reductionism" is especially prevalent in the scientific community in Europe and America, but it won't bring us even one step closer to finding the cause of atherosclerosis. So when I moved to RCAST, I stopped doing "reductionism"-type research and began taking a careful look at the relationships between elements. That's how I hit on the notion of "systems."

When I moved to RCAST, I stopped using mice on my research. In one sense, my experiments on altering mouse genes create black boxes. I wasn't satisfied with just focusing on one gene by itself-I really wanted to see all the genes together. As luck would have it, in 2000 the human genome was decoded and mapped, giving us a picture of all the genes in the human body. The total number is huge but at least it's finite, so we set about collecting exhaustive data that would hopefully enable us to identify systems.

Around the same time, Systems Biology was developed-mostly from research done in the US-which allows scientists to make biological predictions using computer models. Now, this is a kind of high-level reductionism, which basically says that everything can be solved in terms of differential equations. My own view is that it seems strange to talk about solving differential equations when you don't even know whether what you want to study can be expressed as differential equations. The popular view now on dealing with any complex system is to choose a model and then plug whatever phenomenon you're dealing with into it. Now, in contrast to this, what we at RCAST call Systems Biology and Medicine is not simply about positing systems. The system as a whole may be too complex to fully understand, but by intervening in a biological process, we can see whether a given phenomenon occurs the way we predicted it would, or whether something different happens. If you're just going to look at phenomena and create models, you could come up with any number of different models. What we do instead is to intervene in an actual biological process and then alter and adapt our model based on the results, or identify a system-we take that as the most important first step.

What I've become most acutely aware of is that the special characteristic of a system is that in it, biological phenomena are repeated over and over with incremental changes. When an activating stimulus is introduced, it triggers a signal that terminates it-what we call an autotermination signal. Usually activation will cease at that point, but if a continuous stimulus is introduced, it sets up a cycle; when the stimulus is repeated for a certain amount of time, the number of activated factors increases. This increases the number of inactivating factors, and activating effect is suppressed. But when activated factors become fewer, the number of inactivating factors also decreases. If the stimulus continues, then the number of activated factors increases again-this is what we call oscillation. Oscillation is a basic biological principle. For example, it determines the number of our vertebrae-if a certain protein activates and inactivates itself seven times-seven oscillations-then we get seven vertebrae. That is to say, the periodicity of oscillation functions like a clock.

In the case of atherosclerosis, the problem is that white blood cells adhere to the arterial wall and pass into it. Superficially, the way they adhere appears the same, but we've discovered that the phenomena occurring during the first oscillation and second oscillation are in fact different. We believe that when genes are activated during the first oscillation, what's more important than the fact that the oscillation occurs, is that a system of oscillation with incremental change is established. Cholesterol lowering medications are designed on the presumption that when a gene oscillates, you can introduce a certain stimulus and such-and-such will occur. But what we're doing is taking it one step further. In other words, we're trying to predict what phenomena will occur after the gene is activated, and design a medication accordingly. This could reduce side effects. We believe such drugs would have a high social value, so we're trying to use this process to predict mechanisms one step beyond gene activation to create new medications.

Simplifying one's predictions is fine, but in fact medications work by activating genes, so if its effect is prolonged, genes will continue to oscillate and trigger various phenomena. In such a case, the first wave might well occur just as predicted, but the second wave will occur differently. You end up considering such things when you try to understand the relationships between elements. The problem is, with the third wave and beyond, so many elements come into play that it's impossible to calculate or predict-right now two waves are the limit, I think.

The impetus behind the book was what you might call "pop evolutionary biology"-for example, when people talk about things like the "love gene" or the "criminal gene." The public tends to think genes are responsible for everything. But as one should realize from the fact that we talk about "gene activation," genes are in fact passive, not active, and are only one part of a system that is subject to control. There may be such a thing as an "intelligent system," but there's no such thing as an "intelligent gene"-that's what we believe. But in society at large it's exactly the opposite; genes are the active agent and people are just vehicles. I've even heard people theorizing about "selfish genes" and that sort of thing. Anyway, a while back I was talking to my old friend Kaneko (Professors Kodama and Kaneko attended junior high and high school together). He was saying that even in economics people hold some very simplistic views, such as the performance principle. He argued that once you boil down a complex process to one mechanism or element, you threaten the integrity and sustainability of the entire organization.