A Conversation with Allan Balmain

December 10, 2015, by NCI Staff

Editor’s Note: Allan Balmain is the Distinguished Professor of Cancer Genetics at the University of California, San Francisco and a Fellow of the Royal Society. He has been publishing important papers on RAS biology for almost 30 years. RAS Central editors recently interviewed Dr. Balmain about his personal history and scientific career.

Can we ask you how you came to be at the UCSF? What was your path? Are you a Scot originally, is that correct? I am indeed, I was born in Scotland in a fishing village right up in the north, and grew up in Glasgow. It was quite a long tortuous route to get to UCSF. I did my Ph.D. in Glasgow in physical organic chemistry, and then went to Strasbourg to do a postdoc, so I lived in France for a while. That also was physical organic chemistry. We were irradiating molecules and they would undergo these rearrangements and strange reactions, and we'd use mass spectrometry, NMR, various chemical techniques and physical techniques to elucidate the structures. Anyway it was great but I think the start of the long journey toward UCSF was when I found out that some of the chemicals I was working with as a postdoc in France turned out to be related to TPA, a tumor promoting agent for which tetradecanoyl phorbol acetate is the chemical name. I worked on molecules related to TPA that were tumor promoters and I got a lot more interested then in why these compounds could cause cancer, or be involved in causing cancer. So that was the main impetus for leaving chemistry and going into biochemistry. I went to Heidelberg and worked in the Institute for Biochemistry at the German Cancer Research Center in Heidelberg for almost five years.

Did you see an advertisement for a postdoc that interested you, is that how it worked?

No, as usual with these things, it came through contacts. My first postdoc in Strasbourg was in a chemistry lab, with a wonderful French chemist called Guy Ourisson who had an ongoing collaboration with Erich Hecker, a German chemist in Heidelberg. Hecker's lab was actually the one that isolated TPA, back in the 60s or the very early 70s. So I went from Ourisson's lab to Hecker's lab. Hecker was doing biochemistry in addition to doing the chemistry, so that was why I went to his lab, and I got a fellowship from the German Alexander von Humboldt Stiftung to transition from one lab to the other. I did biochemistry there, but kept coming up against roadblocks that convinced me that I had to learn molecular biology and genetics.

I then went back to Glasgow as a group leader at the Beatson Institute, basically to learn molecular biology. They had recently cloned the globin gene and showed for the first time that globin was encoded by a single gene in the genome. Then in the early 80s RAS came to prominence as a human oncogene after the publication of the transfection data from Weinberg and Barbacid and Wigler. In Hecker's lab I had learned about mouse models of cancer induced by chemical mutagens such as dimethylbenzanthracene (DMBA) and tumor promoters such as TPA. At that point we hardly knew what a gene was, let alone how this chemical carcinogen interacted with genes to cause mutations. At the Beatson I started to try and understand how initiating mutations in cancer could be caused by exposure to mutagenic chemicals that are in the environment. Using the techniques pioneered by others, we found that the DMBA caused specific mutations in one of the RAS genes, establishing a connection between mutagen exposure and activation of RAS oncogenes.

I was at the Beatson, and then went to California to work with Onyx Pharmaceuticals after their Founder, Frank McCormick, left to become Cancer Center Director at UCSF. To be honest I was like a fish out of water, and it was not a very pleasant period because it rapidly became clear that there were too many restrictions and I would not be able to pursue the longer term goal of using mouse genetics to understand cancer progression. After that I came to UCSF and I've been there since 1999.

In the urethane story [urethane treatment of mice with one active KRAS allele caused lung tumors with Q61L mutations, while treatment of mice with two KRAS alleles caused tumors with Q61R mutations], it seems there must be an extremely powerful selection going on in almost the first cell division of these tumors. The presence of a wild type RAS allele seems to provide a tremendous selection away from when you just have one allele that's mutated.

That's exactly right and we are working really hard on trying to sort that out. But most of what we do is based on genetics and we haven't really pursued the biochemical aspects in as much detail as I would like. Presumably there must be a biochemical explanation ultimately but we don't know what that is. We were hoping that some of the people working on RAS dimerization might come up with something because that's a potential explanation, if RAS really does dimerize, because it provides a mechanism by which the wild type allele could affect signaling through the mutant allele. If there is no wild type allele and the only dimerization can be between mutant alleles, that may deliver a different strength of signaling and may contribute to this selection.

There is one other twist which hasn't been published which we are in the process of writing up just now, which is that when you treat the mice with urethane you would think that the chemical is the major determinant of the kinds of mutations you get. But what we've seen now is that if you take different strains of mice and you treat them with the same chemical, the tumors that arise in those different strains can have different mutations.

My graduate student Peter Westcott did a back cross between two strains, Mus musculus and Mus spretus, and looked at a lot of lung tumors in the back crossed population. In this population each mouse is genetically unique and it has different combinations of genes it has inherited from the Musculus (FVB) parent or the Spretus parent. Depending on which alleles the mouse has inherited it can push the mutations in the lung tumors in one direction or the other. In other words, with a pure FVB mouse treated with urethane almost all the mutations are Arginines. However in a Kras heterozygous mouse of the same strain, almost all the mutations are Leucines. That's published data.

In the back cross between FVB and Spretus, some individual mice have tumors mainly with Arginine mutations, whereas others may have primarily Leucines, although every mouse has 2 copies of wild type Kras to start with. The result depends on the intrinsic genetic background of each animal. We think each animal is inheriting polymorphisms that may control the processing or splicing of Kras in some way, so it's mimicking the effect of screwing up one allele, and perhaps resulting in a lower level of signaling or altered splicing. We don't know how it works but we've identified a couple of genetic loci, one of which is Kras itself, so depending which allele of Kras the mouse inherits it's more likely to get Arginine or Leucine mutations. But also there's another stronger locus on chromosome 10, not near any of the Ras genes, that can also contribute to mutation selection in each mouse. So it's actually much more complex and sophisticated than we had ever thought.

It's a challenge to convey the depth of these complexities to non-biologists, even very smart ones.

Yes - biochemists like to do things in tubes and they mix proteins and they get interactions, or they analyze cells growing on plastic dishes in vitro, and this approach is what has given us many of the signal transduction diagrams. But in vivo, while these interactions can happen, they may happen in one cell type but not in another, and they may go backwards in some cells rather than forwards, as the signaling is context dependent. That's where we really do need more complex in vivo models to begin to address these questions. You can clearly find something very important in vitro but when you go to the in vivo situation it can be really quite different.

They might reply "If we can find something that will save some lives we don't necessarily have to understand everything about how it works."

I fully agree, I certainly subscribe to that, but it really depends on where the individual feels he or she wants to be on that spectrum from basic science to patients. Each individual has to decide where they're comfortable. Some people are driven to get drugs and they want to do it in the quickest possible way and I understand that. But I think what we've learned from the last 5-10 years in particular is just how complex it all is. There is not a single targeted therapy that works on its own, as they all lead to resistance. That could have been predicted based on the network architecture of signaling, which is something that we will have to work on a lot more if we are going appreciate the complexity in signaling pathways in cells and in whole organisms.

Gleevec sometimes cures, does it not?

Gleevec was the first targeted therapy for chronic myeloid leukemia, and it is a wonderful drug, although some patients develop resistance. What I would argue though is that leukemias in general are very different diseases from, for example, metastatic pancreatic cancer. These two types of cancer are very different genetically, and many solid tumors have loads of point mutations and genetic events that we don't necessarily find at any significant level in most leukemias, so these are very different diseases.

What are the motivations that drive you in research?

The main incentive I've always felt is simple curiosity. I always want to know what the next step is and what's driving progression. For some people the incentive is "I want to find the magic bullet for cancer” for others people it's money, or being in a powerful position. When I am interviewing students I always try to distinguish between the students who want to do something and the students who want to be something. And I'm less impressed if someone says “I want to be a director”, or “I want to be a CEO” when asked what their goals are, and more impressed by the student who can tell me what biological problem she or he wants to solve and why and how they want to do it. So there are these different kinds of incentives that different people have. My own philosophy is to use the available information from different fields, starting with chemistry and physical chemistry and molecular biology and genetics and then computational biology. We try and use all of the tools that are available to try to understand complex cancer questions.

Tell us more about your computational biology activity. It doesn't seem like there would be much overlap with mouse models.

It actually follows on fairly naturally from experiments we did many years ago. When we started making mouse models what we found is that if we treat mice of a particular strain with carcinogens, they will get a certain number of characteristic tumors. However if you change the strain of mouse the results can be very different. Some strains just simply do not get any or only very few tumors, as they are genetically resistant. We started working with Mus spretus 30 years ago in Glasgow. It's a wild derived strain of mouse and genetically it's like a super mouse, as it is genetically resistant to many types of cancer. It can be irradiated and swim in carcinogens and they really don't get many tumors.

But the reason we got interested in computational questions came from the genetics. We were very naïve and we thought that if we breed these sensitive and resistant mice together and make some crosses that we'd very easily find the genes that make the Spretus mouse resistant or other genes that make the FVB mouse susceptible. It turns out that this is very difficult, because there are many genes that work together in combinatorial networks to induce susceptibility or resistance. It's very difficult to find them one by one so this is where we got interested in algorithms that can enable us to look at the combinatorial effects of different genes. We had a paper in 2009 where we developed network approaches to finding pathways linked to susceptibility, and this is still one of the main directions of the lab. Kras is in a very different computational network from Hras, in completely normal cells and normal tissues, and we are using these approaches to try and predict the function of genes. We published a paper in 2014 on using computational networks from completely normal tissue to predict gene function. It's an area that most people don't work in but there's a gold mine of information that I think has to be tapped.

Do you spend most of your lab time breeding mice and making new mouse models?

Well no, we do use mouse models but we also culture cells from tumors and have a large panel of cell lines representing multiple stages of cancers, including normal cells, benign tumors with RAS mutations, malignant tumors with the same RAS mutations that have progressed through other genetic alterations, and metastatic cell lines. We constantly flip from in vivo data to in vitro data, and the combination is very useful. So for example if we are looking at gene expression in tumors, often we do not know which cell types are expressing the gene - it could be stromal cells, it could be infiltrating blood cells, all kinds of cells that are inside tumors in vivo. When you purify cell lines and clone them you get rid of all of that, so there are no stromal cells, there are no immune infiltrating cells, and you can look at the picture specifically within the tumor cells. We flip back and forth between the data sets to ensure that the gene of interest is actually expressed in the cloned epithelial cells of the tumor. So we go back and forth all the time, depending on the question being asked.

Under the heading of advice for young scientists, if you were a new postdoc in a new institution and looking to make the biggest difference in your field, do you have any particular advice to give to a young scientist?

I feel that for a young scientist just going into the start of their career they should really try and get into the best place possible where they're going to be challenged. Don't be conservative, don't stick with what you did in your PhD, get out and do something different, but try to use the early post-doc years to identify a biological system that is going to be around for a while, and pick the big questions. The people I admire the most are those who have identified a system that they can work on for years and that enables them to tackle some of the big questions in cancer biology and cancer genetics.

Also, I try to make sure that postdocs coming into my own lab are not stuck in one very narrow field, that they have a fair bit of flexibility and freedom to look at different questions within a certain area, within a system. But I always try to get them to develop an interest and some expertise in computational biology. They don't necessarily have to be inventors of algorithms, but they should know how to use the information that's out there. A big difference between now and five or 10 years ago is the huge amount of data that are available, and most of the analyses that have been done are relatively superficial. There are answers to many interesting biological questions if they learn how to ask the questions properly and develop a basic understanding of computational biology and bioinformatics to apply to these large data sets. It will be a tremendous boost no matter which direction they go in.

Did you have a lucky break? Was there a particular interview that changed the course of your career?

I was very lucky in the first postdoc that I took. I went to France largely because I wanted to make sure I could learn French. I loved the French culture, so it was a non-chemistry decision to go to France. That being said I got into the lab of Guy Ourisson, a brilliant chemist, and he made a tremendous impression on me. I was only in his lab for 15 or 18 months before I moved to Heidelberg to follow up on the biochemistry, but he was a real inspiration. He was a classic old-style European professor in many ways, head of a huge lab, a huge department, but he was not the typical autocratic type that you used to find in many of these situations. I did a project with him that led to a couple of papers. The second of the papers, I had some ideas about the photochemical properties of a particular molecule, and did some theoretical work and a few experiments which worked out quite well. I wrote a paper with both of us as co-authors, and he was really supportive, loved the paper, but he decided not to put his name on the paper because he felt he hadn't contributed enough to the story. That was the first paper I had as a single author. There are not many people who would have done that in that situation. He was my postdoc supervisor, the project was started in his lab, although I developed it and wrote the paper. I just thought it was very generous and I have tried to maintain that kind of attitude in my own lab. It was a lucky choice of first post-doc.

How do you like to spend time when you're not in the lab?

I do a lot of biking ... I have a pretty fancy bike, a carbon fiber frame, the Madone Trek series, pretty nice lightweight bike, it's about 16 pounds, it's sitting in my office right now.

Do you ride it to work?

I ride it to work as often as possible. I park at the north side of the Golden Gate bridge and bike over the bridge and down to the bay and along the bay and into the lab. There are quite a lot of bike paths on that route, and I get time to think about science and the day ahead.

Are you on the road a lot, giving talks or consulting, that sort of thing?

I've cut down on travel. I have had four children, but there's a 20 year time span between the oldest and youngest, so I've had children to deal with throughout my whole career. You don't get a lot of time to do many other things. I always wanted to spend more time playing golf but I never quite made it, first of all because I'm just not very good but also because I never had enough time. Weekends were taken up with the family and their activities so getting that balance is always challenging. Science is demanding, and you have to be passionate about it, but if it has come down to a crunch the family comes first and sometimes things have to take a back seat on the work front. That's a big challenge, but it’s even more of a challenge I think for women, trying to get that balance right. My wife is also a scientist, she's also a professor at UCSF, so trying to juggle everything, her responsibilities and career aspirations with my own and with the kids demands ... We're never bored.