If I could summarize UCLA professor Dr. Alexander Hoffmann’s seminar at the Buck Institute in one sentence, it would be: “signaling within a cell is a dynamic process, therefore to truly understand signaling we need to understand its dynamics.”
At any given time, every cell in our body is constantly being bombarded with signals, and these signals are generating responses in these cells. To make matters more complicated, these signals can come from within the cell (intracellular) or from outside the cell (extracellular). Signals tell the cell to divide, to grow, to notice something strange in their environment, to change their shape and, depending the situation, to die. These signals don’t work as typical on/off switches. The process is much more complex. Signals are dynamic, in the sense that it’s not just the presence or absence, but the strength, duration and frequency of signals that generates a particular response in a cell. These dynamic changes are so important that altering their normal course of events has detrimental effects on a cell’s health and consequently on the whole organism.
In 2002, during his postdoctoral training in the lab of the Nobel Prize winner Dr. David Baltimore, Dr. Hoffmann together with Dr. Andre Levchenko as co-first authors published a Science article in which they generated a mathematical model describing the temporal control of NF-κB activation (Hoffmann et al. (2002) Science 298: 1241-1245). The NF-κB signaling pathway plays multiple roles in the cell such as controlling growth, survival and apoptosis; however, its most commonly studied role is in the immune system. Dr. Hoffmann and colleagues visualized this signaling pathway as a model in which an input (signal coming from outside the cell through its cell-surface receptors) was generating an output (changes in gene expression). Using this mathematical model combined with experimental data from mouse fibroblasts, they could decipher which components of the pathway were needed for the output they observed and this way understand the dynamics of the NF-κB signaling module.
Currently, Dr. Hoffmann runs the Signaling Systems Laboratory at UCLA where he and his team are taking a systems biology approach to understand signaling dynamics of NF-κB in different cell types, disease, stress and innate immunity response (among many other interesting projects). The first part of his talk at the Buck Institute was focused on the work done by former postdoctoral student Marcelo Behar and was published in Cell (Behar et al. (2013) Cell 155: 448-461). For this project they use their knowledge of the NF-κB signaling module to theoretically predict how different drug targets modulate this pathway. The relevance of this study is that by understanding the dynamic changes you can then select specific drugs targets and optimal time for administration to minimize pleiotropic and undesired effects.
Next he described the work done by a former graduate student Paul Loriaux focused on using signaling dynamics to identify biomarkers. Dr. Hoffmann’s lab’s findings suggests that the identification of biomarkers is more promising if we consider the dynamic changes of these molecules, such as protein abundance and protein turnover. Hence his suggestion that biomarker identification will be more productive if searched in unfixed tissues and using an experimental set-up that considers taking into account their dynamics.
SAGE sat down with Dr. Hoffmann to learn more about his research and to ask him about his interesting career path.
Q: Before constructing your model, do you do any filtering of the data?
AH: Absolutely. For this kind of modeling is a very detailed representation of the reactions of the molecular mechanism of the cell. The information that we use comes from two kinds of data. One type of data is from biochemical analysis where people have measured the affinity of two proteins interacting or the kinase activity, which is all in the literature. This is not high-throughput data. It is data that is very detailed experimentally. So that gives us many of the kinetic rate constants. If we are not considering genetic variation those constants will be the same in every cell. What is different in every cell is the abundance of proteins and the expression levels. So which are the particular cells that we are interested in? We then have to go and make to measurements ourselves. So in the filtering you have to evaluate whether the published experiment was done properly and interpreted correctly.
Q: Are you looking into NF-κB dynamics in different model organisms such as fruit flies, worms, etc.?
AH: For NF-κB, we looked at Drosophila, and signaling in this organism is not as dynamic as in the case of mammals. The system works differently in flies. NF-κB is a negative regulator but doesn’t provide very strong negative feedback, so we are wondering why that is the case. Why in vertebrates do you have very dynamic NF-κB signaling? Even though NF-κB is very important in Drosophila, it’s just not as dynamic. We thought that has to do with the role of NF-κB in the cell. In vertebrates, you have the innate and adaptive immune system. The adaptive system relies on cytokine communication. Therefore mammalian cells have to respond to two kind of signals: pathogens, which are the “enemies”, and cytokines, which come from neighboring cells “friends”. Yet, both signals activate the NF-κB pathway. The cell has to be capable of distinguishing between the two signals so there could be two different dynamics: the pathogen dynamics and the cytokine dynamics. That’s a very simplified view but you can imagine that when NF-κB is transient, it’s probably cytokines, and when is long-term, it’s probably pathogens. Drosophila doesn’t need that distinction. That’s one of the ideas that we discussed in a recent paper. My goal in the future is to study the physiological relevance of the dynamical code to humans.
Q: Are you planning to do primary cell cultures to study NF-κB signaling dynamics instead of using cell cultures of transformed cells?
AH: Absolutely. To probe the dynamics of a molecular network, you need live cells. The first step that we took in this project was to make retroviruses encoding fluorescent reporters to infect cell lines and select the cell lines to study the dynamics. We found that the dynamics were really interesting. Thus, we made the investment and made mice that are knock-ins for all these different regulators. Now we can do primary cell lines from different cell types (e.g. macrophages, endothelial cells, hepatocytes, glial cells, among others) and look at the different conditions to understand how the dynamics change in health and disease, old and young, etc. So now that we have these mice is going to be very exciting!
Q: About your future research directions. Your research has shown very nicely how dynamic the NF-κB pathway is. Are you thinking of studying/modeling the interactions between pathways?
AH: That’s the job of field not only my lab (laughs). We can only do so much. We hope that the approach that we take engages other people so they focus on the other pathways. We do want to expand, and there’s also more work to do with NF-κB because so far we have only shown the dynamics of one NF-κB family member. However, there are 15 NF-κB’s, and they are all interrelated. And that’s ongoing.
The other major pathway involved in the immune response is the interferon network. The interferon response is similarly interesting in that it has several regulators and these activate coordinately or even the same genes as NF-κB. However there’s a clear distinction, as far as I can tell, which is that NF-κB has numerous negative-feedback loops to shut the signaling down. Interferon signaling contains positive-feedback loops, so it’s a completely different dynamic. The way you shut down the interferon response is by killing cells. Even though is a very different response, the pathways are still coordinated. So, that’s something I want to look at.
The other thing I would like to do is to look at the biological consequences of NF-κB. Now we are looking at the pathway upstream NF-κB, but how do NF-κB dynamics gets encoded? So we are looking downstream NF-κB, meaning gene expression because is the gene expression that leads to biology.
Q: About the second part of your talk, showing that dynamic changes need to be considered to identify better biomarkers. How do you visualize this data to be used?
AH: Our work remains theoretical at this time. We all appreciate that biomarkers are useful but their predictive power is limited. For instance, you can have a set of genes, and they have 70% success in making correct predictions. There are two conclusions you can make, either there are other unidentified genes that can explain the other 30%, or maybe you are not looking at the right thing; maybe instead of looking at the abundance of the protein you need to look at the turnover of the protein. The work I presented tries to raise this question.
Q: About your interesting career path…You have a Bachelor’s degree in Zoology and Physics, a PhD in Biochemistry and Molecular Biology researching transcription factors, then postdoctoral training working on HIV and mathematical modeling of NF-κB signaling. Did you wanted to be a scientist working in system biology problems when you were growing up?
AH: I had no idea I wanted to be a scientist. My mom always thought I could be an architect because I was always building things with LEGO blocks (laughs). I just enjoyed science, and I am excited about new problems. New problems force me to think about new things. I did my Bachelor’s in physics because I had a great high school teacher, so I was very excited about physics, and I still am. But as an undergrad, I spent a summer in a biology lab, and I didn’t know anything about biology at that time. Within 10 weeks, I went from making Sodium Chloride solutions to cloning a gene. I was so excited about the fact that you needed to know so little to do so much in Biology. Whereas in physics, it seemed like I needed to study another 8 years to come up with something creative. So I switched fields.
During my postdoc, I had this dynamic data set that I didn’t understand how to go further with it. Luckily, I went to a seminar taught by John Tyson [a computational biologist] who was studying the cell cycle using differential equations, and I thought “learned that 10 years ago, but I still remember” maybe I could use differential equations too!
I then found a purpose for my circuitous educational history. The combination of physics and biology gave me my own approach to problem solving. Anyone still in the training process that has different experiences and training from different mentors, should find the opportunity to combine some of these experiences in a unique way to make a unique contribution. To be able to do this is very gratifying.
Q: Should people in biology learn bioinformatics even when they are not doing research in this area?
AH: I think it is critical for any biologist to be able to deal with data because biologists are better at interpreting the data than a professional bioinformatician. We are always looking for biologists who are strong on the experimental side and also strong in computational analysis methods.
The reality is that we generate much more data now than when I was in grad school. So the skill set that today’s biologists require are very different. In my mind there’s a shift. The molecular biology revolution was about developing tools to measure, and the challenge was to get the data. Now it’s not so much about getting data. You can get data using kits you can buy, the challenge is to analyze it.
For more on Dr. Alexander Hoffman and his work check out the UCLA Signaling Systems Lab website.