Center for Cancer Research, NCI
Name: Louis Staudt
Position: Senior investigator, Center for Cancer Research, NCI
Background: Postdoc, Whitehead Institute — 1984-1987
MD, PhD, immunology, University of Pennsylvania — 1982
BA, biochemistry, Harvard University — 1976
Though trained as a physician, Louis Staudt "got the science bug" during his postdoctoral fellowship with David Baltimore at the Whitehead Institute and chose to pursue a career in medical research. That career brought him to the National Cancer Institute, where he studies the molecular basis of human lymphoid malignancies.
Last month, he and his colleagues at the NCI published a paper in Nature describing a loss-of-function RNAi screen that uncovered a new target for a subgroup of diffuse large B-cell lymphoma and may prove useful in identifying new kinds of cancer targets.
Recently, Staudt spoke with RNAi News about his research.
Could you talk about your lab at NCI and its focus?
We've been working for several years, since the late '90s, on the molecular diagnosis of human lymphomas. The impetus was that these are a heterogeneous set of diseases and the current diagnostic methods, which were largely microscopy at the time, were inadequate to sort these out. The idea then was to use gene-expression profiling as a technology that could give us clearer diagnosis and definition of these diseases. So a large part of my lab over the last eight years or so has been focused on that issue.
What we found out of that approach is that certain diagnostic categories of lymphoma are actually composed of several molecularly distinct diseases that nonetheless look identical through the microscope. These diseases arise from different stages of normal B-cell development; they utilize different oncogenic pathways and different oncogenic events, and they have quite different clinical behaviors — some being more curable and some being very much less curable.
Out of that general finding we became interested in the molecular targets that might be specific for each of these types of lymphoma [and] that could be potentially attacked therapeutically. A lot of our work has focused on the most common type of non-Hodgkin's lymphoma called diffuse large B-cell lymphoma, or DLBCL. In that diagnostic category we think there are a minimum of three different diseases that can be easily dissected out by gene-expression profiling. [The first is] called germinal centre B-cell-like DLBCL, and that's the most common type. The second type is called activated B-cell-like DLBCL, or ABC, and that's the second most common type. Finally, there is the primary mediastinal B-cell lymphoma, or PMBL, which is [the least common].
We found that two of these types of lymphoma, the ABC type and PMBL type, had activity of a signaling pathway, the NF-kappaB pathway, that was constitutive — that is, it was on all the time. We could show that if we interrupted that pathway we killed the cells. The same pathway had absolutely no influence on the survival of the GCB type of DLBCL. So we began to get the idea that each of these types of lymphomas has very unique signaling pathways that regulate survival and proliferation. We wanted to take a more general approach to identifying these pathways. Furthermore … we were interested in the mechanisms that gave rise to this constitutive NF-kappaB activation because the NF-kappaB pathway is very promiscuous — many things can activate it and we didn't know exactly [what] was at work in our lymphomas. That was the starting place for our Nature paper.
Over the last two years or so, we've embarked upon an effort to use an RNA interference library approach to identify molecular targets in cancer, which we euphemistically call the Achilles' heel screen because the aim is to find that aspect of the cancer cell that is most vulnerable to attack. In this approach, we've made a library of retroviral vectors that express small-hairpin RNAs that can mediate RNA interference when introduced into the cells.
Thus far, our library is targeting 2,500 human genes with three shRNA vectors for each gene — so roughly 7,500 vectors have been made. We then used those vectors … which are inducible … to identify pathways for proliferation and survival in cancer cells. [An] important feature of the vector is a way of monitoring each vector in a population of cells. That is a so-called molecular barcode where a 60-basepair random sequence has been inserted into each of the 7,500 different vectors, which we've sequence to know exactly which random sequence is associated with which shRNA. We can use a DNA microarray [with] the barcode sequences to monitor the abundance of each vector.
How did you select the genes that you targeted with the shRNAs?
Obviously, this is a sub-genomic library that represents roughly one-tenth of the human genome. We wanted to make a lean, mean version of the genome, and therefore we targeted all protein kinase genes, all PI3 kinase genes, all deubiquinating enzymes, a vast number of NF-kappaB pathway regulators, and many known oncogenes and tumor suppressors, as well as a discovery component where we have looked at a variety of genes that are simply differentially regulated among lymphoma types in the hopes that some of them may give us specific targets. It's got a good portion of what we might call the druggable genome. And we're intent on building it larger over the next couple of years.
What were the key findings of the screen?
Immediately we found that shRNAs that targeted genes in the NF-kappaB pathway were selectively toxic for the ABC type of DLBCL and not the GCB type of DLBCL. Of course, in part we knew that — we knew one of the shRNAs targeted the I-kappaB kinase, which is the key kinase in the pathway. But the new finding in the paper, at least in respect to our lymphomas, was that there was an upstream pathway that was responsible for the activity of the NF-kappaB pathway called the CARD11 pathway. That CARD11 pathway encompasses three of our shRNAs: one for CARD11, one for MALT1, and one for BCL10 — they all function in a very well worked-out pathway that in normal lymphocytes lead to activation of the NF-kappaB pathway following antigen stimulation of the lymphocytes.
That was a pleasing sort of biological outcome, but I think the reason [this research] was published in Nature was there are some broader implications of the work. The screens could probably be productively done in almost any cancer cell line, to start with, and we intend to do many cancer cell lines. The idea would be to identify which pathways control proliferation and survival in each type of cancer. In that way [they] could yield a sort of functional taxonomy of cancer that would reveal the drug targets for each type of cancer. In fact, we think some of this functional taxonomy would cut across current diagnostic categories, and this might be the type of molecular knowledge you would need to know if you wanted to choose the appropriate therapy.
The second broad conclusion that is a little less obvious is that we believe we can identify with this method a new set of targets in cancer that are beyond those that are traditionally considered — the oncogenes and the tumor-suppressor pathways. The reason we believe that is that it could be the case, and we think it may be the case, that the cancer cell may rely on a pathway and genes for survival that are not themselves mutated or otherwise disregulated, but simply are a feature of the type of cell from which the cancer derived. The normal mechanism that controls apoptosis and proliferation in the normal cell may also control those in the malignant cells. These never would have been picked up by normal methods in cancer biology because they, by definition, don't have mutations in the genes, they are not translocations of the genes, or amplifications of the genes. But this RNA interference technology can pick them up because it simply asks, 'What happens if I knock down the expression of a particular gene to the survival or proliferation of the cancer cell?' So we believe that there may be a new set of targets for therapeutic development.
So what is next for you? To do wider screens?
I'd like to do screens in several cell lines representing each type of human lymphoma, to begin with, but also extend it to some of the solid tumors just to see what kind of common features there could be and what kind of new features there may be. I think we will find pathways that people may already have known about. On the other hand, I think there is a discovery potential here where we could find new things that simply haven't been revealed before by other methods.