Basic Medical Research Award
Interview by James Darnell
Lasker Laureate James Darnell of The Rockefeller University interviews Robert G. Roeder, the 2003 Albert Lasker Basic Medical Research Award winner for his pioneering studies on eukaryotic RNA polymerases and the general transcriptional machinery which opened gene expression in animal cells to biochemical analysis.
Date of interview: September 19, 2003
Darnell: I'm Jim Darnell, professor at The Rockefeller University, and I have the distinct pleasure and privilege this morning of having a chat with my colleague, Bob Roeder, who is the winner of this year's Lasker Award for scientific achievement. The special friendship with my colleague, Roeder, began a number of years before he moved to Rockefeller 20-odd years ago.
We're going to have a chance to hear him tell us how he began his scientific career, and we will introduce you through his eyes to some of the events that have occurred that he feels are significant during his more than 30 years of studying how cells make RNA.
Bob, I know that we've shared a lot. I've been diddling with RNA ten years longer than you have. It was a very important time in the late '60s, when your first important publication came out. And we certainly want to get to that, but we share something else: we both came from very small towns, as we've talked about many times, and I think people who may see this might be interested to know how a young fellow born in Boonville, Indiana, got started in major medical research. What happened in Boonville that interested you in science?
Roeder: Well, actually, nothing happened in Boonville that interested me in science because I moved from there at the beginning of high school. But I will tell you a little bit about Boonville. It was a farming community founded, not by the famous Boone, but by his brother Radcliff, and I had the standard education. Actually, the hardest education was on my father's farm, where I learned how to work hard, but I did enjoy school. After my freshman year in high school, we moved north to a town called Jasper, Indiana where I completed high school. And that was where I began to take an interest in some aspects of science. I was interested in mathematics and chemistry. Biology wasn't then of much interest because it was so descriptive and involved too much memorization. I much preferred chemistry — doing things with my hands and mixing things together. But that basically was my first exposure to science.
In terms of a future career in science, I hadn't thought much about it until in the end of high school, when I was struggling to think about something to do. My father was very determined that I should stay on the farm, so much so that he had nothing to do with my college education. In fact, he did his best to keep me on the farm. But I escaped to Wabash College, which was then, and is now, a men's school and had fabulous science departments in biology and chemistry in particular, and in math. I started out as a math major, but it became clear I wasn't a genius in math. Many of the kids, from private schools particularly, had advanced training in mathematics. Fortunately, I developed an interest in biology and chemistry as a result of the excellent teaching by people there.
Darnell: Well, we don't want to make a sociological tract out of this but I'll ask you one question. What do you think would happen to a youngster these days, who was born in Boonville and moved to Jasper for high school? Would he or she end up in science? I've often wondered this from my own background also. What's happening in mid-America, small-town America, in terms of education, do you know?
Roeder: Well, regarding my own high school, I know that it's much more modernized in terms of teaching biology and molecular biology. In fact, one of the people I played football with in high school — and I can tell you I was better known for playing football in high school than I was as a scholar, even though I was at the head of my class — ended up getting a degree in biology and teaching biology/biochemistry in our high school, and was very interested in having me come back and talk on occasion to the students. I only wish they had been so up-to-date when I was there.
Darnell: Good so we're not going to lose the next generation of Roeders. Good.
Roeder: I hope not.
Darnell: All right, let's move on beyond your successful time at Wabash College to the time when you started to get your Ph.D. I know that you went to Illinois to enroll in biochemistry to begin with. And you've told me that there were famous professors that you had a chance to meet there. But let's skip that for a second, because it was after your mentor, Bill Rutter, moved to the University of Washington that things began which are relevant to what we'll talk about the rest of our time.
Darnell: Tell me what it was like in Rutter's laboratory. Reflect on what you and Rutter and others knew and thought about how animal cells make RNA at that time.
Roeder: Well, Bill's lab was in a transition at that time. He was better known as a classical enzymologist, particularly for his work on aldolase. In the last year at Illinois, when I joined his lab, he had just come back from a sabbatical in Cliff Grobstein's laboratory at Stanford and had begun to work on the pancreas system as a developmental model. In terms of looking at developmental programs, he began by developing sensitive assays to measure the levels of all of the pancreatic enzymes and secreted proteins. He also had visions of studying RNA synthesis as well, although nothing was going on in the laboratory at the time.
However, thinking that down the track he would study transcription problems in the pancreas. Bill gave me an opportunity to join the lab and, hopefully, to find a problem that would ultimately end up at that junction. The lab was a very exciting place. Bill was one of the most dynamic and stimulating people I've ever encountered in science. Everyone essentially had the freedom to do as they wished, although there were regular group meetings and periodic meetings with him as well. But in the animal cell world, this was a time when your own lab was really making seminal observations about heterogeneous nuclear RNA (HnRNA).
The messenger concept was well established in bacteria, but obviously no messenger RNAs were yet defined in animal cells — only the major classes of RNA, including HnRNA. We had meetings with others at Seattle who were interested in RNA. In particular, Ben Hall, who had developed some of the hybridization assays with Spiegelman, was using that technology mainly in phage systems. Brian McCarthy had come from Terrestrial Magnetism to Seattle — to the University of Washington — and was analyzing populations of RNA by DNA-RNA hybridization, mainly in sea urchin embryos with the Whiteleys.
I remember a particular meeting between the Rutter, Hall and McCarthy labs — those interested in the somewhat new field of gene expression in animal cells. I presented the work of Don Brown on ribosomal RNA gene structure and populations of RNA in animal cells, in this case during developmental transitions in Xenopus. And I recall one of the aforementioned individuals making the remark at the end, "Is that all that's known?" It literally was the Dark Ages. We were all anticipating, of course, that the messenger hypothesis would be validated in animal cells and were waiting for people like you to solve some of those problems.
But the field was limited by not having tractable genetics in animal cells and by the complexity of the genome, such that the DNA-RNA hybridization experiments were really not measuring what people wanted them to measure.
Darnell: Of course, that's right. As any scientist who may look at this interview may know, what you decided to do and successfully did was to study the enzymatic synthesis of RNA. Can you remember a time when it was decisively decided that that was what you were going to attack?
Roeder: I can, because initially there were a number of diverse and disappointing studies. Some of them involved the use of isolated nuclei from animal tissue, such as rat liver, to measure changes in the overall levels of RNA synthesis and in RNA populations during hormonal responses. I remember, in that case, the only thing I could conclude that was induced in this response was ribosomal RNA synthesis.
But in terms of thinking about the enzymology, the studies in isolated nuclei were useful but not giving me any insights. I admit some bias by being in an enzymology laboratory where people talked a lot about enzymes and purified them. And there had been some speculation from studies by Widnell and Tata who had looked at RNA synthesis in isolated nuclei under different divalent metal ion and salt concentrations. Under one set of conditions, the RNA had a ribosomal RNA-like base composition, and, in another case, it had a DNA-like or an HnRNA-like base composition. So the speculation was that there might be two different RNA polymerases operating.
This was intriguing to me. And since the things I was doing were really not leading to an understanding either of a transcription mechanism or its regulation, I began to think that looking at the enzyme or the enzymes was the way to go. At that time, purified ribosomal RNA genes were available in the Brown and Birnstiel laboratories, and I certainly thought that down the track there would be an opportunity to reconstruct a specific transcription event in a test tube.
Darnell: At that point, were you, or was the group, reflecting on any of the successes —by that time not complete but beginning successes —in studying DNA replication? In other words, that's a complicated, multi-enzyme, multi-protein task also. Was that important in your thinking?
Roeder: I think there wasn't much discussion about that in the Rutter laboratory, but in those days, much more so than now, I was conscientious about covering the literature. I had other things to do, but I did follow the nucleic acid enzymology literature, particularly the work from the Kornberg laboratory, where it was clear you could study enzymatic reactions relating to transactions on DNA. And again the enzymology was the key there. The difference, of course, was that the enzymology was carried out in bacterial systems where the enzymes were much more abundant than in animal cells.
Darnell: Right, right. So, finally came the time when you got solubilized enzymes. Tell us about the first year or two after those discoveries.
Roeder: Well, the discovery of a method to quantitatively solubilize the enzyme(s) occurred in January 1969. I had spent a little time trying the classical approach that the bacterial people used — which was to grind up cells in a low salt buffer, sonicate them or run them through a French press, and spin out the debris. This yielded a soluble bacterial enzyme that was one part in 50 and had to be purified only 50- or 100-fold. That approach did not work well in animal cells, although there were at least a half a dozen reports in the literature of a soluble mammalian enzyme. However, as it turns out, this was only a few percent of the total RNA polymerase, because most was bound to DNA and engaged in transcription complexes. That of course is the basis of the nuclear run-on assay that you and I appreciate as the major means to measure an absolute rate of transcription.
So after fussing around with related techniques and having so little soluble enzyme, I decided there had to be another method. In reading the literature about nuclei, it was clear that the RNA polymerase was bound to chromatin or to DNA in the form of chromatin, and chromatin was sort of a dirty word in those days. There was a lot of messy work on chromatin, but it was clearly something that had to be dealt with and I realized that I had to get rid of the histones and the DNA to get the enzyme solubilized. Fortunately, I ran across a 1942 paper by Allfrey and Mirsky, from The Rockefeller University, on the properties of nucleohistones. It was shown that high salt would readily dissociate the histones and that upon rapid dialysis or dilution, the DNA and the histone would rapidly precipitate from solution.
Of course they weren't studying RNA polymerases, but that particular insight was the key to getting the enzyme solubilized. The approach that I used was to dissolve nuclei in high salt (0.4 molar ammonium sulfate) and then to sonicate, which disrupted the DNA and reduced the viscosity to produce a solution that one could work with. Everything was soluble in 0.4 molar ammonium sulfate, and then the trick was to dilute rapidly to 0.1 molar ammonium sulfate, whereupon DNA and histones rapidly aggregated and the polymerase remained in solution.
Darnell: And so the discovery that there were several polymerase activities followed the solubilization, and there's a famous chromatogram of three different polymerases which I've taken advantage of, as you know, and put in a textbook. And was that a eureka moment, for you, when you saw the three polymerases coming off the chromatogram?
Roeder: That definitely was a eureka moment — not the last, but certainly the first and perhaps the most exciting because it really launched, in my view, the whole field. Again, a few people had been studying animal cell RNA polymerases, but there was really not enough soluble activity to work with. At the time there were many ongoing studies of bacterial RNA polymerase, even though, ironically, the first report of a bona fide RNA polymerase activity, period, was Sam Weiss' description in 1959 — a report in the Journal of the American Chemical Society— of an aggregate (chromatin-bound) enzyme from rat liver. It wasn't a soluble enzyme but it was an active aggregate enzyme. So the chromatogram you referred to was really a turning point for me, and happily it happened very quickly after I had a soluble enzyme.
And I have to admit that the trick to seeing three RNA polymerases on the particular ion exchange column that I used was a result of my knowing that Sephadex A-25 had a very high resolving power. My good friend Ed Penhoet had used it to resolve five aldolase isozymes, or hybrids of two different aldolases, and had recommended it.
Darnell: So just to remind anyone who may eventually be looking on here, we're still, I won't say stuck, but we have spent a considerable time reflecting on the bench activity when Bob Roeder was a graduate student. This all occurred, and the discovery of the enzymes — three enzymes — occurred when he was a graduate student. Let's move on, skipping a bit, because there's something that I have always thought was of great importance in the program that has led you to where you are now. And that is the ability not just to observe the incorporation of ribonucleotides into RNA, but to make certain that the RNA that was being formed in the reaction started at the right place. Initiation demands that you start at the right place. And that happened first with polymerase III, but then it also eventually happened with polymerase II.
I think it would be worthwhile if you would just recount for us the Correct Start Site thesis and why this is so important for subsequent work.
Roeder: Right, but let me interject a moment to cover two years of my life as a postdoc. Having three enzymes was exciting, and the next step was to try to do what you suggested — to get the enzyme to recognize, if there existed, a specific start site. The best candidates of the day were the ribosomal genes because they had been purified and fairly well characterized — especially by Don Brown, where I went for a postdoc. I suspected strongly that one of the enzymes, RNA polymerase I, was the correct polymerase because it was localized in the nucleolus — the site of ribosomal RNA synthesis. So I packed my bags, went to Don Brown's lab, and purified the Xenopus enzymes — not to homogeneity but adequately, I thought, for transcription studies.
One of my great surprises and disappointments was finding that purified RNA polymerase I did not work correctly on purified ribosomal RNA genes. This led me to suspect either that there wasn't a really specific initiation site, although that really didn't seem probable, or that something might be missing or, given a lot of attention to chromatin in those days, that maybe one needed a chromatin template to partially restrict the activity of the enzyme. So that simple experiment did not work but, at Washington University, once we had established through studies with the mushroom toxin alpha manathin that the three enzymes really did have three different functions, we could then focus on transcription of ribosomal RNA genes by pol I, transcription of mRNA-producing genes by pol II, and transcription of small structural RNA genes by pol III with much more confidence.
And there really was a question in those days about whether there was a specific initiation site. Of course, nothing was known early on about genes for proteins — that is, genes producing proteins through messenger RNAs — although we did try to look at some of those initially. However, we also began to study the 5S RNA genes because Don Brown had isolated them as a pure population even before the cloning days. The gene product was a short RNA, the gene being only a hundred or so nucleotides long. But the key feature of 5S RNA was that it had a 5' nucleoside triphosphate at the end, which meant that this was the initiating nucleotide and that the RNA wasn't likely to be processed. And of course processing was that we worried about — that the stable RNA species were derived from processing of larger RNAs with unknown start sites. We would never know where to look for, or how to ascertain, specific initiation.
So it is true, as you hinted, that we got initiation to work first on these small genes. But it wasn't a pure enzyme on a pure DNA. It was a pure enzyme on a chromatin preparation from oocytes which had 30,000 active 5S RNA genes. And the polymerase initiated and terminated correctly, but it depended on chromosomal bound proteins that we later characterized.
Darnell: However, it was a formal proof that you could, in a broken cell preparation, correctly start and stop an RNA.
Roeder: It was key because it gave us confidence. The work proceeded from there to showing that cloned genes could be transcribed in extracts. Thus, after our initial description of success with a purified enzyme, three laboratories basically showed that cloned genes encoding small structural RNAs could be accurately transcribed by RNA polymerase III in unfractionated cellular extracts — Don Brown's lab, Guang Jer Wu's lab and our own lab. In our special case, we were able to show that the isolated RNA polymerase III would transcribe the gene, or initiate at the promoter, but only with additional soluble proteins that we had purified from the extract.
But that definitely gave us confidence that the more interesting and diverse class of genes — those encoding proteins — might show the same success, given an ability to know what to look for. And that's where your own work was so pivotal for us.
Darnell: Why don't you just briefly go over the experiment that Tony Weil in your laboratory made on RNA polymerase II-transcribed genes, the genes that encode proteins?
Roeder: So here again, the key question was, "What is a start site?" We weren't so concerned about termination sites, but "is there a specific start site that we could hope to use to reconstruct initiation?" Adenovirus provided the genome of the day because it was a viral genome that could be chopped up with restriction enzymes into pieces that could be cloned and characterized. Again, the real question was, "Is there a specific initiation site?" And, of course, your own studies, using a variety of techniques — nuclear run-on assays, UV mapping, and so forth — pinpointed several initiation sites on the adenovirus genome.
But the one that's near and dear to us, the adenovirus major late promoter that you mapped to position thirty or so...gave us clear picture of where to look for a specific initiation event. And I have to tell you, one of the other eureka moments, of course, was our seeing that initiation event established. That was the work of Tony Weil, who used a system comprised of purified DNA fragments, purified RNA polymerase II, and extracts from the cell.
The initial strategy, a simple sort of low-resolution analysis for specific initiation, was to look for an RNA product corresponding in size to a transcript extending from the initiation site to the termination site, which in this case was the end of the DNA restriction fragment — a so-called "run-off product." I remember coming back from one of many seminar trips to an excited Tony Weil, who anxiously showed me the run-off transcript. In fact, that event convinced me that I should travel more — because it was really an exciting event. This is what everybody in the field was waiting for.
We were all interested in the small RNA genes as models but, of course, the diversity of the genes encoded in proteins made them of more intrinsic interest.
Darnell: Well, you've explained it clearly, but in the possibility that there are students that may look at this, let me reiterate what Dr. Roeder said. At this point, total extracts of nuclei plus purified polymerase plus pure DNA allowed correct starts. And those, you will recognize, are the ingredients for all of the purification and identification of factors that has followed. Now, we've taken a good bit, perhaps close to most of our time to get to this stage. The events after this accelerated. More labs began to work on the problem. And now we have reached the point where a great many, perhaps we will never know all, but a great many of the proteins that participate in starting RNA synthesis have been discovered, a large fraction of them in your laboratory.
We certainly don't have time to go through, step by step, all of these. But I wonder if you would like to try to characterize this era in which so many different proteins have been shown to play a role in getting this great machine to make RNA, messenger RNA in particular, which turns out to be a key enzymatic process in controlling all of life's diverse events. Go from the time when you could start correctly....
Roeder: There are several levels to that question, but the first one of course was what to do when we had specific initiation working with the purified polymerase. The first question of course is "what are the other factors?" And it was clear from those initial experiments, by the way, that those factors that allowed the polymerase to work correctly were general initiation factors. Because remember, the adenovirus major late promoter is only expressed late in adenovirus infection, after other viral proteins have carried out certain functions.
And we were seeing initiation in extracts from uninfected cells. So we were pretty confident that these were simply general initiation factors — recognizing what we've come to call the "common core" promoters. There were two routes to go. One was to use that system to map promoter elements, which most labs did because it was an easy assay for those good with recombinant DNA. We chose the more challenging and harder route, and to me the more important one at the time for us, of going to the cold room and further fractionating the extracts to define the factors. So that led to the definition of the general initiation factors.
We started the process. Many people followed suit — a lot of former post-docs, as well as other laboratory members — in an effort to elaborate and elucidate the nature of the general initiation factors, which of course are a large number. Our own studies had also identified the factor that recognizes the core promoter — the TFIID that recognizes the TATA box in the case of RNA polymerase II-transcribed genes. So that was really a start, and said well, there's a second level of control beyond the RNA polymerase.
Darnell: So by this time you had how many polypeptides on your hands? Fifty at that point?
Roeder: Well...[it] includes the rest of the field for the simple reason that the pre-initiation complex for a gene encoding a protein is about 44 polypeptides if you include the polymerase. And we already knew about complexity from the structure of the enzymes. We had witnessed, in 1974, the first complete comparative analysis of polymerases I, II, and III. All these were found to contain ten or more subunits at the time and were far more complex than the bacterial polymerase. So the enzymes themselves said life is going to be more complicated in animal cells than in bacteria. With the advent of the general initiation factors, another layer of complexity was evident.
But this is all the basic machinery without regulation.
Darnell: But you and many, many other people around the world have, since the 1980s at least, concentrated on proteins that have some role to play in regulation. And just to be telegraphic about it, there are proteins that largely direct the rate of synthesis, and there are proteins that connect the machinery — the 44 polypeptides that you just talked about — with the regulatory factors. That has been an especially rewarding chapter for you in the last five years or so. And perhaps one that you could tell us about because of its great importance and general importance.
Roeder: Right, right. A word on the history of that. We certainly assumed when we had polymerases and general initiation factors working on DNA that we could now look at the issue of regulatory factors. And although we had succeeded in that area with the 5S gene-specific factor TFIIIA, which was the first of the many hundreds or thousands of DNA-binding regulatory factors that are now known, most in the field also began isolating and characterizing such regulatory factors because this could be done easily through DNA binding assays and various cloning methods. But analyzing those factors in the purified systems, in the case of RNA polymerase II, did not give the elevated or regulated transcription that we saw in crude systems such as nuclear extracts. And that simply said, there was another level of complexity that we certainly had not anticipated.
That came in two forms in terms of the types of factors involved. But in relation to what you alluded to, namely the mediator in animal cells, as early as 1991 we had seen an activity that was needed in addition to the 44 polypeptides for any of several activators to work. A similar activity had also been seen in yeast systems by Roger Kornberg and Rick Young. We didn't know at the time that a major component of our coactivator activity would turn out to be the same as the yeast mediator. But, indeed, this proved to be the case as we identified a complex of 25 or so polypeptides that was needed for a given DNA binding activator to communicate with the basic machinery to make it work more efficiently. And as you know, this has turned out to involve interactions of specific DNA-binding activators or repressors with specific mediator subunits and interactions of the mediator with RNA polymerase and the basic machinery.
And this complex, I think, is the major means for communication by activators, albeit there are other somewhat more specialized coactivators as well. So these were cofactors that were found to operate in our purified systems with DNA templates, but another large group of cofactors that are equally important are those that modify chromatin and that operate (largely) upstream. Although we've been involved with those, they were really discovered by other people.
Darnell: So we've ended up in this odyssey with four groups of proteins, the 44 or so general transcription factors, a major coactivator that's 20 proteins, the dozens of proteins involved in chromatin remodeling, and then the thousands of regulatory proteins that are involved in directing the whole show. We can't go on forever, but there are a couple of more things I would like to hear you out on, and I'm sure anybody who is watching this would like to hear it also. You, and all of the other colleagues around the world, have uncovered what is the most complex machine so far uncovered in all of biology: the machine that makes RNA in a specific way at a specific time. Where do you think the biochemists of the future, or the molecular geneticists of the future, can look for some relief in this complication, so that we can try to tell the world how control is brought about? What can we hope to see in the coming years that will simplify things sufficiently so that we don't have to deal with thousands of proteins to understand things? Is there any hope for that?
Roeder: Well, I think one of the important questions, given this incredible complexity of general factors and cofactors, is whether there is any redundancy in the system. Do we need all of those factors on any given gene? So I think we still have some bookkeeping to do. I think one of the important approaches now is to try to couple our in vitro mechanistic studies with complementary characterization of genes of interest and importance in the cell — to look in the cell and say what factors are on that gene during an activation process. And this of course involves the common Chip, or chromatin immuno-precipitation assay.
That, I think, will tell us at least what the likely players are and whether these factors are really there en masse, or whether there is some specialization going on. Certainly in terms of the chromatin remodeling factors, I don't think you're going to find all of them there at once. But the Chip assay can also give you some idea of the kinetics of their appearance. One of the things we'd like to do, and we've talked about this between ourselves, is to look at natural complex enhansosomes — that is, aggregates of DNA-binding regulatory factors that function in a cooperative manner — to really understand the mechanism, because I don't think we know that yet. And the question is, "Can we reproduce those functions in our test-tube assays, beginning with chromatin templates?"
So I think there are two issues there. One is, can we reproduce the phenomenon on a more natural template? And, since I mentioned enhansosomes, this brings up the issue of enhancers. Of course, many of these are not just a few hundred nucleotides away from the promoter but a thousand or 10,000 or 100,000 nucleotides in some cases. I don't think we know how distal enhancers work, and I don't know whether we could ever reproduce their function in a test tube, but I would like to try with some moderately far-removed enhancers to see if we can really understand the mechanisms — including the dynamics of the formation and the function of the pre-initiation complex. So that's part of your question in terms of what biochemistry remains. I think there's still a major challenge for the biochemistry.
Of course, I think another important issue concerns biophysical or structural analysis. We've proceeded from initial, very beautiful studies by our former colleague and my collaborator, Stephen Burley, on the pre-initiation complex with some of the early events to a point where Roger Kornberg in yeast and Seth Darst in bacterial systems have provided exquisite pictures of the large polymerases in action on DNA. And the question is, "Can we extend this to a structure, and an analysis of the dynamics, of the pre-initiation complex?"
Darnell: We've covered a lot of territory, and tomorrow, we get the chance to applaud you and to listen to your response to winning the 2003 Lasker Award. Congratulations. It's been a pleasure to be with you this morning. Thanks.
Roeder: Thank you, Jim.