Michael Hall (University of Basel): The Story of TOR (Target of Rapamycin)
Good morning. My name is Michael Hall. I’m a professor of Biochemistry at the Biozentrum of the University of Basel. And I work on… on cell growth and iBiology has asked me to tell you the story of the discovery of… of TOR. Now, TOR is a is a kinase, it’s a highly conserved kinase, and more importantly it’s a central controller of cell growth, and this is the story I’ll tell you today. So, this is the timeline of the TOR field. The story actually starts in 1965 and that’s when a group of scientists from Montreal set out for Easter Island, known as Rapa Nui to the locals, and they were prospecting for… for exotic microbes which produced natural products or secondary metabolites, which they could then develop into a… an antifungal. They did isolate a bacterium from the soil; it was a Streptomyces from which they isolated a compound which they called rapamycin, after Rapa Nui. However, when they started to develop this compound as an antifungal, they realized it had the undesirable side effect of suppressing the immune system. So, it was then dropped and was never further developed as an antifungal. Years later, when immunosuppressive therapy came into existence, this drug was then rediscovered, but now for use as an immunosuppressive. So, it was now developed for the very reason for which it had originally been rejected. So, we started our studies on… on rapamycin, trying to lose state the molecular mechanism of rapamycin action. And this story was started by… by an outstanding postdoc who joined the lab in 1989, Joe Heitman. He was a MD/PhD who just finished his PhD studies at… in New York at Rockefeller University and, given his medical background, was interested in understanding how… how drugs functioned. And we also… another fortuitous circumstance was an… an ongoing collaboration with Rao Movva, who was a group leader at a local Basel company by the name of Sandoz, which is now known as Novartis. Now, Joe and Rao had the brilliant idea to study the mechanism of action of rapamycin by taking an approach, a genetic approach, with the simple eukaryote yeast — Saccharomyces cerevisiae, in particular. Now, this was a brilliant idea because not only did it allow one to use genetics to go after the target of rapamycin, but it was also an unusual idea, because, at that time, rapamycin was being developed for a… as a drug for use on humans, and everybody else was of course working on mammalian cells to understand how this drug worked. But Joe and Rao remembered that rapamycin had originally been isolated as an antifungal, and you could therefore use… use yeast, and yeast genetics in particular, to… to study rapamycin action. So, what Joe did was he grew up a culture of our… of our standard lab yeast strain and then plated it on… on solid media containing rapamycin. And the rapamycin killed most of the cells which… which were plated on this… on this petri dish, but a few spontaneous mutants arose. He then picked these mutants, these and several others. He characterized about a total of 20 different rapamycin-resistant mutants, at least in the early phases of the study, and he then purified each one of these individually… mutants, and then characterized them. More particularly, he… he determined the number of complementation groups and what genes were mutated in these… in these… in these strains to… to confer rapamycin resistance. So, Joe found that these uhh… rapamycin-resistant yeast mutants were defective in any one of three different genes. The vast majority of the mutations were recessive and in a gene called fpr1. This is actually a gene which Joe had already characterized in the lab in the context of his earlier studies on a… on a rapamycin-like drug called FK506. He also obtained mutants in two novel genes — these mutations were extremely rare and, unlike the fpr1 mutations, were dominant. So, this was the first… the first appearance in the literature, in 1991, of… of TOR, so this was the original identification of TOR, in this case as genetic loci in the yeast genome. But we still didn’t know what the TOR genes encoded, nor did we know why these mutations were dominant and… and rare. Now, to answer these questions, two new students joined the project. One was a Jeanette Kunz, an extremely bright and hardworking Swiss student, and then Stephen Helli… Helliwell, a British student who was one of the more colorful members of our… of our lab, as you can probably guess from this photograph. So, what Jeanette and Stephen did was they cloned the… the TOR genes from our rapamycin-resistant yeast mutants and sequenced these… these… these genes, both the wild-type version and the rapamycin-resistance-conferring version. And this allowed us to understand why the TOR mutants were… were dominant and rare, and the fpr1 mutants were… were recessive and… and common. And the uhh.. and the reason relates the mechanism of action of rapamycin — and that’s shown on this slide. This is rapamycin, drawn in red, here. This is a very lipophilic molecule, so it simply diffuses across the plasma membrane. It does not require any kind of transporter, which when mutated could confer rapamycin resistance. And, once inside the cell, it then forms a complex with another very small, 12 kiloDalton protein called FKBP, which is absolutely essential for drug action. The… the toxic agent is the rapamycin/FKBP complex — the drug alone does nothing. And the fpr1, or the FKBP, protein is not only completely essential for drug action but it’s not required for cell viability. So, any simple loss-of-function mutation in the gene encoding this FKBP protein confers complete rapamycin resistance, and such mutations would be recessive. That’s why most mutations were in the FKBP-encoding gene and why these mutations were recessive. TOR, on the other hand, and this is just a very small portion of TOR. TOR is about a 300 kiloDalton protein, as compared to this 12 kiloDalton protein, here, so you can see this is indeed a very small part of the… of the TOR protein. This is the FKBP/rapamycin binding site in… in TOR. And, unlike FKBP, TOR is completely essential for cell viability. And therefore it can tolerate very little mutation. In fact, every mutation we obtained which fell into the TOR gene modified a… one residue — they all fell in the same codon — which modified a residue in this alpha helix, which was a key contact site between TOR and rapamycin. So, the effect of these mutations was to prevent the binding of rapamycin to… to TOR, without otherwise affecting TOR activity. And, given that these mutations were confined to this single codon, this is why the… these mutations were extremely rare, and of course they conferred rapamycin resistance in the presence of a… of a wild-type copy of the TOR gene, which is why they were dominant. So, these mutations were extremely informative. They not only helped elucidate the mechanism action of rapamycin, it forms… the fact it forms a complex of FKBP, it also led to the identification of the binding sites in… in the TOR protein, and, most importantly, led to the identification of TOR itself. So, once we cloned the… the TOR genes and sequenced them, we found, first of all, that the… the two TOR genes encode very similar proteins — these proteins are 70% identical. And they also turned out to be the founding members of this class of kinases, protein kinases, called PI kinase-related protein kinases. And the reason for that is they… they… they… all the members of this class of atypical protein kinases resembled lipid kinases, PI kinases in particular. Since its early discovery of TOR in yeast, TOR has been found, now, in all eukaryotes, from yeast all the way to human. And the name TOR — from yeast — has been kept for all these organisms. In the case of human TOR, or mammalian TOR, call it mTOR for mammalian TOR. So, this is where we were in 1993 after the cloning and sequencing of our… of our TOR mutants. We knew that TOR existed and it was the copy of the… the target of a FKBP/rapamycin complex. And, in this model, you can actually see… this is the time where we still thought TOR was a lipid kinase, a PI kinase in particular, as indicated by the PI, the phosphoinositol, in this figure. We had no idea at the time what was upstream and what was actually downstream of TOR. We also incorrectly thought the role of TOR was to control G1 progression, in other words to control cell division. But this turned out to be wrong. And, in fact, the role of TOR is to control cell growth. And here I have to make an important distinction between cell growth, which is an increase in cell size or cell mass, versus cell division, which is an increase in cell number. We never expected that the TOR would be a controller of cell growth because, as hard as it might be to believe from the perspective of today, in those days nobody thought cell growth was actively controlled. In other words, there would be no system to control cell growth; it was thought to be a spontaneous process that just happens when nutrients are… or building blocks are available. So, we had to do additional research to eventually arrive at the conclusion that the role of TOR2 is to control cell growth. And we came to this… to this conclusion based on a great deal of work — not only from our lab, but a large number of other largely yeast labs in this in those times — which showed that TOR controls a large number of cellular processes. Now, these cellular processes can be subdivided into two groups: the anabolic processes which TOR activates, and the catabolic processes which TOR inhibits. So, TOR balances these opposing forces of synthesis and degradation in response to, and this is important… in response to nutrients. And this is also something we discovered in the mid-1990s, that the… TOR is activated by… by nutrients. So… and then this led to the model that TOR is a central controller of cell growth. This is most photogenically illustrated in experiments done in Drosophila by Tom Neufeld and then Bruce Edgar. On the far side, here, we have an experiment done by Tom Neufeld. This is the fat body of a fruit fly and what Tom did was he selectively inactivated TOR in these two GFP-expressing cells. And, as you can see, these cells are smaller than their neighboring wild-type cells. Bruce Edgar did the inverse experiment. What he did was he hyperactivated TOR signaling in these GFP-expressing cells, and he observed the opposite effect — the cells became much larger. So, the obvious conclusion from these experiments, which were largely confirmation of work done earlier in yeast, was that the role of TOR is to… is to control cell growth and thereby cell size. Now, Tom Neufeld took this experiment one step further by, now, isolating a mutant fly, a TOR mutant fly in which all cells of the entire animal are defective for TOR signaling. And it’s important to point out that this TOR mutant flies a so-called hypomorph, so it’s a partial loss-of-function. If it had been a… a complete loss-of-function, this fly never would have been born given that TOR is absolutely essential for viability. So, when Tom Neufeld isolated this fly, he then asked the question, why is this fly smaller? Is it smaller because it has fewer cells? Or is it smaller because it has the normal number of cells but now each individual cell is slightly smaller? And he could answer this question by… by looking in the wing, where he could… at the wing is a plane of cells, so he could relatively easily measure the sizes of individual cells and count the number of cells, and from that he could extrapolate to the whole animal. And what he found was this TOR mutant fly had more or less the normal number of cells, but each individual cell was slightly smaller. So, the conclusion from this is… is that the role of TOR is not only to control cell size, but thereby also to control size of individual tissues and ultimately size of the whole animal. So, TOR controls growth very broadly — broadly in terms of organism and broadly in terms of physiological context, essentially any… any situation in eukaryotic biology where you see a change in cell size, there’s a high likelihood that TOR will be somehow involved. So, here we had another conundrum. We had known earlier from our… from our genetic studies in yeast that there are two TOR genes, two TOR proteins in yeast, and whereas these two proteins were similar, 70% identical, they were not functionally identical. And this came from the observation that if we knocked out the TOR1 gene nothing happened, so we proposed that… that the TOR2 then has some redundant function with… with TOR1. If we knocked out TOR2 alone the cells died, but they didn’t die in the G1 phase of the cell cycle. And, finally, if we knocked out both TOR genes simultaneously, the cells of course still died, but now they died in the G1 phase of the cell cycle. This told us that TOR2 has two functions, TOR1 has one function, and that one function of TOR1 overlaps with one of the two functions of TOR2. So, we didn’t know what these two so-called functions were and how they controlled cell growth. Well, again, a great deal of work from our and a number of other labs doing yeast genetics found that these two separate functions of TOR are in fact two separate signaling pathways. One is the… what we call the TOR2-unique pathway, and this signals to the actin cytoskeleton. And… and we therefore view this as a pathway which mediates spatial control of cell growth. The other function of TOR, the one that either TOR1 or TOR2 can perform, this controls all these processes which lead to mass accumulation in response to nutrients, and we view this… this signaling pathway… this signaling branch of TOR as mediating temporal control of cell growth. So, we think the logic of these two different TOR signaling pathways is that they integrate spatial and temporal control of cell growth. Of course, these two aspects of cell growth need to be integrated. So, at this stage, we had another problem and that is, what determines the specificity of the… of the TORs? Why can TOR2 signal through both pathways but TOR1 only through one? And we also didn’t know why the TOR-shared branch was rapamycin sensitive, whereas the TOR2-unique branch was rapamycin resistant. And to tackle these problems, another postdoc joined the lab, Canadian postdoc Robbie Loewith, who teamed up with my long-term technician, Wolfgang Oppliger, and they decided… unlike our previous work, decided to take a biochemical approach to understanding TOR signaling. So, Robbie did this, he purified the two TOR proteins, and what he found was that the TORs did in fact purify… co-purify with other proteins, and we had these purified or what we now call TOR complexes… TOR complex 1 and TOR complex 2… when he had these two purified TORCs in the test tube, we sequenced the proteins. This all… allowed us isolate the genes encoding all these subunits of the two TOR complexes and, once we had the… the genes isolated, he was able to knock these genes out. And when he did this, he found that these two TOR complexes corresponded to these two previously identified TOR signaling pathways. So, at this stage, we started wondering whether the TOR complexes, like TOR itself, might be conserved all the way to human. But, to address this question, another postdoc joined the lab, Estela Jacinto. She joined the lab from Michael Karin’s lab and was the first person in my lab to start studying TOR in mammalian cells to ask this specific question, is… is TOR or are the TOR complexes conserved in… in mammals? So, she and… and notably other members of the TOR field at this time, most importantly David Sabatini, Kazu Yonezawa, and Kun-Liang Guan, showed that the two complexes indeed are conserved in humans. They are made up of the same proteins, or orthologues of the proteins, that were originally discovered in yeast. They phosphorylate or act on many of the same downstream proteins to control the same cellular processes involved in cell growth. So, the picture that emerges of the TOR signaling network… and we call it a network because it’s more than a single pathway… the TOR signaling network is a primordial or ancestral signaling network which has been conserved all the way from yeast to human to control this very fundamental process of cell growth. Now, the sole exception to that statement I just made is this part of the network up here, which is the growth factor or insulin signaling pathway. This evolved much later. This evolved with multicellularity and was then grafted on to the more primordial TOR signaling pathway which already existed in unicellular yeast. And the reason for this is that cell growth control in metazoans is more complex than unicellular organisms, because in metazoans it’s critical to control the growth of every cell in the body with every other cell in the body, so you need an input which controls growth over a whole body plan. And this is then what the growth factor signaling input provides, this control of cell growth over a whole body plan. Now, I’d like to end with a note on the clinical relevance of… of… of TOR. We now know, based on work over the last 15 years, that TOR signaling is… is functionally linked to a large number of diseases, all of which are characterized by inappropriate or ectopic cell growth. These can be malignant forms of cell growth such as cancer or more benign forms of cell growth like cardiac hypertrophy, but not… but in all cases cell growth, which also underscores the… the… the function of TOR as a controller of cell growth. More recently, TOR has been implicated in another set of disorders, the so-called metabolic disorders, and we know that high… chronically high circulating levels of nutrients can upregulate TOR even in the absence of growth factors. This will lead to adipogenesis and obesity, which of course can lead to type 2 diabetes. But we have a more direct link to type 2 diabetes in the TOR signaling network, and that is that TOR complex 1 can negatively feedback on the insulin signaling pathway by phosphorylating IRS. As a result of this, high levels of TOR can lead to insulin insensitivity, which is one of the hallmarks of type 2 diabetes. In this context, some have actually proposed using rapamycin as an anti-diabetic drug, so this would be a fourth major therapeutic area for this remarkable drug. However, I don’t think anybody’s actually seriously developing rapamycin as an anti-diabetic because there are a number of other complications. But at least one finding, or one… one thing that gives credence to this notion that rapamycin could be developed into an anti-diabetic is that the world’s most commonly prescribed anti-diabetic, metformin, works at least in part by down-regulating the TOR signaling pathway. So, I’ve come to the end of the story of the discovery of TOR, or of these three sub-discoveries of TOR signaling. The story I told you spanned the years of 1990 to 2004. A lot has happened since 2004. The size of the field has grown even further. There’s still a great deal of exciting biology, but that will have to be a story for another day.