Initially? I was at Johns Hopkins in an M.D.-Ph.D. program, and I did my thesis in a lab with a person who had many different interests, one of which was pharmacology. He had access to some potent molecules, one of which was called rapamycin. I did my thesis on that and have been working on the mechanism of this small molecule effectively ever since. I can’t claim much forethought in picking what I ended up working on.

Because I had a medical background, my first exposure to rapamycin was because it was considered an up-and-coming immunosuppressant. It was projected for use in organ transplantation. I’d been interested in this subject, this medical problem of transplanting an organ and keeping it alive in the host. When I was a student, rapamycin was in clinical trials and that was my entrée into it. Here’s this immunosuppressant; it’s not known how it works. It was already appreciated at the time, although not so much by me, that it had anti-fungal effects, as well as potential anticancer effects. This molecule had a lot of potentially interesting properties.

“The fact that we have a drug that can inhibit the activation of Akt, that can inhibit the upstream activator, and that clinically is already known to apparently work best in cancers with hyperactive Akt, tells me, at least, that the drug is inhibiting Akt through complex mechanisms.”

Rapamycin is a small molecule and nobody was sure how it worked. We now understand that it works by interfering with the function of a very large protein, mTOR, which stands for "mammalian target of rapamycin." In essence, after the discovery of mTOR in 1994, our research switched from studying rapamycin itself to trying to understand the function of mTOR. And now we’re studying the way mTOR is connected to Akt. We and other groups had identified mTOR as part of protein complexes with the cell. mTOR lives bound to other proteins; it doesn’t live by itself.

Yes, and this is how this kind of research works. You have a drug that does some interesting thing. You identify what that drug binds to, which in this case turned out to be mTOR. At the time, we didn’t know anything about mTOR. So we looked at the sequence, which told us that it likely was a kinase; it could transfer a phosphate to other proteins, a process called phosphorylation. Then what we and other people did was to try to identify other interacting proteins of mTOR, hoping that would help us understand what was going on. This led to the discovery that mTOR is part of two distinct complexes that are mutually exclusive. Now we call one mTOR complex one, and the other is mTOR complex two. They’re abbreviated mTORC1 and mTORC2.

At the time we started that work the function of mTORC2 wasn’t understood. At the same time there was a kinase, known as Akt/PKB, which is now very famous because it is known to be involved in diseases like cancer and diabetes. There had been a lot of work suggesting that was true and a lot of drug development was aimed at trying to inhibit this kinase. But there had been this missing piece in understanding how Akt is activated. In that Science paper we showed that mTORC2 was the missing kinase activating Akt. And that’s why that paper has been so highly cited. This particular pathway is hyperactive in almost all cancers and so it is considered one of the big drug targets for cancer drugs.

The funny thing is that there had been other kinases proposed as the one that phosphorylates Akt. So one of the things we had to do in that paper is argue against some of this previous work. There was no reason to propose yet another kinase, if the ones already proposed actually did the job. This made that paper a little controversial, and people were somewhat upset about that.

But we were quite convinced that what we were saying was true and so important. And it’s now been proven to be true by knocking out components of mTORC2 in laboratory animals. We were quite convinced it was true and so, yes, we thought that was an important paper at the time, that it resolved this long-standing issue in the field. And that’s why people have appreciated that paper since then.

One thing we did was prove this absolutely, so we generated knockout animals of some of the mTORC2 components, as did other people. These experiments were strongly suggestive that what we said was correct. We also continued working on trying to identify other components of these complexes. And then we did something that has proven to be really controversial, but I think, also really important.

Now I have to backtrack because I over-simplified the story a little when I first told it. There are these two complexes that bind mTOR, but only one of them, mTORC1, seemed to be sensitive to rapamycin. mTORC2 seemed to be insensitive. We realized, based on how we knew this drug worked and how it bound to mTOR, that eventually it should affect the function of mTORC2, as well.

Rapamycin works by binding to mTOR. So we hypothesized that when it bound, it might eventually interfere with the assembly of mTORC2—in other words, with the assembly of other compounds that had to fit in that binding site. We thought that over a long enough period of time it would work to inhibit the assembly of mTORC2 and so eventually that particular complex would only form at very low levels because rapamycin was blocking its assembly. We’ve now shown that this appears to be the case.

One reason we find this interesting is that many of the diseases for which rapamycin is indicated or for which it’s in trials have hyperactive Akt activity. The fact that we have a drug that can inhibit the activation of Akt, that can inhibit the upstream activator, and that clinically is already known to apparently work best in cancers with hyperactive Akt, tells me, at least, that the drug is inhibiting Akt through complex mechanisms. So we proposed in this recent paper that this inhibition of mTORC2 formation was an important clinical attribute of this drug.

Well, some people believe it and others don’t.

It’s a tough problem because when you add rapamycin you inhibit mTORC1, and now we’re claiming that you also inhibit mTORC2. Somehow you have to find a way to just inhibit mTORC2, and that’s very difficult. We don’t have any sure ways to do that. We might potentially do it genetically, by deleting just components of mTORC2 and not mTORC1 and ask what happens. But it’s a tough problem.

Some of it was just very technical in nature. It turned out that purifying complexes was very tricky for very simple reasons. These were things that you don’t typically think about. For example, the types of detergents you use to lyse cells. It turned out that those detergents were rupturing these complexes, and it took us forever to find that out. A lot of these little ridiculous technical issues came into play in these experiments. You never really think about these and they can determine everything.

Yes, a lot of these things depended on luck. In the Akt story, for example, there was really just the chance observation that we didn’t understand and that turned out to be the whole key for unraveling that story. I think, at least for me, there is a tremendous amount of luck in a lot of the things in which I’ve been involved. This was no exception.

Well, it turns out that Akt regulates this mTORC1 complex early. It’s a positive regulator of mTORC1, but then there’s a feedback loop. Once mTORC1 is activated, it turns around and inhibits Akt. That had been quite well established. We were doing some important experiments with RNAi, knocking down different components of mTORC1, and when we knocked down a protein called raptor, we saw that mTORC1 activity was suppressed, And, as we expected Akt activity went up because there was less mTORC1 to feedback and suppress it. That made a lot of sense.

But then when we did RNAi to mTORC1 itself, we actually saw that Akt wasn’t activated; in fact, it was somewhat inhibited. This didn’t make any sense in the current model of how things work. It made us realize that mTORC1 had to have some other function in this pathway, and that’s how we eventually came upon this connection with mTORC2. It was a bit convoluted, but it was one of these observations that just kept nagging at us for a while. We could explain it away for technical reasons—maybe the RNAi against mTORC1 just didn’t work very well. But it turned out there was a real biological reason for why we were seeing this unexpected phenomenon, and that turned out to be key.

This is one of the fascinating aspects of biology. I think to succeed you have to be lucky but to get some of that luck, the key is to try a lot of different things and work quite hard.

  Where do you see the research on this Akt pathway going in the next five years?

I think the era of figuring out pieces of this pathway will come to an end. We will understand what all the components are. So the directions we will have to go in will be to understand which pathways have medical relevance and how to perturb them. That will have to be done physiologically, on animals, and eventually on human patients. You can focus so much on physiology and cell biology that you lose track of the facts that cells don’t live in plastic dishes, but in animals, and ultimately you have to understand what they do there. That to me is the exciting area to think about now.

I’d like them to understand and appreciate this connection between the clinical world, starting with a drug, and then figuring out how it works, and how that leads to new hypotheses about how a disease might work. To me that’s been the way to opportunity. Start with clinical science, go to basic science, and then, hopefully go back to the clinic to apply what you’ve learned.

David M. Sabatini, M.D., Ph.D.
Whitehead Institute
MIT
Cambridge, MA, USA