Initiative Spotlight: A Conversation with Frank McCormick

March 6, 2019, by Jim Hartley

Hartley: One of the major goals of the RAS Initiative is to hit RAS protein directly. How would you characterize our progress, especially in light of the progress of the G12C covalent inhibitors being developed by a number of pharmas?

McCormick: We have two programs actually at the Frederick National Lab, which target other residues on KRAS [for covalent modification]. One targets the site of prenylation on KRAS 4b which is cysteine-185. We’ve been able to make compounds which are on target and effective at knocking down KRAS at submicromolar concentrations. We hope those compounds mature enough to be tested in animal models during 2019, and we’ll see if they have the legs to become preclinical compounds. So that is moving forward. The other targets histidine-95. The idea of attacking histidine-95 is very attractive because it’s a unique residue in KRAS, not in H and N, but it is chemically challenging. We are trying very hard to find compounds that engage histidine-95 and either target KRAS for degradation or interfere with its function somehow. Those compounds and that whole program is a year or two behind the cysteine-185 approach. And we know other people, Steve Fesik has a big program using NMR-based fragment screens that find compounds that bind at different parts on the RAS protein, Kevan Shokat is trying different approaches also, so I think we can expect 2019 to see multiple other ways of hitting KRAS that may or may not be allele specific but at least hit KRAS with some selectivity.

Hartley: The C185 targeting is predicated on there being enough of a window between the time the full length protein is synthesized and the time at which it has been processed at its C terminus, at which point it’s going to be immune to any interference.

McCormick: Obviously we didn’t know that when we started off but we’re now sure that we can prevent KRAS processing and deplete KRAS just by preventing its getting farnesylated. We don’t know precisely how efficient that process is and how much escapes, but we can knock down KRAS effectively with compounds that are relatively early in the whole development process. We don’t think that that window is too short, or that there is anything intrinsically preventing that approach from going forward, it’s just a question of getting compounds which bind efficiently enough to KRAS to prevent processing happening. And we’ve seen in the last year a big increase in potency of the compounds we’ve been developing, and now we have NMR structures on the new pocket it forms, where the new compounds bind, and we’re starting to accelerate the process of finding compounds that interact through that mechanism. It’s a tough process but it is going in the right direction.

Hartley: The paper by Kevan Shokat’s group that invigorated the covalent inhibitor approach was published in 2013, and I think I’ve heard him say that there were seven years of work that went into that paper. Clinical trials in humans are just getting underway, correct? Are other oncogenic mutants being targeted besides G12C?

McCormick: Well as of today, Amgen and Mirati are recruiting for trials of their G12C inhibitors. We also know of efforts going on at Wellspring, at Pfizer, Lilly, Novartis,  all of which have G12C programs coming down the pipeline. So we imagine over the next year or so there will be quite a few different compounds targeting G12C tested in the clinic. It will be extremely interesting to see how they pan out as single agents and whether they require other combinations with other compounds. So that will play out over the next couple of years, for sure. This is a new type of drug and there are obviously technical challenges in developing a covalent inhibitor for this type of protein. It is not very well established exactly what kind of off-target effects we’ll see and how the drugs will behave, so having more compounds with different chemotypes and so on will be part of the whole process. So we’ll learn a lot from it in terms of the benefits of targeting G12C directly, and also a lot about the requirements for the properties of the compounds that make it successful. And that will play out over time.

As for other [oncogenic] mutations though, I haven’t seen any evidence that someone has a compound that targets other mutations in the same way. Because it’s much more challenging to target G12D or G12V or other alleles. I don’t know exactly where people are in attempts to do just that, we know lots of people are trying, but there’s nothing that I know of that is anywhere close to being a compound heading for the clinic for the other alleles.

Hartley: To a novice it’s extraordinary how much time and effort is involved.

McCormick: Drug development is always a long process, to find a compound that hits the target cleanly, doesn’t hit other targets, and distributes properly in the body and is safe and has good bioavailability. That’s a long process regardless of how many resources are thrown at a problem, it’s just a very long and difficult process. I don’t think it seems slow at all, really. Compounds are already in the clinic 5 years after the original paper, that’s pretty good. And we’re talking about a new class of compounds which have unpredictable PK [pharmacokinetics] and properties in vivo, so I think that’s gone as fast as you can expect for any kind of new approach for a target.

Hartley: The classic biochemical screens for non-covalent inhibitors have not had much success, have they?

McCormick: The whole field has sort of tried the obvious interactions like RAS-RAF binding, and overall has failed to find compounds that disrupt that very tight interaction, not a surprise to me because the interaction itself consists of two anti-parallel beta strands which don’t present any kind of pockets for compounds to bind to. Busting up that interaction is desirable but technically seems to be impossible. I think pharma companies that are interested in RAS have tried most obvious biochemical screens for things that affect mutant RAS by direct, medicinal chemistry based like high-throughput screens. So I think we’re not going to do much more of that unless we find new pockets and new approaches to which high throughput screens could be applied.

However, another of our major goals, besides hitting RAS directly, is to understand precisely how RAS activates its effectors in the plasma membrane. RAF kinase obviously is the best understood in that respect, although PI3 kinase also is well validated as a target. RAS binds to both of those proteins and activates them only in the context of the plasma membrane. At the moment we don’t understand precisely how RAS activates either of those enzymes, because we don’t understand exactly how the proteins insert in the membranes, and how RAS induces conformational changes and rearrangements in the proteins as they get recruited into the plasma membrane. But our hope is, just like solving the structures of the proteins for the first time, when we understand the nature of the activation process in the plasma membrane, we will find new opportunities to intervene, because we’ll find biochemical steps in the whole activation process which can be blocked by small molecule attacks. So that’s the premise, and we have some evidence that actually already might be true, we’ve learned a little bit more about how RAS activates RAF, and which parts of RAS are essential for the activation process, beyond just the binding of RAS to RAF at the membrane, there are other sequences which are involved in activation, and interfering with those sequences presents a new opportunity to target preventing RAS activation of RAF kinase, which again was not on the table until quite recently. Just by studying RAS and its effector interactions in the membrane we’ve already seen a new opportunity to intervene. 

Hartley: What are your views on the models we use to work with RAS? Cell lines, organoids, mice?

McCormick: I think each of these models have their pros and cons, it depends what you’re trying to measure. The fundamental fact is that RAS activates the MAP kinase pathway in all cells, including flies and worms, and the pathways are the same, literally, in flies and worms. So getting hung up on what cell type we’re using, for that particular pathway, is not very helpful. But it may be true that if you want to ask other questions, does RAS activate different pathways in epithelial cells, then obviously you have to use the right cell type. If we’re just looking at activation of MAP kinase as a measure of RAS activity, you can do it in any cell which is convenient, including Drosophila, which is where the pathway was first defined, obviously. People have a knee-jerk response that all RAS experiments should be done in epithelial-derived tumor cell lines, but even there, in a tumor cell line you have a spectrum of cells which range from full mesenchyme-like to full epithelial-like, depending on how you sort them, so it’s not like they are homogeneously epithelial. And RAS mutations don’t originate in differentiated epithelial cells, generally, they seem to occur in either stem cells or progenitor cells. Those are the cell types which one should be interested in if you’re looking at initiation of cancer, not at fully differentiated epithelial cells that have to de-differentiate essentially to become a cancer cell. Choosing the right cell model depends on what question you’re trying to address.

Hartley: I was struck by the recent Wellspring paper where one of their G12C drugs was dramatically more effective in spheroids than in monolayers.

McCormick: That’s a very good point, and if you go back to the history of where RAS oncogenes were discovered and what the whole thing was about, the classic Weinberg papers looked at the effect of RAS oncogenes on 3T3 cells, right? And the assay was whether they grow or not, it’s whether they form foci or not, which is overgrowth on top of a stationary cell culture. So the property that they impart is not just proliferation, it’s proliferation under conditions were normal cells are arrested. That assay became the workhorse of the oncogene field, focus formation, not proliferation. That’s a very different feature of RAS than just the ability to support growth of cells in 2d. Any cell will grow in 2d if you give it enough growth factors. In the old days there was a lot of debate about the right assay for transformation, so people argued that 3d cells are more reflective of malignancy in vivo than 2d cells, which I think is true, it’s a different aspect of RAS signaling, the ability to sustain survival in the absence of attachment is a different assay than just going through the cell cycle in the presence of low serum. So I think that growth in 3d reflects more of the malignant features of RAS than does growth in 2d. And then growth in vivo raises the bar even further.

Hartley: It’s interesting that focus formation is a type of 3d growth.

McCormick: In those days it was really hard to make monolayers of 3T3 cells which behaved well, so that all the normal cells would stop completely when it gets confluent. Stu Aaronson and those guys at the NCI in Frederick at some point figured out a protocol which would give no background. Cells tended to degenerate over time and you started getting random foci. But if you took low passage 3T3 cells, they stop dead when they reach confluence, and you can add as much growth factor as you like and they won’t activate MAP kinase. But mutant RAS just keeps on going and then you get these morphologically different cells which can pile up and grow on top of the monolayer, that’s the feature which defines oncogenesis, not just growth. It’s a spectrum, I think, from that all the way up to growth in vivo. I’m sure the Wellspring people found in some cells if you have enough serum around they’ll keep going, and the RAS dependence is very hard to see when the cells are happily growing anyway.