Posted 9 July 2004
Martin Citron is the Associate Director of Research at Amgen.
Interviewed by Gabrielle Strobel
GS: What is the hypothesis driving the research in your lab?
MC: I'm working on Parkinson's and Alzheimer's disease. In Alzheimer's, we are pursuing treatment approaches that would lower amyloid, the hope is to come up with compounds or molecules that could be used to lower brain amyloid.
GS: We'll get to your work with α-synuclein later on. To start, I'll put one of our Alzforum poll questions to you. You may know the box on the right-hand side of our home page where we post a new question every month. Recently, it was: "Do you think Aβ has a normal function?"
MC: I haven't seen strong evidence that Aβ has a normal function, except for some electrophysiological data. I personally have no hard data to prove this. Frankly, I believe that Aβ is a piece of junk.
GS: You think it's junk?
MC: Yes. My hypothesis is that there are probably a whole bunch of peptides like Aβ floating around that are degradation products of membrane proteins and exist in small amounts. Nobody has cared about them because they were not involved in any serious disease. Aβ, because of its involvement in pathology, has been studied in great depth, and people have come up with the idea that it should have some function. But the default, or the easiest, assumption is that it has none.
GS. That puts you in the minority. In our poll of 109 respondents, 20 percent agreed with you and 17 percent were undecided, but a whopping 62 percent said they did think it had a physiological function. Why would that be?
MC: We can say one thing for sure: It doesn't have a vital role, because the APP knockouts, while they have some problems, are viable. The β-secretase knockouts are viable. There's no way that Aβ can have a central and critical role. That's very clear from the available data.
Then the question becomes: Does it have some subtle role? It's very hard to exclude this possibility. There might be some minor problem in neurochemistry or behavior, and more work is needed to sort this out completely.
GS: I think you alluded to Roberto Malinow's research suggesting Aβ exerts a negative feedback control on synaptic transmission (see ARF related news story). This has generated buzz in the community because synaptic dysfunction has become a focus of attention in the search for mechanisms of early cognitive symptoms. Yet I've seen few, if any, studies along this line to follow up on Malinow's original work. Have you read it?
MC: I have. Even so, from a drug development perspective, I doubt it would have negative implications if we reduce Aβ by, say, 50 percent. So for me, this is a theoretical concern at this point.
GS: One other hypothesis on Aβ function would impinge on drug development. Some groups propose that Aβ is a vascular sealant that separates the brain from the rest of the body and protects against hemorrhage. By this reasoning, amyloid is a good thing to have, and therefore, amyloid-removing therapies could do damage.
MC: I think that doesn't hold a lot of strength because the β-secretase knockout mice are fine and the APP knockout mice are fine. If Aβ were needed to protect from catastrophic hemorrhages, then the mice should be dead or at least have gross pathology. We have analyzed the β-secretase knockout mice at two years of age and we didn't see any (Luo et al., 2003). I can't exclude subtle neurochemical changes, or subtle behavioral changes because we haven't done a whole behavior battery on the mice, but if those mice had hemorrhages due to the absence of Aβ, we would have picked that up.
GS: Then why do most tissues of the body make Aβ? Hidden in this is a question about the function of APP and its cleavage. It just moves the question of function back to APP or its other fragments.
MC: In the simple model, where Aβ is a piece of junk, APP indeed has some physiological role but we don't understand it fully. APP can be cleaved by α-secretase and by β-secretase. In normal circumstances, the vast majority of APP is cleaved by α-secretase and you produce this secreted piece, APPSα. APPSα has some physiological role that is not fully understood, and I'm convinced that full-length APP also has a physiological role.
A small proportion of APP gets cleaved by β-secretase. Then the remaining C99 stub happens to be cleaved by γ-secretase almost as an afterthought, and you produce small amounts of Aβ.
If you look at β-secretase, APP is a poor substrate for it. β-secretase has some other physiological role that we don't understand yet, but which has nothing to do with APP cleavage. I think Aβ production is essentially an accident of nature, and nobody cared about it until people started to get 80 years old.
GS: Should you give more thought to the differences between mouse and human? Perhaps in humans, the biological circumstances under which APP gets cleaved, under which BACE is active, under which γ-secretase is active, differ considerably from the mouse?
MC: I don't think so. The only major difference is that mouse APP has some mutations that make it an even worse substrate for β-secretase than APP. Otherwise, the fundamentals of processing are very similar. For example, a major portion of APP goes into the α-secretase pathway in mouse cells as well as human.
GS: To pick up on your thought on "junk" peptides: Are there other examples of fairly ubiquitous peptides that have no function? Maybe knowing of other such breakdown products would make this idea that their production is really just an accident more palatable.
MC: I should dig into the literature on that. I don't know any off the top of my head. You can't easily publish a paper describing a junk peptide, and given the way science works, people don't go into cataloguing peptides that they can't ascribe any function to. Go back to the history of APP processing in cells in the early nineties, and you'll see it took a lot of effort until people could visualize the Aβ peptide and see that it was made from intact cells. This required generating high-affinity antibodies to detect the very small amounts that are being made, and nobody would go to that length to find some peptide of unknown function.
GS: Does the concept of ectodomain shedding come in here? If people studied that more, perhaps they would come across more examples of junk peptides.
MC: People study the large secreted N-terminal piece, and they don't care very much about the membrane stub anymore. The α-secretase has various substrates other than APP, and BACE may, too. One could, in principle, try to find the peptide analogues from these but it's technically quite an effort to raise specific antibodies to small peptides.
GS: You are best known for identifying BACE, almost five years ago now. Given what you've learned since then, what do you think its normal function might be?
MC: I still don't know. People are starting to propose other substrates for BACE, and some of those proposals are then confirmed with either our knockout mice or knockout mice that other people have made. The two examples that look most convincing to me are P-selectin glycoprotein ligand-1, (Lichtenthaler et al., 2003 and the sialyl-transferase ST6Gal1 (Kitazume et al., 2002). Beyond that, I don't know yet. The fact that the knockout mice seem very healthy makes it difficult to assign a function to this enzyme.
GS: In your mind, is BACE still a great drug target?
MC: Yes. So as far as the biology goes it's very promising. Major efforts in the field are directed to it.
GS: The BACE active site is wide and difficult to inhibit with a small-molecule drug. Is it still true that none of the inhibitors found so far are realistic drug candidates, merely experimental tools, or has this changed?
MC: That's hard to answer because this is mostly happening in companies, and at least the big ones won't tell you where they are in their programs. It's quite doable to make specific inhibitors to BACE, what's challenging is then making small molecules that have the right pharmacological properties. I can't say exactly where everybody is in this race.
GS: There was a spate of papers on BACE inhibitors around the turn of the year, a half-dozen within two months. They all looked like experimental compounds to me. Many of them were from company labs, and the mere fact that these companies released them in the public literature rather than keeping them as a trade secret reinforced my sense that they were not serious candidates for drug discovery. Everyone is waiting for BACE inhibitors that could move into the clinic!
MC: I've read all these papers, and your view that these are tools rather than real drug candidates is right. We need to be patient here. If you look at the history of HIV protease inhibitors, it's also taken quite some time from the concept that inhibition of this protease might be useful to seeing drugs good enough for the clinic. In that case, you didn't even have to get them to the brain. It's going to take a while. But I personally think that at some point those compounds will enter the clinic.
GS: Let's follow up on the BACE knockout mice. Initially, they seemed normal and that was deemed great news. Now, the picture is getting more complicated. For one, the knockout mice may have a subtle learning deficit, as you and collaborators have reported (see ARF related news story). Others think that unexpected phenotypes will show up once the mice are stressed. What do you know beyond absence of a developmental phenotype?
MC: As far as pathology goes, we've looked in old mice and haven't seen anything. Regarding learning and memory, it's too early to come to a final conclusion. The major finding of the Ohno paper is that BACE deficiency can solve some of the memory problems that you see with the Tg2576 transgene, so that's the good news. And there's the paper by the GlaxoSmithKline group, which suggests a role in anxiety (Harrison et al., 2003). But even if there's a behavioral phenotype in a knockout or transgenic, a pharmacological intervention wouldn't mimic a knockout or a transgenic. These genetics experiments are not directly applicable to drug therapy but instead guide you to the areas you have to study. We've seen for years that removing BACE doesn't kill the animals or doesn't do major damage.
GS: The paper you just mentioned, I found it interesting. They report that mice overexpressing BACE in neurons were bold and exploratory, while mice not expressing it in neurons were timid. Other scientists have said privately that they expected to see a behavioral phenotype in BACE knockouts once the mice are old enough to test. In general, do you prefer to look at the overt symptoms (viable: yes-no, gross pathology: yes-no), and perhaps consider this current work as harping on small points? You think that despite these complexities a pharmacological inhibition of BACE still makes sense?
MC: Yes. It's true that knockouts can raise red flags about pharmacological intervention. Let's say, hypothetically, if BACE knockouts develop colon cancer, we would be very concerned for drug development. With a knockout, the gene is absent throughout development, so any developmental effects of the gene come into play and the knockout could have a more severe phenotype than pharmacological intervention.
Bart de Strooper's group has shown at a meeting that BACE expression is particularly high during development in the brain, so anything you see in the brain in the knockout mice could really be a consequence of development that you wouldn't necessarily see with a pharmacological intervention.
On the other hand, because the gene is absent all along, you could see compensation in these animals that you might not necessarily see with a more subtle pharmacological intervention. In that respect, the knockout phenotype might actually be milder than that of the pharmacological intervention. I think we must be careful when interpreting knockout results for drug targets.
GS: Diana Dominguez in Bart de Strooper's lab has made BACE1 and 2 double knockouts. Half of them died by three week of age, perhaps because they're much more sensitive to common infections in the mouse house there than mice typically are. This is preliminary data presented at a meeting. Does this surprise you, given that some of the other BACE targets appear to play a role in the immune system?
MC: I haven't seen the original data yet, but it would certainly be interesting. Maybe it's got something to do with compensation and BACE2. Even less is known about BACE2 and its substrates.
GS: Let's talk about the APP intracellular domain (AICD). It is generating much interest these days. What is the role of BACE in making it, and do you think this is an important event?
MC: Once you get APP's C-terminal fragment formed, γ-secretase cleavage will release the AICD to the cell. You can also get there via α-secretase cleavage, so BACE cleavage is not the only pathway to generate the AICD. How important a role it plays is hard to say at this point. I haven't seen a paper yet that demonstrates that AICD generation is critical.
GS: As an editor I observe that many people are on the edge of their seats to learn which genes AICD regulates. This follows the finding that it goes to the nucleus and acts to activate some genes and repress others. A number of labs are working on identifying these genes, and it would open up the signaling role of APP.
MC: The rationale for this followed from the Notch intracellular domain. It's such a striking parallel that you tend to think, well, if that happens with NICD, maybe AICD also does something important. But I haven't seen anything in vivo yet.
GS: Are you following the research on lipid rafts?
GS: I don't understand it well, but suspect they will prove important. Trafficking of APP processing components into these membrane domains, signal transduction across them, even cytoskeletal dynamics of either side of them are questions of interest here. With age, as various lipids go up or down in concentration, perhaps the rafts' composition and fluidity changes and so does APP processing as a consequence. You can tell I'm waving hands, but I sense it will be important. What's your take on this?
MC: You are probably right from a cell-biological perspective. I just don't know whether it will translate into therapeutic approaches. If you start to deal with the fundamentals of cell biology, then I'm not sure how I would intervene therapeutically.
GS: The Holy Grail in AD diagnostics right now is finding a robust, predictive marker for presymptomatic AD. This will be even more important once experimental treatments look like they might work. What, to you, are the most promising leads?
MC: The brain imaging field, particular amyloid imaging and brain volume measurements at this point. I haven't come across anything in terms of straightforward serum markers other than those things that have been in discussion for many years, each of which has problems.
GS: I want to ask you a general question about AD research. In April, two columns by Sharon Begley in the Wall Street Journal brought into public view what some scientists in the field have been saying for years. She charged that the amyloid hypothesis has monopolized AD research to the exclusion of all other approaches. She writes that this slows down progress, has put all therapeutic eggs in one basket, and that the amyloid hypothesis is wrong, to boot. Have you read this?
GS: What's your view?
MC: I personally think the columns misrepresent the big picture. They remind me a bit of the controversy on whether HIV is the cause of AIDS or not. I recall in the early days, when there were no HIV drugs, this discussion was very active. There were people who had quite some arguments against the hypothesis that HIV causes AIDS, for example the fact that you couldn't detect significant amounts of the virus in the blood. While all these people were articulate and vociferous about the shortcomings of the HIV hypothesis, they didn't come up with fully testable alternative hypotheses, and now that the drugs are known to work the issue has died down.
Everybody in our field knows the shortcomings of the amyloid hypothesis. I find that these two articles in the Wall Street Journal miss the point on several angles.
GS: Which ones?
MC: For example, it cites the work of Ashley Bush and the clioquinol trial as an example of something that's outside the amyloid field (see ARF Live Discussion). But if one reads his papers, one of the major assumptions is that clioquinol reduces amyloid, so I think his work takes a special angle on the amyloid field but is within it.
The other story was Jie Shen's presenilin mutations as loss of function (see ARF related news story). Jie knocked out all four presenilin alleles—PS1, PS2—and gets a dramatic phenotype. In the Alzheimer's brain, you may have partial loss of function of one of the presenilin alleles plus you have the other three present, too. Now, does her mouse totally model what's happening in the Alzheimer's brain? I don't think so, and in her paper Jie does not claim it does. Jie's point can fairly be made, but the WSJ article portrays it out of balance.
There is this argument that the presenilin mutation works via a loss of function and the amyloid hypothesis is wrong. I have yet to see an explanation for the fact that APP mutations that cause Alzheimer's disease are either within the Aβ peptide or just around the cleavage sites. APP is 700 amino acids long, and if the mutations that cause AD work through a loss-of-function, then you would expect that they would be all over the place. You should have nonsense mutations throughout, not just these very selective few mutations that all drive up Aβ or change its aggregation properties. I think before claiming that there's a sort of conspiracy of the establishment going on, one should look at the whole field in balance.
GS: Your invoking the HIV debate strikes me as coincidental. We recently organized a Live Discussion around an article in the Journal of Medical Ethics by a South African ethicist. He asked whether scientists have a responsibility to resist taking their internal disagreements to the general public when there's a public health interest at stake. He quoted an example that has led to disastrous public health policy in South Africa. This is an extreme case, and drugs proved clearly who was right and who was wrong, but we wanted to invite the AD field to use it as an opportunity for some navel-gazing (see ARF Live Discussion).
MC: The major difference is that for HIV, it's clear who is wrong, whereas with amyloid there is no proof. My main concern is that if you think the amyloid hypothesis is wrong, then you should come up with an alternative, testable idea. Not something vague like Alzheimer's is a disease of aging, or you have too much or too little oxygen. It must be testable.
GS: Do you consider any areas outside of the amyloid hypothesis particularly promising?
MC: Yes. For one, the whole tau area. If one were able to block tangle formation or resolve existing tangles, that would be great. Second, the entire area of inflammation in Alzheimer's disease. That's a convoluted field, but it may play a role. Third, interventions at the level of cholesterol. Both with inflammation and cholesterol, there may be reasons to tie them into the amyloid pathway. But then again, maybe they're not tied into the amyloid pathway at all. I think either could potentially lead to treatment breakthroughs even before any anti-amyloid drug has been tested in phase 3.
GS: Because drugs already exist?
MC: Yes. Another area—and I don't know how fruitful it is at this point—is ApoE. My personal impression is that more than 10 years after it's role in AD has been discovered, we still don't understand at the molecular level how ApoE4 increases your risk of getting the disease. I think that would be an area to learn something.
GS: I've covered ApoE extensively, and there are almost as many suggested mechanisms as labs working on it, which is not very many (see, for example, ARF conference story). Like with inflammation or cholesterol changes, some labs study the role of ApoE within the context of the amyloid hypothesis, and others look at it outside of it, but none has really hit a home run yet in any way that I can see.
MC: Yes. It is hard to prove anything in this field. This is why the antiinflammatories or the cholesterol-lowering drugs are so attractive. Once you can show these agents work in a clinical trial, then there will be intense motivation to find out what exactly they are doing. We have no pharmacological tool to turn ApoE4 into ApoE3. That's one of the problems in the ApoE effort. You can see a variety of effects, but to conclude which one is the key important one is very difficult.
GS: Which facet of AD research should receive more funding than it does now?
MC: Biomarker research. That's something companies won't touch in a major way, I think.
GS: Why not?
MC: Because they are focused on moving a drug forward. Researching biomarkers doesn't give you a drug per se, it just helps to make the testing of some future drugs easier. Few companies will stretch themselves for that. The likelihood of success is relatively low.
GS: Plus companies want a drug that people take for years. Not a one-time sale.
MC: I think the one-time diagnostic test would be interesting, but it's not a major product in itself for a company focused on drug development.
GS: What, to you, is the single most vexing question in Alzheimer's research?
MC: One thing I can't get my arms around is the role of inflammation in the disease. Exactly which aspects are productive and which aspects might be worth suppressing pharmacologically? Some of this is moving very actively right now, and I think we'll know more in a few years, but right now this is confusing.
Another question is what happens downstream of Aβ production? I'm personally convinced that Aβ42 becomes harmful when it's either overproduced or not cleared enough. But there are still a lot of holes in the exact mechanism downstream from Aβ.
GS: What kind of therapy is likely to be most effective? You'll say BACE inhibitors! (I shouldn't second-guess you.)
MC: I won't say BACE inhibitors would be more effective than any other way of lowering secreted Aβ. Immunotherapy might work. The antiinflammatories might still work, it's too early to write them off. It's hard to predict!
GS: The amyloid hypothesis has shifted away from plaques as the primary toxic agent and toward the view that oligomers, protofibrils, the whole continuum of species coexist and each is bad in its own way. From a drugs perspective, is it important to distinguish between these? Some say, "Yes. Dissolving plaques does no good when there's insufficient capacity for clearing Aβ, it will only generate more toxic oligomers." Others say, "No. Just dissolve the plaques, the rest will be degraded or carried away." Are these fine points, or is this therapeutically important?
MC: They are probably fine points. From a secretase inhibitor perspective, it doesn't matter because you are upstream. From an immunotherapy perspective, it would matter if your antibody does not recognize one species and is exclusive for another species. Again, it's too early to tell.
Evidence has accumulated that the oligomers are harmful. But I haven't seen data to suggest that full-length fibrils are harmless or that the plaques are harmless. So I agree that probably they are all bad. There is limited data that monomers are bad, (but see Hayashi et al., 2004). It appears to me that everything downstream from the monomer is harmful.
GS: That seems to become a consensus view.
What causes sporadic AD?
MC: I don't know. It does not appear that massive overproduction of Aβ is a major contributor. Whether there are deficiencies in clearance or enhanced aggregation, or somehow increased susceptibility to the toxic effects of Aβ or its oligomers or fibrils, is hard to say at this point.
GS: I see a dichotomy in the field between scientists who consider Alzheimer's a heterogeneous syndrome, with different predisposing and acquired factors converging on a similar outcome in different people, and others who think there is going to be one key pathway that will explain the disease and make good drug targets. Where do you come down?
MC: For every disease, you have the splitters and the lumpers, right? From a therapeutic perspective, you want to be on the lumpers' side. Otherwise you get to levels of complexity that, in the end, you can't do anything about. I would say that AD has multiple causes, but I think the likelihood that Aβ plays a critical role in all forms is pretty high, so one should go after Aβ.
It's a bit like with heart disease, where cholesterol-lowering treatments seem to be beneficial regardless of the exact cause of the heart disease.
GS: You also work on α-synuclein fibrillization. What do you think is the critical step there? And how do you explain α-synuclein's role in Parkinson's.
MC: I'm not sure α-synuclein is as critical in Parkinson's disease as amyloid is in Alzheimer's. It appears that enhanced aggregation of α-synuclein may be a major pathway towards Parkinson's disease. That's based on the mutations and also on the recent finding that if you increase the gene dose of α-synuclein, you also get PD (see ARF related news story).
GS: How about the dimerization you describe?
MC: John Carpenter has been the major driver of this work. He published in collaboration with us that the oxidative dimer formation is the critical rate-limiting step (Krishnan et al., 2003).
GS: Why the dimerization? It caught my eye because every protein that aggregates must, by definition, go through smaller steps and I am interested in where the similarities are with other neurodegenerative diseases that involve protein aggregates. I wonder whether it will be possible to find what the rate-limiting step or most critical step is in each of these diseases.
MC: That will vary from protein to protein. We and others have found that it takes extremely long to aggregate a-synuclein without drastic measures, and it's the dimer formation that's really rate-limiting. The kinetics of that are different from, let's say, the Aβ case, but at least in these two cases, you have oligomerization of what appears to be toxic entities from smaller monomers.
In both cases, you have mutations that enhance this, and in both cases, you have the increased gene dosage, which also gives you the disease (Down's for AD), so there are some parallels.
GS: That puts you in company with John Hardy. He also proposes that a gradual rise, sometimes accelerated by genetics, in the concentration of these peptides leads to formation of the noxious aggregates across these diseases (see ARF conference story).
How do the A53T and A30P mutations of α-synuclein fit into this picture?
MC: They both accelerate the formation of this oxidative dimer, and thus the overall aggregation process.
GS: Where in the cell would this happen?
MC: I am not sure. The protein is cytoplasmic, so it might happen there. Our studies are strictly in vitro, so I can't say exactly what would happen in the neuron. The in-vitro studies we've done jibe with the in-vitro studies that were done with Aβ.
GS: That's a criticism often leveled at oligomer and aggregation studies: they're almost all in vitro.
MC: True, but there's a large literature out there on the other effects that these mutations have. It may be too early to completely decide which effects are relevant, but if you go for, let's say again, partial loss of function, then why would triplication of α-synuclein give you Parkinson's disease?
GS: How does the finding about α-synuclein fibrillization tie in oxidative stress?
MC: This dimer formation is oxidation-dependent. It's tempting to tie the in-vitro aggregation together with all the literature that oxidative damage may be involved in Parkinson's disease, but we haven't produced any in-vivo data to further prove this.
GS: Are there mouse models prone to oxidative stress that could be crossed with, say an a-synuclein- overexpressing strain or other models to bring this about in vivo?
MC: Some of the chemical models, for example the rotenone model, ultimately come down to oxidative stress. Yes, this may be an area worth looking into.
GS: Can antioxidants treat or prevent neurodegenerative disease? It would be a logical conclusion.
MC: Absolutely, there may be some effect there. Whether that would hold up for diseases other than Parkinson's disease I don't know. You need to get those into the brain and into the right cells at the necessary doses, and I don't know what the side effects would be. There are practical problems.
GS: Peter Lansbury wrote that in ALS, dimer formation of SOD is a normal step and actually prevents its aggregation into other, more sinister SOD oligomers (see ARF related news story). By that reasoning, in ALS with SOD, the dimer is what you would want to promote, maybe, with a drug, whereas with α-synuclein, you would want to prevent dimers from forming.
MC: I don't know how strong the case is that aggregates drive ALS, it is not exactly my field. I thought there was still uncertainty about whether you find those aggregates in all cases of disease pathology. In AD, the mutant forms have amyloid pathology, and the sporadic cases have amyloid pathology, as well. We therefore say that anti-amyloid therapy is worth pursuing for everyone. In ALS, I'm not sure whether the sporadic cases show aggregates of SOD.
GS: What can you tell us about Amgen's AD program? Your name is well-known in the AD field but Amgen is not nearly as high-profile an AD drug developer as, say, Elan, Lilly, or even small companies like Neurochem, to pick just one example.
MC: One reason is that while Amgen has efforts in the area, their role in its overall portfolio is smaller than in a company like Neurochem, which essentially owes its existence to the amyloid drugs that are moving forward right now. The second reason is that Amgen tends to be silent about things that are in early stages of development not only for AD, but also for other indications. For example, in neuroscience we have two molecules with novel mechanism of action in development for pain, which we have not widely discussed.
What I can say is that we are pursuing several areas of investigation in Alzheimer's disease with focus on amyloid and that Amgen is committed to the neuroscience therapeutic area and to expanding the neuroscience research effort.
GS: What drew you to Amgen?
MC: If drug development is your main interest, you can pursue that better in the pharmaceutical industry than in academia. You can do more discovery research in academia, but to really try to translate that into therapies you almost need an industry environment. I would really like to see the amyloid treatments come to pass.
As in every place, there are advantages and disadvantages. You can get a broader view of therapeutic areas and you get to know more treatment approaches. You can't drill as deep as you could in academia. You can't work on one molecule for 20 years, but at the same time, you get a more comprehensive view on treatment of neurologic disease and you focus on what you think is most relevant for treatment. That's always been my favorite aspect. I personally never cared that much about what the normal function of APP might be, for example.
GS: How much of what you actually study can you publish?
MC: We can publish quite a bit, as you've seen with the β-secretase. I think many people in academia think that we have a very controlling environment where they forbid you to publish great data that you've produced, and you have to put them into a drawer. For me it's not so. The limitation comes more in terms of the time you have for basic, publishable discoveries. Much of the time is spent to develop drugs, let's say, for a target like β-secretase, and that work wouldn't necessarily be of broad, general interest.
GS: We thank you for this interview.
MC: My pleasure.