Skip to main content
Premium Trial:

Request an Annual Quote

Harvard s Timothy Mitchison on Cytoskeletal Proteins as Drug Targets

Premium

At A Glance

Name: Timothy Mitchison

Position: Professor of Systems Biology, Harvard Medical School; Co-director, Institute of Chemistry and Cell Biology, Harvard

Background: Assistant/Full Professor, Pharmacology and Biochemistry, University of California, San Francisco — 1987-1997; Independent Research Fellow, National Institute for Medical Research, London — 1985-1986; PhD and Postdoc, Biochemistry, UCSF — 1980-1985

Timothy Mitchison was one of the initial faculty members of Harvard’s department of systems biology. Created last year, it was the school’s first completely new department in 20 years. Mitchison’s lab focuses on the dynamics of the cytoskeleton, and uses imaging-based assays in living cells and in vitro extracts, among other molecular biology techniques. Inside Bioassays caught up with Mitchison to discuss his work in this area.

How did you originally become interested in the dynamics of the cytoskeleton?

In my PhD [in San Francisco] with Marc Kirschner I started working on it, and remained interested in it ever since.

You published a paper last year in Combinatorial Chemistry and High-Throughput Screening [2003 Jun; 6(4): 279-86] on high-throughput fluorescence cell imaging for phenotypic screens, and you also work on assays for cytoskeletal function. Tell me a little about these projects and if they are related.

Yes — I moved to Harvard to try to do small-molecule discovery in the area of the cytoskeleton, with the idea of getting research reagents, but also with the idea of trying to help jump start discovery of therapeutic drugs in the area. The microtubule cytoskeleton is a fairly important target in cancer therapy. There’s a drug called Taxol that gets used a lot, and Taxol’s a horrible drug — we really ought to be able to do better than that. So I wanted to participate in that. We wanted to essentially do drug discovery on the cytoskeleton, so we needed methods to look at the cytoskeleton in high-throughput so we could do screening. And so we’ve been developing automated microscopy for that application.

I’m pleased to say that not one of our drugs, but one of the proteins I’ve been working on — my first student worked on a motor protein called Eg5 where we got some compounds that hit it back in 1999. It’s a kinesin motor protein that plays an essential role in mitosis. And we got a good small molecule that poisoned Eg5 back in 1999, and now various biotechs and pharmas [have] started chasing it as a cancer target. People have compounds now in Phase II trials, so if that pans out, that would be one example of this kind of research leading to real progress in cancer therapy. There are a lot of “if’s” though — things fail for all sorts of reasons.

What sorts of reasons?

Drugs fail in clinical trials because they either don’t work or are too poisonous. I don’t know what’s happening with the Eg5 [drug]. Academics can help point the way towards drug discovery, but it’s difficult to do real drug discovery.

I’ve heard of Taxol, and there are a few companies focused on drugs targeting the cytoskeleton, but has that been a huge drug target [for drugs other than] Taxol?

No, not really. There are a few other microtubule poisons — the Vinca alkaloids — but no, there aren’t really many.

What makes the cytoskeleton a good target, in your opinion? Where does the promise lie in using the cytoskeleton as a drug target?

Well, I didn’t say anything about good. I don’t think Taxol is a good drug. The hope would be to get something as effective as Taxol, with a lot less toxicity — that still won’t be a great drug, but it will be a lot better than Taxol.

I’m not actually arguing that the cytoskeleton is a fantastic target in cancer. A lot of people want to target the causes of cancer rather than the consequences of cancer, like cell division, and maybe that will be possible. You basically just need to try everything. Taxol does work, it’s just rather poisonous and you get a lot of resistance developing. If you could do a bit better or as good, but different, then that would be really useful.

The flip side to that question, then, would be what are the major difficulties in developing drugs against the cytoskeleton? What roadblocks need to be overcome?

The major obstacle, I think, is there’s always a problem when you have whole new classes of proteins that have never been drugged before, you just don’t know how easy it is going to be to get good compounds.

I think conceptually, though, the bigger problem is that the cytoskeleton is quite similar in all the cells in our body, so getting the specificity to treat cancer without poisoning normal cells is the problem. That’s why Taxol is a problem. And so I think even if you could target Eg5 successfully, the problem is that the stem cells in your bone marrow also use Eg5, so you would still have bone marrow toxicity, presumably. That’s really the problem — the cytoskeleton is involved in all sorts of basic processes. In general, that doesn’t make for a good drug target.

Can you tell me a little more about the imaging platform you’ve developed?

We haven’t put much money into instrumentation — we’ve purchased instruments from companies that are building high-throughput microscopes. We’ve tried to keep on the cutting edge of that technology. In fact, we just had a new one delivered today, and I think we’ve become known as one of the academic centers doing that.

I’d say we’ve put our energy more onto the application — we’ve really shown that you really can use high-throughput microscopy to screen drug libraries if you want to. We have a big study that we’re getting ready to publish using high-throughput microscopy in what I would call a profiling application; that is, not high-throughput screening, but gaining a huge amount of information, kind of analogous to a DNA chip.

More and more I think where I want to put our efforts is in data analysis — software for taking these huge numbers of images, and extracting the information out of them. [Regarding] that review chapter that I wrote [in Comb. Chem. High Throughput Screen] — what’s noticeable in that chapter is that there is no automated analysis. All that early work was done by just looking at the images and writing something in your notebook, and clearly that’s not acceptable. In the long term, it’s too labor-intensive, you’re going to miss a lot, it’s not reliable, et cetera. We need fully automated ways of extracting mechanistic data from huge numbers of images. For example, this new paper we’re preparing is all about that — there’s nothing special about the microscopy, really. It’s about taking thousands upon thousands of images and extracting a lot of data out of them.

What types of vendors are you using for your imaging platforms? It seems that high-throughput imaging is still relatively young.

It is, and I think it’s only going to get bigger and bigger. We used two generations of instruments from Universal Imaging — the MetaMorph platform. They’ve been very good, but now there’s a third generation of instruments coming out.

There’s an excellent microscope that’s been built by Axon, which I haven’t used, but I’ve heard good things about. We just got a beta-site instrument from a company called [Applied Precision], which I think is going to beat the pants off the rest of them. I’m very excited about it. Up until now, people have been essentially taking microscopes and running them automatically — what Applied Precision has done I think is something that is fundamentally different. They build microscopes, but they also build a very good DNA chip scanner. And they’ve now really built an instrument that’s a hybrid between a microscope and a chip scanner. It’s using technology from their chip scanner, which is really superb, but it’s too expensive for most people. It doesn’t compete with the cheap laser-scanning chip scanners. Anyway, this new instrument that we have, we’re just putting it through its paces now, is kind of a cross, and built by people who really know automation.

But I’m not knocking UIC — we’ve gotten really excellent work out of them. The major company in the field until recently was Cellomics, and the only reason we never went with them is that their instruments are really out of our price range. I’m always looking for instruments that can be acquired for under $200,000 — that’s sort of my upper limit. People make these million-dollar instruments, and I don’t know who they’re going to sell them to — frankly, not us academics.

What’s next for your lab and do you see any commercial potential being realized?

I’m very excited about the use of this high-throughput microscopy for profiling and for doing the things that people are trying to do with DNA chips. That is, taking a smaller number of samples — let’s say hundreds instead of tens of thousands — and getting a great deal of information on them that’s relevant to mechanism, and this can be for drug discovery or another context. And this sort of extracting information relative to drug discovery, I think, is going to be very interesting.

On the commercial side of things, the work that you cited is more primary hit discovery — taking a chemical library and looking for hits. I don’t think microscopy is going to be used for that, for various reasons. Basically in pharma or biotech, people prefer pure protein targets if they can get them. Where I see this microscopy, and in particular, this in-depth information being really important, is in what people call the “hit-to-lead” phase, where you have maybe a thousand molecules and you need a lot more information. Which ones really do what you want? Which ones are toxic? In what way are they toxic? I’m not in professional drug discovery, but I’ve heard that hit-to-lead is a major bottleneck for them. And the more you can predict behavior in animals and people, the faster you can go, and I see a huge role for microscopy in that area — not higher throughput necessarily, but more information.

The Scan

Shape of Them All

According to BBC News, researchers have developed a protein structure database that includes much of the human proteome.

For Flu and More

The Wall Street Journal reports that several vaccine developers are working on mRNA-based vaccines for influenza.

To Boost Women

China's Ministry of Science and Technology aims to boost the number of female researchers through a new policy, reports the South China Morning Post.

Science Papers Describe Approach to Predict Chemotherapeutic Response, Role of Transcriptional Noise

In Science this week: neural network to predict chemotherapeutic response in cancer patients, and more.