Professor, department of cell biology and chemistry
Scripps Research Institute
At A Glance
Name: Benjamin Cravatt
Position: Professor, department of cell biology and chemistry, Skaggs Institute for Chemical Biology, the Scripps Research Institute, since 2004. Assistant, associate professor since 1996.
Background: PhD in macromolecular and cellular structure and chemistry, Scripps Research Institute, 1996.
BA in biological sciences, Stanford University, 1992.
In 2002, ProteoMonitor interviewed Benjamin Cravatt to find out about his chemical probes, which label only active enzymes (see ProteoMonitor 8/12/2002). Next month, Cravatt is scheduled to give a talk on "activity-based proteomics" at the Swiss Proteomics Society Congress in Zurich. ProteoMonitor decided to talk with Cravatt to find out more about what the term means, and to get an update on his work.
What have you done with those enzymes that you targeted [using your chemical probes]?
The types of experiments we would perform were relatively standard. You're still doing comparative systems biology profiling, so you're taking healthy and diseased cells, or whatever your two comparatives are, and looking for differentially regulated in this case enzyme activity. So instead of doing a change of experiment or a 2D gel, you're looking for changes in protein function with the hypothesis being that if you can look at the functional state of proteins, you might be able to gain a deeper appreciation for the molecular elements of cells that are actually contributing to the phenotype that you're examining. So if it's a cancer cell, normal cell, fatty liver or normal liver, you're looking for enzymes that are involved. You use these probes to identify enzyme activities that are differentially expressed in those two samples, and those become the hypothesis-generating experiments that lead you to focus on specific enzymes to study in more detail.
What are the main applications of your work in this area?
We spend most of our time in the cancer research area, because I think it's well suited for looking at elevated protein activities, given that you have so many model systems that you can look at. Cancer is some sort of aberration of protein expression and activity that lead to various properties that make them pathogenic invasion, migration, metastasis. So we've been looking for enzyme activities that might be linked to those phenotypes.
Now we have a couple of nice candidate enzymes that we're studying in more detail. So it really is a tool that allows us to focus on specific proteins that are consistently elevated. For example, in invasive cancer cells, if we see enzymes that are elevated in invasive cancer cells across a variety of tumor types, that becomes a high-priority target to study in more detail.
Anyone who tries to convince you that they can actually solve problems with proteomics or genomics is probably pulling your leg. All these tools can really do is focus you on new hypotheses, and then you have to take more classical approaches to try to understand what those proteins are actually doing. Our lab is very committed to doing that as well not just throwing out large datasets to the community that nobody ever follows up on.
Can you describe what 'activity-based proteomics' is?
Well, that's basically the method. Activity-based protein profiling is basically the method of taking chemical probes, throwing them into a proteome and looking for changes in activity. What I've just described is basically what we call activity-based proteomics.
What enzymes have you found that are particularly promising?
We've found some very interesting enzymes. We're actually quite interested in the enzymes that we've found that have no known function. We've found an uncharacterized membrane hydrolase that's highly elevated in cancer cells that are invasive for many different tumor types. We actually have a decent idea of what that enzyme does, but it's not published yet. The identification of that enzyme as being elevated in cancer cells was published in our 2002 Proceedings of the National Academy of Sciences paper, and we've been studying it ever since. It's taken a few years to figure out exactly what it's doing, but it's quite interesting, and I don't think anyone would have found that any other way.
The whole point of the method is that we're not trying to find proteins that are already well characterized. We're trying to find proteins that no one knows what they do. So they're not going to have names because no one knows what the heck they do.
We also found a lot of enzymes that others have seen as potentially being involved in cancer, so that's some cross validation.
Would you say this is a high-throughput technique?
I don't think we're trying to solve the problem of throughput. That's not our interest. We're trying to solve the problem of content. So I consider most proteomics efforts to be low content. They provide you with an abundance-based measurement of a change in a protein, and considering that most enzymes are regulated post translationally, you really don't have any idea whether the protein that you're measuring is active or not. So our goal is to understand whether proteins are active, and which proteins are active. I don't care if it takes us five times as long to do that. I want the content to be as enriched as possible. So we're not trying to things faster and smaller, we're trying to do things better.
The probes don't label enzymes that are in an inactive form. They don't label zymogen forms of proteases; they don't label inhibitor-bound forms of enzymes. By the sheer fact that they label the proteins, it provides you with a read out of the state of activity of those enzymes. So we're not just profiling enzymes based on their abundance. As a matter of fact, if there's an enzyme that's present in huge concentrations, but it's not active, we won't see it. We won't label it all, because we only label active species.
Are you working mainly with human systems for this?
We've worked with mouse systems as well, but obviously, the closer we can get to the humans, the better off we are, because it's more closely related to the human disease. So yes, we try to work with human cell lines and primary human tumors if we can. If we need to use mouse models as well, we will.
Do you think this will lead to a diagnostic or a therapeutic?
We're hopeful, yes. I mean, diagnostic minimally that definitely is a possibility. The targets we're identifying are novel markers of cancer phenotypes, so in principle they could be diagnostics. We're hoping they'll also be targets. That's the ultimate goal to try to find new therapeutic targets, and to validate them.
Now that you've identified this one enzyme that's very active in cancer, what's your next step in taking this forward?
Well, I don't think there's anything too fancy in what one has to do next. It's just like with any protein. You find a way to disrupt the protein specifically and see what the consequences are on cancer cell biology. That's the direction that we're going. Hopefully if you disrupt the activity, it disrupts some part of cancer cell biology migration, invasion, tumor growth then it becomes more of a potentially validated target. If it's a validated target, you try to develop inhibitors and use those things to treat cancer.
Again, I don't think genomics and proteomics is too complicated after the stage of identifying interesting expression patterns of proteins or activities. From there on out, it's just more classical biochemistry and cell biology to sort out what those targets are doing.
Do you work with other researchers as kind of a service?
Yes, we obviously have tons of collaborations with cancer researchers. We're helping them to identify active enzymes in their biological systems. We have to do that within reason because we're a small academic lab, so we can't form a service industry. But we have 20-plus ongoing collaborations.
Have you thought about turning this into a commercial activity, where you provide a service for researchers?
I think realistically that's just not a viable business plan. You're not going to be funded to do that. Unless the NIH funds some kind of center to do it, which they haven't done yet, my impression is that it's a noble endeavor which is in reality not going to happen. There's no money. I don't have the money to set that up, and I don't know where to get the money to set that up. But it's a great idea. I think it would be fantastic. It would definitely relieve some of the burden on us.
Are you working on developing any other kind of technology?
The other thing that we've done, which is quite nice, is complement this approach with metabolite-profiling efforts to try to understand what the endogenous biochemical function is of the enzymes we identify. So if you get a novel enzyme out of a cancer cell, and you have no idea what it does, we can use RNAi and knock it down, or inhibitors and try [to see] what it does at a cellular level, but understanding what its endogenous substrates are has been a real challenge. So we've developed this metabolite-profiling effort so we can now look at the effects of inhibition of an enzyme at a global small-molecule level to see [which] substrates for the enzyme might be changing in response to its inhibition. That's been really helpful. That's definitely helped us identify the endogenous substrates for uncharacterized enzymes. We like that technology a lot.
Are there other agents out there that do the same thing as these enzyme active-site probes?
I think there's a whole slew of chemical probes that label enzyme active sites now since this technology emerged in the late 90s, but if you're asking me whether there's a complementary technology that can capture the same type of information that this whole set of chemical probes does, I don't really think so. I don't think it's too easy to do these types of experiments with classical biological approaches. This is really a good example of where chemistry has a unique impact on biological systems that would be hard to achieve with molecular biology techniques. So I think that's why there's a lot of interest in the field, because it's hard for biologists to gain access to this level of information content using kits that you can buy from Qiagen.