Skip to main content
Premium Trial:

Request an Annual Quote

Stephen Michnick Discusses the PCA Assay and its Role in Drug Discovery

Premium

At A Glance

Name: Stephen Michnick

Position: Associate Professor, biochemistry department, Université de Montréal — 1995-present; Chair, Scientific Advisory Board, Odyssey Thera

Background: Postdoc, chemistry department, Harvard University; PhD, University of Toronto

Stephen Michnick is well-known on the commercial and academic drug discovery scene primarily as the developer of a technique called protein-fragment complementation assays (PCA), in which fragments of a reporter can be separately expressed with proteins that are known to interact to assay whether a particular compound is having an effect on the given pathway. The powerful technique is the basis of the drug discovery strategy used by drug development firm Odyssey Thera, and has also found significant application in basic cell biology research. Michnick spoke with Inside Bioassays last week to weigh in on PCA and cell-based assays, in general, as well as their roles in modern drug discovery.

You’re probably most well-known for developing PCA strategy. Can you describe the evolution of that technique?

It came from convergent directions. I was trained as a structural biologist, and the last project I did before I changed my direction was studying a protein called FK506 binding protein. The notion — which has now become extremely popular — that you can use natural products or chemicals to probe biochemical pathways and determine functions of genes, was emerging at the time I was doing this work, and was in fact emerging in my lab. The central question that came to me was that it was very nice to think you might be able to do this with chemicals. But chemicals that do very specific and interesting things are not that common. So the question I asked was: How could you, at will, try to connect up the products of genes, and try to figure out what they’re doing with each other in the cell? So you needed a very generic way of doing that, if you wanted to do that for any kind of biochemical process. It’s not like you can measure everything in the universe. My view was that, if you could measure anything, you could measure one thing that would be a proxy for everything else that’s going on, and that would obviously be protein interactions.

Then the question was: How do you measure protein interactions and specifically, how do you do it in living cells? This was another aspect of what I wanted to do — to do this in living, intact cells. I knew as a protein chemist that doing anything in vitro would get you nowhere. Well, not nowhere, but it’s an exceedingly slow process, even with the sorts of technologies that people use now, like protein chips, and so on. It was sort of a conceptual thing, too: Does it really look in vivo what it looks like in vitro? At the same time, I had started the program in protein folding. And it occurred to me that protein folding was a perfect way to measure things. If you could couple protein folding to a process that you want to measure, it would make a terrific probe. So then the question was simply: How do you do that? Then the idea of fragment complementation evolved from that. If there was some way that I could just attach a protein that folds, or unfolds, to one protein, and then, as a result of another protein interacting with it, I could induce the folding of the reporter protein, then that would be great. I couldn’t conceive of any way of doing that, but the fragment complementation made perfect sense.

You touched on this briefly, but how do you see cell-based assay approaches such as this leading to better or more efficient drug discovery than biochemical assays?

Let’s start out with this idea of being able to measure things generically in intact cells. By generically, I mean let’s say you want to probe any specific pathway. How are you going to do it? You can develop a specific assay for that pathway, but what if I want to look at 20, 30, 40, or 100 simultaneously. You’re not going to have that many different assays — certainly not cell-based assays. So you need to measure something generically. The central premise of what we do here, is that virtually every biochemical process in the cell can be monitored, within the hierarchy of pathways, by probing specific protein interactions or protein complexes. The idea is that some perturbation of a pathway is somehow going to impinge on that interaction, and tell you something has happened in this pathway, as a result of that thing — whether it be an siRNA, or a hormone, or a drug. What you can imagine is as follows: You have a drug. You think it’s a very specific drug. You think it’s extremely specific. You’ve shown with every biochemical assay that you can, for instance, that this thing binds only to this target that I’m interested in, and only very weakly to what’s homologous to it. Well, that doesn’t tell you what happens in the cell. It doesn’t tell you that in the universe of thousands of proteins and their variants in the cell, that nothing else will happen. So it’s the old ‘looking for the keys under the street lamp, because that’s where the light is’ problem. So you can imagine that by introducing what are called ‘sentinels’ — assays that report not only on every specific pathway, but on different levels of hierarchy on a pathway — that if you threw one compound on that cell, and only the pathway that you would predict lights up, you would say: Wow, that’s a really good thing. If some other pathways lit up, you might look at them and say: That’s not good, that’s a pro-apoptotic pathway, or that signal’s telling us that you’re turning off DNA repair pathways. So you could set up a panel of assays and do this in one fell swoop on any lead compound, and you would have a significant contribution to drug discovery.

There are two sides to this — the assay technology and the instrumentation, or readout, technology. Do you see these evolving together, or is one limiting the other?

From our point of view, measurement technology doesn’t limit us. In fact, another aspect of the PCA approach is that it’s not an assay [per se], it’s a family of assays based on a concept. And there is a reason that generality is important. Let’s say that a perfectly reasonable way to measure what you want to measure is with an enzyme that simply allows for survival under some selective condition — an antibiotic marker. Survival selection assays from an instrumentation standpoint is about one of the unsexiest things that you can imagine. But it works. So for instance, for doing protein engineering stuff, which we do a lot of, that’s the method of choice because it’s fast, efficient, and the only instrument that it requires is the human eye.

Anyway, we developed assays that are based on fluorescent proteins, luminescent proteins, enzymes that produce a fluorescent product — and you can measure them any way that you want. You don’t choose the instrumentation; you choose the assay that’s appropriate to the problem you want to solve.

Do you endorse or prefer any particular instrumentation to conduct these assays?

If I said we’re using a particular manufacturer’s device, that would be unfair. Any instrument you buy, you do so for a purpose — especially plate readers. I’m not going to say that so-and-so’s device is better than another device; it just does something that we need to do.

The only thing I would say — which I think is very obvious, and a lot of people will tell you — more and more, imaging is becoming the method of choice in cell biology analysis. And quantitation is something that is starting to be revived. There’s always been interest in it, of course, but there’s new interest in using imaging as a way to deal with things. And there are good reasons for that. You can develop a lot of interesting cell-based assays that have nothing to do with, say, simple intensity. Very obvious ones being things like nuclear translocation assays, where it’s obvious you need imaging technology to do something like that. But it can even be more subtle. Sometimes our response is a change in a signal going from diffuse, with no change in intensity, to [concentrated] with the same intensity — another thing you’ll never measure with a simple spectroscopic approach. So all developments in imaging technology [are] desirable, and I should add, all the new stuff going on in in vivo imaging is really important, too.

The IP protecting these assays is the basis of Odyssey Thera’s drug discovery platform. Do they own the IP?

Yes.

Are you still involved with Odyssey?

I’m the chairman of the scientific advisory board, and I collaborate with them. And they also support research in my lab. Full disclosure. And I’m a shareholder. Full, full disclosure.

You still work with the technique in your own research. Can you tell me a little bit about your current research?

Globally what we’re trying to understand in my lab is how biochemical pathways are actually organized in the cell. It’s starting to become a hot topic, and it’s something we’ve been thinking about for a long time. It gets back to this idea of the generic nature of protein interactions as probes of biochemical pathway organization. We do this in a number of ways. We have projects that range from a very simple, but grandiose one, which is that we’re mapping all the interactions in yeast, just using a simple survival selection assay. And the reason for that is just to ask: Can we capture a single snapshot of how a proteome is organized? Obviously people have already done this with other approaches, such as mass spectroscopic approaches, and yeast two-hybrid approaches. We think we can go beyond what has been done, and complement what has been done, so that’s why we’re doing that. We’re obviously also thinking of doing this in higher organisms, including humans. But from there, there are really two aspects of doing this approach. One is to actually detect interactions. I always tell people that they should always remember what the real genius of the two-hybrid strategy was, at its inception, which was it completely generalized the whole problem of expression cloning by saying: OK, forget about what specifically a protein does. Just associate it with another protein — something physical that is common to what proteins do. And that will give you your first inference, from which you can then test a hypothesis that the gene does something. What we did is to take that kernel of genius about the two-hybrid approach, and expand it. What you can do with PCA is, you can say whether an interaction occurs or doesn’t occur. But what PCA provides you with is a direct measurement of what happens to that complex, in the context of which it is normally expressed, when you perturb the pathway that you think a gene is involved in.

So for example, our most recent work was a paper in Nature Cell Biology, from April [2004 Apr;6(4):358-65], where we reported on an essential link between two signal transduction pathways that are normally opposed to each other. What we found was that there was a very simple link between the TGF-ß signaling pathway and the general tyrosine kinase receptor signaling pathways. We discovered this link in a screen, and then we followed it up with biological experiments. Once we found the link, via just the screen, we then perturbed both these pathways to ask: Well, is the complex responding to these perturbations? Because if it’s just a constitutive, non-specific interaction, it won’t do that. And it did respond, so that gave us the next clues to go forward and do the biological experiments that would validate all this. So what PCA does is provide for a first-pass validation that an interaction is biologically meaningful.

We also study protein folding in and of itself, but it’s also something that evolved parallel to all this, because the assay strategy is a protein engineering and protein folding problem, and we have to understand that in order to do it well.

Where do you see the overarching field of cell-based assays going? Do you see any major trends developing?

Sure, just from the point of view of cell biology. We think we understand biochemical processes now from a chemical perspective. We understand a lot of them, but we don’t understand enough of them. The real question is, at the end of the day, what’s happening in a cell? What can we measure in a cell that’s quantitative, without cracking it open? And does what we see follow from what we know happens in a test tube? You can put it as simply as that. From a practical point of view — from that of something like drug discovery, the cell-based assay has a number of important ramifications. For one thing, no matter what chemistry and biochemistry you do, chemicals and natural products that have these intricate relationships with molecules that make up a cell, that we can not ascertain easily, those are things we have to know. In the context of drug discovery, it’s to avoid the sorts of things that it typically falls into, such as discovering off-pathway effects that simply would not be predicted. Immediately we ask: If we could do this in a cell — just in cell culture — imagine what it would save us in effort and time; even just an inkling that something is wrong here, or something is right, or there’s something about the way a molecule interacts with the cell that may lead to a problem that we can keep an eye on, or may lead to an alternative therapeutic utility that would be an advantage. If we knew that in the first experiments that we did, imagine how it would change drug discovery. So I think that there’s nothing but cell-based assays, as far as I’m concerned. That’s the only thing worth doing.

I should also add that there’s a whole community of very bright, young bio-organic chemists, who are coming up with all sorts of wonderful ideas. Jim Inglese [of the NIH] has got this really great section of this journal he just started [Assay and Drug Development Technologies] that consists of abstracts from theses — what he thinks might be cool emerging technologies. I commented to him that it’s one of the best parts of the journal. There are all these things that kids are doing that may be complete flops, but they are really great ideas.

 

The Scan

Purnell Choppin Dies

Purnell Choppin, a virologist who led the Howard Hughes Medical Institute, has died at 91, according to the Washington Post.

Effectiveness May Decline, Data From Israel Suggests

The New York Times reports that new Israeli data suggests a decline in Pfizer-BioNTech SARS-CoV-2 vaccine effectiveness against Delta variant infection, though protection against severe disease remains high.

To See Future Risk

Slate looks into the use of polygenic risk scores in embryo screening.

PLOS Papers on Methicillin-Resistant Staphylococcus, Bone Marrow Smear Sequencing, More

In PLOS this week: genomic analysis of methicillin-resistant Staphylococcus pseudintermedius, archived bone marrow sequencing, and more.