AT A GLANCE
Assistant Professor, University Louis Pasteur, Strasbourg, France
PhD, Scripps Research Institute, organic synthesis, and natural product chemistry
In your recent PNAS paper, you describe a small molecule microarray to profile enzymatic activities in cell lysates. How did this project come about?
The idea for this assay came from thinking about combining combinatorial chemistry and microarrays. We can synthesize a library in a strict pool format, which is by far the most efficient synthesis format in terms of generating molecular diversity, but yet you can screen this library in a spatially addressable and a miniaturized format using a microarray. There is a very high level of interest in industry in profiling methodology.
How does the assay work?
We have a library of small molecules which are designed to interrogate the function of certain enzymes in a crude cell lysate. Either we have substrates or we have inhibitors, and basically these test whether an enzyme is functionally active. These small molecules are each tethered to a specific fluorescein-tagged oligonucleotide; we actually use PNA [peptide nucleic acid]. We add them to a crude cell lysate, so the small molecules can bind to the proteins. The way we detect a binding event is by separating the compounds that have bound from the ones that have not by size exclusion filtration. After that, we hybridize the protein-compound complexes to an oligonucleotide array and measure the fluorescence. We have been very fortunate to work with Affymetrix, which has given us GenFlex tag arrays [which can detect up to 2,000 different molecules, each of which is coupled to a unique 20mer oligonucleotide tag].
We were also able to characterize an enzyme that had bound a compound. In this case we used the small molecule inhibitor as an affinity capture probe. We then did a trypsin digest and identified the protein by mass spectrometry. In the paper we restricted ourselves to seven cysteine proteases, but that doesn’t mean that the method is limited to that. We have actually looked at a few other substrates, for different enzyme classes, but this still remains unpublished work.
What are the main applications?
The simplest experiment is to take a disease cell line and compare it to its healthy counterpart and ask ‘What’s the difference?’ The second question after that is, ‘Is the difference the cause of the disease, or is it a downstream casualty?’ The most interesting differences are the ones that are intimately involved with the disease. And that is one of the strengths of the method, that we have a small molecule inhibitor to assess whether the difference that we are observing is relevant or not.
It can also be used as a miniaturized screen. Say you have discovered a new protease, and you would like to find an inhibitor against it, you can use this method. Not only can you find in one shot whether you have an inhibitor for that protease in your library, but you can also tell what the selectivity of your inhibitor for that specific protease is vs. the other proteases.
Another application is in diagnostics. There are a lot of deregulated enzymes that cause a disease. To be able to measure the function of an enzyme in a very miniaturized and parallel format is something that is very desirable in the diagnostic industry. A lot of cancers, for example, overexpress proteases, and a lot of infectious disease agents also utilize proteases.
Why did you choose to work with proteases?
Proteases are one of several classes of enzymes where there is a poor correlation between the presence of the protease and its activity. They are typically expressed as zymogens, meaning enzymes that are not yet active. Probably one of the most famous examples is thrombin, which gets activated during blood coagulation. If we had to wait for the whole gene expression machinery to get turned on, we would all be hemophiliacs! There are a lot of processes in the body where you have proteases that are not active and waiting to be turned on by another protein. It’s also a very druggable class of enzymes; there are in fact already quite a few drugs against proteases.
What were the main technical challenges?
The one thing that required a bit of optimization was the size exclusion purification, but at this point it’s a solved problem. It’s critical that the compounds that are not bound to the protein are removed.
How about losing bound compounds during the filtration step?
At the moment we are using mechanism-based irreversible inhibitors that will form a covalent bond with the enzyme. Thus, we don’t have to be afraid to lose the interaction between the compound and the protein. We are also interested in exploring small molecules which do not need to form a covalent adduct, but we haven’t had the opportunity to do that.
To avoid the problem in the first pass, we just decided to make mechanism-based inhibitors. In retrospect, I think it was a very good idea, because it gives a lot of specificity to the compounds, and we avoid a lot of nonspecific interactions. In serum, for example, there are a lot of sticky proteins. We don’t see those, thanks to the fact that we can wash with fairly stringent conditions in the size exclusion filtration step.
At first sight, one might think there are not many mechanism-based inhibitors, but scanning the literature, you find a lot of these inhibitors for many chemical reactions. Certainly all the proteases have mechanism-based inhibitors, as well as some of the kinases and phosphatases.
How specific are the inhibitors?
They are very specific, but in reality they will be as specific as a protease is. If a protease is not specific, it will react with every inhibitor in there. But proteases tend to have some level of specificity, and we can exploit that.
Could this array be expanded to a larger format?
Absolutely. There can be more than one type of inhibitor in one library, for example. Potentially these chips can be used with a library of 400,000 compounds. Or we could join many libraries onto one chip: Let’s say we have a 10,000-compound library targeted against cysteine proteases. Then we could also have 10,000 against aspartyl protease, and zinc proteases, and then another 10,000 against kinases.
Are you planning to develop the assay commercially?
We come from an academic background, and our primary interest is the development of new tools and using them. It’s not that we are not interested in commercializing this, but we don’t have the same drive as private companies. We have patented the technology, but nobody has approached us yet.