Whitney Laboratory for Marine Bioscience
At A Glance
Name: David Zacharias
Position: Assistant Professor, Whitney Laboratory for Marine Bioscience; Departments of Neuroscience, Physiology, and Pharmacology, College of Medicine, University of Florida, since 2003.
Background: Senior research biologist, Merck Research Labs, San Diego, Calif., 2000-2003; Research associate, Howard Hughes Medical Institute, University of California, San Diego,1997-2000; Postdoc, Mayo Clinic, Scottsdale, Ariz., 1996-1997; PhD, biomedical sciences, Mayo Graduate School, 1996.
David Zacharias has had plenty of experience with automated imaging, having been an early user of Cellomics' ArrayScan while with Merck, and later the Q3DM EIDAQ 100 (now Beckman Coulter's IC 100). Zacharias, who still serves on the scientific advisory board of Vala Sciences, a descendant of Q3DM, recently won an NIH grant to develop a cell-based high-throughput imaging assay for palmitoylation at his lab at the University of Florida. Zacharias took a few moments last week to discuss his work with CBA News.
How did your research lead you to the use of high-throughput cellular imaging and image-analysis technology?
I went to Merck in 2001 with the intention of developing live-cell assays based on GFP. I had been working with Roger Tsien in San Diego on GFP, and I developed a lot of methodologies for measuring protein-protein interactions in a 2-D space using fluorescence resonance energy transfer [FRET]. I had hoped to be able to generate some cell-based assays using similar kinds of technologies for drug discovery, but when I got to Merck, I found that typically the dynamic range of the signal that you get from FRET-based assays using GFP is really too small. There has been one successful GFP/FRET-based assay, and that was for caspase — and that's intramolecular, where you've got a protease site between a CFP and YFP on the same molecule. So that's the simplest possible and most robust system you can develop. When you're doing intermolecular FRET, the dynamic range is typically a lot lower and it's a lot harder to control the stoichiometry of the donor and the acceptor.
I was hoping to use this FRET-based technology to monitor small molecule interactions with proteins using CFP and YFP, and watching to see what happens when we screen a library of a million compounds. But as I said, the dynamic range ended up being insufficient for high-throughput screening, and that's something I only came to learn after some time in industry. I think most academics don't understand how robust an assay has to be to get to a high-throughput screen. As an academic who made that transition to industry, that was an interesting fact to learn. But I still was really keen on using fluorescent proteins to develop assays.
To do that, I visited Merck in Terlings Park, [UK], where I was kind of a guest of the entire site, but visited Peter Simpson's lab and saw for the first time this instrument that could make morphometric measurements of cellular phenomena — the Cellomics ArrayScan. I was absolutely stunned that you could measure the length of neurites in millions of cells on a multi-well plate in a few minutes. During my graduate work I took pictures of neurons and measured neurites with a yardstick on slides up on a screen. I could get about 10 in a week if I was lucky. So at the time [at Merck], the NF?B translocation assay was out, as well, and they explained that to me. I thought: If I'm ever going to use fluorescent proteins for drug discovery, this kind of technology is my outlet. It was about that time that Merck started to bring in different platforms and develop those platforms to the point where they could be used for high-throughput screening. In 2001, Cellomics' ArrayScan was the king of the hill. We wanted to bring in new technologies, and there were some technologies in development that used confocal, and some that had incubator chambers — some really fancy stuff was coming out. So I was part of a committee to explore these new technologies, and the committee chose to bring on two or three new platforms, and the Q3DM instrument was one of those platforms. I was absolutely thrilled that I would be able to continue using fluorescent proteins to do these drug-discovery assays, and that's really how I got to learn about this technology.
So you developed a relationship with Q3DM …
I worked with Q3DM since my time with Merck to develop the EIDAQ 100, which is now the [Beckman Coulter] IC 100. They were a local company, just a few miles away from our Merck site, and I developed a close relationship with all of those guys. They really tailored an instrument to the needs we had, and specifically for fluorescent-protein based assays.
How did you start using this for your current research?
I also had some interest in palmitoylation — some work that I had done in Tsien's lab caused me to become really interested in the mechanisms that regulate the palmitoylation of proteins. I could see that this was by far the best method available to study palmitoylation, but at the time, there wasn't even an algorithm that existed to quantify plasma membrane localization of a fluorophore. So I started to harass my friends at Q3DM, now Vala Sciences, to develop an algorithm that would allow me to measure this distribution — not just how much is on the plasma membrane, but to be able to follow incrementally the departure of the fluorophore from the plasma membrane, or a plasma membrane-to-cytoplasm translocation assay. During the development of that algorithm, they were bought by Beckman Coulter, and a few of them went on to make Vala Sciences. I could see that this technology wasn't something only for industry, but that it would be an enormous boon to cell biology studies in academia. I had always been frustrated by the genuine lack of quantitative analysis that existed in most cell biology studies. This was the way to get truly hardcore quantitative data that was not biased by the user. That was one of the primary reasons I left Merck — my lust for this ability to do really quantitative cell biology in an academic center.
The nice thing is that the NIH under the brilliant leadership of people such as Linda Brady, Jim Inglese, and Chris Austin, the NIH has taken strong notice of this technology and incorporated it into the NIH roadmap. All of the assays that I develop for my basic cell biology are immediately translatable to high-throughput screens that fit into the imaging component of these Molecular Libraries Screening Centers. It's actually a seamless translation from my work into this area. There is also an increase in the outlet for the publication for this kind of data, and that's specifically the journal Assay and Drug Discovery Technology. This is really kind of a one-stop place where an academic can come to understand the rigor of producing a high-throughput assay, because academics have not traditionally thought in terms of millions of data points. But to get really quantitative cell biology measurements, it takes those sorts of numbers. As an academic, I'm interested in making millions of measurements on a single compound rather than millions of compounds with a single measurement.
Are you still using the Beckman Coulter platform now?
Yes, but the model I have in my lab is a late-generation EIDAQ 100 — it's still all open architecture, which is what I prefer, because then I can really control the light path in ways that you can't now that Beckman has stuck it into that mauve kitchen cabinet. Friends of mine down at Scripps at Florida have this IC 100, and I'm not sure what to think about it. To me it looks less user-friendly than the older version, but the way [Beckman] has it set up now is probably better suited for industry.
The image-analysis algorithms that you're now developing — is it in collaboration with Vala?
Yes, I do collaborate directly with them, and in the interest of full disclosure, I'm a member of their scientific advisory board and a consultant for the company.
We've been focusing primarily on the development of that plasma membrane-to-cytoplasm translocation assay, and that's complete. It's really a brilliant piece of work that's allowed me to quantitatively measure palmitoylation. There will be some work that I'll be able to publish next month: One of the first proofs of concept that this algorithm works was to determine the residence half-life of palmitates on a fluorescent protein-based substrate for palmitoylation. It's essentially the end terminus of GAP43 fused to the N-terminus of GFP. I was able to inhibit palmitoylation with 2-bromopalmitate, which is the common inhibitor that people use, and then over time follow the translocation from the palmitoylated form at the plasma membrane, to the non-palmitoylated form in the cytoplasm. The residence half-life turned out to be within minutes of the same measurement that was made biochemically by two other groups in the past.
So is the Vala software open to tailoring by a researcher for their specific needs?
I haven't tried to tinker with it myself. I let the experts do that. I think that a person with skills that lean in that direction greater than my skills could probably tinker with that algorithm and fine-tune it for their own interests.
Do you know if others are attempting to look at palmitoylation with high-content screening?
I don't know for sure, but I think it's a safe assumption that people are. It's recently become really hot. We've identified druggable targets now in this pathway.
What is the importance of the palmitoylation process?
It is a reversible post-translational modification, much like phosphorylation. It's been fairly difficult to study, and early on there was a bias among many scientists that the process was auto-catalytic, implying that there were no enzymes that actually did the job of adding palmitate to proteins. That attitude really inhibited the development of the field for a very long time, and that's one reason that I went into it. That was the advice of Roger Tsien, when I was in his lab: Find out what everybody else is doing, and run the other way as fast as you can. Six or eight years ago, when I became interested in this, it appeared to be a field that wasn't overcrowded by enormous egos and personalities and budgets, but I think that's about to change. Now, a family of enzymes called palmitoyl transferases, or PATs, have been identified, and to some extent, characterized. There are two or three of these that have been implicated in disease states — one of these is HIP-14, and the evidence is tenuous, but one of them is probably involved in one form of schizophrenia, and that's the one that's most interesting to me.
Do you have specific things in mind for the NIH grant money?
It's not a huge budget — it's a one-year R21. But it allows me to develop and validate an assay for palmitoylation. The grant was written to validate an assay for antagonists of palmitoylation, but in the end, the more important mode is probably an agonist assay. We'll do both of these, and we're developing probes that will allow us to develop and validate an agonist assay, which is a hell of a lot harder than the antagonist assay. But the point was to develop an antagonist assay to the point where the Z-score was sufficient that it could move forward to a high-throughput screen.
So you do have plans to do large-scale screening of modulators of this process?
Absolutely. I've talked to Linda Brady and Jim Inglese several times about where to move this assay once it's developed. Jim has some interest in palmitoylation, as well. Eventually this assay will move to either the intramural program or to one of the Molecular Library Screening Centers, like at the Burnham Institute, where [Vala Science's] Jeff Price is. Likewise, it could move to Scripps Florida, because there is a relationship between the University of Florida and Scripps. There are plenty of outlets for this, and I think there is a big interest, because so many proteins in the synapse are palmitoylated that it's got to have a huge impact. The regulation of this on protein trafficking and protein function is going to be enormous — more than people imagined.