On Friday, Scripps Research Institute will hold a ribbon-cutting ceremony for its new temporary home at Florida Atlantic University in Jupiter, Fla. Gov. Jeb Bush, one of the prime movers in getting the La Jolla, Calif.-based institution to set up separate operations in southern Florida, will be in attendance.
Over the next two years, an all-star group of scientists that Scripps has enticed to Florida will work in the building while a battle wages over the exact location for the permanent facility in northern Palm Beach County and how to spend upwards of $800 million in public money earmarked for the institution with the hopes of promoting economic development for the area.
Already the facility has been gearing up for research activities in drug discovery and development. This summer, it will install a $10 million Kalypsys robotic system for high-throughput screening.
Leading the installation of the institution's core lab facilities are John Hogenesch, who is a professor and head of biomedical sciences and genome technology and formerly program manager of genomics for the Genomics Institute of the Novartis Research Foundation in La Jolla, Calif.; and Pat Griffin, a professor of drug discovery, and formerly senior director for basic chemistry and molecular profiling for proteomics with Merck.
BioCommerce Week spoke to both managers about the facilities being developed at Scripps Florida and the tools that the scientists will use. BCW spoke to Hogenesch about the genomics and HTS facilities and then with Griffin about the proteomics and small-molecule labs.
Head of Biomedical Sciences, Genome Technology
One of the trends in laboratory construction seems to be the idea that instruments need to be more mobile. You seem to be following that model as you have moved from widely separated temporary facilities, to new temporary facilities.
John Hogenesch: We were at Boca, and at West Palm, and there were some people who were waiting for this temporary site to open. Now everybody has moved in. We are setting our facilities up in temporary space; and we're just going to move them when we have the permanent facility.
What are the temporary quarters like?
We are in a 40,000 square-foot facility at Jupiter, at the John D. MacArthur campus at Florida Atlantic University. We have 100 employees, 80 on-site and another 20 in the pipeline. We should be at 150 employees by the end of this year in this building. The employees really comprise three separate areas. The first area is basic biomedical research, R01 academic science. Those include Nagi Ayad [assistant professor, biomedical Science], from Marc Kirschner's group [Department of Cell Biology, Harvard Medical School] at Harvard, and Teresa Reyes [assistant professor, biomedical science] from the Salk Institute. I also have a lab in basic science for my work on circadian biology.
A second component is a technology component, and I think it is one of the areas that makes Scripps Florida somewhat unique. We have a group of laboratories that are meant to provide core services in genomics, informatics, engineering, and proteomics, and will be available to researchers in academia as well as in drug discovery.
The third component is drug discovery with the Kalypsys robotic machine. We are going to do full-bore drug discovery in an academic/industrial atmosphere with representatives from both the pure academic world and the pharmaceutical world. We have people like Chris Liang [associate director, medicinal chemistry], who was at Pfizer Sugen, and Claes Wahlestedt [professor and director of pharmacogenomics], who was at Karolinska and Aztra-Zeneca, and Nixon Ramos, who was at Merck Rosetta, and a lot of people with significant industry experience working on site.
The idea is for them to come together, for people who sit right next door to each other, to talk and to begin to strike up collaborations with the idea that real leaps in knowledge are going to be made via collaboration.
We have a really excellent chemistry department headed by Bill Roush, who was former head of chemistry at the University of Michigan. They are building a medicinal chemistry group, so when we do a screen on the Kalypsys robot and we have a list of hits and have selected several scaffolds to investigate further as lead series, we have a team of medicinal chemists upstairs who can put resources on those scaffolds and screens and begin to build more drug-like molecules.
In genomics, what are the tools you are using?
We have a microarray platform, Affymetrix, which I have worked with extensively and [I'm] very comfortable with. We will have a core that will perform arrays. Hand the core a sample and you get your RNA expression dynamics back. We have been doing two things with that platform. We are doing resequencing, basically tiling down genomic regions, and amplifying them with PCR, hybridizing the arrays looking for polymorphisms a la Perlegen. We are using Affymetrix arrays. We are not going after the whole genome, we are basically targeting specific classes of proteins. The second related technology is high-density genotyping with the 100K chip set and the soon-to-be 500K chipset from Affymetrix. We will offer those as core services internally.
Are you looking at a secondary array platform to use?
Yes. I was at the Genomics Institute [of the Novartis Research Foundation] of La Jolla for the last five years and we looked at a number of other platforms. And we will probably use a number of other platforms — because not all problems fall into those limited sets. For example, another technology we have to do genotyping is the Sequenom system — principally because of its flexibility. But if you want to do 100,000 SNPs on a single sample, there's not many better ways to do it than Affy.
It really makes sense for us to concentrate our resources in one flexible platform that can accomplish a lot of our goals rather than support multiple platforms on site.
We try to get the best price possible and volume does help. The decisions are not always made on a purely cost basis, though. Sometimes the devil you know is better than new devils that pop up now and again, even if they are cheaper devils. I'll give you an example: We really like a particular primer vendor, IDT. They don't give us necessarily the lowest price. We can probably do better price-wise with a lot of other vendors. But over the years, we have used them for long oligos, short oligos, oligos just for genotyping, oligos to tack bases onto genes, lots of different applications. They are very high quality for many of these applications. After a while, when you keep coming back and finding that their oligos are performing at the level that you need them to be again, and again, and again, it becomes a reflex to buy them. And, even if someone else comes in a couple of cents lower per base, we are not going to risk something simply for a few pennies.
Any thoughts about high throughput?
Not really. Our group [at Genomics Institute of Novartis] developed the array of arrays. After a while, we saw two things occurring. One was that chip usage went down over the years, and the second thing was that technicians who were performing these things didn't like to do experiments where if it went wrong, $50,000 would be lost. There was a pretty big psychological force being applied there. It wasn't like it was coming out of their pocket — if something goes wrong, it goes wrong. They just didn't like the idea that a botched experiment was going to cost that kind of money. So, they would largely steer clear of them.
Doesn't that speak to an advantage for automation?
We tried that extensively, and we were unable to get the automation to work as robustly as a couple of key technicians. Ninety percent success was not good enough.
What other equipment are you using?
We have offline stuff, shared between genomics and lead discovery for doing siRNA screens and cDNA screens. Those would include pieces like the PerkinElmer ViewLux and the Beckman-Coulter Q3DM high content imaging machine, and the Molecular Dynamics analysis machine, which basically reads luminescence, fluorescence, and [fluorescence polarization].
What are the capabilities of the high-throughput screening system?
The Kalypsys robotics fully automates the process of small-molecule screening. The system that we purchased can screen approximately 500,000 to 750,000 small molecules in a day in 1,536-well format, in volumes between half a microliter and 5 microliters. So because it can screen in that small volume, it cuts costs down significantly per assay, maybe by a factor of 20.
The system is about 40 feet by 20 feet, and that includes compound management. One of the really unique things about the system is that compounds are stored online and they are constantly accessible. They get switched out every six months. We have continuously available compounds. There is no carting it off to the freezer. You run your scripts for data analysis, you have it pick your 1,000 hits from your assay. You can have it rack out those hits, set up to do an 8-point dose response curve on all 1,000 molecules and generate a lot of high-quality data for the medicinal chemists to look at and make the best decisions.
What kind of computation hardware do you have?
The IT group here is setting up a 100-node, Opteron processor Sun Linux Beowulf cluster.
What are the challenges in the Florida environment?
There is not an installed base of technicians in Palm Beach County. In San Diego, there are more than 300 biotech companies and three major universities, so there is a constant flux of technicians. Here there isn't that installed base.
Will there be training?
Yes, but that is a half-decade to a decade away. It will take a while to get senior technicians trained and in here.
So, you will be looking to draw technicians from other areas. At least the weather in Florida is somewhat comparable to San Diego.
Well, for half a year it is. It's a lot hotter here.
How about other things, like real estate, and salaries?
Real estate is about half the price of San Diego, and salaries will be roughly comparable.
Some of the large tool vendors are offering to come in and manage all the instruments in a lab. Is that of interest to you?
That is really akin to the IBM service model, the Lou Gerstner switch in 1990ish. I guess if we were confident that they could run those pieces of equipment, we would probably be interested. But, there are only five or six Kalypsys systems installed in the world, so we would have to have evidence that [they] knew what they were doing, like they used to run one.
How about other instruments, like microarrays, mass specs?
I think that would be attractive if it was attractive from a pricing point of view. But, I think you are going to have to have multiple installed bases in an area, so you could have someone you could flip between institute one and institute two and do the QC on a weekly or monthly basis. I don't think they are going to have a lot of action there.
What should the vendors know about you as a potential customer?
Both Pat and I have run groups at Merck or at GNF at La Jolla and we know a lot of these guys pretty well and we know when they are giving us their best price and we know when they are not giving us their best price. I guess the one big thing is, one of the things people are making their decisions heavily based on is service. We recently made a decision to go with a microscope that was probably inferior to one we could have purchased. But the decision was based primarily on the local service and our experience with the local service guys.
So, the service business in Florida is different than what you experienced in San Diego?
It is clear they are not sending their best people to low-traffic areas. You would probably make the same decision, too, if you had huge areas like Boston or San Diego or Seattle. You would probably send your best teams there. That is one thing we have instantly noticed, is the level of services. Some of the imaging companies like PerkinElmer and Beckman Coulter have operations in Miami and that has been good. Software vendors are no problem. But others, they don't do so well. I'm not going to name names but some of them don't have the best people down here. And, another question, why do they lump us with South Carolina? Some of the guys cover both states.
What equipment is on your wish list?
Pat Griffin: My group is focused on proteomics and we also support pharmacokinetics, which is more of a drug-discovery technology for drug development. On the proteomics side, we are primarily working with nanoelectrospray, linear ion trap mass spectrometers. We are in the process of securing funds for a hybrid linear ion trap FT ICR [Fourier transform ion cyclotron resonance]. The only one that is out there is Thermo Finnigan. We are seeking funds for that. Ideally, if we can get that, we can pump up our capabilities for protein identification, especially low-abundant proteins. The instrument is about $850,000. The other instruments we have average about $300,000 each.
Why are you interested in this hybrid?
The reason for the interest in the hybrid is doing efficient MS/MS and then transmitting those fragment ions into the ICR. To try to do everything within the ICR cell is difficult, but not impossible. There are other options. The type of applications we are doing are mostly differential analysis — comparing a proteome under two different conditions and identifying which proteins are differentially expressed under those conditions. The sample prep right now is multidimensional chromatography. For possible peptide analysis we use metal affinity chromatography.
What other instruments do you have?
We have four instruments in place — two linear ion-trap mass spectrometers — Thermo Finnegan. And, we have an ABI 4000Q triple quad ion trap, and then we have a Thermo Finnegan quantum high-res triple quad.
One of the triple quads, the ABI, is specially for small-molecule quantitation out of biofluids. It can be used for other applications. We basically have a core of instruments, and they each have their primary objectives. But if samples come up that need to be analyzed on the ABI 4000Q, then that is what we will do. That instrument is initially dedicated to supporting the pharmacokinetic analysis of compounds coming out of the medicinal chemistry group. We were interested in having the Q capability for metabolite identification. The 5000 has a limited mass range. That doesn't affect the small molecule part, it does impact if you are going to use the instrument for other applications.
What are the plans for throughput?
It's not geared up as a high-throughput operation. We are establishing the core tech as investigators come on board, and [as they] have experiments that need support by the group, we will supply that support. As the demand increases, we know how to go to higher-throughput processes on sample prep and we will be set to implement that at that time. Right now, the most limiting factor is the computational part — the database searching of all this information. The IT group here is putting together a system to take care of that.