At A Glance
Name: Ralph Garippa
Position: Research leader, Cell-Based High Throughput Screening & Automation, Hoffman-La Roche
Background: Project manager, metabolic diseases, Roche — 1994-1998; Laboratory leader, bronchopulmonary pharmacology, Roche — 1986-1992; PhD, pharmacology, Columbia University; BA, biology, Farleigh Dickinson University
Ralph Garippa has been a pharmaceutical scientist with Hoffman-La Roche since the early 1980s, and has seen drug screening technology steadily progress from nearly 100-percent biochemical end-point assays to an even mixture of both biochemical and high-content cell-based assays. He is one of many in pharma who have embraced the change, and he now heads the cell-based high-throughput screening group at Hoffman-La Roche’s Nutley, NJ-based facility, where he and his colleagues are responsible for assay development, HTS, and automation support required for the therapeutic areas of oncology, inflammation and metabolic diseases. Among other presentations, Garippa led a panel discussion on high-content screening’s role in modern drug discovery at Cambridge Healthtech Institute’s High-Content Analysis held last month in San Francisco. Garippa took a few moments last week to share some of his opinions on the subject with Inside Bioassays.
At the HCA conference last month, you asked the members of a discussion panel whether they thought cell-based assays could ever completely replace biochemical assays in drug screening. What is your opinion?
I think they will never replace biochemical assays, because the two sit so well together side-by-side. One is complementary to the other one. So invariably, when we’re developing new compounds into more and more drug-like characteristics, we will either start with a high-throughput cell-based assay, and immediately go to a follow-up assay in biochemical format; while other times, because the median cost point per well for biochemical [assays] is a few pennies lower, we’ll do the high-throughput and biochemical, then have the cell-based assays waiting right after, to give a more in-depth functional characterization of the compounds that were isolated. So my feeling is no, they won’t totally replace biochemical, and actually, I think an equilibration happened in the last five to seven years, where cell-based assays — at least in my hands — were about one in every five high-throughput screens. Now they’re about 50-50, and from other persons in the industry that I’ve talked to, that trend has either occurred or is occurring.
When you’re referring to new prominence of cell-based assays, are you talking about high-throughput cellular assays or so-called high-content cell-based assays?
Good question. Right now, the high-content, or translocation, or biosensor-type assays are embedded within the 50-50 split of biochemical versus cell-based in high-throughput screening. As more companies get in to it, I think within companies like Roche, AstraZeneca, and Merck, you can further divide the percentages and say, ‘Well, of that 50 percent, about 30 percent of them are translocation-based, and the other 70 percent are more traditional technologies like radiometric binding, FLIPR-based calcium, or cyclic-AMP.’ But as people get more comfortable, and buy the proper equipment to do the translocation or high-content screening assays, I think you’ll see that percentage showing up more as a subdivision of what we’re doing with cell-based assays.
What are some of the most prominent arguments out there that people are making in favor of replacing biochemical assays with cell-based assays?
Again, I wouldn’t say replacement, but it’s a choice, where in the past, certain labs might have only ever considered a biochemical assay, but now when it comes to assays development, they will give equal weighting as to whether they want to put up a cell-based or biochemical assay. The three main reasons are still the ones that held ten years ago: You’re going to get a functional readout; you’ll get — if you’re looking for it — some index of early cytotoxicity, which can be a liability; and third, if you have an intracellular target, you’ll get some measure of the penetration of that compound within the cell through the lipid bilayer. You don’t get that kind of information with a biochemical assay as performed in a test tube or microtiter well. You can get it later, once you look at the hit list, but some people prefer to have that higher-density data right off the bat.
On the flip side, what do biochemical assays offer that make them unable to be replaced by biochemical assays?
I think for the most part, up until now they’ve been easier, because companies have kept the cell passaging, the cell fermentation, and the cell work separate from the screening. In other words, people were always thinking, ‘If I run cell-based assays, I’ve got to make and characterize a cell line, isolate a clone or pool of clones to use; I’ve got to spend money on media, serum, flasks for plating, [and] robotic systems for carrying out the plating.’ But what we’re seeing now — at least the early phases of it — is that cells can be used simply as reagents. So there are companies like Cell and Molecular Technologies, CMT, in Phillipsburg, NJ — they’re offering cell supply as an outservice. And they also have cells that are under division arrest, meaning that if you bring them to your site and you plate them, they won’t overgrow. In the next few days, you can use them at a certain density. People like Tom Kost [of GlaxoSmithKline] and Conrad Cowen, who is at Norak [Biosciences] — they’ve been pushing forward measures to just use frozen cells. We’ve got frozen stocks of cells that we use; we thaw them, and then we start passaging. But they’re saying that you can use these cells right out of the cryotube a day later, and that the shock that the freeze-thaw causes on the cells is no worse than the shock of trypsinizing — lifting those cells from passage right off the plate, and putting them on microtiter wells the next day and assaying them. So more and more, the thinking and the methodology is coming about that cells are just another assay reagent, and you can treat them like the reagents in a biochemical assay — whether that’s a luminescent agent, whether it’s an antibody or buffer — any other reagent that goes into the mixture.
Roche obviously uses a variety of cell-based assays in its drug-discovery research. Are there any general technology platforms or assay types that you’ve settled on, or do you tailor it to the specific information you need?
There are always going to be some go-to assays. FLIPR from Molecular Devices, for calcium flux, has been a tremendously useful tool — both in high-throughput and follow-up assays. We use a number of different formats for cyclic-AMP. But we also have some very specialized cell-based assays in the therapeutic areas, and some of those are amenable to higher-throughput, or robotics. We just did some work last year with Dr. Grace Ju here at Roche, in order to automate the pipetting of cells in oncology into soft agar. If we can do this in higher throughput, it means we can test more compounds for their ability to inhibit the three-dimensional growth of cancer cells in soft agar; whereas when we were doing that with hand pipetting, and trying to carefully control the temperatures within beakers and so forth, it was very hard to do that — to work in that temperature and time window. Now we’ve worked with Titertek [of Huntsville, Ala.] — they have an instrument, the MAP-C Agar Dispensing Unit, that will dispense the soft agar, and they have reservoirs and fluid lines that are controlled temperature-wise, so we can do a lot more before the agar hardens and before we kill the cells. So that’s just one example of taking a low, slow assay, and making it higher throughput in a meaningful way to the therapeutic area. We also have therapeutic areas such as inflammation and metabolic disease, and within metabolic disease, there are some assays now that we’re doing around very specific cell types, like adipocytes or muscle cells or hepatocytes, because those are the cell types that we want to query based on the targets that we’re doing.
What do pharmaceutical researchers still want to see from cell-based assays technology vendors, either in terms of instrumentation, assay technology, or other things?
It’s pretty good [right now] in that people can go so far, but we do need the front advanced. If you just take high-content screening, or high-content imaging, there are some very good instruments out there by a dozen different manufacturers, so that gives you the choice between using a laser line-scanning instrument, a conventional microscope, or a 3-D confocal microscope, either in low or high throughput. But we’re just now getting better with all of the reagents and tools associated — the fluorescent tags, the fluorescent proteins, antibodies, things like quantum dots, and further on, it’s a different data set. So in the past, we only had endpoint numerical data, but now we’ve got image data. And those data files are much bigger, and you sort them differently. You can’t really sort an image file based on ascending or descending order; you have to choose an object characteristic to sort. Now what we really need are the images associated with the numerical data in ways that are compound-centric in the retrieval. In biotech and pharma, everything depends on the compound. We want to look up the data based on what that compound did, and then construct a time profile of all the assays and all the activities associated with that compound. So the data sets, meta tables, data management, and data mining tools really have to catch up with the rest of the query technology in imaging to really make it most useful. I chose HCS as an example for improvements, but we could go right through the industry naming things that could be better. [For example], these are the early days of high-throughput ion-channel work, and there are only a handful of companies. But I think they’re at the stage where high-content companies were a few years ago, and we’re going to see some really seminal advancements there, to make HCS for ion channels into HTS, and then get better definition, more like the gold standard of microelectrode physiology, in an automated way.