Breakthroughs are making cell-based screening faster, easier, more powerful.
By Malorye A. Branca
December 15, 2003 | David Zacharias remembers how things used to be all too well. Ten years ago, he was taking photographs of brain cells and projecting the images up onto a screen. Next, ruler in hand, he would measure the length of each tiny twig-like neurite extending from each cell body — neurite by neurite, cell by cell.
Not only was it tedious work, it was "incredibly open to the individual user's biases," says Zacharias, assistant professor at the University of Florida College of Medicine, and formerly in the molecular neuroscience division at Merck. "Each time, I would just decide where 'the body' of the cell started, and measure out from there."
That kind of task is a drag for a graduate student, which Zacharias was at the time, but it's an anathema to pharmaceutical companies, where everything needs to be done large scale. And yet, companies routinely rely on studies using living cells to select those candidate drugs that will receive hundreds of millions of dollars of investment.
It's one of the most critical junctures in the drug development process — and one of the most problematic. Delays and bad choices fuel extra costs and countless missed opportunities.
| "Seek and find: Cells once hidden are easily identified today. Using Q3DM's EIDAQ system, three nucleated red blood cells from a male fetus stand out thanks to staining (Y chromosome is red).
Red light, green light: Her2 and nuclear antigens in SK-BR-3 cancer cells are wearing green Quantum Dot (Qdot) and red Qdot conjugates, respectively.
CNS Drug Filter: Different drugs elicit different gene-expression patterns in cultured human neurons. CuraGen uses such "genomic efficacy profiles" to detect which drugs have antidepressant (red), antipsychotic (blue), and opioid (green) characteristics.
That's why automated, high-throughput (and even "ultra-high-throughput"), cell-based screening has become such a bustling field. "There has been a surge of interest in cellular analysis," says Anne Jones, product director for cell biology at Amersham Biosciences. Microarrays, fluorescent microscopy, RNAi, next-generation flow cytometry, and a bevy of related tools are affording researchers a much clearer view of cellular activity and a better way to select new drugs.
Typically, the first step in going from biological target to drug is high-throughput screening (HTS), by which many thousands of compounds are tested simultaneously against a target molecule. In principle, companies end up with a bounty of drug leads, improving their ultimate chances of success.
But HTS has been one of the major disappointments of modern drug discovery. "HTS has not filled the pipelines like it was supposed to," Zacharias says. A typical complaint about traditional HTS is that it provides too little information. Knowing a drug binds tightly to one molecule says little about other proteins it might interact with. As a result, many compounds promoted to "hit" status, and further development, must later be dropped because of toxic side effects: They may hit the target, but they hit something else important as well.
Those side effects have become the pharmaceutical industry's Achilles' heel. Hence, scientists are desperate to cull problem molecules from their lead collections as quickly as possible. One approach is to conduct screens in more "human-like" systems sooner and to gather information about all of a drug's biological effects. To achieve this, companies are turning to cellular assays earlier in the screening process.
"About three years ago, some of the big pharmas started moving toward using more cellular assays in HTS," says Oren Beske, head of cell biology at Vitra Bioscience. "Now, some companies do 100 percent of their screening in cells."
Drug developers are also trying to use cells that are more "biologically correct" in their assays, and they are collecting many more data during the screens. These trends are fed by a growing number of breakthrough tools aimed at helping drug makers get a better feel for how good compounds really are, before they sink hundreds of millions more dollars into them.
High on High Content
The biggest change is the embrace of high-content screening (HCS). Cellomics pioneered this field in the late 1990s, bringing together living cells, fluorescent tags, micro-imaging, and new algorithms. Combining these into a seamless system has been a tall order, but it's working. Today, beyond the simple binding or activation of one chemical to one biological molecule, scientists can track changes in cell shape, movement, differentiation, adhesion, and even the inner trafficking of proteins from one compartment to another.
|Cell Screen Team
|Suppliers specializing in cell-based screening:
· Acumen Bioscience (TTP LabTech)
· Amersham Biosciences
· Atto Bioscience
· Axon Instruments
· Norak Biosciences
· Novasite Pharmaceuticals
· Quantum Dot
In recent years, HCS has become rife with new players that see this as a hot opportunity (see "Cell Screen Team"). Users also seem convinced. "People have really dived into this field," says Susan Catalano, senior scientist and principal consultant at Drug Discovery Imaging, which specializes in HCS. "These instruments are in place in most of the arenas I'm working in."
A group leader at the Genomics Institute of the Novartis Research Foundation (GNF), Sumit Chanda began using Q3DM's EIDAQ 100 to study gene function using cDNAs and siRNAs (small interfering RNAs). The high-throughput microscopy system provided "better information" than traditional methods. His team is now collaborating with GNF's internal engineering department, which includes some former Saturn Corp. (a subsidiary of General Motors) engineers, to develop a robotic system that would integrate high-content imaging. "GNF's compound library is a log greater in complexity than any nucleic acid collection," Chanda says. But having been able to identify hundreds of novel gene activities with the EIDAQ, they've decided it's "time to invest more in automation and stir things up in compound screening."
Cellomics recently combined siRNA and HCS in an intriguing study of camptothecin, a cancer drug. First, the company used its ArrayScan platform to verify that p53 was knocked down in some cells and that camptothecin alone could make cells stall in an early phase of reproduction. When cells treated with the p53-targeting siRNA also received camptothecin, they did not show the same physical effects as seen with the drug alone. In this study, HCS was used for everything from measuring multiple knockdowns and tracking off-target effects, to monitoring delivery into the cells.
HCS is a veritable revelation. "With a system like Q3DM's, you can choose from more than 70 different metrics to examine," Zacharias says. "And much more subtle differences can be picked out."
High-throughput microscopy is not the only way to get more content from cellular assays. CompuCyte's system uses laser scanning to produce a "light-microscopy-like image," says Elena Holden, president and CEO. The instrument has been available since 1996, but an automated version was only just introduced. "The key attribute is its great flexibility," Holden says. The iCyte Automated Imaging Cytometer, and related software, can be used for target validation as well as predictive toxicology. The basic assays quantify amounts of DNA and record cell cycle- and morphological-markers.
Bionaut, meanwhile, engineers proprietary Sentinel Lines using a variety of read-out systems and tags incorporated at specific, disease-related genetic sites. "It's like a light bulb that goes off if a particular regulatory pathway is being hit," says Mehran Khodadoust, Bionaut's president and chief scientific officer. These uniquely engineered cell lines can help characterize a compound's activity or find optimal compounds in primary screens. "We drop the genes into genetic sites that are regulated by the gene of interest," Khodadoust says. This leaves the pathway comparatively undisturbed, but makes it easy to record its activity.
| Cell-based solution: Novasite's John Ransom (director of flow cytometry and screening) and Juan Ballesteros hope to use advanced cell-based screening methods to jump-start drug discovery. Their proprietary system provides data about multiple cellular characteristics simultaneously.
Novasite Pharmaceuticals' proprietary system uses the highest-speed flow cytometer available, novel automation, and bioinformatics to examine thousands of cells per second, each cell independently. Different cell types can be labeled with one of 20 different colors, mixed with the compound of interest, and fed into the system. The result is a vast amount of data about how a compound affects multiple types of cells.
"We look at about four compounds per minute," says Juan Ballesteros, Novasite's vice president of research. "That might sound like a little, compared to a FLIPR that can do 100 per minute, but we get more information because we are looking at multiple receptors in parallel."
Union Biometrica, a division of Harvard Biosciences, has given the term "high content" a whole new meaning with its COPAS system. COPAS is also based on flow cytometry, but it has been adapted so that (very small) whole animals can be analyzed. "There are great new opportunities when you are looking at intact, multicellular tissue," says Rock Pulak, the company's director of biology. After all, animals such as Caenorhabditis elegans, Drosophila, and zebrafish have emerged as indispensable models for certain diseases and embryonic development. "It would be unfair to say you can't do high content on a multicellular system," he says.
Cellular assays have some serious limitations, of course. For example, cultured cell lines, which are easiest to come by, are not always exactly what they seem. Because they are grown under artificial conditions for long periods, they don't always accurately match the natural, or "primary," cells found in the body.
Researchers at Pfizer experienced this problem when they compared gene expression profiles from xenografts in mice with cell lines derived from those tumors. "We could tell by matching up the profiles these were not the same as the tumor," says senior scientist Michael Gieseg.
That's nerve-wracking to scientists who are depending on these cell lines to validate targets or test new compounds. "Nobody really knows how much things are drifting over time," Gieseg says, who recently moved from oncology to inflammatory diseases, where he's again doing expression profiling. Luckily for the Pfizer researchers, "the Affymetrix platform is very robust in our hands, and the patterns have been highly predictive," he says. Cost is a consideration that might give pause to many academic groups, but for Pfizer, Gieseg says, "It's a high priority to make sure we can trust the cell lines 100 percent."
CuraGen has also used microarrays to improve cellular assays. "There are some areas in the central nervous system (CNS) for which no good assays exist," says Richard Shimkets, vice president of scientific development. The company has been one of the pioneers in the use of microarrays to predict which compounds have the greatest risk of side effects. Now it has applied that same strategy to develop an assay that screens CNS compounds for specific kinds of activity.
Researchers first produced a series of gene expression profiles by treating cells with chemicals known to act on the CNS. A small number of genes were found to be good markers of antipsychotic, antidepressant, or opiate effects (Gunther, E.C. et al. "Prediction of clinical drug efficacy by classification of drug-induced genomic expression profiles in vitro," PNAS 100: 9608-9613; 2003).
Now, instead of scrutinizing many individual cells or performing multiple animal studies, "which we couldn't always believe anyway," Shimkets says, the researchers simply test for the markers. "We found clear, distinct patterns. Applied to new compounds, these help us figure out what class they fall into."
Doing cellular assays can also be expensive, that's why Guava Technologies is working on "making sophisticated cellular assays more accessible," says marketing manager Glenn Terashita. The company offers miniaturized, microliter-scaled, "benchtop" instruments, along with high-quality assays for cell counting, viability, and apoptosis. The Guava PCA-96 can analyze up to a couple thousand cells per second (see New Products, Aug. 2003 Bio·IT World, page 56). Each of Guava's assays also comes with corresponding software. "We take a total turnkey approach," Terashita says. "Customers can avoid developing the actual assays if they want to."
That Cost Issue
Affymetrix is also looking for ways to make its chips more accessible. The company is developing a HighThroughputArray (HTA), with a tiny U133A GeneChip in each of 96 wells, for automated processing. Because of time, reagents, etc., "the HTA format instantly, and dramatically, reduces the cost of processing any microarray experiments," says Steven Lombardi, Affymetrix's vice president of corporate development. The company is also working on shrinking the chips, which should lower costs further.
|There are some data issues brewing in the high-content screening (HCS) arena.
Microarrays could be useful for a variety of drug development tasks, such as generating profiles for structure activity relationships. That's the kind of information that is useful over and over again, but researchers still have a hard time accepting the cost at current rates. "People would like to do it, but they can't afford it," Lombardi says. Hence, the development of the HTA. Johnson & Johnson has gained early access to the HTA system and will help the chipmaker validate the system.
Activity in this field is still rising, overall. "Broadly, we expect the cellular analysis market to grow tenfold," Amersham's Jones says. "There is a buzz, and now everyone wants to do it." The rapid introduction of robust tools for RNAi had a particular impact, leading to multiple deals, such as Akceli's with Dharmacon and Invitrogen's acquisition of Sequitur. "You have to be doing something with RNAi," Jones says.
Labeling is another area that's hopping. Researchers desire multiple colors to track multiple activities, and they TMTM want labels that won't mess up the biology either. Recently introduced Qdots are nanocrystals that take on different colors at slightly different sizes. Made by Quantum Dot, these labeling agents are also very stable, and can emit for days, or longer, as evidenced by a somewhat macabre photo of a rat whose interior was still glowing days after a dose of Qdots.
Fluorobodies are an even newer labeling advance (Zeytun, A. et al. "Fluorobodies combine GFP fluorescence with the binding characteristics of antibodies," Nature Biotechnology, 9 Nov. 2003). These molecules are formed by combining what is already a favorite tool for cellular assays — green fluorescent protein (GFP) — with some antibody parts. Co-developer Andrew Bradbury, at Los Alamos National Laboratory, says some of the key features of fluorobodies are that they are "simple to prepare, simple to use, and their fluorescence is directly linked to function — i.e., if the fluorobody is not fluorescent, it can't bind."
Many researchers would also like to do more with primary cells, even though these are much harder to come by. In working with these samples, it will be important to have techniques that make the most out of every single cell. Vitra Bioscience believes its new CellCard System is ideal for use with primary cells.
The company's CellPlex system tests compounds against up to ten different cell lines in a 96-well microtitre plate system. "We have a novel encoded-particle technology, called CellCard," Vitra's Beske says. This codes each cell line. Vitra's system thus "reduces the number of cells you need to generate each data point a hundredfold," he says.
Down the line, it's hoped that other alternatives to traditional cultured cells will arise. "Everyone is trying to find a cell line that is more like a primary cell," Catalano says. Here, she says, "stem cells offer the greatest promise."
The tools, of course are constantly being refined and expanded. For example, Amersham is planning to launch more than 20 assays spanning "a complete range of signaling proteins." Q3DM is trying to accelerate the speed with which its instrument collects images. "The physical motion of the camera across the sample is one of the chief bottlenecks," says Edward Hunter, chief technology officer at Q3DM. It plans to have an EIDAQ1000 out within a year. Bionaut is combining its Sentinel Lines approach with high-throughput microscopy using an Acumen Assay Explorer to generate "data on both the activity of a regulatory pathway, and to correlate that with phenotypic changes," Khodadoust says.
Screening for single parameters still fills a critical role, but even that's changing. "We've seen a consolidation around a couple of key target areas — G-protein coupled receptors and kinases," says Michael Biros, director of marketing for Molecular Devices, maker of the seemingly ubiquitous FLIPR HTS system.
And, while the shift to high content is real, pharmaceutical companies are more cautious today, having just worked through a pretty traumatic genomic technology explosion. Now, Biros sees many researchers taking a long, hard look at how they are going to use the new higher-content systems, such as his company's Discovery-1 system. He likes that; after all, just getting more information isn't always a better thing.
"Companies want right content, not just high content," Biros says. With that type of sensible approach, the whole mix of tools might just come together and deliver some much-needed relief in the lead selection arena.
PHOTOS CREDITS: RED BLOOD CELLS, ANTIGENS, AND CNS DRUG FILTER: REPRODUCED WITH PERMISSION FROM WU ET AL. NAT. BIOTECHNOL. (2003) MACMILLAN MAGAZINES LTD.; RANSOM AND BALLESTEROS BY FRANK ROGOZIENSKI