Drug Discovery & Development

Drug Discovery & Development
Reed Business Information
Rockaway, NJ, 07866

Go With the Flow: Sorting Cells and Other Small Things

As a tool for selecting cells against which to screen drug candidates, flow cytometry is now also used to sort beads, tissue pieces, and even tiny organisms

Angelo DePalma, PhD
DePalma is a freelance writer based in Newton, N.J.

Long a mainstay analyzer in medical research and diagnostics, flow cytometry now serves broad life science markets, including drug discovery. For years, flow cytometry suffered from the reputation of high cost and inaccessibility. Expensive, sprawling cell sorters, much like early mass spectrometers, were relegated to central locations at large institutions, constantly attended by a team of technicians. While flow instruments are still hardly an impulse purchase, benchtop instruments costing about $100,000 have become accessible to almost any biologist. Today, industrial research groups requiring flow cytometry probably already have an instrument. According to vendor DakoCytomation, Copenhagen, Denmark, the worldwide flow cytometry market is valued at about $600 million annually.

click the image to enlarge

Union Biometrica’s COPAS sorting. Objects flow from a continuously mixed sample cup nto the flow cell where they are illuminated by two low energy lasers: a red diode laser (670 nm) used to measure the axial length and the optical density of the object; a multi-line argon laser (488 nm/ 514 nm) used to excite multiple fluorescence excitation and emission wavelengths. A pneumatic sorting mechanism located directly after the flow cell, then sorts objects by temporarily switching an air diverter off and on. The selected objects are dispensed, and those objects not meeting the sort criteria are gently passed along to a sample container.
(Source: Union Biometrica)

In drug discovery, flow cytometry is used primarily to analyze and isolate cells, particularly rare engineered cells expressing specific phenotypes, or scarce primary cells from complex mixtures. Thus, flow cytometry supports and underscores drug discovery’s increasing need for target specificity.

“Tissues contain lots of different cells,” says John Dunne, PhD, scientific director at BD Biosciences Immunocytometry Systems, San Jose, Calif. “Often, only one cell type is relevant to characterize drug effect on an appropriate drug target. Flow cytometry allows you to separate cell types for whole cell functional analyses, and/or fractionate those cells for more specific biochemical readouts like proteomics arrays or mass [spectrometric] techniques,” says Dunne.

Although microscopy still plays a role in drug discovery and even scale-up and manufacturing, especially in biotech, flow methods have replaced microscopy for large-population sorting tasks. For example, Philip Marder, principal investigator at Eli Lilly and Co., Indianapolis, Ind., uses BD’s and others’ flow cytometers for early-phase drug discovery using calcium flux, or “flip” assays. Marder claims an improvement of signal-to-noise from about three to 30 using cell sorting.

Marder’s group looks for subpopulations of cells that engage in calcium flux, which is measured by labeling cells with the calcium-sensitive dye Indo I. “We sort and isolate cells with higher calcium response and grow them up, creating populations of high responders for subsequent compound screening,” says Marder.

click the image to enlarge

Process flow for high-throughput screening of a diverse human antibody library. (Source: Diversa Corp.)

“It would be possible to do this work with a microscope, but there you can get one or two cells per minute, at most. Here, you can find a very minor subpopulation and be sure of collecting lots of cells, because the machine is sorting 10 to 30,000 cells per minute,” says Marder.

Not just for cells
As its name implies, flow cytometry is about counting cells. While “cytometry” also applies to bead-counting and bead-sorting, the term “flow sorting” may be more accurate. Bead-sorting and collection, which uses instrumentation similar or identical to that used to manipulate cells, offers the benefits of bead-based chemistry, including great multiplicity, high throughput, parallelism, and the potential for combinatorial work, enhanced by identification and collection capabilities.

Bead sorting comes in as many varieties as there are beads and unique surface chemistries. Diversa Corp., San Diego, uses sorting of bead-like microbe envelopes to discover, develop, and optimize antibody products for industrial, agricultural, animal health, and pharmaceutical markets. The company’s preclinical programs include an antifungal product acquired from GlaxoSmithKline, plus other small-molecule and protein therapeutics.

One of Diversa’s antibody-generating and antibody-selection technologies uses spherical microcapsules that envelop individual bacteria. Made from agarose and other gel-like materials, microcapsules form uniform, porous, spheres around their antibody-generating cargo. Diversa tags the microcapsules fluorescently, then recovers specific microcapsulated bacteria carrying the desired antibody, using flow cytometry.

OK, Now You Have Your Cells

Sorting extremely rare cell subpopulations expressing drug targets is satisfyingto a point. While cytometry addresses the cells’ identity, the technique is not of much help when you need a few micrograms of DNA or RNA, and all you have are a few thousand cells.

“People think PCR is super-sensitive, but to carry out just one real-time quantitative PCR reaction requires 10 to 20 nanograms or more RNA in a reaction well,” says John Todd, PhD, vice president of marketing at NuGen Technologies Inc., San Carlos, Calif. “Scientists typically don’t just do a single well, they often run everything in triplicate. And they may be interested in studying multiple gene transcripts in a given tissue. In effect, they need not just nanogram quantities of RNA, but micrograms, which requires collecting millions of cells from a single sample.”

The problem with needing millions of specialized cells is that often you can’t collect them from small samples. NuGen Technologies, which specializes in identification and amplification of biomolecules, has devised a kind of “back end” amplification for flow cytometry and other genomics work, which amplifies as little as 5 ng of messenger RNA 10,000-fold in one shot with no temperature recycling.

The three-step process takes just four hours. Step 1 synthesizes first strand cDNA complementary to mRNA through a chimeric DNA/RNA primer. Step 2 creates double-stranded cDNA, which serves as the template for isothermal, linear amplification in step 3. Two NuGen products incorporate Ribo-SPIA for microarray applications: Ovation Aminoally System for spotted arrays, and Ovation Biotin System for GeneChip arrays. A third product, Ovation 3-Prime Amplification System, will be introduced in Q3 2004 and additional gene expression applications are planned for next year. NuGen also is developing a method that provides 100,000-fold amplification, enabling array experiments from just a few cells, but it’s still at the research stage.

New technologies like RiboSPIA address the need to obtain more information from ever-smaller samples. For example, investigators interested in inflammatory diseases could work with all white cells from a blood sample, but might only be interested in one subpopulation of eosinophils. An assay using all white blood cells will give rise to a lot of non-eosinophil noise, where subpopulation collection might not yield enough material to perform the desired study.

“All molecular techniques require significant amounts of material which is in very short supply. It’s a perennial problem,” says NuGen CEO Jan D’Alvise. “For example an Affymetrix array requires between two and five micrograms of RNA. If you only have 200 to 10,000 cells you can’t get that much RNA from the specimen.”

RNA amplification based on T7 techniques take two to five days of intense work. “Amplification at this level makes flow cytometry more accessible,” Todd says.

Ribo-SPIA employs linear amplification and generic primers to amplify all gene cripts uniformly, so that the amplified transcripts closely resemble the makeup of the starting material. By contrast, exponential amplification like PCR significantly skew transcript representation during amplification and are not easily multiplexed. Consequently, PCR amplifications less closely resemble represent biological distribution of the original sample.

Each bacterium in the collection expresses one novel antibody, in FAb form, per microcapsule. Diversa uses DakoCytomation MoFlo cell sorters, which analyze up to 50,000 cells per second. Because the microcapsules are much larger than the bacteria they carry, “only” 5,000 are analyzed per second. Slow by cell standards, perhaps, but still fast enough to process roughly 100 million microcapsules per day per machine.

In flow methods, analysis speed is limited not by detection, but by the size of micro-droplets containing the analyte (cell or bead). “Since the microcapsules are quite large compared with cells they require larger droplets, which are formed more slowly than if the capsules were smaller,” says senior staff scientist Eileen Tozer, PhD, Diversa’s sorting technique allows it to discover and collect as few as one antibody from a population of 109 .

Cells externalize the antibodies, which are captured within the microcapsule. Antigen binding, detected by a reporter fluorescent tag, allows isolation of all bacteria expressing the desired FAb. Usually several such bacteria are found, which is when the real fun starts. Individual clones are expanded, and the antibodies they express are tested for binding and other properties. Selectivity, for example, is determined by binding against proteins related to the target and/or random proteins. After finding the lead FAb, if one or more parameters require improvement (e.g., affinity, specificity, stability, expression, and so forth), Diversa optimizes it through a proprietary mutagenesis technique, Gene Site Saturation Mutagenesis, which is capable of replacing any and all amino acid variants of the original antibody.

Sorting instruments from Union Biometrica, Somerville, Mass., also work with beads, particularly large hydrophilic polyethylene glycol (PEG)-based microspheres. More typical polystyrene beads are hard and extremely hydrophobic. “PEG sucks up water,” says Rock Pulak, PhD, director of biology at Union Biometrica, “which can be exploited as an environment in which compounds synthesized on the bead surface can be studied in an aqueous environment similar to that found in the body.”

Because Union Biometrica’s atypical instruments work with analytes that are a much larger than cells, end-users who do bead-based chemistry can operate at a much more convenient scale than with “small bore” cytometers. Customers have synthesized peptides on hydrophilic bead surfaces, then subjected libraries of such beads to various assays and fished out beads covered with peptides that exhibit the desired binding. A single bead provides enough material for mass spectroscopy or sequencing.

End-users interested in creating customizable beads enjoy many options. Luminex, Austin, Texas, sells smallish 5.6-micron polystyrene beads, each bearing 100 million carboxyl groups on their surface. “Our beads are a completely open platform,” says Christoph Cordes, Central Region Manager. Beads accept peptides, proteins, receptors, ligandsany molecule that can be attached through a carboxylate. Cordes claims that flow counters can distinguish up to 100 different analytes in a reaction volume using the fluorescent ratio method.

Movin’ on up
One rationale for flow cytometry in discovery is that drugs work on cells and not, technically speaking, on tissues. Another school of thought recognizes the importance of cellular context: While cellular response is probably the best indicator of a drug’s inherent efficacy, it says nothing about pharmacokinetics and bioavailability in the context of tissues, organs and whole organisms, thus the basis for renewed interest in assays involving tissues, even intact animals. Flow sorting methods also support this approach.

Union Biometrica has developed a line of instruments, COPAS (complex object parametric analysis and sorting) which addresses separation and harvest of small multicellular organisms such as C. elegans, as well as cell clusters and tissues. Where traditional flow cytometry is designed for individual cells at sizes of up to 20 micrometers, Union Biometrica’s instruments can handle delicate groups of cells, or even organisms measuring up to 1.5 millimeters in diameter. For example the company’s first instrument, initially developed for the 1,000-cell worm C. elegans, pushes the 150 micrometer-wide organism through the detection aperture lengthwise, oriented with the flow.

“If you’re working with 100 or so worms at a time, you’re better off using a microscope,” says Rock Pulak. “Our instrument offers a precise way to analyze and dispense tens or hundreds of thousands of organisms.”

When Flow Methods Don’t Cut It

Although flow cytometry has become a common tool in various disciplines of cell biology it does not address every cell analysis need, or handle every type of experiment. Flow methods are limited to analyzing cells in fluids. Cell-based assays requiring time resolution, such as enzyme kinetics or drug uptake, make no sense on individual cells. Flow cytometry cannot analyze subcellular location of fluorochromes, and cells may not be re-analyzed. Similarly the technique works poorly with adherent cells, which must be removed from their millieu. Finally, as has been pointed out, flow cytometry is impractical for small sample sizes and low-cell number specimens.

Laser scanning cytometry (LSC), which combines some aspects of flow cytometry and some of image analysis, is viewed as a technique complementary to both methods and overcomes many of the drawbacks of flow cytometry. In a way LSC resembles flow cytometry, in that it measures light scatter and fluorescence resulting from interaction of the cells with a laser beam. In LSC analysis, cells may be live or fixed, adherent or in solution, but whatever is holding themstandard slides, microtiter plates, petri dishesis moved incrementally along the x-y plane, much as in standard microscopy. Photons emitted from fluorescent events on cells then travel back through the light path and are collected by wavelength-specific photomultiplier tubes. Applications of LSC include cell cycle and DNA content, apoptosis, DNA damage, intracellular translocation and compartmentalization assays, tissue and tissue microarray analysis, immunophenotyping and many others. These applications have been applied throughout the drug discovery landscape for lead optimization, pre-clinical drug safety studies, functional genomics and biomarker discovery.

CompuCyte has three cytometers on the market: LSC, the fully automated iCyte, and iCys. iCyte may be interfaced with a robotic sample loader and is suitable for drug and biomarker discovery. iCys is more appropriate for interactive research applications.

“LSC overcomes what I see as serious drawbacks of flow cytometry, namely lack of visualization and the requirement that cells be suspended,” says Elena Holden, MD, CEO of CompuCyte Corp., Cambridge, Mass. “With LSC, users can quantify fluorescent signals and show the relationships between the quantitative assessment and visualized morphology on adherent cells, tissue samples or even whole organisms like zebra fish.”

Union Biometrica designed three other instruments, each with a flowcell a bit wider than the next. Together, the four flow counters analyze particles ranging from single cells up to 1.2 mm-diameter Xenopus oocytes used to study ion channels, and similarly sized zebrafish embryos. “The instrument doesn’t care what kind of material it’s processing, as long as it fits,” says Pulak. “This allows allows experimentation on Drosophila embryos and larvae, mosquito larvae, Arabdopsis seeds, as well as other organisms.”

Such “model organisms” (and others) are in great demand for drug discovery because their genomes are mapped and the organisms themselves are accessible with genetics, molecular biology, and tools like RNA interference. “You can do things with model organisms that you simply cannot do with largeror certain other smallerorganisms,” says Pulak.

Concerned that shear forces could rip tissues apart during analysis, Union Biometrica designed instruments to minimize shear and operate at fairly low speed. Shear isn’t much of a problem with C. elegans, which sport a tough epidermis. Mammalian tissues pose shear-damage concerns, however, especially for clusters of embryonic stem cells or embryoid bodies.

Cell clusters represent the best way to study certain cell types, for example the insulin-producing beta cells within pancreatic islets. Sorting and harvesting beta cells from surrounding structures and cells could provide curative islet transplantation for type I diabetics or a platform for developing insulin-inducing agents for type II diabetics.
Pulak admits that drug discovery has not yet latched on to the idea of cell cluster experiments. “There’s not as great a demand for sorting clusters as for single cells. Although they don’t substitute for tissue, cell clusters or matrices more closely resemble tissue than do single cells. So you can think of cluster analysis as a bridge between single cells and more complex tissues.”

Past, present, future
Drug discovery has employed flow cytometry to varying degrees for at least 20 years. Yet, says BD’s Dunne, the technique has been underused as a core competency in high-throughput drug discovery, largely because of cytometry’s reputation as a complex, “scientific” method. “Cell-based assays generally have lagged behind soluble chemistry, mostly because of their practical complexity and perhaps partly because pharma R&D senior management tends to have a chemistry background. As a result cell biology has experienced a difficult adoption path. Flow cytometry and microscopy were viewed as tricky, so they’ve been only very hesitantly adopted. But today it’s more common to see both flow cytometry and microscopy in mainstream drug discovery labs.”

To overcome some of the “science” objection to flow methods, vendors must make their methods amenable to the current drug discovery infrastructure, which Dunne says BD has succeeded in doing with its BD FACSAria cell sorter. “A combination of auto-regulation and high performance in the BD FACSAria cell sorter altered the adoption curve for flow cytometry in drug discovery and life sciences research.”

As with other high-throughput methods, it may be too early to tell whether flow cytometry will help fill pipelines new compounds. “It may be a bit premature to decide on the value of new technologies like flow cytometry,” cautions BD’s Dunne. “While there’s a big scientific drive to publicize successes, there are proprietary interests at stake as well,” says Dunne, “so we might not hear about all the strategies that are being implemented, and how well they work.”