Drug Testers Think Small

by William Wells

If you are looking for a new drug, unwieldy flasks and tubes are out, but worms, single cells and glass chips are in.

Finding a drug used to be simple. Test some chemicals against protein targets thought to be involved in a disease; find the chemical that jams the works of the protein.

Conceptually, drug hunting still works that way. But the low-tech approach to testing was also a time-consuming approach. One-at-a-time test-tube science was not fast enough to keep up with the stream of new protein targets and chemicals. And the drug-discovery pipeline behaves like any other pipeline: a hold-up at one point negates any gain in speed made at another point. With combinatorial chemistry and genomics in place, Roger Tsien of the University of California, San Diego, says that, "there was a gap in the next stage of the food chain." So researchers started thinking creatively, and came up with some strange and unusual ways to plug that gap.

Replacing Lab Mice with Worms

The tiny worm Caenorhabditis elegans only lays eggs when it is well fed. But give it some Prozac, or any number of similar anti-depressants, and these worms will lay eggs even when the cupboard is bare (or, in their case, when there are no bacteria to eat).

To Carl Johnson this is more than a pharmacological curiosity. Giving a chemical to a worm and looking for egg laying is a whole lot easier and safer than giving the same chemical to a human and testing for a reduction in depression. If he wants to find more chemicals that act like Prozac on humans, the worm test is a good place to start. Based on this concept, Johnson helped form NemaPharm, Inc., which is now part of AXYS Pharmaceuticals (South San Francisco, Calif.). Devgen N.V. (Ghent, Belgium) is using a similar approach.

The Ultimate
Worm Dispenser

NemaPharm’s testing tool is its unique worm dispenser. Fast-moving fluid straightens the worms out so a light-based detector can measure their length and therefore their age. The machine spits out worms of the appropriate age into a single compartment of a dish with 1536 such compartments. A different chemical is added to each compartment (or well) and the effect of the chemical is noted. Thousands of these assays make up a chemical screen.

The worms are handy because they are small (a few millimeters long), fast-growing (in a nutritious liquid they can produce grandchildren in a single day), easily stored (freezing and thawing doesn’t bother them), and they have been well studied by biologists interested in simple behaviors.

Worms as Workhorse and Test Subjects

For these reasons they have become the NemaPharm workhorse and test subject. "The essence of what NemaPharm is doing is to establish surrogates for human diseases [in worms] and then establish selections for drugs against those diseases," says Johnson. As the Prozac example shows, the surrogate may not be perfect, but that isn’t necessarily a problem. The protein target of Prozac is involved in different behaviors in humans and worms, but what matters is that the same protein target is present in worms, and interfering with it causes some measurable symptom. Explains Johnson: "We are trying to use the same molecules and the pathways in C. elegans, not to reproduce the same behaviors."

An extreme example is the ras mutation, a genetic change that in humans can lead to cancer. In worms an over-active ras produces extra female reproductive structures, which are visible as bumps on the outside of the worm. The current screen for anti-ras chemicals involves looking for the disappearance of these bumps. Johnson hopes that he can automate this process soon.

Worms are not the ultimate solution for everything. They have a tremendous capacity for modifying and throwing out foreign chemicals, which may mask the effects of potentially useful drugs. And the differences between worm and human proteins are likely to be crucial in determining which is the perfect final drug.

But for the earlier stages of drug discovery, worms look like an excellent tool for picking the best target, and for finding a starting chemical for researchers to work on. Not all proteins are equally amenable to being shut off with a chemical. When Johnson looks for chemicals that shut off the over-active ras, he is really testing for a chemical that turns off either ras itself, or anything in its chain of command. If he finds that a lot of chemicals turn off another protein in this molecular pathway, that protein may make a better target for future screens.

Glowing Cells

The testing system devised by Tsien also targets entire molecular pathways, but here the test unit is a single cell. "Ultimately we are trying to cure whole organisms, and the cell is the smallest unit that is genetically optimizable," says Tsien.

A company called Aurora Biosciences Corporation (La Jolla, Calif.) has licensed and patented many of Tsien’s inventions. Aurora’s aim is to test whether certain molecular pathways in the cell are on or off, by detecting whether the genes at the end of those pathways are on or off. The part of a cancer-causing gene that turns the gene on or off can be linked to any piece of DNA, so the new composite gene makes something that is easily detected. In this case that something is called beta-lactamase. Now if a drug turns off the cancer-causing pathway in the Aurora cells, it also turns off the production of beta-lactamase.

To detect beta-lactamase, Aurora uses Tsien’s inventions: custom-designed fluorescent molecules (or fluorophores). These molecules absorb high energy light waves (such as violet), and re-emit lower energy light (such as blue). The amount of blue light coming out can indicate the amount of fluorophore present.

A Molecular Test System

Aurora’s standard test system uses a molecule made of two fluorophores. The first absorbs violet light and would, by itself, emit blue light. But if the two-part molecule is intact, the light energy is transferred to the second fluorophore, which emits green light. It stays intact unless beta-lactamase is around, in which case the molecule is split in two, no energy transfer takes place, and there is no green light.

Green Fluorescent
Protein (GFP) in
Aequorea victoria

The basis for the second set of fluorophores has been provided by evolution. Green fluorescent protein (GFP) is used by the Pacific Northwest jellyfish Aequorea victoria to make a green flash when it is agitated. This is presumably an attempt to confuse its enemies and thus defend itself.

But GFP can also be used in drug hunting. GFP makes its own fluorophore by joining together three of its building blocks in a ring. By changing building blocks in and around the fluorophore, Tsien has improved the strength of the green fluorescence, and modified it to produce blue, cyan and yellow variants.

The union of GFP and BFP (the blue variant of GFP) yields a protease sensor. Proteases chop up proteins, and are important in many diseases. HIV protease, for example, has been a prime target for anti-AIDS drugs. If GFP and BFP are linked, energy can pass between them as with the chemical fluorophores. A protease separates the two proteins, stops the energy transfer, and changes the color of the glow. An anti-protease drug should jam the protease and preserve the GFP-BFP link.

Cellular Signaling

GFP Tracks Proteins
into the Nucleus

GFP and its relatives are also good for seeing where proteins are in the cell. Some of the signals that tell a cell to grow either normally (as in wound repair) or aberrantly (as in cancer) exert their effect by sending a protein into the nucleus, the compartment of the cell that stores the DNA. Only in the nucleus can the protein switch on genes to get the cell growing. Attaching GFP to the protein doesn’t stop the protein from finding its way to the nucleus, but now it’s movement can be tracked.

About half of the technical staff at Aurora are engineers, a good indicator that machinery is as important as fluorophores. In about two years, Aurora aims to have its system fully operational, screening 100,000 compounds a day using 3456-well "nanoplates" that are the same size as conventional dishes that have 96 wells. Each well of a nanoplate only holds ~1 µl (10-6 liters, including tens to hundreds of cells). A pizoelectric device delivers volumes as small as 200 pl (where 1 pl is 10-12 liters) by using high frequency voltage pulses to create 2000 drops per second. The tiny volumes mean that chemicals can be delivered without the need for dilution beforehand.

In the meantime, Aurora is using prototype versions for the screening of ~10,000 single compounds a day, using 96-well and 384-well plates. Already the throughput is high enough to dispense with the complications of testing mixtures of several chemicals, which was done to cope with the flood of chemicals being made by combinatorial chemists. "Mixtures seemed like a good idea early on [in the history of combinatorial chemistry]," says Gordon Foulkes, until recently Aurora’s Chief Technical Officer. "But by the time you can screen as fast as we can, single compounds make a lot more sense."

The Lab-On-A-Chip

Layout of a
Assay Chip

As companies screen more chemicals in more assays, the pressure for miniaturization is intense. Smaller assays take up less space, run faster, and use less chemicals and proteins. Most engineers design the newer, smaller assay equipment by modifying existing equipment, but this approach can only be pushed so far.

Caliper Technologies Corporation (Palo Alto, Calif.) has taken their lead from a very different source: the computer industry. Caliper etches tiny channels on glass chips and uses electricity to shunt liquids around, thus achieving, in miniature, the measuring and mixing that is at the center of all drug assays.

The contrast with current laboratory hardware is striking. "If you go through a standard laboratory it is unbelievably primitive," says Michael Knapp, Vice President for Science and Technology at Caliper. Where once there were test tubes and beakers, the Caliper chips, just 2 cm on each side, slot into a machine with the color, size and lines of a Macintosh computer. "You can imagine putting it on your desktop and a scientist will look like any other office worker," says George Church (Harvard Medical School, Boston, Mass., and a member of Caliper’s scientific advisory board).

The Macarena

The box -- dubbed the Macarena -- contains the electronics and a sensitive microscope that detects the changes in fluorescence that signal the outcome of the experiment. "It becomes a personal laboratory," says Knapp.

Caliper's machine
delivers multiple tiny
fluid samples

The experimental design is defined by the layout of the chip. "You create a fluidic logic that is not unlike the logic of electrons flowing in computer chips," says Knapp. Each chip has a unique layout of reservoirs and channels. Electrical pulses bring fluids together in the channels, mix them together, and separate the dissolved proteins and chemicals based on how fast they move in an electric field.

The Macarena, which will be launched later this year, is aimed at the research scientist doing handfuls of tests at a time. But in the next few months Caliper will begin testing large numbers of potential drugs on custom-built machines, which can deliver multiple samples of fluid as small as 16 picoliters. A different chemical is sent into the system for testing every few seconds. "You could assemble warehouses of equipment and get to that throughput, but it would cost too much," says Knapp.

Drug Hunting

Where Knapp hopes this is all heading is to a new way of hunting for drugs. The old approach involved an exhaustive search for the single, perfect disease target to put in your screen -- the protein found by biochemistry to be involved in allergic reactions, or the gene that predisposed certain families to cancer. Now that DNA sequencing is fast and cheap, random sequencing plus a hint of relevance can land a gene and its associated protein in a screening program. But, says Knapp, "defining a valid target is very vague." If the Caliper system gets fast enough Knapp wants to dispense with this imprecise filtering step, discard all pretense of scientific hypothesis and reasoning, and go for the ultimate testing system: one that tests every chemical against every protein target.

"People are going to hate this idea," he says. "I already know people who hate this idea, because they think ‘I’m a scientist, I’m smarter than that.’ But this is the killer app of the new laboratory."

Originally published in the web magazine Access Excellence.