Characterizing the glycome is a daunting task, but there are some new high throughput tools to tackle it.
When newspaper headlines hailed the “completion” of the first draft of the human genome sequence in 2000, drug developers understood that the really difficult work was just about to begin. That’s because genes themselves are not particularly good drug targets; the really interesting molecules are the proteins and their post-translational modifications. In a characteristically cruel twist, though, nature has made these structures considerably harder to work with than DNA.
Moving from DNA to proteins to carbohydrates, biological complexity skyrockets while screening capacity plummets. Robots routinely array the 20,000 or so genes in the human genome into silicon chips, but capturing just a few thousand proteins remains daunting. The problem gets far worse at the carbohydrate level.
“The numbers of oligosaccharides are … unfathomably large, huge, because of all of the variations,” says Professor Ten Feizi, Director of the Glycosciences Laboratory, Division of Medicine at Imperial College Northwick Park Campus, Middlesex, UK. With conservative estimates of the glycome running in the millions of molecules, the best available screening tools probe just a tiny portion. “As of this week, we’re up to 600 probes, which represent oligosaccharide sequences from two to 20 monosaccharide units long,” says Feizi, adding “That’s miniscule compared to the gene arrays, but 600 for carbohydrate arrays is quite a large number, I think.”
Gimme some sugar
One of the biggest challenges in screening for carbohydrate-binding drugs is simply obtaining the carbohydrates. Because they cannot be cloned like genes, or transferred to conventional overexpression systems like proteins, sugars in the lab generally come from either their natural source or a chemical synthesis. Neither approach is especially user-friendly, but researchers have gradually made progress on both.
Isolating carbohydrates from natural sources is an especially good strategy when the specific target is unknown. “If there’s a … cell, which you believe is expressing a carbohydrate determinant, and you don’t know what it is, we have the possibility to make designer arrays from them,” says Feizi. She and her colleagues pioneered the use of such glycoarrays, beginning with a handful of spots on nitrocellulose membranes and working up to the hundreds of features they can now put onto glass slides.
The key is a chemical modification called the neoglycolytic technology, which couples each oligosaccharide to a lipid molecule. “Unlike some of the other conjugation procedures, where multiple oligosaccharides are placed onto large molecules such as albumin, they ionize extremely well, so they lend themselves to mass spectrometry,” says Feizi. The technique works for synthetic oligosaccharides as well as naturally-derived ones, allowing the researchers to combine the two in a single screen.
Researchers who already know their target’s structure may choose to synthesize it by conventional organic chemistry instead of using a natural source. However, even hard-core synthetic chemists often like to include naturally-derived sugars in the screen. “With synthetically-defined sugars, you know exactly what’s there; with natural ones … you know those sugars are presented in natural ways,” says Jeff Gildersleeve, PhD, head of the chemical biology section in the laboratory of medicinal chemistry at the National Cancer Institute in Frederick, Md.
Sweetening the pot
The other big advantage of naturally-derived sugars is that nature has already done the hard work of generating diversity. While a synthetic chemist can take weeks to make a single new variation on a polysaccharide chain, a cell can generate hundreds of variants each minute.
Unsurprisingly, many researchers are now trying to combine the diversity of natural carbohydrates with the predictability of synthetic ones, using techniques such as combinatorial chemistry, automated synthesis, and enzymatic reactions. “I think all three of those ways are really improving, and over the next few years there will be significant advances that will really increase the diversity on arrays at least ten-fold,” says Gildersleeve.
Xi Chen, PhD, assistant professor of chemistry at the University of California in Davis, Calif., is one of the scientists developing new carbohydrate synthesis techniques. Chen, whose research focuses on characterizing sialic acid-binding proteins and inhibitors, favors an approach she calls chemoenzymatic synthesis. By feeding synthetic precursor molecules to commercially-available enzymes, she and her colleagues are able to generate a wide range of structures quickly.
“We don’t want to complicate the process by purifying the intermediate, [so] we kind of put three enzymes in one pot, mix everything, then … after several hours, overnight, we get our final product,” says Chen. The team has since extended this “one pot” synthesis by combining multiple precursors and enzyme variants across a series of reactions, all in a 96-well plate. “Start with fifteen types of sialic acid precursor [and] two different kinds of linker, you will get a thousand compounds at the end,” she says.
By altering the available starting materials, the researchers can also create mixtures of natural and wholly artificial structures. That allows them to probe the normal binding activity of a carbohydrate-binding protein, as well as its tolerance for variations that would never occur in nature. The result is a system that can yield fine-grained information on protein-carbohydrate binding, and also help illuminate the best targets for inhibitors or activators.
The long haul
Regardless of the synthesis strategy, yields are usually small, which is why almost all carbohydrate screeners rely on microarrays. “With milligrams of material, you can do hundreds of thousands of studies, and that’s where the glycoarrays are very attractive to glycoscientists,” says Sabine Flitsch, PhD, professor of chemical biology at the University of Manchester in Manchester, UK.
Indeed, Flitsch and her colleagues in the UK Glycoarray Consortium, which she coordinates, are now focusing on making glycoarrays even more versatile. Current glycoarrays, such as Feizi’s, use a simple fluorescent readout to determine which proteins bind the carbohydrates on the array. While that information is certainly necessary, it is far from sufficient for drug development. “[We’re] trying to make slightly more advanced chips, which maybe have slightly less throughput, but where we can actually measure binding constants using surface plasmon resonance … or where we can look at enzymatic reactions, protein-carbohydrate interactions not only in terms of binding, but also in terms of alterations,” says Flitsch.
Even with a new generation of tools, scientific understanding of the glycome remains rudimentary. Worse, some of it appears to be wrong, as Gildersleeve recently discovered. Screening a large panel of anti-carbohydrate monoclonal antibodies, he and his colleagues found that many cross-react with multiple antigens. Previous work is built on the assumption that these antibodies are specific. The bad news is still sinking in, Gildersleeve says, but there are ways to adapt to it: “Part of the problem is thinking that an antibody is specific and having it not be [specific]. If you know that the antibody cross-reacts with two or three things, then you can draw your conclusions based on that information.”
The result provides an important cautionary note for drug developers, who will also have to aim carefully to avoid hitting too many targets. “There are many instances of physiological carbohydrate-protein interactions, but we need to know about cross-reactions, and then design drugs which are much more specific,” says Feizi. She adds that “we need a database of these, the recognition systems within glycomes, and then we can think about targeting them. It’s a very long haul.”
About the Author
Originally trained as a microbiologist, Alan Dove has been writing about science and its interfaces with industry and government for more than a decade.
This article was published in Drug Discovery & Development magazine: Vol. 11, No. 6, June, 2008, pp. 26-28.
Filed Under: Drug Discovery