NEW YORK, Feb 15 – With scores of businesses looking to develop protein chips, it would seem a company with years of experience mass-producing microarrays might have a leg up on the competition.
But Affymetrix, maker of the widely used GeneChip DNA microarrays, has no immediate plans to invest in protein chip research. “It’s just too far in the future,” a spokeswoman said.
Affymetrix’s reluctance to enter the burgeoning market underscores the significant hurdles researchers face as they try to produce equivalent tools for studying proteins. Manufacturing a reproducible microarray of proteins will likely require advancements in surface chemistry, detection systems, and protein synthesis, among others, and, scientists say, knowing how to make a DNA array won’t be much help.
However, with potential users hungering for an easy-to-use protein chip and market researchers such as BioInsights predicting that the market for these tools will grow to an annual $490 million by 2006, there are plenty of players trying to overcome the obstacles. In fact, there are at least two early versions of a protein chip already on the market.
Ciphergen, in Fremont, Calif., sells an affinity chip with a surface that weakly binds proteins screened from a sample, and uses mass spectrometry to identify the protein profile. “Its main utility is discovery, especially if you’re trying to find a biomarker and don’t know the antigen or antibody,” said Chris Pohl, vice president for research and development at Ciphergen.
Another type of chip—available since 1990—comes from Biacore, in Uppsala, Sweden. Instead of screening protein biomarkers, Biacore’s chip uses a technique called surface plasmon resonance to investigate protein binding kinetics. The technique is good for measuring protein-protein interactions because no labeling of the proteins is required, a step that might influence how the proteins interact, said Ruedi Aebersold, a cellular biologist at the Institute for Systems Biology in Seattle.
Ultimately, however, scientists would like more than what either Ciphergen, Biacore, or any other company currently offers. Ideally, said Aebersold, one type of protein chip—a protein microarray—would identify the proteins in a sample, and quantify how much of each protein is present. This type of high-throughput chip would have applications for basic research, allowing users to determine what proteins a gene expresses in a sample, and perhaps also in clinical diagnostics, by helping doctors gather protein profiles from patients.
HOW TO BUILD A PROTEIN CHIP
In order for protein microarrays to fulfill their promise, however, researchers must first determine which proteins are important or interesting enough to place on a chip.
Large Scale Proteomicics, a subsidiary of Large Scale Biology, is hoping to leverage its proteomics expertise—namely its index of protein expression levels from 157 tissues—to get a head start in the protein chip market. In January, LSP, in Germantown, Md., hired San Diego-based Biosite to manufacture between 2,000 and 5,000 antibodies for the proteins contained in LSP’s index. Once LSP and Biosite have created their library of protein antibodies, the companies will either manufacture the chip themselves, using Biosite’s existing platform for making specialized diagnostic chips, or find a partner, said Biosite vice president for research Ken Buechler.
Other potential chipmakers are focusing less on the content, and more on improving the techniques for attaching proteins to surfaces. Unlike DNA, proteins do not have an exact complement with which they will readily hybridize. Thus, researchers have devised a number of methods for capturing and binding proteins to the chip.
Biosite, for example, is creating binders by manufacturing full antibodies to proteins, using a technique called phage display technology. In Biosite’s version of this technique, mice are injected with a human protein, causing the mice to produce the desired antibodies. To get the antibodies out, scientists remove the spleens of the mice, extract the antibodies, and clone the antibodies in a bacterial phage, giving the phage a protein “coat” in the shape of the antibody. Researchers then multiply the number of phages by infecting E. coli or some other organism with the phage, and allowing the organism to replicate. Repeating this process with different proteins results in a library of different proteins.
Buechler said the phage display technique takes two to three months to create one antibody, but that the technique was much faster than the traditional approach to making antibodies, called hybridoma, that can take up to a year. Furthermore, there are benefits to using the diversity of an animals immune system to create the antibody, he said, because you can isolate antibodies with affinity to more than one site on the protein.
Alternatively, Phylos, based in Lexington, Mass., is proposing a method for binding specific proteins that employs an artificial binder—a kind of pseudo-antibody—using the scaffold of a human protein called fibronectin. The artificial binders are about one quarter the size of regular antibodies, allowing more to be packed on a chip, said Phylos CEO Gustaf Christensen.
Artificial binders take only “days to a few weeks” to manufacture, giving Phylos a relatively fast response time, added Christensen. And because Phylos can isolate multiple binders for the same protein, he said, they should be as versatile as full antibodies.
Meanwhile, scientists at Somalogic, based in Boulder, Colorado, have patented a method for using single strands of DNA to mimick antibodies. The technique takes advantage of oligonucleotides’ ability to assume three dimensional shapes, said Todd Gander, a [spokesman] at Somalogic, and relies on a patented [trial-and-error] approach to find strands of nucleic acid that—when folded—bind with a specific protein.
Somalogic’s binding technique also involves photochemistry. When a protein initially binds with the oligonucleotide, only the protein’s affinity for that shape keeps it there. But when the surface is exposed to UV light, active compounds placed in the oligonucleotide strand form crosslinks with electron-rich areas of the protein, said Gander. This is only happens, he said, if the protein is a perfect match. “It’s analogous to having two antibodies,” he said.
GETTING PROTEINS TO STAY PUT
Finding a good binder for a protein, however, is complicated by the tendency of proteins to denature in less-than-optimal environments—such as near chip surfaces. Scientists at Zyomyx, in Hayward, Calif., are attempting to make proteins feel more at home when brought near to the surface of a chip, so that they will stay in their natural conformation and not denature.
To accomplish this, researchers at Zyomyx are applying surface chemistry derived from microscopy experiments carried out by the chief technical officer, Peter Wagner. The technique immobilizes proteins on the surface of the chip while surrounding them with a hydrophilic environment, said Zyomyx CEO Lawrence Cohen. “We provide the sticky points on a shag carpet,” Cohen said. “We’re interested in the binding sites and how the neutral sites contribute to the [protein’s] environment.”
Other chip technologies, such as the antibody approach of LSB and Biosite, can get away with some surface inactivation because their arrays contain a relatively large number of antibodies, and some are bound to be in their natural conformation, Cohen added. “The problem comes with miniaturizing the devices,” he said. “That’s when you have to start thinking about surface chemistry.”
While Zyomyx perfects its surface chemistry, others are focusing their attention on detection technologies. HTS Biosciences, an 80:20 joint venture between Applied Biosystems and Quantech, hopes to bring to market three technology platforms: a surface plasmon resonance system similar to that of Biacore, a phase fluorescence platform, and an array-based chemiluminescence detection technology.
Researchers will want to use the SPR platform for drug target discovery, said HTS CEO Greg Freitag, and the phase fluorescence and chemiluminesence techniques for applications requiring greater sensitivity, such as identifying where proteins have bound with the surface and how much of that protein is present.
“Most people have one technology that they have to develop into a system that can be all things for all people. Instead we’ve been able to acquire a number of synergistic technologies and focus each detection system to its own use,” said Freitag.
With many companies hoping to release beta versions of their systems by the end of this year, it may be only a matter of time before a shakeout proves which technologies will be successful.
“The proof is in the pudding,” said Felicia Gentile, CEO of Bioinsights. “Can any of these guys make a product anyone will buy at a reasonable cost?”
As with any new technology, you’ll have two or three companies with people with the right expertise, added Phil Andrews, a biochemist at the University of Michigan. “There’s a lot of ideas and it’s a very exciting time, but we don’t know exactly how good they’re going to be.”