| Functional protein microarrays are critical for the next phase of proteomics research. Like DNA chips, protein chips will be able to analyze thousands of samples simultaneously, leading the way towards a complete map of the entire complement of human proteins. But, unlike DNA, proteins are not so easy to attach to chips. Where DNA is robust, and able to withstand harsh experimental conditions, proteins are fragile and will denature if they aren't treated gently. While there are good methods for amplifying DNA, so that even very tiny amounts can be detected, there are none available for proteins. It's a whole new set of constraints, and protein biochip companies are coming up with some interesting -- and very different -- ways to solve the dilemma. |
The use of biochips in scientific research and drug discovery has become so widespread that it's almost impossible now to imagine how we got along without them. Biochips -- common parlance for DNA microarrays -- have taken their place in experimental biology's lineup of fundamental enabling technologies (earlier breakthroughs include cloning, automated DNA sequencing and the polymerase chain reaction). Biochips have allowed researchers to detect thousands of genes simultaneously from a very tiny sample. They've also proved invaluable for analyzing the expression of those genes, and for rapidly identifying new potential drug targets.
And, in the not-too-distant future, DNA microarrays should enable the introduction of personalized medicine (i.e., pharmacogenomics) into clinical practice, diagnostic labs and doctor's offices -- that is, if these miniaturized assays and their accompanying reagents and instrumentation can be manufactured cheaply enough for everyday use. With the recent entry into the arena of heavy-weight tech giants Motorola, Agilent, Corning and Mitsubishi Rayon, this, too, may come to pass sooner than we thought.
But biochips -- officially biological microchips -- come in other flavors, too. For one, microchips don't necessarily have to be silicon wafers. They can be made of aluminum, for instance, or even glass. And they don't have to be wafers; slides are also used. For another, some companies are developing microfluidic chips, on which it's possible to perform lab-type assays in situ. And, most importantly, DNA is not the only biological entity that can be arrayed or spotted onto the surface of these "chips": It's also possible to configure arrays of proteins.
Possible, but difficult. Proteins present their own daunting challenges -- which are now being tackled by the growing list of companies that are rising to meet the demands presented by the newly burgeoning field of proteomics. For, while the human genome may contain 100,000 genes, post-translational modifications and RNA splicing events result in far greater than 100,000 proteins. Thus, the entire complement of human proteins (the proteome) could number in the hundreds of millions, perhaps even billions. Protein microarrays, which will be able to measure thousands of proteins simultaneously, are the only way to go. Ideally, such arrays will be able to measure not only protein-protein interactions, but also protein-small molecule interactions and enzyme-substrate reactions. Eventually, it's hoped, they'll also be used for differential profiling -- for instance, to distinguish the proteins of healthy cells from those of diseased cells. But first, it's necessary to devise the chips themselves.
One thing that the DNA microarrays have going for them is the robust nature of the nucleic acids themselves. Oligonucleotides (oligos) can be absorbed onto various organic solid phase materials, dried and re-hydrated without any loss of activity. Proteins, on the other hand, can't be immobilized on the same types of materials, for they are extremely sensitive to the physical and chemical properties of the particular substrate. And some substrates will work for one class of proteins but not another, making it difficult to come up with a so-called generic solid phase material.
As well, proteins cannot be dried; they must remain in a liquid environment to retain their activity. Proteins are so sensitive to their environment that they will denature at solid-liquid and liquid-air interfaces (which become considerable as assays are made ever smaller, because at the same time, surface-to-volume ratios increase.)
One doesn't have to be concerned with the structure of an oligo, either: It only has to hybridize with its complement (in true zipper-like fashion). Proteins, however, have one-, two- and three-dimensional configurations as they transform from a straight chain of amino acids into a functional unit. Since measuring function is what it's all about, this aspect is critical to creating a bona fide protein microarray. To top it off, these microarrays must also be able to stand up to high-speed processing and analysis.
Functional protein microarrays are considered critical to progress in proteomics research. And, not surprisingly, academic and industrial researchers alike are expending considerable time and energy to come up with an operable system.
One group of university researchers, in particular, has already come up with a solution that works -- in the lab. Gavin MacBeath and Stuart Schreiber, both affiliated with Harvard University, worked out a method to immobilize proteins by covalently attaching them to glass microscope slides. (You can read the details in their Science paper: MacBeath and Schreiber. Printing Proteins As Microarrays For High-Throughput Function Determination. Science 289: 1760; 8 Sept. 2000). The scientists also tested their protein microarrays, which contain about 10,000 spots per slide, and demonstrated that they were capable of detecting interactions between one protein and another, a small molecule and a protein and an enzyme and its substrate. Best of all, MacBeath and Schreiber used standard lab equipment and materials in their experiments: As they stated in their paper, "One of our primary objectives in pursing this approach was to make the technology easily accessible and compatible with standard instrumentation."
While academics might thus have their hands on an inexpensive way to create protein microarrays in the lab, the biotech firms that are tackling this challenge are more concerned with large-scale, standardized methods for producing protein biochips.
Privately held Zyomyx Inc. could be the first of these young biotech firms to put a product on the market. It's lining up partners and has just raised $20 million in the first tranche of a series D financing, giving it the extra fuel to carry its R&D programs forward towards commercial introduction.
The Hayward, CA firm, which was founded in 1998, is exploiting its expertise in surface chemistry to create arrays of proteins that are oriented correctly on a solid base. It does this in a modular fashion: The chip's base (or device) material is covered with a multi-component organic thin film. The idea here is to eliminate any surface defects on the device so that the immobilized proteins will be arrayed uniformly, as well as to insulate the proteins from the support material so they aren't denatured. As well, Zyomyx says that this approach helps to reduce non-specific protein binding. "A major bottleneck in the construction of high-density protein microarrays is a limited understanding of protein interaction with engineered materials to prevent surface-induced inactivation," explained Lawrence Cohen, Zyomyx's COO.
After the device's surface has been prepared, an attachment tag is added to the top layer, and a protein capture agent (such as an antibody fragment, a peptide or even a novel scaffold of antibody-like molecules) gets bound to its free end. These protein capture agents can be tailored specifically to snag the proteins of interest during the actual experiment. The protein arrays are integrated into flow chambers, so that the proteins are always in aqueous solution and won't be denatured. Target compounds or proteins introduced into the chambers interact with the immobilized proteins and their binding can be detected by various methods, including but not limited to fluorescence. The company is also developing ultra-high throughput protein dispensers and a biochip reader. It intends to sell an integrated analytical system, but it should also be possible for researchers to modify their conventional confocal readers for DNA chips to work with Zyomyx's chip holders.
Although Zyomyx is aiming to produce very dense microarrays (currently, the capacity is 10,000 monoclonal antibodies per square centimeter), the company used a much smaller array (a 5X5 array, with 25 monoclonals) to demonstrate proof-of-concept. "We concluded the proof-of-concept stage at the end of 1999," Cohen explained. "We're in the phase now where we're starting to get data on biology and performance…We're doing relatively complicated arrays, and the sensitivity levels are as good as the best [standard] immunological assays. There's no compromise [in sensitivity] due to size." As well, Zyomyx is conducting several pilot studies with biotech firms, in which it is collecting generalized performance data using different antibodies.
Cohen said that the first commercial application of Zyomyx's chips will be for drug discovery research, particularly in protein profiling. "The chips will contain arrays of antibodies to known proteins, and will be used to profile their expression in various tissues." These might include assays to detect cytokines, transcription factors, or the products of oncogenes. "In general, we feel that the core technology platform allows the development of protein biochips to miniaturize many types of protein measurements including (in addition to protein profiling) protein discovery, protein activities, protein structure and the identification of protein-protein and protein-small molecule interactions," Cohen added.
While Zyomyx has devised methods to immobilize proteins on chip surfaces, CombiMatrix Corp. has worked out the means to synthesize the materials right there on the chip. This small private company, currently nestled in the foothills outside Seattle, uses electrochemistry and semiconductor technology to synthesize biological materials on a three-dimensional active chip surface. This chip consists of thousands upon thousands of tiny "virtual flasks" arranged in a grid pattern on the surface of a semiconductor wafer. Potentially, these chips can support very high density arrays, up to one million sites per square centimeter.
Each virtual flask is about the diameter of a human hair; each is separated from the others by chemical solutions; and each is wired to a computer, which can direct it to construct a specific chemical compound. "The device is an array of electrodes, each of which is individually addressed," explained Donald Montgomery, co-founder, senior VP and chief technical officer. "There's a porous reaction layer overlaying the electrodes, which gives a high surface area; the molecules are synthesized in the interstitial spaces."
For DNA microarrays, the process involves synthesizing oligonucleotides on the chip. After the first base is covalently bound to the chip, the electrode is turned on, producing protons (acid), which deprotect the nucleotide so that one floating in solution above it can attach. The reaction is stopped, and then this process is repeated, thus building an oligonucleotide of specified sequence onto the chip. Thus, it's easy to customize the arrays on these chips. In fact, according to Siavash Ghazvini, CombiMatrix's VP of business development, customers will be able to access this computer from the Web; they'll order their chips simply by entering the desired sequence online. And, the customized DNA chips can be turned around in 48 hours.
CombiMatrix claims its customized chips will not only be faster to produce than the competition's, but cheaper, too. That's because the firm "lets the semiconducter industry make the chips," Montgomery explained. "We use these to produce arrays of biomolecules." And, as those chips get denser, so, too, can CombiMatrix's arrays. "Our first device contained 1,170 sites per square centimeter; newer-technology chips allow us to get 471,000 sites per square centimeter. Also available now are chips that will allow 1 million sites per square centimeter, and in 18 months it will be 4 million."
And though the company's major thrust has been in the DNA chip arena, it's also developing protein microarrays, everything from antibodies to peptides. Peptides are apparently built on the chip one amino acid at a time, though the company is not releasing any details yet on its exact process. As well, CombiMatrix is devising ways to create antibody arrays that depend on antibodies in solution finding their targets on the chip.
There are lots of challenges involved in creating protein microarrays, Ghazvini pointed out. "How do you get antibodies onto a chip in a specific orientation? How do you attach them to the surface? How do you maintain their structure and function once they're on the chip? And, how do you prevent cross-hybridization?" Antibodies are more stable than globular proteins, he continued, but they're still affected by temperature and pH.
Challenges abound, but the future applications of protein biochips make this a compelling project. "We have worked out some of the characteristics [for a protein chip]," Ghazvini said. "Our 3-D matrix keeps them hydrated and functional. They're also in a biologically friendly milieu, instead of on glass." In the near-term, CombiMatrix (which is a majority-owned affiliate of Acacia Research Corp.) is also propelled forward through its Phase II SBIR grant, which it received from the Department of Defense in January 2000 to develop immunochemical assays for chemical and biological warfare agents. "We're making antibody chips now," he added.
Packard BioScience Co. has come up with yet a third method for creating protein microarrays. In fact, concurrent with its acquisition of GSLI Life Sciences (with expertise in laser detection and data analysis), Packard created a new division, Packard BioChip Technologies LLC, just to develop its technology.
Packard BioChip's twist is to use hydrogel chips: In this method, the probe molecules are immobilized within a three-dimensional hydrophilic polymer matrix on a glass microscope slide. The matrix also contains a coupling reagent that reacts with amine groups on the probe molecules, attaching them covalently and immobilizing them within the matrix. The gel is aqueous, porous (creating a large surface area) and dimensional (eliminating solid-liquid interfaces). And, although the company currently uses glass slides as the solid support for its chips, "We have the ability to be pretty flexible," said Jocelyn Burke, VP and general manager of the biochip ventures division. "We see the possibility of putting our hydrogel matrix on other types of supports."
Thus, although Packard has concentrated on DNA chips so far, the hydrogels seem to provide an ideal environment for proteins. "The gel substrate is truly three-dimensional and porous," Burke explained. "We speculate that the gel provides a hydrophilic environment that lends itself to protein-protein interactions, and that these proteins behave more naturally." The company also has some experimental evidence, on both antibodies and enzymes, to support this theory, she said.
There's another critical element to Packard BioChip's approach, however, and that concerns its method for spotting samples onto the chips. The company recently received a patent on its inkjet printing method for dispensing liquid samples, which permits the accurate dispensing of multiple droplets of DNA, proteins or even cells in a single spot on a microarray. Importantly, this is non-contact dispensing, which means that "you can put a lot more protein down [in a single spot] and still have a good spot," Burke explained. Moreover, "you know how much protein you've put down. You know how many droplets you've dispersed [on one spot] and you can calculate the actual amount of material deposited." According to Burke, with contact methods for spotting chips, not only can there be physical damage to the substrate when contact is made, but also "it's very difficult to know how much sample you actually put down." Knowing the exact amount of material is an important aspect of achieving the high levels of sensitivity and specificity that protein microarrays will require. "You can't amplify proteins like you can nucleic acids [to increase the sensitivity]," Burke said. So, improvements must come in other areas, which may also include new imaging technologies or labels for detecting reactions.
Packard BioScience, which has several (undisclosed) chip partnerships in the diagnostics arena, also recently signed a collaboration with proteomics firm Oxford GlycoSciences plc (OGS) to develop protein microarrays.
"We tried quite a few formats for biochips that could detect proteins in a sample," explained Raj Parekh, OGS' chief scientific officer. The intent is to prepare highly specific antibodies, and screen them against a large library of human proteins, he continued. "We needed a chemistry that would allow us to deposit the antibodies [on the chip] while still retaining their binding affinity." However, OGS researchers found that the two relatively common methods for doing this -- direct chemical attachment or attaching to the surface via protein G -- were "not as efficient as they should be," Parekh said. "This is where Packard comes in." Packard's hydrogel-based platform allows the creation of antibody arrays without chemical attachment, he explained. "There are still challenges, however," Parekh said. "We have to be able to use these antibody arrays in a productive manner."
The Packard alliance is only part of OGS' overall industrial-scale proteomics strategy, however. Within days of announcing its Packard deal, OGS had also signed on two more partners -- PE Biosystems Group and Cambridge Antibody Technology Group plc (CAT). In the PE deal, OGS will become an early access customer for Applied Biosystems' next-generation MALDI TOF/TOF mass spectrometer, which promises to enable high-throughput analysis of proteins -- necessary for the large-scale research efforts that proteomics demands. In the CAT deal, OGS has signed another partner to develop protein chip technology, this one specifically employing antibody-based microarrays.
"We want to marry our protein library with CAT's human antibody libraries to develop highly specific, high affinity arrays," Parekh explained. The protein-antibody pairs selected will be used to develop a new generation of microarrays for detecting proteins. "For instance, imagine creating biochips for membrane proteins," he continued. "We'll use our knowledge of proteins to select antibodies from CAT's libraries and array and deposit them on chips according to Packard's technology."
OGS already has the prototype microarrays; now it intends to scale them up for industrial applications, including drug discovery, screening and diagnostics. But these commercial microarrays will probably be a diverse lot, themselves. "Proteins are so diverse physically," Parekh said, that he doesn't forsee "global" biochips but rather "panels of biochips focused around specific applications." This strategy fits right in with current research efforts in proteomics, too, for most scientists today concentrate on families of proteins.
Protein families are certainly the core of AxCell Biosciences Corp.'s business strategy. And, the Princeton, NJ subsidiary of Cytogen Corp. intends to sell defined sets of known protein families, as arrays, for use in lead optimization in drug discovery programs. It's building an inter-functional proteomic database of human protein signaling pathways, which can be used to correlate protein pathway data with sequence, expression, tissue distribution, structural and bibliographic information that exists for that particular protein and pathway.
To measure protein interactions on a chip, AxCell recently joined forces with privately held Molecular Staging Inc. MSI, of Guilford, CT, brings to the alliance its rolling circle amplification technology, a nucleic acid amplification technology for detecting the presence of target molecules arrayed on a chip. Parent company Cytogen already has a separate, ongoing collaboration with MSI to develop ultra-sensitive diagnostic assays for prostate cancer; according to John Rodwell, AxCell's acting president and chief technical officer, the companies have already initiated work on protein chip technology. Now, they intend to develop that into a high-throughput technology for measuring protein-protein interactions.
Newly-public Ciphergen Biosystems Inc. is developing a multi-tasking chip that should allow researchers to capture, separate and quantitatively analyze proteins right on the chip. This system integrates mass spectrometry (in particular, surface enhanced laser desorption/ionization [SELDI]) and biochip technology on a single chip.
Ciphergen's ProteinChip System uses small arrays with chemically (cationic, hydrophobic, etc.) or biologically (antibody, receptor, DNA, etc.) treated surfaces to interact with proteins. The chips, which are made of aluminum (although they could be made of some other metal), are about three inches long and one centimeter wide; each contains eight sites and a group of 12 can be processed in the equivalent of a 96-well format, explained Christopher Pohl, Ciphergen's VP of R&D.
"We use two different methods to attach proteins to the chips," he continued. But in both cases, which employ different chemistries, the company first coats the chip surface with a film of silicon dioxide. Then the molecules of interest are attached covalently via a silane coupling reagent. Ciphergen offers ready-to-use chips as well as pre-activated chips. In the latter case, "It's up to the customer to apply their antibodies and block the remaining [reactive] groups," he continued. "Both [types of coated surfaces] will react readily with antibodies; choosing one over the other depends on the antibody."
The arrays select unknown proteins from the experimental sample (which can even be a complex sample such as whole blood or tissue) via affinity capture, which takes advantage of the characteristics of the pre-treated surface. After the chip is washed to remove excess sample, dried, and the surface treated with a surfactant to mask any remaining non-specifically bound sample, the captured proteins are desorbed and ionized by laser excitation, and then detected according to molecular weight. Known proteins are analyzed using on-chip functional assays.
Although one of the fundamental challenges of most protein microarrays is to keep those proteins from being denatured, Ciphergen's system largely bypasses this constraint. Because it's designed to measure the mass of the captured proteins (rather than their activity), once the initial binding reaction has occurred, "it's unimportant whether the captured proteins are exposed to denaturing conditions. It doesn't make a difference because we're interested in mass," Pohl said.
Ciphergen's system can be used to identify and characterize proteins, but it's also amenable to discovery-based research. For instance, it's possible to process samples from a normal cell and a diseased one and then compare their protein expression profiles. The proteins that turn out to be present in the diseased cell, but not the normal one, become potentially relevant biomarkers. "Two-thirds of the applications our customers are using center on profiling experiments," he said. "They're screening for biological markers where the protein is unknown."
Once protein microarrays are perfected, researchers will be able to do this, and a lot more, on a routine basis. That includes measuring the functions of thousands upon thousands of proteins simultaneously, and rapidly screening them all for new drug targets. It also will be (relatively) easy to take a potential drug candidate and screen all the metabolic pathways to check for unwanted reactions. But that's the future -- and it's barely begun. |