As a genomics-based biotech company, Millennium Pharmaceuticals has built some of its largest collaborations to date — with companies such as Bayer and Aventis — on the basis of its microarray and other gene expression technologies. But in recent months the Cambridge, Mass.-based company has indicated it is expanding its efforts in proteomics to augment its drug discovery arsenal.
Last month, Millennium signed on to become the first customer for a protein bioarray platform manufactured by Biacore, a company based in Uppsala, Sweden. And unlike some established pharmaceutical companies, which tend not to develop new technology independently, Millennium is developing its own platform for separating proteins using multidimensional liquid chromatography, combined with tandem mass spectrometry, a senior Millennium scientist said at a protoemics conference earlier this month.
The protein group at Millennium has been around since shortly after the company’s founding in 1993, but has grown sizeably since then to encompass “just about a little of everything,” including yeast two-hybrid and phage display studies to predict protein interactions, and 2D gel electrophoresis for separating and analyzing proteins, said Brad Guild, director of protein biochemistry at Millennium.
The fundamental driver for the expansion of the proteomics effort, said Guild, is to identify a series of protein markers that will help diagnose and monitor patients with certain diseases, and ultimately help match those patients with drug therapies that fit their protein or gene profile.
A year ago, the company entered a collaboration with Biacore to design a platform to study interactions between proteins and other molecules using surface plasmon resonance. Millennium was not the only pharmaceutical company to participate in the technology development program — SmithKline Beecham and Pharmacia also took part — but last month Millennium became the first to use Biacore’s S51 system as part of its drug discovery platform.
“We’re really building [the S51 system] into a high-density, multi-spot approach to protein-protein interactions,” said Guild. “It’s being developed more as a process technology central to all technology development within Millennium as it relates to proteomics.”
More recently, the company began building up its capabilities in multidimensional liquid chromatography, as an addition to its 2D gel separation equipment. Guild’s group uses both offline chromatography systems — those that aren’t directly coupled with mass spectrometers — as well as online chromatography methods for separating tryptic peptide digests before introducing them into the mass spectrometer. For these types of protein discovery applications, the Guild group uses Thermo Finnigan LCQ ion trap spectrometers.
“It’s fair to say that we have favored multi-dimensional chromatography over 2D gel electrophoresis,” Guild admitted, although he said Millennium also uses some 2D gels. “The advantage from my perspective is very simple: Whatever spot you can see at a minimum concentration on a 2D gel, you can see it at one tenth that concentration by using online multidimensional chromatography techniques before you put it into the mass spec.”
The higher sensitivity of the chromatography approach also means that smaller amounts of sample are required for analysis, Guild added. “When you’re working with very valuable, very limited patient samples, like a serum sample, or synovial fluid, or even cerebral spinal fluid, you don’t have a lot of that material to work with,” he said. “You can put a lot less of that on a series of online columns and still get information out the back end.”
Hands on technology development
Millennium hasn’t completely determined how it will quantify differences in protein expression between separate samples, but Steven Carr, senior director of discovery technologies, said that the company is considering a number of approaches similar to the ICAT reagent technology developed by Ruedi Aebersold of the Institute for Systems Biology. The ICAT reagents, which are used to isotopically label cysteine residues in peptide fragments, comprise only one of many methods for labeling protein fragments, Carr said, and Millennium is weighing whether to develop an analogous technique.
“We’re currently able to measure things semi-quantitatively” when comparing diseased with normal tissue or fluid samples, Carr said, “but we can only see five-fold or greater [changes in protein abundance], not one and half to two-fold changes.”
To process the “tens to hundreds of thousands” of spectra that come out of Millennium’s mass spectrometers each day, the company has designed a software suite that picks out the best spectra and inputs the data into both proprietary and public protein sequence databases. The software, called SpectraMill, then compares protein sequences to predict which gene or parent molecule the protein fragment is associated with.
Once the group discovers a protein that may indicate a specific disease state, Guild’s group then uses triple quadrupole mass spectrometers to validate that a candidate protein marker found in synovial fluid, for example, is also found in blood serum, where it may be more easily extracted from a patient in diagnostic applications.
Guild said the proteomics efforts at Millennium have already paid off in contributions to drug discovery. As an example, Guild cited a protein target that his group has studied to determine how post-translational modifications to the target affect drug delivery and toxicity. Guild hopes that his candidate proteins will ultimately find use as disease-monitoring markers during drug administration, but, he added, “that is yet to be proven.”
Willingness to take risks on new technology may determine the impact of proteomics within Millennium, said Guild. “We have a group here that has done some phenomenal things with respect to taking what is available commercially and then building on that,” he said. “If there is nothing commercially available, we build it ourselves and integrate it into the drug discovery platform.” — JSM