Drawing researchers from the UK, continental Europe, and the US, last week’s IBC conference “Proteomics and the Proteome” in Geneva, Switzerland, painted a picture of proteomics in its most cosmopolitan of guises. The three-day event featured talks on new protein separation technology, protein and antibody microarrays, as well as the potential for large-scale coordinated proteomics research efforts. Here are a few highlights from the meeting:
In a provocative talk that stimulated incredulous responses from the 100-odd attendees, Marc Reymond, chief medical officer of Hennigsdorf, Germany-based Europroteome, described his efforts to combine clinical and pathology data on colorectal cancer with molecular information obtained from 2D gel electrophoresis analysis of patient samples. To the surprise of conference attendees — Reymond himself was also initially taken aback, he said — his unpublished data seemed to indicate that clinical and pathological data alone could predict colorectal cancer patients’ likelihood of survival. “This is hot stuff,” Reymond said.
The findings, illustrated by Reymond’s graph of patients’ likelihood of survival versus age, are a thorn in the side of proteomics researchers who believe that basic biochemical data can serve as a reliable tool for diagnosing patient susceptibility and the progression of disease — with or without clinical or pathology data. But Reymond was direct: “Those who don’t consider pathology and clinical data will fail,” he said.
To generate his data, Reymond and his 40-odd collaborators from research hospitals across Europe compiled pathology and clinical information from 135 patients — both living and deceased — with phase I colorectal cancer, and enlisted a German bioinformatics company to write a rule-based algorithm to predict a given patient’s likelihood of survival at a specific age. Reymond’s collaborators trained their algorithm using data taken from 25 percent of the patients in the study, and then successfully predicted the likelihood that the tumors in the other 75 percent of the patients would metastasize.
The results are also provocative because the ethical guidelines of the study prevented Reymond from informing the living patients of their chances for survival — much to the dismay of some who heard the talk. Reymond agreed that his results should stimulate discussion of how to construct similar studies in the future.
But not everyone in attendance was convinced of the validity of the results. “We need to go back and look at what’s in that list [of pathology and clinical data],” said Sam Hanash, a proteomics researcher at the University of Michigan and president of HUPO. “At the moment it’s not very clear.”
Others also took a calculated view, noting that pathology and clinical information could be considered “higher level” data. “If we knew more about biochemical mechanisms [proteomics and other basic biological disciplines] we could do this too,” said Brian Davis, director of innovation research at Incyte Genomics.
The value of such studies may come from applying the data to patient cohort studies, enabling pharmaceutical companies to identify patient populations that may respond positively to a given treatment, said Ian Humphery-Smith, managing director and chief operating officer of Glaucus Proteomics in the Netherlands.
Sense Proteomic Launches p53 Protein Microarray
In the field of protein microarrays, most companies are still in the process of developing products. Sense Proteomic, based in Cambridge, UK, is one exception. CEO Roland Kozlowski said the company began marketing in late January a protein microarray containing 49 p53 mutant proteins, plus the wild-type, in a format that allows researchers to study the interactions of the tumor suppressor with other proteins, DNA, and small molecules.
Sense’s microarray platform consists of an E. coli-based expression system that attaches a peptide or protein domain to the N- or C-terminus of each protein. This tag allows the company to attach the proteins to a dextrin-coated glass plate using the streptavidin-biotin interaction, Kozlowski said, and to generate spots containing 200 picograms of protein. To detect an interaction, Sense has developed a label-free mass spectrometry technique on which the company has recently filed a patent, he said.
Together with Genetix, its manufacturing partner, Sense is also planning to launch similar microarrays containing p450 proteins for use in toxicology studies by the end of the year or early 2003, Kozlowski said. In addition, the company is preparing microarrays of breast cancer-related proteins. Kozlowski added that his company’s most pressing current need is to acquire capabilities in marketing and distribution.
Innovation Must Succeed in the Marketplace
Contrasting with the majority of technology-oriented presentations, Brian Davis, director of innovation research at Incyte, spoke more abstractly about his company’s attempt to shape its information products to appeal to the evolving nature of the pharmaceutical business. “To be innovative, one must succeed in the marketplace,” he paraphrased from Edward Tufte, author of the 1982 book The Visual Display of Quantitative Information.
Davis, formerly with Proteome, which was acquired by Incyte in December 2000, stressed that truly valuable information on biochemical pathways should be closely tied in with clinical disease and drug data, although he added that his talk should not be construed as indicating Incyte had plans to launch new products incorporating these ideas.
As an example of this approach, Davis showed a putative biochemical pathway for the mechanism of Crohn’s disease, an inflammatory bowel condition, overlaid with information on the known effects of the drugs currently on the market. The graphical interface allowed the user to select only certain classes of proteins, such as kinases, and view their potential interactions with the disease pathway.
Glaucus Proteomics Takes Steps Towards Proteome Chip
In describing his company’s three-year effort to develop a human proteome “look-alike” chip, Glaucus Proteomics’ Ian Humphery-Smith said his company — in collaboration with antibody manufacturer Medarex — is planning to release within the next three to four months an alpha version of a microarray containing up to 200 antibodies.
As part of its longer-term goal, Glaucus is putting together a platform capable of expressing proteins in multiple organisms, both bacterial and mammalian, Humphery-Smith said. Although expressing proteins through non-mammalian vectors would deprive them of posttranslational modifications, the potential up-side to this approach, he said, is that removing modifications would diminish the potential for proteins to non-specifically interact.
In January, Glaucus, which is based in Brunnik, the Netherlands, opened a 27,000-square-foot production facility in nearby Odijk, Humphery-Smith said. Since December the company has had access to the SARA supercomputing facility in Amsterdam, through a partnership with the Dutch government. Humphery-Smith said Glaucus plans to use the supercomputer to process the data from protein microarray experiments, and create tools for pattern recognition and data visualization.
As a challenge to the scientific community, Humphery-Smith pinpointed the pressing need for researchers to develop highly specific affinity ligands — antibody or otherwise — as a crucial ingredient to producing effective protein and antibody microarrays. In the face of the potential 400,000 to 500,000 proteins, including their isoforms and modified versions, in the human proteome, he said the discipline is “shying away” from the challenge.