While a number of proteomics firms continue to knock on the doors of big pharma for business, many pharmaceutical companies have chosen to integrate proteomics into their own R&D pipeline. In a morning session at the HUPO meeting in Versailles last November, scientists from Bristol-Myers Squibb, Novartis, Millennium Pharmaceuticals, and Celera gave participants a glance of the way they are using proteomics techniques to discover new drug targets and biomarkers for specific diseases.
Scientists at Bristol-Myers Squibb, for example, have worked to find new drug targets for osteoporosis by comparing proteins in stimulated and unstimulated osteoclasts, which degrade bone. According to Stanley Hefta, director of proteomics at BMS, the goal of the project was to identify intermediates in signal transduction pathways that are involved in osteoclast differentiation. After pulling down protein complexes with an anti-phospho-tyrosine antibody, digesting the proteins, enriching for phospho-peptides, and analyzing them by LC/LC-MS/MS, the researchers found 320 proteins that were unique to the stimulated cells. These included specific kinases and their putative substrates. Further insight came from combining these results with gene expression profiles and from using advanced bioinformatics analysis, Hefta said.
To validate the significance of a specific kinase, the scientists tested its ability to phosphorylate substrates in vitro and in cell extracts. They also constructed a constitutively active kinase: transgenic mice carrying this modified kinase showed a decrease in bone density. Likewise, the bone density of mice in which the kinase was knocked out increased. The next step, creating and testing specific inhibitors, is on its way, Hefta said.
Combining transcriptional and protein profiling also proved successful in the discovery of biomarkers. Using a human hepatocyte cell line as a model system for hepatotoxicity, BMS scientists tested the effect of compounds on gene expression and on proteins found in the supernatant medium.
Correlating gene and protein expression levels with toxicity will allow BMS scientists to integrate this data into clinical trials shortly, Hefta said. Markers for efficacy, he added, will eventually help determine the right dose for a new drug in clinical trials.
Continuing the theme of biomarkers, Steven Carr, senior director for discovery technologies at Millennium Pharmaceuticals, presented the results from a two-year-old collaboration with Roche Diagnostics to develop prognostic markers for an aggressive form of rheumatoid arthritis. This project involved analyzing pre-fractionated synovial fluid samples from patients and controls by LC/LC-MS/MS, validating candidate markers in serum, and developing antibody-based tests for larger validation studies. According to Carr, the mass spec operation was the bottleneck of this study. The researchers identified more than 650 unique proteins in synovial fluid samples, of which 30 were candidate markers, including known markers for inflammation. They were able to quantify six of these proteins in the serum of patients and healthy controls, using isotopically labeled synthetic peptides as standards, and found that three proteins were specific for the aggressive form of the disease.
According to Carr, a remaining technical challenge is the (lack of) depth of coverage of proteins, but still “there is a fair amount of low-hanging fruit.” Moreover, validating biomarkers is a slow process, he said, since it takes about six months to develop an ELISA-type assay.
Scott Patterson, vice president for proteomics at Celera, gave an example of how his company uses proteomics to find new cell-surface targets for therapeutic antibodies. His group has been comparing the expression of cell surface proteins on a pancreatic cancer cell line and a control cell line. After modifying the surface proteins, the scientists were able to isolate and digest them and measure peptides expressed differentially in two samples using the ICAT approach. While the vast majority of peptides showed no difference, the researchers zoomed in on those that did, and identified them by MS/MS. These differentially expressed proteins then need to be confirmed by other techniques, Patterson said, and their functional significance must be validated, for example by generating inhibitory antibodies.
If any of the conference participants thought at this point that drug companies had shunned 2D gel electrophoresis completely, Johannes Voshol from Novartis proved them wrong. His group analyzed the effect of a natural anti-tumor compound on protein expression in a tumor cell line using 2D gels. One protein, a potential biomarker, was shown to be acetylated only in the untreated cell line. But the study also revealed the target of the compound, an enzyme. A recombinant form of this enzyme was inhibited by the compound, which is now being used as a lead to develop an anti-cancer drug.
Novartis researchers have also been using 2D gel electrophoresis to study differences in protein expression in postmortem brain samples from schizophrenic patients and healthy individuals. Validation is difficult, Voshol said, since no cell-line based model system exists for schizophrenia. The main technical limitation for 2D gels, he said, is currently the automated image pattern analysis. Also, he said, “proteomics is slow,” and results cannot be expected overnight.