Last week, the proteomics crowd flocked to sunny San Diego for some new-year PepTalk — at Cambridge Health Institute's second annual Pep Talk — The Protein Information Week conference. During the first two days of the meeting, about 600 participants and speakers listened to talks on protein array platforms, protein informatics solutions, and protein expression strategies. Following is a cross-section of the presentations.
A number of talks in the protein informatics track dealt with the blooming field of multiplexed protein disease markers. Arguably, the area received a push last year with the publication on ovarian cancer by Chip Petricoin and Lance Liotta in The Lancet.
Describing his group's approach to generating serum protein patterns in cancer, Petricoin, co-director of the FDA-NCI clinical proteomics program, remarked that the fact they are not identifying the proteins in the pattern generated has been “completely anxiety causing” to both researchers and the FDA. Since that study, Petricoin's group has switched from using a Ciphergen SELDI-TOF instrument to an ABI Q-Star, which he said offers higher resolution and mass accuracy, but has retained the SELDI protein chips. However, the group is also collaborating with both ABI and Advion Biosciences to try different separation and mass spectrometry platforms. In addition to its work on ovarian cancer, the group has been creating serum protein pattern-based diagnostic models — using Correlogic's genetic algorithm — for breast cancer, lung cancer, pancreatic cancer, and prostate cancer. Moreover, studies on cardiovascular and inflammatory diseases are planned or ongoing. A clinical trial for ovarian cancer detection will start in about three months at an NCI proteomic pattern diagnostic reference lab that was founded in November, Petricoin said. This trial, which will monitor disease recurrence, will involve about 1,000 samples and last 6-12 months, Petricoin told ProteoMonitor. The aim is to obtain FDA approval for a process that is equivalent to or better than the currently used biomarker, which has a high false positive rate. Correlogic would license this test to reference labs. The mass spectrom -etry data of this trial will become publicly available on an ongoing basis, Petricon said.
Daniel Chan, professor at Johns Hopkins Medical Institutions, has also been looking for SELDI-generated mass peaks that can serve as disease biomarkers — for example in breast cancer. In his presentation, he described a software package developed by his group to compute the contribution of each peak to the separation of two diagnostic groups. He pointed out potential sources of bias, for example, biological variation in subgroups, study protocols, or instrumentation.
Protein Chips Still In FLUX
Conference presentations on protein microarrays showed that the field is still in flux. No dominant format has established itself, judging by the plethora of platforms — including substrates, content, and detection methods — presented in the array track.
Andrew Bradbury, staff scientist at Los Alamos National Laboratory, gave one of the few talks that dealt with creating new binding agents — a pressing need for antibody arrays. Bradbury talked about fluorobodies, antibody mimics which join the likes of affibodies, fibronectin domains, or single-chain antibodies, and use mutated green fluorescent protein as a scaffold in which binding regions from antibodies are inserted. More than half of the resulting proteins — a library of 107 molecules — retain fluorescence and can be used to probe protein arrays without the need for labels.
Michael Fiechtner of Genicon Sciences presented on the use of the company's resonance light scattering particles as a binding detection mechanism in protein arrays. According to the company, this RLS technology promises increased sensitivity and dynamic range as well as archivable results. Genicon already has a detection instrument and a toolkit for DNA microarrays — distributed by Qiagen — on the market. Fiechtner told ProteoMonitor that Genicon is planning to launch a protein microarray toolkit during the first half of this year, probably distributed by a company other than Qiagen. Long term, the company is hoping to work with companies that can provide array content, he said.
An example of how researchers might integrate the Genicon technology into their arrays was provided by Bernhard Geierstanger, a group leader at the Genomics Institute of the Novartis Research Foundation. His group has printed 48 arrays, each with 144 spots, on a glass slide, separating them spatially by a rubber gasket. Spot morphology is one of the major limitations of their performance, he said, besides the lack of specific antibodies that do not cross-react. A prototype array for 84 human disease markers has achieved a sensitivity of 1 pg/ml, using Genicon's detection method, Geierstanger said.
Microarrays with a bend were at the center of a talk by Arun Majumdar from the nanoengineering laboratory at the University of California, Berkeley. Arrayed microcantilevers, each covered with a probe molecule, bend up or down in response to a conformational change that occurs when a specific target molecule binds. This movement can be measured optically without the need for a label. So far, Majumdar has been able to measure ten cantilevers in parallel, and has achieved a sensitivity of 2 pg/ml. The next step will be to increase the number of cantilevers to 1,000. One application, he said, would be the measurement of tumor markers in serum. Majumdar told ProteoMonitor that he was planning to commercialize the technology by launching a startup company later this year, depending on his ability to raise funding.
James Wang of Hypromatrix discussed dissociable antibody arrays, a label-free array technology that turns conventional antibody arrays upside down. Instead of probing an array of antibodies with a sample, the company probes an immobilized sample with an antibody array, and allows antibodies that have found a binding partner to dissociate off their support onto the sample. The system does not require the sample proteins to be labeled or solubilized. Moreover, it can detect the subcellular localization of proteins.
The protein expression track focused on the bottleneck of many structural proteomics projects: how to express and purify thousands of different biologically active proteins in sufficient quantities for x-ray crystallography or NMR, especially such “problem children” as membrane proteins. “This is a very tough problem, and we don't have a solution for it,” said Tauseef Butt, vice president for research and development at Life Sensors, opening the first morning session.
With a general solution still at some distance, researchers pres-ented strategies that have shown some success. The overarching theme in these approaches: trying as many different combinations of vectors, tags, organisms, strains, growth, and extraction conditions as possible. Almost every speaker used Invitrogen's Gateway system for cloning the gene of interest into a variety of expression vectors bearing different tags.
Jim Hartley from the NCI/SAIC Protein Expression Laboratory showed that four “troublesome” proteins had good and soluble expression in E. coli when the temperature was reduced to 16°C after induction. Proteolysis could be prevented in some cases using shorter induction periods. Proteins with predicted transmembrane domains, however, tended to be poorly expressed.
These talks showed that proteomics is spanning more diverse areas of research.