While the study of cells at the genomic level is quite complex, it gets even more complex at the protein level. At the post-translational level, modifications such as phosphorylation, acetylation and others that come into play cause changes in both structure and function of proteins. Hence a comprehensive study of both genes and proteins becomes important when trying to understand the cellular basis of disease onset and progression.

An image of EGFR and beta-actin expression in 90 different cancer cell lines detected in a protein lysate array assay. (Source, Wei Zhang).

“Probably 30% of the data obtained at the transcript level does not correlate with the data at the protein level,” says Wei Zhang, Ph.D., director of the M.D. Anderson Cancer Center’s Genomics Core laboratory and an associate professor in the department of pathology at the University of Texas in Houston. “The transcript studies cannot provide all the information about what happens at the protein level and they do not reflect post-translational modifications.” Though his work at the genomics center mostly involves gene expression profiling, studying gene copy number changes and real time PCR validation, Zhang says that his ultimate goal in obtaining these various types of genomic data is to understand cellular function. “We like to call ourselves functional genomics researchers and not just genomic researchers,” he says. “I know many people think of genomics and proteomics as separate disciplines but when you study the functional aspects of cancer biology they are really the same.”

While genomic techniques such as, cDNA microarrays and real-time PCR have matured over the years, proteomic tools are still struggling to meet the high-throughput (HT) demands of researchers. Several labs have tried to emulate the success of DNA microarrays by designing HT protein arrays for proteomics. Zhang’s group has also worked for many years custom designing reverse-phase protein lysate arrays for HT protein profiling. A paper recently published by Zhang and his colleagues in the Journal of Proteome Research (6:2753-2767, 2007) discusses the use of these protein lysate arrays to profile proteins and compare signaling pathways in different types of cancers. “The idea was to make a protein lysate array with 90 different cancer cell lines and then provide it to anyone in the cancer center looking to screen for a specific protein in their cancer model,” says Zhang. The commercially available cell lines used for the study comprised 12 different types of cancer cells including breast, colon, lung, kidney, ovarian, pancreatic, prostate and others.

Creating the protein array involved extracting cell lysates, spotting them on polyvinylidene fluoride (PVDF)-coated glass slides and then probing them with several antibodies. The array consisted of 96 samples, in multiple dilutions, spotted in triplicate. “The power of this technology is that we can process multiple samples at the same time,” says Zhang. “We process each sample by making six two-fold dilutions and then spot triplicates. So for each sample we have 18 data points.” The array design, sample preparation and dilutions are made less labor intensive by the use of robotic liquid handlers and array spotters.

The need for minimal amounts of sample, considerable savings in time and labor and the potential for automation make protein arrays very conducive for large scale quantitative proteomics. The use of arrays also helps normalize the data and increases reproducibility since all samples are examined on one slide rather than on different Western blots. The major limitation however, is the dot-blot nature of the technology which demands that only high quality antibodies with little or no non-specific binding be used. “Those antibodies that may work on Western Blots may not work with these arrays,” says Zhang.

“What I envision is to use this technology to evaluate how signaling pathways respond to specific treatments and to look at cross-talks between different pathways,” says Zhang. For instance, the recently published study found that epidermal growth factor receptor (EGFR) was the most heterogeneously expressed protein among the various cancer cell lines. “50% of breast cancer cell lines expressed EGFR while the others did not,” says Zhang. “So you can clearly see different patterns of major signaling pathways.”

Zhang does point out that the technique is only powerful when many samples are processed together. “Only when you process multiple samples can you gain an idea about the tumor heterogeneity, something that you cannot appreciate when you have limited samples. That’s when you can look at the big picture.”

Wei Zhang