by Gina Shaw

Somewhere, deep within a complex cell extract sample, sit the potential biomarker proteins Dr. Pierre Thibault is hoping to profile and characterize. The barrier to finding them? Everything else in the sample. Thibault, principal investigator and director of the proteomics facility at the Institute for Research in Immunology and Cancer at the University of Montreal, is particularly focused on characterizing low-level proteins - exactly the kind of target that is often missed with the traditional "shotgun proteomics" approach to protein characterization.

"Any sort of protein extract coming from different biomarker studies will contain a diverse population of proteins, a diversity that is naturally selected by the number of different entities, but also by the dynamic range and protein abundance," says Thibault. "So the question becomes, how do you decide what is statistically significant? Where should you focus your attention? What protein is actually representing the change in your sample?"

Data-dependent MS/MS acquisition is one way to answer those questions, though it can identify only a fraction of the proteins present in the sample. Usually, and confoundingly, these are proteins of higher abundance. But abundance does not necessarily mean significance with regard to identifying disease biomarkers.

The success of the profiling approach relies on the reproducibility of retention time, accurate mass and ion abundance to track molecular features across data sets. Retention time reproducibility for the HPLC-chip system was around 0.5% RSD. A comparison of ion profiles across LC-TOF replicate analyses enabled the reproducible tracking of over 1500 unique peptide ion clusters. These analyses revealed that 95% of reproducibly detected peptide ions across replicate injections showed less than 35% variation in intensity across three orders of magnitude over 3 days.

"Typically less than 5% of your sample will be of interest, so if you're doing a comprehensive approach, you're spending 95% of your time identifying things that are of no significant value to you," Thibault says. "We envisioned, instead, a sort of profiling approach. Using 2-D liquid chromatography with mass spec, and separating tens of thousands of peptide peaks from their sample, we wondered if we could actually identify those peptides that change in abundance, using them as surrogates of the original proteins."

Using Agilent's HPLC chip/6210 TOF LC/MS system, he was able to do just that. His TOF profiling strategy identified the proteins that changed in abundance - for example, in monocyte cells stimulated with tumor-promoting agents. "As in most cases, the majority of proteins do not change, but we were able to find both those that were up- and down-regulated," Thibault explains.

Once Thibault's team has a list of the differentially-expressed proteins, they're targeted for MS/MS identification using an ion trap mass spectrometer. "The HPLC-Chip system is totally transferable between our ESI-TOF MS and our ion trap MS, so we can use essentially the same methods and HPLC separations for profiling and identification," he says. "It's a reproducible, consistent platform that allows us to mine complex samples and get reliable results. It's also advantageous in that it allows you to re-interrogate the sample at a later time should you not have sufficient data for confirmation."

Of course, this process isn't as simple as that summary makes it sound. For starters, it requires finely-calibrated software that has high sensitivity and specificity for efficient peptide detection. That software was at least three years in the making in Thibault's lab. "That was the first step. You have to have a good dynamic range of detection so that you understand the error rate, which can be due to sample complexity or sample abundance."

And while the fine-tuned profile-directed approach is a cost-effective way of identifying candidate proteins that change in abundance across conditions, Thibault isn't completely satisfied yet. "This approach puts a lot of pressure on our capability to obtain more peak capacity and resolution power in all separations. I could envision another platform, where the fractionation process could be offline."

Ultimately, he says, the future of protein biomarker discovery lies in the capacity to see more from the sample, something that requires additional dimensions of separation. "That, in turn, requires that you improve your peak capacity somehow. The ability to combine dimensions of separation - that can be done in numerous ways, and these other aspects are currently in development by manufacturers."