NEW YORK (GenomeWeb) – A team led by researchers from Johannes-Gutenberg University Mainz has completed an in-depth proteomic profile of the mouse hippocampus postsynaptic density.
Presented in a paper published last month in Proteomics, the study identified a set of 879 high-confidence PSD proteins. Moreover, the researchers quantified the proteins present at each step of the PSD purification process, allowing them to better separate out true PSD proteins from contaminants, Stefan Tenzer, a JGU researcher and senior author on the paper, told ProteoMonitor.
The study represents the first published application to a large-scale proteome profiling experiment of the UDMSE technique, a data-independent acquisition mass spec approach developed by Tenzer's lab. Use of the approach, Tenzer said, was key to obtaining the reproducible quantitation across multiple purified fractions required for the study.
Tenzer also noted ongoing work on a new version of the method that will use spectral libraries as opposed to conventional databases for making peptide IDs. This will allow for MS2-based quantitation using the technique, which, he said, could improve its dynamic range.
Published last year in a study in Nature Methods, the UDMSE uses the specific behavior of ions in the upfront ion mobility system of Waters' Synapt G2-S mass spec to improve the fragmentation efficiency of the company's traditional MSE DIA workflow.
Launched in 2006, Waters' MSE workflow was the first commercially available DIA method. Unlike AB Sciex's Swath or DIA approaches being developed for Thermo Fisher instruments that divide samples into m/z-based fragmentation windows, the Waters method uses the co-elution times to match precursor ions to the fragment ions generated, allowing them to be searched against databases to make peptide IDs.
In this approach, all co-eluting ions of both low and high m/z are fragmented in parallel. However, low m/z ions typically fragment best at low collision energies, while high m/z ions fragment best at high collision energies. This means that in order to fragment both types of ions, the instrument must ramp the collision energy over the course of the scan cycle so that it fragments both at low and high collision energy.
Such ramping, however, leaves at every part of the scan a large portion of ions being fragmented inefficiently. And so, to get around this problem, Tenzer and his colleagues used the instrument's IMS technology to establish the approximate m/z of the ions being fragmented at a given time. Small ions travel faster through the IMS than large ions, making it possible to predict when ions of a particular m/z will reach the instrument's collision cell and, therefore, making it possible to adjust the collision energies for optimal fragmentation.
Tenzer and his colleagues also presented in the Nature Methods paper an open source software program, called ISOQuant, for analysis of DIA data generated using the method. The package uses cross-annotation of mass spec ion signals from different runs on the basis of mass, retention time, and drift time to improve the reproducibility of peptide IDs and quantitative data across runs.
In an analysis included in last year's study of UDMSE from three technical replicates, the researchers found that applying the ISOQuant package improved the overlap of proteins identified in all three replicates from 48.5 percent to 97.8 percent.
The high reproducibility of the UDMSE technique was key to enabling the JGU team to determine with high confidence the proteins belonging to the mouse PSD in the recent Proteomics study, Tenzer said. Specifically, it allowed them to reproducibly quantify the proteins present at each stage of the purification process, which let them determine which proteins were being winnowed by the purification.
Past efforts have typically performed mass spec analysis of their samples only at the end of the purification process, Tenzer said, adding that this meant that contaminant proteins that were still detectable – though at diminished levels – at the end of the purification were included as PSD components.
"I think the strength of that project is that we really classified in detail all of the different purification steps and can really say in detail that, for instance, at this step we find these proteins," he said. "We quantify every protein at every purification step, and this allows us to say, 'this is a PSD protein,' because it is retained in all the purification steps."
For instance, "we still see a lot of mitochondrial proteins in the purified fractions," Tenzer said. "But when we look at the abundance profile we clearly see they basically are diminished in the purified fractions even if they are still detectable."
He added that "using the UDMSE DIA approach lets us be highly reproducible in [quantifying] proteins in all of the replicates because we have the same information on all of them."
Tenzer noted that his team is currently working on an alternate version of UDMSE that, like other DIA methods like Swath, will use spectral libraries for making peptide IDs.
"Swath relies on the [QTOF's] quadrupole for picking mass ranges, but actually you could imagine doing something like that on the ion mobility instead," he said. "You could interrogate our data with a spectral library, not using quadrupole selection but using the ion mobility as a filter [to cycle through mass ranges]."
Using spectral libraries would allow the researchers to do UDMSE quantitation on fragment ions instead of precursor ions, Tenzer said, noting that this could improve the method's dynamic range.
"I think the spectral library-based approaches have an advantage with lower-[abundance] signals because they are looking at multiple fragments at one time and not just one ion," he said. "So we could be able to identify more [proteins] and more reliably in the lower orders of magnitude."