By Tony Fong
A team of German and Swiss researchers led by Ruedi Aebersold has devised a new method for developing single-reaction monitoring assays that they said eliminates a significant bottleneck associated with the process.
In a study published Dec. 7 in the online edition of Nature Methods, the researchers also demonstrated the use of their method against all known yeast kinases and phosphatases.
In the study, Aebersold and his colleagues said that SRM has several attractive features, including its high sensitivity — capable of detecting at the low attomole level — a broad dynamic range of up to five orders of magnitude, and quantitative capability.
According to the authors, existing methods for developing a high-quality SRM assay for a protein is a tedious process. First, the assay has to be validated to ensure that it selectively monitors the analyte of interest. It also must be optimized, which calls for determining the most suitable SRM transitions for each target peptide and other associated LC-MS parameters — "a lengthy and iterative process," the researchers said.
Assay validation typically relies on acquiring full-scan MS/MS spectra for the targeted peptide on the same mass spec that will be used to deploy the assay. However, "complex" backgrounds that obscure the fragmentation patterns and limit the dynamic range compromises the acquisition of reliable MS/MS spectra of peptides, "thus making the validation of transitions for low-abundance peptides extremely challenging," the authors said.
In total, SRM assays have to be developed and validated "by a method that has a substantially lower sensitivity and dynamic range than the SRM assay itself, which has prevented the routine development of SRM assays for low-abundance proteins," they added.
The method described in Nature Methods overcomes these limitations, its developers said. The technique is based on using low-cost libraries of crude, unpurified synthetic peptides as a reference for validating and optimizing SRM assays, and on a mass spec method for generating the assays at a throughput of more than 100 per hour.
The method consists of six steps: First, using data from existing proteomics databases or bioinformatics prediction, a set of proteotypic peptides is selected for each target protein, followed by synthesis of the selected peptides by spot synthesis on a microscale. The peptides are recovered from the synthesis support in a crude, unpurified form.
Pools of about 100 such synthesis products are analyzed by an SRM-triggered MS/MS method, "whereby the detection of any of a few anticipated transitions for each peptide triggers the acquisition of a full MS/MS spectrum for the target peptide."
Validation of the assays and extraction of the "most favorable" SRM coordinates for each peptide are performed by using "consensus MS/MS spectra in which multiple spectra per peptide are acquired," the authors said.
Next, the sensitivity of the SRM assays can be optionally increased by optimizing the transition- or peptide-specific mass spec parameters in an additional MS/SRM run, where the top transitions are measured at different parameter steps.
Lastly, the optimized and validated assays can be used to detect and quantify the target proteins in a biological sample and entered into a publicly accessible repository for SRM assays. "The elution times observed for the synthetic products can be used to schedule acquisition of the SRM traces, thus drastically increasing the number of measurements per analysis in a biological sample," the researchers said.
To test whether their method could generate validated SRM transitions using crude synthetic peptide preparations at high throughput, they randomly chose 125 S. cerevisiae proteins and synthesized 480 proteotypic peptide sequences.
[ pagebreak ]
Peptide aliquots were combined to create five samples, each containing 96 to 140 peptides. The samples were analyzed on a triple-quadrupole mass spec, "using predicted SRM transitions to trigger acquisition of full MS/MS spectra for doubly and triply charged form of each peptide," they said.
They developed SRM assays for 432 of the 480 peptides in less than six hours of instrument time. A success rate of peptide identification of 89 percent on average was achieved, with a false positive error rate of 1 percent. Saturating the capabilities of the acquisition software by targeting 150 peptides per run lowered the success rate to 83 percent, which the authors said was not a substantial decrease.
Proteotypic peptides selected based on empirical evidence or by bioinformatics prediction resulted in success rates of 86 percent and 91 percent, respectively. Overall, the team of researchers was able to develop SRM assays with at least one proteotypic peptide per protein for 124 of the 125 targets.
They then demonstrated the capability of the method by applying it on yeast kinases and phosphatases. Using a protein set of all known kinases and phosphatases as well as hypothetical proteins with putative kinase and phosphatase activity — a total of 156 proteins — they "unambiguously" detected 84 kinases and 26 phosphatases, 71 percent of the target protein set, using their SRM assays.
Included were previously undetected proteins with abundances of about 20,000 to 112 copies per cell. No bias toward medium-high abundance proteins was observed, they reported, and by performing off-gel electrophoresis fractions of the same sample, an additional six kinases and phosphatases were detected.
"The kinase and phosphatase coverage achieved by SRM using an unfractionated yeast total proteome digest was almost as high as that obtained in a recent yeast proteomic analysis based on a large-scale sample fractionation," the authors said.
They hypothesized that some of the target proteins were not detected due to absence of the protein in the sample; "low abundance and selection of proteotypic peptides with suboptimal MS signal response;" or modification of the targeted protein in the selected proteotypic peptide.
Their technique "eliminates a substantial bottleneck in SRM assay development, the generation of full-scan MS/MS spectra from low-abundance peptides in a" triple-quad, Aebersold and his colleagues said. At $5 to $15 for each crude peptide generated, their method is considerably cheaper than methods that depend on classical peptide synthesis of purified proteins, about $500 each. And crude peptide libraries can be generated at a "very high rate" of greater than 50,000 peptides per month.
"This opens exciting possibilities in biotechnological, biomedical, pharmaceutical, and biological applications, and makes the quantitative analysis of a whole proteome by SRM a concrete possibility," they said.