By Tony Fong
While isobaric labeling methods have become a popular strategy for quantitative proteomics, they have limitations when used on ion trap mass spectrometers and hybrid mass specs such as the LTQ Orbitrap.
Researchers in Austria, however, have developed a method that they say provides sensitive and accurate protein identification as well as isobaric-labeling-based quantitation, specifically iTRAQ labeling, on such instruments. In a study describing the method, the researchers add that in a comparative analysis, their method outperformed two recently developed strategies for use of iTRAQ on Orbitraps.
The study was published Aug. 11 in the online edition of the Journal of Proteome Research. In it, the authors write that as mass spec-based proteomics shifts its attention away from identification to relative and absolute quantification of proteins, several techniques have been developed to allow such work to be done.
Among the "most elegant and precise" methods for quantitation are those based on isotopic labeling, in which proteins are quantified based on the ratios of the intensities of differentially labeled but chemically identical proteolytic peptides. In recent years, the authors write, iTRAQ — which stands for isobaric tags for relative and absolute quantification — has become an especially popular method of isobaric-labeling-based quantitation because the technology can be used for all protein sources for which metabolic labeling is not possible, such as human protein samples.
Applied Biosystems, a division of Life Technologies, developed, manufactures, and sells iTRAQ reagents.
The iTRAQ method is based on the collision-induced dissociation method of fragmentation, and while the technology works well for a broad range of mass specs, iTRAQ-based quantitation cannot be done on ion traps or hybrid instruments containing ion traps for fragmentation such as Orbitraps, manufactured by Thermo Fisher Scientific, the authors of the JPR study wrote.
This limitation is particularly troublesome, they added, because Orbitraps have become the "instrumentation of choice" for many proteomics labs in the last two years.
iTRAQ-based quantitation under standard CID conditions "is not feasible [on ion traps and Orbitraps] because of their low mass cut-off limitation, prohibiting the analysis of product ions with [mass-to-charge] values less than 25 to 30 percent of the precursor ion," they said.
Last year, Thermo Fisher in-licensed an isobaric mass tagging technology called TMT, which competes with iTRAQ. Thomas Köcher, a researcher at the Research Institute of Molecular Pathology and the first author on the JPR study, said via e-mail that the limitations of iTRAQ on the Orbitrap also apply to TMT. With both TMT and iTRAQ generated reporter ions have "very low mass" and the "usual strategy" on the Orbitrap is to fragment the peptide ions in the linear ion trap part of the instrument and then measured.
"However, for physical limitations (so-called low mass cut-off value), these [reporter ions from both iTRAQ and TMT] cannot be analyzed," Köcher said. "The method we described is just a method [to increase] the capabilities of the LTQ-Orbitrap to quantify peptides labeled with either TMT or iTRAQ."
For now, at least, iTRAQ is the market leader in isobaric mass tag technology and because "mass spec people tend to be rather conservative" TMT will be a "serious competitor" to iTRAQ only if it is "much cheaper," Köcher said. "You can do some experiments with TMT you cannot do with iTRAQ, but most people will not need these possibilities."
In an e-mail Andreas Huhmer, director of proteomics marketing for Thermo Fisher, said that hybrid mass spectrometry "has made it possible to combine the advantages of different detectors and fragmentation techniques regardless of the mass tag chemistry," and added that the ideas presented in the JPR study is "novel and extremely useful for a broader audience."
Best of Both Worlds
To address the limitations of isobaric mass tagging on Orbitraps, two technologies were recently developed that allow analysis of the low m/z region of iTRAQ reporter ions: pulsed Q dissociation, or PQD, and higher energy C-trap dissociation, or CID. Both techniques, while allowing for iTRAQ quantitation in an Orbitrap, have limitations, however. PDQ has "poor fragmentation efficiency" and low product ion counts, compared to CID "and is naturally less suited for precise quantitation," the researchers said.
Meanwhile, HCD offers high fragmentation efficiency, but in the Orbitrap the technology is limited by a "lower duty cycle," because the fragment ions generated have to be detected in the Orbitrap part of the instrument, rather than in the linear ion trap.
[ pagebreak ]
Recent studies, they added, suggest PQD is a better technique for protein identification and quantitation than HCD "with respect to the achievable sensitivity." And while HCD and PQD can generate b- and y-ions for peptide identification, they both suffer in terms of sensitivity when compared to CID performed and detected in the linear ion trap and Orbitrap, according to the researchers.
To overcome these limitations, the researchers combine the strengths of CID with HCD and come up with a method they call CID-HCD, an analytical strategy that consists of alternating CID and HCD MS/MS experiments.
In brief, for each precursor ion a CID spectrum is generated and recorded in the linear ion trap. An HCD spectrum is then generated from the same peptide ion. The CID spectrum is used for peptide identification, and the HCD spectrum is used "solely for the reporter ion-based quantification," the researchers said.
They also created a software tool for merging and processing the CID-HCD dataset. The iTRAQ reporter ions are extracted from each HCD spectrum and the intensity levels are normalized to one. And "if present, the tool removes the reporter ion-specific m/z region from the corresponding CID spectra." Finally, the extracted values of the four iTRAQ channels are merged with the respective CID data.
To test their method, the researchers compared it with PQD and HCD in terms of detection and quantification limits by investigating the minimum amount of tryptically digested bovine serum albumin needed in order to identify proteins with at least two matching peptides with a Mascot score greater than 25 and at least one of the peptides containing all four iTRAQ reporter ions.
For both PQD and CID-HCD, the limit of quantification was determined to 100 amole iTRAQ-labeled bovine serum albumin in a 100:100:200:200 amol peptide mixture. With similar time demands across all three methods, PQD outperformed HCD "in all aspects," the researchers said. But CID-HCD resulted in "significantly better figures of merit than with PQD," according to the researchers.
Next, an analysis of all three methods was done using a complex protein mixture generated from mouse hearts after either transaortic constriction or sham surgery. "Data interpretation of the three triplicates clearly showed that CID-HCD outclassed the other two methods" in the number of proteins and peptides identified as well as quantified, the researchers said.
Based on 1,033 peptides, 985 with all four reporter ions present, the HCD dataset resulted in the identification of 217 protein groups and the quantification of 168 protein groups, on average. PQD was able to identify 170 protein groups and quantify 108 protein groups, based on an average of 576 peptides identified, 443 of them with all four reporter ions present.
The CID-HCD approach identified an average of 282 proteins and quantified on average 237 proteins, with an average of 1,813 matching peptides, of which 1,728 contained all iTRAQ channels.
In an evaluation of the accuracy of quantitation of the three methods, the authors determined that all three "led to a geometric mean of the complete set of iTRAQ-labeled peptides, which was close to one, the expected value, with similar standard deviations within each set of technical replicates."
In addition, they noted that HCD and CID-HCD had identical geometric standard deviations of about 1.2, "suggesting that the high collisional energy used in the CID-HCD approach did not distort the accuracy of HCD-based quantification."
PDQ achieved a standard deviation of 2.0, "pointing to a much lower precision."
Finally, they used their approach to perform a two-dimensional LC-MS/MS experiment to compare the protein expression in a mouse heart after transverse aortic constriction, an in vivo model of cardiac stress, to a control sample from an animal that received sham treatment.
The combined dataset resulted in the identification of 1,733 protein groups, based on 8,321 unique peptides with a Mascot ion score of 24 and a significance threshold of lower than .01. Of the 8,321 peptides, 7,815 contained quantitative information from all iTRAQ reporter ions, the researchers said.
The 1,733 protein groups were also analyzed with respect to their gene ontology annotation. Based on the GO subcellular localization annotations, 1,315 of the identified proteins originate in the cytoplasm and 816 are localized in membranes. The most common molecular functions of the GO-annotated protein groups were binding activities such as catalytic activity, protein binding, and metal ion binding. Metabolic processes were the most frequent GO-annotated biological process, followed by cell organization and biogenesis, regulation of biological processes, and transport.