Abstract and Introduction
New mass spectrometry (MS) methods, collectively known as data independent analysis and hyper reaction monitoring, have recently emerged. These methods hold promises to address the shortcomings of data-dependent analysis and selected reaction monitoring (SRM) employed in shotgun and targeted proteomics, respectively. They allow MS analyses of all species in a complex sample indiscriminately, or permit SRM-like experiments conducted with full high-resolution product ion spectra, potentially leading to higher sequence coverage or analytical selectivity. These methods include MSE, all-ion fragmentation, Fourier transform-all reaction monitoring, SWATH Acquisition, multiplexed MS/MS, pseudo-SRM (pSRM) and parallel reaction monitoring (PRM). In this review, the strengths and pitfalls of these methods are discussed and illustrated with examples. In essence, the suitability of the use of each method is contingent on the biological questions posed. Although these methods do not fundamentally change the shape of proteomics, they are useful additional tools that should expedite biological discoveries.
The analysis of peptides generated by proteolytic digestion of proteins, known as bottom-up proteomics, serves as the basis for many of the protein research undertaken by mass spectrometry (MS) laboratories today. In bottom-up proteomics, three different approaches are commonly used: discovery-based approach (or shotgun proteomics); directed approach; and targeted approach (or targeted proteomics).
Discovery-based or shotgun proteomics employs data-dependent acquisition (DDA). Herein, a hybrid mass spectrometer first performs a survey scan, from which the peptide ions with the intensity above a predefined threshold value, are stochastically selected, isolated and sequenced by product ion scanning. In selecting the precursor ions, there is a preference toward the ions having the highest ion intensity. Other additional selection criteria, such as dynamic exclusion, background subtraction, charge state selection, etc. are also used to prevent redundant acquisition of the most abundant peptides, or to avoid acquiring product ion spectra of the interferences. Recently, precursor ion selection is also used to determine the most appropriate fragmentation techniques that are accessible on the same instrumental platform.
In direct approach, besides the signal intensity of the ions, certain characteristic fragmented ions produced are chosen as prerequisites, to trigger product ion scanning. This approach is generally executed using either precursor ion scan or neutral loss scan. Phosphorylated serine and tyrosine containing peptides, and acetylated peptides are examples of molecules that are often monitored by such kind of approaches.[3–5]
In targeted proteomics, selected reaction monitoring (SRM), also known as multiple reaction monitoring (MRM), is used to monitor a number of selected precursor-fragment transitions of the targeted peptides. The selection of the SRM transitions is normally calculated on the basis of the data acquired previously by product ion scanning, repository data in the public databases or based on a series of empirical rules predicting the enzymatic cleavage sites. The latter is referred as in silico digestion. The signal of the SRM transitions can further initiate a product ion scan when the SRM experiment is performed on hybrid quadrupole linear ion trap (Q-TRAP) or other fast scanning tandem quadrupole instruments to provide a confirmatory product ion spectrum. This method was first reported by Cox et al. and Uwin et al. and was termed by the authors MRM-initiated detection and sequencing (MIDAS).
In the past decade, most of the MS-based proteomic studies were carried out using shotgun proteomics to maximize the amount of information acquiring in an experiment. However, it is now apparent that DDA has a number of limitations including instrumental scanning speed, stochastic selection of ions for fragmentation and poor repeatability, a relatively narrow dynamic range and the issues of chimericy (co-fragmentation of two or more ions),[12,13] etc. Furthermore, the number of peptides presented in a biological digest may be many times larger than the number of ions that can be sequenced in an experiment despite the recent advances in instrumentation. Consequently, and as shown by the work of Michalski et al., most of the information would still be inaccessible (referred as under-sampling) even though the experiment was performed on a LTQ-Orbitrap Velos mass spectrometer. Due to the bias nature of DDA for the most abundant species, low abundance peptides are unlikely being sequenced in a complex biological sample. Similarly, closely eluted isobaric species, such as phosphorylated peptides that differ only in the sites of phosphorylation, may not be sequenced completely owing to the typical dynamic exclusion setting used in DDA. On the other hand, precursor ion scan and neutral loss scan have a limited applicability and they are normally used as complimentary methods.
Recently, there has been a renaissance of targeted proteomics using SRM method. It is because SRM offers several advantages, such as specificity, reproducibility, sensitivity, linearity and it ideally suits for quantitative analysis. Furthermore, SRM-based quantification can be coupled to different strategies for relative (differential) or absolute protein quantification. One of the absolute approaches is termed absolute quantification (AQUA) of proteins. In this strategy, synthetic peptides incorporated stable isotopes are spiked to a protein digest as internal standards to mimic the native peptides formed by proteolysis. This method allows accurate quantification of a network of proteins in a biological system. A similar method, stable isotope dilution (SID), is routinely being used in quantification of pharmaceutical compounds and small molecules. An additional advantage of SRM is that the experiment can be conducted on relatively low-cost triple quadrupole-type instruments.
Despite the advantages, targeted proteomics has not been the preferred method by many proteomic researchers. Given that it is a targeted approach, a prior knowledge of the targeted proteins in the sample is a requisite. Arguably, up to 6000 transitions can be monitored by an SRM experiment using triggered or intelligent selected reaction monitoring (iSRM) on the latest triple quadrupole mass spectrometer (e.g., TSQ Vantage). However, only a relatively small number of proteins (up to 100) are monitored by a typical SRM experiment in practise. The method also requires lengthy and labor-intensive development and optimization process. Compared to SRM, other tag-based quantitative proteomic methods, such as isotope-coded affinity tags (ICATs),[19,20] isobaric tags for relative and absolute quantification (iTRAQ),[21–23] tandem mass tags (TMT),[24,25] isotope-coded protein label (ICPL),[26,27] etc. offer a much greater flexibility (in terms of what proteins and the number of proteins being measured in an experiment), whereas SILAC[28,29] offers good reproducibility and accuracy.
Furthermore, in SRM, the detection of a chromatographic peak, even with all the predicted SRM transitions detected, does not confirm the identity of the peptide. This is because the mass of interfering ions could fall within the tolerance of both quadrupoles and leads to a false positive identification. The problem of interfering or isobaric ions can be alleviated with the use of differential mobility separation (e.g., DMS or FAIMS) or MRM3 (Box 1). Additionally, both of these techniques lowers the lower detection limit and increases the dynamic range. However, since they either lead to ion neutralization or increase the length of duty cycle, one may have to either reduce the number of analytes that can be measured concomitantly in a complex sample or to have a lower absolute sensitivity of the measurement.[32,33] On the other hand, even if sequence information is acquired using the MIDAS approach, the product ion spectrum is usually of low quality (low mass resolution/accuracy or high interferences) for confident assignment of the peptide identity. The data generated will have to be validated with relatively expensive reference or isotopically labeled peptide standards. In short, and as shown by the Aebersold lab, it is paradoxical that highly sensitive 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.
However, the situation for targeted proteomics might be changing. A major driving force is that MS-based proteomics is in a transitional phase from being largely a discovery-based analysis to emphasizing more on hypothesis driven analysis. Hence there is an impeccable need for high analytical precision, accuracy and wide dynamic range of targeted proteomics. Much of the biomarker candidates identified by shotgun proteomics in the past were not being followed-up or validated. Our laboratory (and many other) performs validation and characterization of these candidates with biochemical techniques, such as western blotting. However, the required antibodies are often unavailable. Many commercially available antibodies do not work effectively, and the associated cost is normally very high. Should there be a large number of potential biomarkers, multiplexed targeted proteomic methods would be more time and cost efficient than biochemical investigations to rationale the biomarker candidates identified in the discovery-based approaches. But the restraints and difficulties associated with SRM have not made the objectives of targeted proteomics a reality and this has fuelled the recent advances of MS methods and instrumentation that permit data-independent analysis (DIA) or hyper reaction monitoring (HRM) on high-resolution and accurate-mass hybrid systems. This potentially allows one to obtain sequence information for peptide identification while the SRM-like ion chromatographs can still be subsequently extracted or reconstructed from the raw data. This does not only allow comprehensive qualitative assay of the species in the sample but the method can also provide quantitative information.
In this review, a number of selected DIA and HRM methods reported, to date, are introduced and discussed with respect to how these new MS approaches could be used in to advanced protein biomarker discovery and validation. Their advantages, limitations and potential impacts on proteomics are being considered.
Expert Rev Proteomics © 2013 Expert Reviews Ltd.