Acquisition of protein signatures using SWATH-MS

Protein signatures are becoming increasingly prevalent in hospitals for the routine diagnosis and prognosis of diseases. Protein signatures are quantitative patterns of groups of proteins that distinguish biological samples. For example, the occurrence of prostate cancer is nowadays routinely screened by measuring the concentration of prostate-specific antigen (PSA), and a new multi-protein signature of glycosylated plasma proteins has been recently described that diagnoses and grades prostate cancer more sensitively than PSA alone.(1)

However, the relatively low penetration of the most advanced proteomic technologies into the clinic severely delays the adoption of such successful candidates into clinical tests. Currently, protein quantification in clinical samples is most frequently performed by antibody-based assays such as immunohistochemistry staining techniques or enzyme-linked immunosorbent assay.

Although routine, reliable and sensitive, these antibody-based assays have several shortcomings. The main one is related to the availability of the antibody itself. When a novel protein signature for a disease is discovered in a proteomics study, high-quality antibodies need to be raised against several antigens and reliable assays based on these antibodies need to be established and validated. This process is lengthy, costly and not always successful.

The antibodies might suffer from large batch-to-batch variability or from signal interference that complicate the development and validation procedures. There are also known issues of cross-reactivity when several antibodies are used simultaneously, limiting the possibilities in multiplexing large number of assays. In this respect, some recent developments in liquid chromatography coupled to mass spectrometry (LC-MS/MS) may be well worth considering in a clinical environment as a valuable alternative or complement to antibody-based assays. 

Discovery or shotgun proteomics is the most widely used mass spectrometric method for routine biological sample analysis. The whole protein content of a sample (for example, tissue, biopsy or bodily fluid) is digested by a sequence- specific endopeptidase (most frequently trypsin), yielding a complex peptide sample that is then analysed by LC-MS/MS.(2) The LC-MS/MS analysis consists of separating the peptide mixture resulting from the digestion on a chromatographic column (for example, according to their hydrophobicity) while the MS instrument alternately records the mass-to-charge ratio (m/z) of the eluting peptides (MS spectra; see Box 1), and their corresponding fragment ion spectra (MS/MS spectra)(3) (Table 1). 

The main advantage of this technique is that m/z of the sample peptides match best the mass range observable by current MS instruments.

Also, liquid chromatography offers efficient separation of the peptides and enables direct online analysis with MS instruments using electrospray ionisation. Progress in bioinformatics has also contributed to an increase in the throughput and reliability in peptide identification and quantification by automating fragment ion mass spectra annotations as well as qualitative and quantitative data analysis overall. On most current MS instruments, thousands of peptides, and therefore proteins, can typically be identified in a single LC-MS/MS analysis. However, even though this number significantly exceeds the number of proteins typically quantified by antibody-based assays, this powerful mass spectrometric technique has enjoyed very slow adoption in clinical research.

The main reason for this is the perceived data consistency achieved on complex peptides samples, which arises from the stochastic sampling of the peptides during typical shotgun LC-MS/MS analyses (Table 1). Discovery proteomics remains the best-established method to rapidly identify large number of peptides and proteins in a sample. However, in cases when the requirements of a research project call for the reproducible and accurate quantification of a pre-determined set of proteins across multiple samples, targeted LC-MS/MS proteomic methods, exemplified by multiple or selected reaction monitoring mass spectrometry (MRM or SRM MS), have been increasingly and successfully applied. 

The consistency of the SRM MS approach is achieved by forcing the MS instrument to monitor a pre-defined set of peptides regardless of their abundance in a sample (Table 1). In practice, the instrument sequentially records the signals for a pre-programmed series of fragment ions for each targeted peptide precursor of interest during the chromatographic separation. This MS mode enables the concomitant detection and quantification of specific peptides in complex mixtures with exquisite sensitivity and dynamic range.(4) Several recent cross-validation studies have shown the robustness of the SRM-MS methodology with coefficient of variation of <20% across multiple laboratories,(5,6) demonstrating that SRM could be the MS method of choice in the clinics for robust and high-throughput disease biomarker monitoring. However, despite the advances that SRM promises, the method presents a few shortcomings.

Because the MS instrument records only fragment ion signals that were pre-programmed for that sample injection, the quantification of additional peptides that were not included in the original list necessarily requires re-programming the acquisition method and re-injecting the sample. Also, because even on the fastest available instruments SRM can quantify at most a few hundred peptides, this method lacks the speed and the dynamics that would be required to large numbers of proteins across large numbers of samples.

To overcome the limitations of shotgun and SRM-based targeted proteomic strategies, we have recently introduced a technique – SWATH-MS – that combines data independent MS/MS acquisition and in silico-targeted analysis of the thus acquired datasets.(7) In practice, this technique consists of two sequential steps: data acquisition and data analysis. The first step, the data acquisition, consists of generating complete and time-resolved MS/MS records for all the species detectable in a sample. To achieve this, a highly sensitive fast scanning MS instrument acquires fragment ion spectra by cycling sequentially through the mass range of interest (for example, 400–1200 m/z) in 32 discrete ‘swaths’ of 25 mass units (Table 1).(8)

This ensures that the fragment ion signals of any peptide present in the sample are recorded in a time-resolved fashion in the data. The second step, targeted data analysis, consists of mining the generated data by extracting, a posteriori, the fragment ion signals for any peptides of interest. In other words, this step virtually transposes, at the data analysis level, the concepts of the targeted data acquisition of the same fragment ion signals, which underlies the consistency and reproducibility of the SRM measurements across sample cohorts. In addition, because of the completeness of the fragment ion data acquired, the method offers almost unlimited possibilities to perform in silico data re-analysis, ranging from dynamic quantification refinement to the expansion of the list of peptides to quantify, depending on prior results. 

In its current implementation, and with its promising future developments, this platform seems capable of addressing the current limitations and questions in the field of high-throughput studies of biomarkers and may prove a proteomic platform of choice for the clinics when considering the simplicity, robustness, consistency and completeness of the MS data acquisition.

To exemplify the potential of SWATH-MS in clinical biomarker studies, we recently reported how the method can be used to perform quantitative measurement of the N-linked glycoproteins in human plasma.(9) Blood plasma is the body fluid of choice for the detection of disease biomarker candidates. However, the high complexity and dynamic range of total plasma proteome make those samples particularly challenging for sensitive biomarker profiling using traditional proteomic techniques. The focus on N-glycosites is based on the fact that most cell surface proteins or proteins secreted/shed from cells are glycosylated and that nearly 80% of the currently used biomarkers in clinics are known to be glycosylated.(10) 

Using dilution series of isotopically labelled peptides representing biomarker candidates, we determined a limit of quantification of 45.6 amol for the SWATH-MS targeted data analysis methodology, after plasma glycopeptides enrichment by solid-phase chemical immobilisation. This corresponds to a protein concentration of 5–10ng/ml in plasma, which places the method in the sensitivity range ready to measure a large set of biomarkers. The quantification of endogenous glycoprotein biomarkers using SWATH-MS also showed a high degree of reproducibility, with a mean coefficient of variation (CV) of 14.90%, and correlated well with SRM results (R2=0.9784). Overall, those results suggest that SWATH-MS targeted data analysis, when combined with N-glycoproteome enrichment from plasma, provides a reproducible, deep and quantitative glycoproteomic reference of human plasma.

It can therefore be seen as a promising integrative proteomic approach for biomarker discovery and a proof of concept for other studies in the field of clinical proteomics. Because SWATH-MS permits the concurrent quantification of a significantly higher number of glycoproteins than SRM, future studies will focus on further applications of this approach, by the extraction of the most complete plasma glycoproteome possible across multiple human samples to address biologically and clinically important questions. Also, because the same SWATH-MS datasets can be easily re-examined for the validation of novel emerging biomarkers, we foresee this technology as a prominent method for future biomarker discovery studies.

In conclusion, the SWATH-MS targeted data extraction methodology appears to be a promising and rational proteomic platform to implement in the clinics. It offers essentially the same consistency and reproducibility as SRM MS, without the hurdles of the MS method pre-programming or the limited number of peptides monitored. Also, the completeness of the fragment ion data generated by SWATH-MS appears as an invaluable clinical resource to mine for biomarker candidates a posteriori, even for those discovered in the future.

1. Cima I et al. Cancer genetics-guided discovery of serum biomarker signatures for diagnosis and prognosis of prostate cancer. Proc Natl Acad Sci USA 2011;108(8):3342–7.
2. Aebersold R, Mann M. Mass spectrometry-based proteomics. Nature 2003;422(6928):198-207.
3. Steen H, Mann M. The ABCs (and XYZs) of peptide sequencing. Nat Rev Mol Cell Biol 2004;5(9):699–711.
4. Huttenhain R et al. Reproducible quantification of cancer-associated proteins in body fluids using targeted proteomics. Sci Transl Med 2012;4(142):142ra94.
5. Addona TA et al. Multi-site assessment of the precision and reproducibility of multiple reaction monitoring-based measurements of proteins in plasma. Nat Biotechnol 2009;27(7):633–41.
6. Prakash A et al. Platform for establishing interlaboratory reproducibility of selected reaction monitoring-based mass spectrometry peptide assays. J Proteome Res 2010;9(12):6678–88.
7. Gillet LC et al. Targeted data extraction of the MS/MS spectra generated by data-independent acquisition: a new concept for consistent and accurate proteome analysis. Mol Cell Proteomics 2012;11(6):O111.016717.
8. Venable JD et al. Automated approach for quantitative analysis of complex peptide mixtures from tandem mass spectra. Nat Methods 2004;1(1):39–45.
9. Yansheng Liu et al. Quantitative measurements of N-linked glycoproteins in human plasma by SWATH-MS. Proteomics 2013; Jan 16 (Epub ahead of print).
10. Schiess R, Wollscheid B, Aebersold R. Targeted proteomic strategy for clinical biomarker discovery. Mol Oncol 2009;3(1):33–44.