Finding the unexpected in SWATH™ Data Sets – Implications for Protein Quantification Ron Bonner; Stephen Tate; Adam Lau AB SCIEX, 71 Four Valley Drive, Concord, ON, L4K 4V8 Canada RESULTS ABSTRACT Data independent acquisition (DIA) methods such as SWATH-MS™ acquisition allow the analysis of many compounds in complex mixtures and have demonstrated a very high degree of quantitation fidelity compared to MRM [1]. For quantitation the major advantages of SWATH acquisition over MRM are the number of analytes and the significant reduction in method development time, but the unbiased acquisition approach also provides the ability to perform additional quantitative and qualitative experiments. In fact, the “digital sample record” can provide a holistic view of the data, and sample, that is not possible by other methods. Here we describe a prototype qualitative browser and illustrate some of the insights that can be obtained. INTRODUCTION To date the main use of SWATH-MS™ acquisition is for reproducibly quantifying known compounds in complex mixtures. Chromatograms of known fragments are extracted from the Swath (25 amu window) expected to contain the precursor ion and the resulting peak groups (peaks appearing in all or several fragment traces) are scored and filtered in comparison to a decoy model. Scoring is based on the similarity of the fragment ion peaks (shape, retention time), their mass accuracy and similar parameters. Qualitative analysis can extract spectra in an untargeted manner or, as described here, can use a semi-targeted approach with the same scoring approach using either predicted precursor and fragments or by searching for the same fragment pattern in different SWATHs (different precursor mass) and at different retention times. Useful insights and capabilities include: Detecting process induced artifacts (miscleavages, deamidation, Met oxidation, etc.) caused by subtle sample preparation differences between laboratories and between researchers in a single lab that can adversely affect quantitation. Confirming protein identification and providing additional peptide matches that may have been missed. Fragment chromatograms are valuable because of the stochastic nature of shotgun ID and the likelihood of mis-identifications at low levels or from mixed peptides when only a single spectrum is available. Detection of unexpected fragments within the instrument that generate a perceived increase in sample complexity and redundant MS/MS spectra. Such in source fragments may also be fragmented but all have the same elution time. Detection of trypsin-induced artifacts, such as truncation, that resemble in source fragmentation but where the products have different retention times Distributing a peptide across several forms will adversely affect its quantitation hence detection is important particularly as the quantitation of compounds in complex samples becomes increasingly important. Similarly, targeting the correct peptides, and hence validation, is critical. The use of MRM is impossible given the large number of possibilities and the time required, but SWATH provides holistic data that can be used to confirm identifications and search for process artifacts. MATERIALS AND METHODS A series of samples were analyzed using a standard SWATH-MS™ acquisition method with 25 amu windows using a Nano LC introduction system and a AB Sciex 5600 TripleTOF® system. A library of peptide spectra was generated from IDA (shotgun) data using the ProteinPilot TM V4.5 software and the SWATH TM acquisition data were processed by extracting fragment chromatograms of confidently identified species in a research version of the PeakView® software with the SWATH Application which performed automated peak detection, transitions/peptides selection, and area extraction. The SWATH-MS data was also used in a prototype qualitative browser that automates the extraction and scoring of target fragments, from shotgun runs or manually specified, in the expected or all Swaths, and at the expected retention time (where known) or across the entire LC run. CONCLUSIONS 1.Extraction of ions for specific peptides from SWATH™-MS acquisition data allows the identity and reproducibility of modified species to be determined 2.Extraction of low confidence identifications, or undetected peptides, improves protein identification confidence and sequence coverage for quantitation 3.SWATH allows holistic review of the sample and identifies issues which could cause problems for quantification, such as the distribution of a peptide across many unexpected forms, and provides a consistent method for monitoring them REFERENCES 1.Gillet LC et al, Mol Cell Proteomics 2012, O Picotti P et al, Mol Cell Proteomics 2007, 6, 1589 TRADEMARKS/LICENSING For Research Use Only. Not for use in diagnostic procedures. The trademarks mentioned herein are the property of AB Sciex Pte. Ltd. or their respective owners. AB SCIEX ™ is being used under license. © 2013 AB SCIEX. Identification of Ladder Sequences Trypsin is expected to produce peptides with a consistent terminal group. While the notion of hindered and missed cleavage is well known, the idea that cleavage can occur at off target sites is poorly understood, poorly recognised and much debated. These peptides have previously been reported in samples at a higher concentration [2] but, as shown here, we also observe them in less concentrated samples. The browser can find “ladder” sequences (truncated forms) by searching for known peptide fragments across all SWATH™ acquisition experiments. If peak groups are found at different retention times they must correspond to distinct forms. Note, however, that this simple approach requires that either the C- or N-terminus is intact in the truncated form. Extending Sequence Coverage The example above shows the investigation of all predicted peptides from one protein in order to assist the identification by extending the sequence coverage. The two peptides shown were not identified by the IDA analysis but are clearly identified in the SWATH™ analysis. The unbiased approach provides spectra for all peptides, even those that would not have been targeted in a shotgun experiment because of mass, charge state, etc. The example below shows another example of this and compares the evidence for a confidently identified peptide to the unidentified peptide of comparable quality. In extreme cases, many ladder sequences are identified in the most abundant proteins. This complicates the identification of peptides suitable for quantification as the reproducibility of these different mis-cleaved forms is unknown. This figure shows all peptide forms detected for a single protein and reveals that only two peptides (green) are represented by a single tryptic peptide; all others have multiple forms which in general will invalidate their use for both normalization and also for quantification. (See also poster TP144). Although it is possible to identify these in shotgun experiments SWATH™acquisition allows determination of their reproducibility and hence their suitability for quantification. 1 The target protein can be transferred from the quantitative browser, entered manually or retrieved from a repository and digested. 2 One or all peptides can be processed by looking for selected fragments in the expected Swath(s) or all. A window around the expected retention time (if known from a shotgun experiment) or the whole run can be processed. 3 After processing the display shows the score (yellow -> red) in each Swath (if processed) and indicates the Swaths expected to contain the precursor (marked 1+, 2+…). 4 Selecting a peptide/Swath generates (lower, left to right): The overlaid fragment chromatograms for the highest scoring Swath. A summary of the scores for all Swaths and retention times; the cross-hairs indicate the one displayed. Here it is clear that several SWATHs have reasonable scores at the same retention time. The MS spectrum for the selected Swath, here showing the 2+ molecular ion of the selected peptide. A background subtracted MS/MS spectrum marked with the sequence ions that match those predicted (top right pane). The browser provides a highly interactive display of information concerning the selected peptides, the scores for the selected fragment ions in all Swaths and their retention time behaviour. The panes are linked so that changing the selected peptide updates all of the others. This example, from the SWATH™ acquisition analysis of an E. coli sample confirms the detection of four truncated versions of the same peptide. Since they are at different retention times they are distinct compounds and not the result of in source fragmentation. Augmenting Protein Identification Results Shotgun protein identification is based on a single snapshot MSMS spectrum of the parent ions isolated by Q1 region which introduces a number of issues. It can be difficult to confirm low level identifications since there is no way to verify that the fragment ions are from the same peptide and not from a mixed spectrum. Using SWATH™ acquisition it is possible to check the presence of the compound by showing that the product ions all elute with a similar profile and thereby increase confidence in the identification. These examples show confirmation of two peptides that were identified but with low confidence values. Extracting the expected fragment chromatograms reveals that they generate peak groups with reasonable scores at the expected retention times.