Denconstruct and Simplify Mega Data from Xenobiotic Metabolite Studies with PCVG
Drug metabolism studies have traditionally relied upon compound-specific LC-MS/MS analyses to quantitate and identify metabolites during the drug discovery and development process. Early stage identification of all metabolites-including low-abundance products-is not always possible, and biotransformation scientists occasionally must backtrack, revisiting samples to unearth information on previously-unidentified compounds. If MS/MS spectra could be acquired upfront for all metabolites, both known and unknown, during the drug discovery process, the chances of overlooking an unpredicted metabolite would decrease, saving time and satisfying regulatory requirements more quickly. Novel, data-independent acquisition (DIA) strategies such as SWATH® Acquisition1,2 now make non-targeted analyses a reality in complex biological matrices and provide an overall snapshot of low-abundance, genotoxic, and major metabolites. Having complete coverage creates a richer, more detailed picture, but wading through the expanse of data-MS/MS spectra for every fragment of every precursor ion-can be daunting and time-consuming. Deconstructing this amassed data into interpretable results requires a powerful algorithm-principal components variable grouping (PCVG)-that effectively filters unabridged MS/MS data to extract comprehensive identification and quantitative information.
Generating complete metabolite fragment ion data sets using SWATH Acquisition
Recently, AB SCIEX scientists showed that PCVG algorithms could rapidly identify and quantify drug metabolites formed in complex biological matrices using DIA methods.1 In these studies, all sample components were acquired in a single injection using SWATH Acquisition, an innovative, non-targeted LC-MS/MS data collection system on TripleTOF® 5600+ system, where an all-inclusive fingerprint is generated from MS/MS scans of every parent ion in the sample (Figure 1).
Figure 1: SWATH Acquisition of all fragment ions for all precursors results in complex data sets. The non-specific fragmentation data collection strategy generates an unbiased record of all fragments of all precursors within each of the multiple SWATH Acquisition isolation windows. Fragments that belong to the same precursor follow the chromatographic profile of that precursor.
SWATH Acquisition permits a full cluster of ions within a wide Q1 mass window to travel concurrently into the collision cell for fragmentation. Subsequent SWATH scans conducted during the same injection sequentially collect fragment ion information on incrementally increasing mass segments across the total mass range of interest (Figure 2). The resulting composite fragment ion data sets for each drug metabolite sample were laden with thousands of MS/MS spectra, rich in metabolic information. Additionally, a given MS/MS spectrum may be a combination of spectra for two or more metabolites, convoluted in such a way that interpretation of these raw, unprocessed spectra can be misleading.
PCVG deconvolution reduces and simplifies complex multivariate data sets
Advances in LC-MS/MS DIA methods have produced fragment ion data sets so vast that discovering the critical connections amongst correlated data points is challenging without further data processing. To simplify multivariate LC/MS data processing within metabolite identification workflows, AB SCIEX scientists have integrated a novel algorithm called principal component variable grouping (PCVG) into a research version of MetabolitePilot™ Software. PCVG reduces the dimensionality of complex data sets by combining correlated variables into new representative groups that are related to a particular peak in the LC/MS chromatogram, delivering data that is easier to manipulate and understand.3 The PCVG algorithm uses an unsupervised method to assign related variables to groups, while also filtering out uncorrelated variables. Deconvolution by the PCVG algorithm proceeds in the following manner:
The pre-processed, aligned, multiplexed fragment ion spectra (i.e., SWATH data) undergo principal component analysis (PCA).
The variables are m/z values, the samples for the initial PCA are the raw, non-specific fragmentation spectra, and the resulting groups are the pure MS/MS spectra.
PCVG analyzes the PCA loadings values to find correlated variables (fragment m/z values).
PCVG automates data reduction, filtering variables that do not correlate with the target LC peak profile.
The smaller, deconvoluted data set facilitates spectral simplification, aiding in data interpretation.
PCVG processing correlates signals across the entire mass range examined, which, in turn, allows researchers to untangle complex relationships from among peaks of interest and to link information on isotopes, adducts, and fragments to related compounds.
Figure 2: SWATH Acquisition sequentially collects MS/MS information for selected mass windows (swaths) across a total mass range of interest. Sequential Q1 isolation was stepped over the mass range of interest (e.g., 25 Da or user defined). The high speed of the TripleTOF® 5600+ System allows for full coverage of the selected mass range in an LC time scale and for high resolution XIC data for all fragment ions.
PCVG-filtering finds and correlates related metabolite peaks
Motivated by the successful reduction of the dimensionality of other LC/MS multivariate data sets obtained for proteomic2 and drug metabolite discoveries,4 AB SCIEX scientists applied PCVG to xenobiotic metabolite data generated by SWATH Acquisition to derive a fingerprint of all parent-related compounds, creating a more easily interpretable data set that equaled and-for some analytes-surpassed the results obtained using information-dependent acquisition (IDA) methods. After deconvoluting the SWATH Acquisition data, researchers confirmed that the PCVG-filtered MS/MS spectra included identical peaks at similar intensities as those obtained by IDA.1 Shown here, SWATH Acquisition of nefazodone metabolites resulted in full retention of isotopic fragment ions and accurate mass information (Figure 3). The PCVG filters maintained fidelity of the raw data and revealed a minor characteristic peak that correlated to the nefazodone (m/z 317) structure. In similar experiments, Biogen Idec scientist, Natasha Penner, used SWATH Acquisition to systematically identify metabolites for a drug candidate5 (Table 1). When comparing SWATH Acquisition data with results obtained using more conventional IDA techniques,
Key challenges of metabolite identification in complex biological matrices
Missing, low-level drug metabolites in complex biological matrices such as bile, plasma, and tissue extracts
Incomplete metabolite information leading to repeated sample analysis and decreased productivity
Non-definitive metabolite identification and characterization due to inadequate MS/MS information
Multiple, non-integrated software platforms complicate data processing, slowing metabolite ID and structure elucidation
Key benefits of SWATH Acquisition and PCVG algorithm for metabolite identification
Comprehensive metabolite fingerprinting of irreplaceable experimental samples using SWATH Acquisition. Having a complete array of spectra (both MS and MS/MS scans) provides a digital archive of all analytes for samples with restricted availability (e.g., pediatric studies, expensive toxicological studies).
The ultimate safety net with 100% MS/MS coverage is realized by capturing structural information for both predicted and unpredicted metabolites, including low-level and genotoxic products.
MetabolitePilot™ is an all-in-one integrated software tool that helps rapidly identify and confirm metabolites with structural elucidation capabilities built-in without the need to switch between multiple software tools.
Easy method development and retrospective data-mining -- Requires no sample-specific method development
Key features of SWATH Acquisition and PCVG algorithm for metabolite Identification
Selective MS/MS quantitation is achieved using single or multiple product ions that are summed from multiple transitions.
A less complex MS/MS spectrum than traditional data-independent acquisition strategies due to PCVG-correlation of related peaks.
Full retention of the isotopic pattern for each fragment due to a wider Q1 selection is ideal for stable-label drug studies (14C- metabolism studies) and 100% MS/MS coverage for low-level metabolite/catabolite ID.
PCVG algorithm enables simplified interpretation and data dimensionality reduction of complex metabolite spectra generated using data-independent acquisition.
PCVG algorithm is a fast, robust, and reliable approach for the deconvolution of multi-component fragment ion spectra that is applied within a research version of MetabolitePilot™ Software.
Figure 3: TOF MS/MS spectra of nefazodone metabolites collected with IDA and SWATH Acquisition methods.1 SWATH Acquisition with PCVG-filtering result in MS/ MS spectra that contain the full isotope pattern for fragment ions. Both background subtract and PCVG-filtering strategies yielded accurate mass information enabling a more confident structure proposal. In the PCVG-filtered TOF MS/MS spectrum, an additional minor peak (m/z 317) that corresponds to a direct substructure of nefazodone was recovered from the raw data.
Dr. Penner observed an increased number of PCVG-filtered metabolite peaks-13 out of the 13 known metabolites-with complete MS/MS coverage when using SWATH Acquisition, surpassing results achieved with generic TOF MS data dependent acquisition. Additionally, each fragment in the MS/MS data filtered by the PCVG algorithm, retained the full isotopic pattern and compared favorably to samples analyzed using traditional MS/MS approaches (Figure 4). Taken together, these data validate and confirm the versatility and accuracy of the PCVG method compared to well-established IDA methods.
Table 1: Metabolite Coverage for Biogen Idec Drug Candidate BIIB021 in Complex Biological Matrix Using Generic TOF MS IDA and SWATH Acquisition5
Figure 4: MS/MS obtained using SWATH Acquision and retention of the full isotopic pattern for a fragment from Biogen Idec drug candidate, BIIB021.5 MS/MS data were deconvoluted (A) with PCVG-filtering and (B) without PCVG-filtering.
Advantages of metabolite discovery with SWATH Acquisition and PCVG filtering
In combination, SWATH Acquisition and the PCVG algorithm provide a powerful method for confident metabolic structure assignment and offer many advantages when processing complex MS/MS fragment ion data sets for quantitation and identification of metabolites. The following benefits allow for improved selectivity and specificity when pinpointing drug related material:
The ultimate safety net is realized by capturing both predicted and unpredicted metabolites. Having a complete array of spectra (both MS and MS/MS scans) provides researchers a digital archive of all analytes, so that data corresponding to unexpected products can be retrospectively probed without having to re-acquire a sample.
Retention of full isotope pattern for each fragment is possible due to the relatively wide SWATH window, which significantly aids in the designation of metabolite structures and elemental composition during metabolite discovery. This isotopic data along with 100% MS/MS spectra provides sufficient structural information for the identification and quantitation of low-abundance metabolites.
A less complex MS/MS spectrum is generated by MS/MSALL with SWATH Acquisition than other DIA techniques. PCVG correlates related peaks, and the relevant spectra make it easier to decide which parent ion goes with which fragment, resulting in higher quality data even with complex data from plasma or bile samples.
Easy method development allows researchers to focus on data analysis instead of compound-specific methods. Because MS/MSALL with SWATH Acquisition is a data independent scan providing quantitative information on all analytes, there is no need to create specialized, data collection strategies for a particular drug candidate; this saves time and unnecessary consumption of limited samples.
Multicomponent quantitation is possible with multiple fragment ion transitions captured simultaneously during SWATH Acquisition in a single injection. This adds an additional layer of confidence to quantitative data by allowing multiple product ions to be summed.
In summary, the PCVG algorithm provides a fast, robust, and reliable approach for deconvoluting non-specific fragmentation data from drug metabolism studies obtained using SWATH Acquisition. PCVG-filtering diminishes the complexity inherent in large, multivariate data sets, while still creating a global picture of xenobiotic drug metabolites and generating highly-interpretable spectra for comprehensive metabolite identification and quantitation. Used during the preliminary stages of the drug discovery process, SWATH Acquisition coupled with the powerful PCVG algorithm (incorporated into MetabolitePilot™ Software) delivers complete metabolite coverage-even of minor products-eliminating the possibility of a missed or underestimated metabolite quantity, thereby streamlining the development of new drug candidates and furthering the understanding of their biotransformation pathways.
Eva Duchoslav1; Gordana Ivosev1; Ignat Shilov2; Hesham Ghobarah1; Lyle Burton1, Suma Ramagiri1 1AB SCIEX, 71 Four Valley Drive, Concord, ON, L4K 4V8 Canada; 2AB SCIEX, Foster City, CA, USA
Special thanks to Natalia Penner, Biogen Idec, DMPK, Cambridge, MA for providing SWATH data examples on Drug Candidate BIIB021
Duchoslav E, Ivosev G, Shilov I, Ghobarah H, Burton L. "Automated metabolite identification and profiling in non-specific fragmentation high-resolution accurate MS data." Poster session presented at: the 61st annual conference of the American Society for Mass Spectromtery; 2013 June 9-13; Minneapolis, MN.
2Gillet LC, Navarro P, Tate S, Röst H, Slevsek N, Reiter L, Bonner R, Aebersold R. "Targeted data extraction of the MS/MS spectra generated by data-independent acquisition: a new concept for consistent and accurate proteome analysis." (June 2012) Mol. Cell Proteomics. DOI 10.1074/mcp.0111.016717
3Ivosev G., Burton L., Bonner R. "Dimensionality reduction and visualization in principal component analysis." (July 2008) Anal. Chem. 80: 4933-4944.
4Hopfgartner, G. "High-resolution mass spectrometry for integrated qualitative and quantitative analysis of pharmaceuticals in biological matrices." (March 2012) Anal Bioanal Chem, 402(8): 2587-96