Article pubs.acs.org/ac
Comparison of Information-Dependent Acquisition, SWATH, and MSAll Techniques in Metabolite Identification Study Employing Ultrahigh-Performance Liquid Chromatography−Quadrupole Timeof-Flight Mass Spectrometry Xiaochun Zhu,* Yuping Chen, and Raju Subramanian Pharmacokinetics and Drug Metabolism, Amgen Inc., One Amgen Center Drive, Thousand Oaks, California 91320, United States S Supporting Information *
ABSTRACT: Sensitive and selective liquid chromatography−mass spectrometry (LC−MS) analysis is a powerful and essential tool for metabolite identification in drug discovery and development. An MS2 (or tandem, MS/MS) mass spectrum is acquired from the fragmentation of a precursor ion by multiple methods including information-dependent acquisition (IDA), SWATH (sequential window acquisition of all theoretical fragment-ion spectra), and MSAll (also called MSE) techniques. We compared these three techniques in their capabilities to produce comprehensive MS2 data by assessing both metabolite MS2 acquisition hit rate and the quality of MS2 spectra. Rat liver microsomal incubations from eight test compounds were analyzed with four methods (IDA, MMDF (multiple mass defect filters)-IDA, SWATH, or MSAll) using an ultrahigh-performance liquid chromatography−qudrupole time-of-flight mass spectrometry (UHPLC−Q-TOF MS) platform. A combined total of 227 drugrelated materials (DRM) were detected from all eight test article incubations, and among those, 5% and 4% of DRM were not triggered for MS2 acquisition with IDA and MMDF-IDA methods, respectively. When the same samples were spiked to an equal volume of blank rat urine (urine sample), the DRM without MS2 acquisition increased to 29% and 18%, correspondingly. In contrast, 100% of DRM in both matrixes were subjected to MS2 acquisition with either the SWATH or MSAll method. However, the quality of the acquired MS2 spectra decreased in the order of IDA, SWATH, and MSAll methods. An average of 10, 9, and 6 out of 10 most abundant ions in MS2 spectra were the real product ions of DRM detected in microsomal samples from IDA, SWATH, and MSAll methods, respectively. The corresponding numbers declined to 9, 6, and 3 in the urine samples. Overall, IDA-based methods acquired qualitatively better MS2 spectra but with a lower MS2 acquisition hit rate than the other two methods. SWATH outperformed the MSAll method given its better quality of MS2 spectra with an identical MS2 acquisition hit rate. of metabolites.1−3 In addition, identification and characterization of in vivo metabolites in nonclinical and clinical excreta is an essential component of an absorption, distribution, metabolism, and excretion (known as ADME) study for any molecule to enable a new drug molecule registration. The predominant tool for identification of drug biotransformation products is liquid chromatography coupled with mass
M
etabolite identification is a crucial and integral part of drug discovery and development. In the drug discovery stage, metabolic profiling is not only used for identification of metabolic soft spots to inform medicinal chemists so that they are able to modify and optimize a lead series to attain the desired pharmacokinetic and drug metabolism properties but also plays a critical role in selection of the animal species used to evaluate the nonclinical safety of the molecule. In the clinical testing of a drug, detection and characterization of circulating metabolites in both animal and human plasma is becoming a prerequisite to satisfy the regulatory guidances on safety testing © 2013 American Chemical Society
Received: October 17, 2013 Accepted: December 18, 2013 Published: December 18, 2013 1202
dx.doi.org/10.1021/ac403385y | Anal. Chem. 2014, 86, 1202−1209
Analytical Chemistry
Article
spectrometry (LC−MS).4−10 Time-of-flight (TOF) mass spectrometry, particularly its hybrid form quadrupole time-offlight (Q-TOF) system,11 is among the top performers for this purpose given its fast acquisition speed, superior sensitivity, high resolution, and excellent mass accuracy.7,12 In a metabolite identification study, m/z of precursor and product ions are recorded in MS and MS2 (also called MS/MS or tandem MS) spectra, respectively, which provide crucial information for the elemental composition analysis and structure elucidation. Using a Q-TOF MS system, high-resolution MS spectra can be collected by the TOF analyzer in the wide pass Q1 mode under low collision energy (CE). In contrast, MS2 data can be acquired in three ways with the Q1 analyzer operated at a narrow, medium, or wide pass mode for precursor ion selection under high CE. The narrow pass Q1 mode is normally employed for information-dependent acquisition (IDA, also called datadependent acquisition, data-directed acquisition, or datadirected analysis). IDA is able to produce MS2 spectra with minimal interference since only ions within the selected narrow m/z window, typically 1−3 Da wide, are transferred to the collision cell to generate product ions. The occurrence of MS2 acquisition for a precursor ion is dependent on preset criteria such as whether this particular ion is among a predefined number of the most abundant ions or in a predefined precursor ion list, or whether a specific neutral loss is detected.13−17 Mass defect filter (MDF) is an efficient data mining tool for highresolution MS data.18,19 The real-time multiple mass defect filters (MMDF) have been successfully applied on the Q-TOF MS to improve the efficiency of the IDA-based method.20 Despite its ability to deliver a high-quality MS2 spectrum, the IDA technique has not been extensively applied to metabolite identification study in the Q-TOF platform, because some ions of interest may not be selected for fragmentation if they do not meet the defined selection criteria.21 Alternatively, the wide pass Q1 mode of a Q-TOF MS for MS2 acquisition has been well-accepted in the drug metabolism community as a technique known as MSAll (also called MSE).22−24 This technique, in contrast to IDA, performs an information-independent acquisition (also called data-independent acquisition) and records all product ions regardless of what the precursor ion is. As all precursor ions formed in the ion source are sent to the collision cell for fragmentation, the resulting nonselective MS2 spectrum may lack specificity if drug-related materials (DRM) coelute with either other metabolites or endogenous components in the matrix. The MSAll technique relies on good chromatographic separation of metabolites to acquire a high-quality MS2 spectrum. Thus, it dramatically benefited from the advancement of modern separation technology, especially the successful development of sub-2 μm column compatible ultrahigh-performance liquid chromatography (UHPLC). Since the UHPLC−Q-TOF system with the MSAll capability was initially reported for metabolism studies,17,25,26 this combined system has been widely used in in vitro and in vivo metabolic profiling.22−24,27−30 The MSAll technique also benefited from data deconvolution algorithms such as retention time and ion mobility drift time alignment, which results in cleaner MS2 spectra.31 Recently, a new technique called SWATH (sequential window acquisition of all theoretical fragment-ion spectra) implemented at the medium pass Q1 mode became available on a Q-TOF system.21,32 Like MSAll, SWATH is also an
information-independent acquisition technique, however, with Q1 ion pass window size in between IDA and MSAll. In this technique, the Q1 is stepped continuously with a medium m/z window, such as 20 or 25 Da, across the whole m/z range of interest. Thus, ions in each medium pass window are transferred into the collision cell and product ions are generated under high CE. The product ions are then analyzed sequentially by the high-resolution TOF analyzer. Compared to MSAll, SWATH obtains MS2 data by fragmentation of a much narrower precursor ion window while retaining the ability to record product ion spectra of all precursor ions within a large mass range. This new technique has been effectively used in the proteomics33,34 and metabolite identification.21 In this study, we compared the capability of the three aforementioned techniques to produce MS2 data for DRM to assess both the number of metabolites whose MS2 spectrum was able to be acquired and the quality of the obtained MS2 spectra. Eight test compounds, bufuralol, chlorpromazine, lansoprazole, midazolam, nefazodone, propranolol, quinidine, and terfenadine (structures shown in Table S1 in the Supporting Information), were chosen, because these compounds were documented to undergo extensive oxidative metabolism and also have diverse structures with a range of molecular weights (259−471 Da) often encountered for smallmolecule drugs.35 For each compound, metabolites were generated in vitro and also spiked into blank (nondosed) urine to simulate an in vivo situation. Samples were analyzed on a UHPLC−Q-TOF system employing four different methods: IDA, MMDF-IDA, SWATH, and MSAll. The first two methods employed IDA, either without or with MMDF enabled. The latter two were embedded with SWATH and MSAll techniques, respectively. Evaluation and comparison were performed in regard to both the MS2 acquisition hit rate and the quality of the MS2 spectra for all DRM detected from these eight test compounds.
■
EXPERIMENTAL SECTION General. Bufuralol, chlorpromazine, lansoprazole, midazolam, nefazodone, propranolol, quinidine, terfenadine, and βnicotinamide adenine dinucleotide 2′-phosphate reduced tetrasodium salt (NADPH) were purchased from SigmaAldrich (St. Louis, MO). APCI positive calibration solution (P/N: 4460136) was purchased from AB Sciex (Concord, Ontario, Canada). Pooled male rat liver microsomes (RLM) were acquired from CellzDirect (Durham, NC). Blank rat urine was collected at Amgen Inc. (Thousand Oaks, CA). Microsomal Incubation. In a final volume of 1 mL, each test compound (10 μM final concentration; bufuralol, chlorpromazine, lansoprazole, midazolam, nefazodone, propranolol, quinidine, and terfenadine) was incubated with RLM (1 mg/mL protein) in potassium phosphate buffer (100 mM, pH 7.4) containing MgCl2 (3 mM) at 37 °C. After preincubation for 3 min, the reaction was initiated by the addition of NADPH (1 mM). The reaction was quenched after 1 h with 1 mL of icecold acetonitrile and then centrifuged at 1400g for 10 min at 4 °C. The supernatants were reduced to approximately 1 mL by evaporating under a stream of nitrogen gas. The concentrated supernatants were either analyzed directly or spiked into an equal volume of blank rat urine (called urine sample hereafter) before analysis by UHPLC−Q-TOF MS. Control samples were incubated without the test compound, which was instead added to the quenched reaction mixture, and processed as described above. 1203
dx.doi.org/10.1021/ac403385y | Anal. Chem. 2014, 86, 1202−1209
Analytical Chemistry
Article
loss. The data was first processed by MetabolitePilot, which automatically generated a list of metabolite candidates. The list was then manually reviewed to remove the false positives. The percentage of total DRM peak area (or percent area) was the ratio of the MS peak area of a DRM to that of all DRM. The MS peak area used in all figures was an average of areas reported by MetabolitePilot for each DRM in RLM data acquired by both IDA and MMDF-IDA methods. MS2 Spectra Quantity (or MS2 Acquisition Hit Rate) Evaluation. The MS2 quantity score (QNS) was defined to measure the MS2 acquisition hit rate of DRM. The QNS was calculated as shown in eq 1,
UHPLC−Q-TOF MS. All samples were subjected to reversed-phase chromatography using a UHPLC system (model 1290; Agilent Technologies Inc., Wilmington, DE) consisting of a binary pump (G4220A), a thermostatted column compartment (G1316C), a diode-array detector (G4212A), and an autosampler (CTC PAL, Leap Technologies, Carrboro, NC). Mobile phases A and B were 0.1% formic acid in water and acetonitrile, respectively. Chromatographic separation was performed at a flow rate of 450 μL/min using a UHPLC column (Acquity HSS T3, 1.8 μm, 2.1 mm × 100 mm; Waters Corp., Milford, MA) maintained at 45 °C. Two solvent gradients were employed. Samples containing quinidine were separated by gradient 1: 0−1 min, 5% B; 1−6 min, 5−25% B; 6−6.5 min, 25−95% B; 6.5−7.5 min, 95% B; 7.5−8 min, 95− 5% B; 8−10 min, 5% B. All other samples were separated by gradient 2, which was the same as gradient 1 except that 65% B was used at 6 min. Injection volume was 10 μL for all samples. Mass spectrometric analysis was performed with a Q-TOF MS (TripleTOF 5600+, AB Sciex) operating in the positive ion mode using a DuoSpray ion source. The instrument was operated at a mass resolution of ∼30 000 for TOF MS scan (at m/z 300) and ∼15 000 for product ion scan in the highsensitivity mode, and automatically calibrated every 10 sample injections using APCI positive calibration solution delivered via a calibration delivery system (AB Sciex). The other experiment parameters were set as follows: curtain gas, 30 (arbitrary units); ion source gas 1, 50 (arbitrary units); ion source gas 2, 50 (arbitrary units); temperature, 500 °C; ion spray voltage floating, 5.5 kV; declustering potential, 80 V. High-resolution MS and MS2 data for each sample were acquired by four different methods: IDA, MMDF-IDA, SWATH, and MSAll method. The IDA method (cycle time 350 ms) was composed of a TOF MS scan (accumulation time, 50 ms; CE, 10 V) and 10 dependent product ion scans (accumulation time, 25 ms each; CE, 45 V) in the highsensitivity mode with dynamic background subtraction. The MMDF-IDA method (cycle time 350 ms) was the same as the IDA method, but with the mass defect filter (MDF) function enabled. The MDF window was set to ±40 mDa, and the mass range was set as ±50 Da around the m/z of each filter template. The MDF template(s) for each compound is listed in Table S1 (in the Supporting Information). The SWATH method (cycle time 380 ms) was composed of a TOF MS scan (accumulation time, 50 ms; CE, 10 V) and a series of product ion scans (accumulation time, 10 ms each; CE, 45 V) of 28 Q1 window of 27 Da from m/z 150−850 in the high-sensitivity mode. The consecutive Q1 windows overlapped by 2 Da. The MSAll method (cycle time 350 ms) was composed of a low-energy TOF MS scan (accumulation time, 150 ms; CE, 10 V) and a high-energy TOF MS scan (accumulation time, 150 ms; CE, 45 V). Mass ranges of TOF MS and product ion scans were both m/z 100−1000 for all methods. IDA, SWATH, and MSAll methods were the same for all test compounds, but the MMDF-IDA method was compound-specific to accommodate their respective MDF settings. The LC−MS data was acquired using Analyst TF 1.6 (AB Sciex). Data Analysis. High-resolution LC−MS data were viewed and processed by PeakView 1.220 (AB Sciex). Screening and identification of metabolites were assisted by MetabolitePilot 1.520 (AB Sciex) employing different peak-finding strategies provided by the software, including predicted metabolites, generic peak finding, multiple mass defect filters, and finding at least one characteristic product ion or characteristic neutral
QNS =
NMS2 × 100 NMS
(1)
where NMS2 and NMS are the number of DRM with an acquired MS2 spectrum and the total number of DRM detected in the MS spectra, respectively, in RLM or urine samples. The QNS ranges from 0 to 100. Higher QNS of a compound means that a higher number of DRM MS ions resulted in MS2 spectra. MS2 Spectra Quality Evaluation. The MS2 quality score (QLS) was calculated by comparing the number of common product ions between the evaluated spectrum and the reference spectrum. MS2 spectra from RLM samples obtained by the MMDF-IDA method were set as reference spectra. To minimize the interference of background peaks (normally
