Anal. Chem. 2010, 82, 794–797
Letters to Analytical Chemistry Ultrathroughput Multiple Reaction Monitoring Mass Spectrometry Xudong Yao,* Bekim Bajrami, and Yu Shi Department of Chemistry, University of Connecticut, Storrs, Connecticut 06269 A novel transformation of the multiplexing potential to the throughput potential of multiple reaction monitoring mass spectrometry is presented for targeted quantitation of proteins. Herein, this method is termed as ultrathroughput multiple reaction monitoring (UMRM) mass spectrometry. It integrates the use of stable isotope dilutionmultiple reaction monitoring mass spectrometry and peptide derivatization with inexpensive and commercial chemicals. One-experiment quantitation of a common signature peptide in 25 different samples demonstrates proof-of-concept for the unprecedented throughput potential of the UMRM technology. Limited throughput is an outstanding bottleneck for unleashing the full potential of contemporary mass spectrometry (MS). The use of stable isotopes and chemical reagents allows one-experiment quantitation of several proteome samples with differential mass codes.1-3 Current MS throughput is low, especially when using low-resolution mass spectrometers like triple quadrupole mass spectrometers, the default mass spectrometers for quantitation of protein targets in biomatrixes. Only very recently has concurrent quantitation of biomarkers in three proteome samples been demonstrated on a triple quadrupole mass spectrometer.4 The conventional approach to increase the MS throughput is to decrease the analysis time for each sample. However, the highly complex nature of proteome samples makes this practice challenging. Reducing proteome sample preparation time is thus considered as an alternative solution for improving the overall analysis throughput,5 but this strategy for the throughput improvement has only limited potential. By comparison, parallel analyses of many samples using non-MS methods, e.g., array* Corresponding author. E-mail:
[email protected]. (1) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R. Nat. Biotechnol. 1999, 17, 994–999. (2) Ross, P. L.; Huang, Y. N.; Marchese, J. N.; Williamson, B.; Parker, K.; Hattan, S.; Khainovski, N.; Pillai, S.; Dey, S.; Daniels, S.; Purkayastha, S.; Juhasz, P.; Martin, S.; Bartlet-Jones, M.; He, F.; Jacobson, A.; Pappin, D. J. Mol. Cell. Proteomics 2004, 3, 1154–1169. (3) Dayon, L.; Hainard, A.; Licker, V.; Turck, N.; Kuhn, K.; Hochstrasser, D. F.; Burkhard, P. R.; Sanchez, J. C. Anal. Chem. 2008, 80, 2921–2931. (4) DeSouza, L. V.; Taylor, A. M.; Li, W.; Minkoff, M. S.; Romaschin, A. D.; Colgan, T. J.; Siu, K. W. J. Proteome Res. 2008, 7, 3525–3534. (5) Anderson, N. L.; Anderson, N. G.; Haines, L. R.; Hardie, D. B.; Olafson, R. W.; Pearson, T. W. J. Proteome Res. 2004, 3, 235–244.
794
Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
based immunoassays of protein biomarkers,6 present aspirational standards for improving the MS throughput potential. The multiplexing potential, on the other hand, is intrinsically very high for MS due to the capability that modern instruments possess. These instruments can efficiently separate analyte ions based on mass-to-charge ratios in the gas phase. In one experiment, MS can quantify signature peptides for up to hundreds of different target proteins in a particular sample.7,8 The MS multiplexing potential refers to one-experiment analysis of different analytes like peptides in a particular sample. The MS throughput potential refers to one-experiment analysis of common analytes in different samples.7 Herein, a novel transformation of the intrinsic multiplexing potential to the throughput potential is proposed for developing a new method of targeted MS quantitation of proteins. This method is termed as ultrathroughput multiple reaction monitoring (UMRM). The use of the “ultrathroughput” term is intended to differentiate the new UMRM strategy for the throughput improvement from the conventional high-throughput MS that mainly focuses on shortening analysis time. The UMRM technology transforms existing approaches of multiple reaction monitoring (MRM) by a unique integration with peptide derivatization. With inexpensive chemicals and commercial mass spectrometers, a single UMRM analysis can quantify common target proteins in tens to hundreds of samples. During the last several years, MRM MS has been significantly expanded in scope, becoming a method of choice for quantifying target proteins in complex biological samples. The scope broadening is mainly attributed to emerging applications of MRM analysis in the protein biomarker area,7 as well as to adaptive improvements in triple quadrupole mass spectrometers by manufacturers. Combined with liquid chromatography (LC) and the method of stable isotope dilution (SID), LC-SID-MRM shows high potential in early phases of developing novel protein biomarkers, in which large numbers of candidates but in small numbers of samples need to be analyzed.7 These candidate biomarker proteins are produced via global quantitative proteomic profiling. The sample number significantly increases, in tandem with the process that encompasses the protein biomarker pipeline from (6) Liu, M. Y.; Xydakis, A. M.; Hoogeveen, R. C.; Jones, P. H.; Smith, E. O.; Nelson, K. W.; Ballantyne, C. M. Clin. Chem. 2005, 51, 1102–1109. (7) Rifai, N.; Gillette, M. A.; Carr, S. A. Nat. Biotechnol. 2006, 24, 971–983. (8) Schiess, R.; Wollscheid, B.; Aebersold, R. Mol. Oncol. 2009, 3, 33–44. 10.1021/ac9026274 2010 American Chemical Society Published on Web 01/11/2010
Figure 1. Perspective comparison of peptide targets for MRM and UMRM measurements of protein biomarkers.
candidate discovery to qualification, verification and validation.7 In later phases of biomarker development, verification and validation of a smaller set of protein biomarkers require quantitative analysis of hundreds and thousands of patient samples. Immunoassays are currently used to meet the throughput requirement. However, development of immunoassays is costly and lengthy for novel protein biomarkers, especially those without existing immunograde antibodies.9 The high specificity of MRM measurements, enhanced by monitoring of selected gas-phase transitions from precursor to fragment ions, allows possible circumvention of the use of immunograde antibodies. The realization of this possibility, however, depends on the significant improvements to the MS throughput so that large numbers of samples can be quantified in a reasonable time period. The rationale for the transformation of the multiplexing potential of MRM methods into the throughput potential for UMRM measurements is illustrated in Figure 1. One MS experiment can quantify up to hundreds of different peptides in a targeted manner. From the multiplexing perspective, these peptides represent signature peptides for different target proteins in one proteome sample. This is commonly practiced in MRM measurements, in which the throughput number (or sample number) is equal to the number of LC runs and the multiplexing number is on the same order of magnitude as that of the number of gas-phase transitions. From the throughput perspective, on the other hand, the different peptides can be considered as a common peptide with varying mass codes specific to hundreds of different samples; the peptide in each sample is chemically derivatized with one sample-specific mass code. The coded peptides can be distinguished by MS. Therefore, all of the derivatized samples can be pooled for one-experiment targeted quantitation of the common peptide but with different mass codes. This is the basic principle for UMRM measurements, for which the throughput number (or sample number) is on the same magnitude of the number of gas-phase transitions monitored. It is important to note that because the basic difference between MRM and UMRM measurements only resides at the selection of peptide targets, both methods are thus executable on the same instruments like (9) Carr, S. A.; Anderson, L. Clin. Chem. 2008, 54, 1749–1752.
Figure 2. General workflow of UMRM experiments. Upper, workflow design; lower, ion chromatograms for a total of 10 transitions monitored for the endogenous signature peptide (DAFLGSFLYEYSR) and the spiked stable isotope reference peptide (DAFL*GSFL*YEYSR) with 5 different derivatizations of peptidyl amines.
quadrupole mass spectrometers. It is certainly feasible to implement customized, targeted peptide quantitation with intermediate throughput and multiplexing numbers, which has a higher throughput number than conventional MRM measurements and a higher multiplexing number than a UMRM experiment quantifying only one signature peptide. In a UMRM experiment (Figure 2, upper), each proteome sample is digested and spiked with known amounts of stable isotope labeled reference peptides, which have the same sequences as those of the counterpart, endogenous signature peptides. Afterward, each spiked digest is subjected to chemical derivatization of the peptidyl carboxylate and/or amine groups. The derivatized samples are pooled and quantified based on peptide gas-phase transitions. Intensities for pairs of derivatized endogenous and reference peptides are used to calculate quantities of the corresponding target proteins in different proteome samples. Demonstrative experiments illustrate the throughput potential of UMRM measurements (Figure 2, lower). A digest of bovine serum albumin (BSA) was spiked with a synthetic BSA peptide with stable isotope labels (DAFL*GSFL*YEYSR; L* was labeled with 13C6 and 15N1; Supporting Information); it is referred to as the spiked labeled reference peptide. The native counterpart peptide resulted from the protein digestion is referred to as the endogenous signature peptide. A total of 10 replicate measurements on the spiked sample, monitoring a gas-phase transition ([M + H]2+ f y9+) of 784.3/1121.5 for the endogenous peptide and 791.3/1128.6 for the labeled reference, gave an average ratio of 0.790 (±0.010) for the endogenous to the labeled Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
795
peptides (Table S1 in the Supporting Information). For the sake of simplicity in demonstration of the UMRM principle, all peptides in this work were measured using a single transition. (In general, 2-3 signature peptides and 2-3 transitions are commonly used for quantifying target proteins in complex samples to ensure the quantitation specificity.) Five aliquots of the spiked sample were derivatized separately for their amine groups, using the Nhydroxysuccinimide (NHS) ester of tert-butoxycarbonyl (t-Boc)protected R-amino isobutyric acid or acetic, propionic, butyric, or succinic anhydrides. These aliquots were pooled for one-experiment quantitation of peak ratios (gas-phase transitions listed in Table S2 in the Supporting Information) for the endogenous to the labeled peptides with five different N-terminal derivatizations, giving an average ratio of 0.879 (±0.035). This ratio showed the high reproducibility of the UMRM method, averaging data obtained from triplicate injections for each derivatization sample and triplicate sample preparations for each derivatization reagent (Table S3 in the Supporting Information). Figure 2 (lower) shows overlaid ion chromatograms for all of these transitions during a single LC run, marked with corresponding derivatization groups. This demonstrates a UMRM experiment with a throughput number of 5, i.e., 5 different samples are combined for a single analysis. Five aliquots of the pooled sample with the five different N-terminal derivatizations were further derivatized with acid catalyzed esterification using methanol, methanol-d4, ethanol, ethanol-d6, and n-propanol, respectively. After the two sequential derivatizations, the endogenous signature peptide and the labeled reference peptide were equally coded with 25 different combinations of peptidyl amine and carboxylate derivatizations. Thus, pooling these samples produced a model mixture of 25 samples with distinct derivatizations. A total of 50 transitions (Table S4 in the Supporting Information) were set for MRM quantitation of the endogenous and labeled peptides during a 30 min gradient. Each pooled sample was analyzed by three runs. Table 1 summarized results of ratios calculated for individual derivatization combinations (Table S5 in the Supporting Information). The average value was 0.827 (±0.089), excluding one sample with weak signals. The remaining two pooled samples for triplicate UMRM experiments gave 0.836 (±0.068) and 0.827 (±0.077), respectively. This demonstrates a UMRM experiment with a throughput number of 25. The high tolerance for sample complexity in the targeted measurements of peptide gas-phase transitions makes it feasible for the UMRM technology to achieve practical applications. The UMRM throughput relies on chemical derivatization. Intrinsic limitations of less-than ideal chemical conversion and sidereactions are associated with chemical derivatization of proteome digests. Therefore, only those reagents with unique application niches have found broad acceptance.2 This stands in striking contrast to the fact that there is a large collection of reports on peptide derivatization for MS analysis.10,11 With high selectivity and large dynamic range, MRM and UMRM measurements target only analytes of interest for analysis and thus are the best MS approaches for mitigating the intrinsic limitations of chemical derivatization. (10) Regnier, F. E.; Julka, S. Proteomics 2006, 6, 3968–3979. (11) Leitner, A.; Lindner, W. Proteomics 2006, 6, 5418–5434.
796
Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
Table 1. UMRM Quantitation of 25 Samples with Different Derivatizations and the Identical Initial Ratio of the Signature to Reference Peptidesa derivatization N1C1 N1C2 N1C3 N1C4 N1C5 N2C1 N2C2 N2C3 N2C4 N2C5 N3C1 N3C2 N3C3 N3C4 N3C5 N4C1 N4C2 N4C3 N4C4 N4C5 N5C1 N5C2 N5C3 N5C4 N5C5 P Average P StDev
sample sample sample pool 1 pool 2 pool 3 0.749 0.783 N/D 0.878 0.840 0.829 0.800 0.777 0.931 0.841 0.896 0.740 0.636 0.710 0.765 0.722 0.978 0.854 0.983 0.905 0.714 0.862 0.840 0.890 0.921 0.827 0.087
0.757 0.804 N/D 0.930 0.955 0.833 0.839 0.785 0.941 0.884 0.817 0.721 0.888 0.744 0.823 0.721 0.838 0.908 0.886 0.834 0.752 0.820 0.826 0.845 0.908 0.836 0.066
0.734 0.903 N/D 0.893 0.915 0.790 0.747 0.772 0.904 0.855 0.860 0.661 0.823 0.718 0.759 0.759 0.894 0.853 0.945 0.887 0.722 0.857 0.828 0.907 0.862 0.827 0.075 SP Average SP StDev
S Average S StDev 0.746 0.830 N/D 0.900 0.904 0.817 0.795 0.778 0.926 0.860 0.858 0.707 0.782 0.724 0.782 0.734 0.904 0.872 0.938 0.875 0.729 0.846 0.831 0.881 0.897
0.012 0.064 N/D 0.027 0.058 0.024 0.046 0.007 0.019 0.022 0.040 0.041 0.131 0.017 0.035 0.022 0.070 0.031 0.049 0.037 0.020 0.023 0.008 0.032 0.031
0.830 0.076
a N/D, not determined. NiCj’s (e.g., N1C2) refer to different derivatization codes (Table S5 in the Supporting Information). P Average and P StDev refer to the average value and the associated standard derivation calculated for different derivatizations in a particular pool. S Average and S StDev refer to the average value and the associated standard derivation calculated for the same derivatization in three different pools. SP Average and SP StDev refer to the overall average value and the associated standard derivation calculated for the triplicate derivatization experiments with triplicate runs. Detailed calculations see Table S5 in the Supporting Information.
Addition of stable isotope reference peptides prior to derivatizations provides the UMRM technology with an extreme flexibility in the reagent selection. This flexibility allows decoupling the coelution requirement for simple MS quantitation12 from the diversity and economical preferences for peptide derivatization. Thus, the rich chemistry for peptide and protein derivatization and the wide range of commercial reagents can readily be exploited to targeted MS quantitation of proteins. In the experiments demonstrated here, both the endogenous signature peptide and the spiked stable isotope reference peptide underwent the same derivatization and sample cleanup procedures. Therefore, the original quantitative information associated with these peptides was carried into the derivatized peptides, allowing unbiased MS quantitation. The reference peptide used 13C and 15N isotopes13 so that the endogenous and reference peptides with the same derivatization coeluted on reversed-phase chromatography. Furthermore, the early addition of the stable isotope reference peptide in the sample preparation workflow (Figure 2, upper) allows the use of nonstable isotope chemicals and inexpensive (12) Zhang, R.; Sioma, C. S.; Wang, S.; Regnier, F. E. Anal. Chem. 2001, 73, 5142–5149. (13) Gerber, S. A.; Rush, J.; Stemman, O.; Kirschner, M. W.; Gygi, S. P. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 6940–6945.
deuterium-based reagents for derivatizations. These reagents have conventionally been considered unfavorable due to the difficulty posed by chromatographic separation of the differentially derivatized peptides.12 In contrast, the chromatographic separation (Figure 2, lower) is desirable for UMRM measurements. This separation is particularly useful when the throughput number of a UMRM experiment is very high and measurements need to be scheduled at different elution times in order to monitor a large number of gas-phase transitions in a single LC run. The flexibility in the derivatization reagent selection allows to deliberately produce derivatized peptides with low elution degeneracy, i.e., small in the number of derivatized peptides coeluting at similar times. Separation of differentially derivatized peptides also lessens the requirement for low mass degeneracy of derivatized peptides, i.e., small in the number of derivatized peptides having similar masses. This further broadens the candidate reagent pool. The extreme flexibility in the reagent selection provides a second channel for mitigating derivatization limitations. The chemical conversion yield of a derivatization is typically less-than ideal and/or derivatizations can be incomplete especially when more than one reactive amine or carboxylate is present on a peptide. However, the signal yield for MS (SYMS) can be well above 100% when comparing a peptide carrying the desired derivatization with the underivatized counterpart peptide. For MRM and UMRM measurements, SYMS for derivatized peptides depends on both the precursor and fragment ion intensities and can be increased by two general ways. One is to increase the (14) Beardsley, R. L.; Reilly, J. P. J. Proteome Res. 2003, 2, 15–21. (15) Mirzaei, H.; Regnier, F. Anal. Chem. 2006, 78, 4175–4183. (16) Williams, D. K., Jr.; Meadows, C. W.; Bori, I. D.; Hawkridge, A. M.; Comins, D. L.; Muddiman, D. C. J. Am. Chem. Soc. 2008, 130, 2122–2123. (17) Diego, P. A.; Bajrami, B.; Jiang, H.; Shi, Y.; Gascon, J. A.; Yao, X. Anal. Chem. 2010, 82, 23–27. (18) Shi, Y.; Bajrami, B.; Yao, X. Anal. Chem. 2009, 81, 6438–6448.
signal intensity for the intact, derivatized peptide; basic and hydrophobic reagents are known to enhance MS signals for derivatized peptides.14-16 The other is to preferentially produce certain fragment ions by active derivatization.17,18 MS signal amplification of peptides via derivatizations is attractive for lowabundance proteins in particular. Unlike nucleic acids, there is no “polymerase chain reaction-like” amplification for proteins. Although chemical derivatization provides an interesting alternative for “amplifying” proteins, derivatization for peptides at low concentrations should have satisfactory reaction kinetics. Focused quantitation of a selected set of target proteins is essential for effective MS investigation of cellular signaling and pathways, as well as fast development of new protein biomarkers and personalized medicine for human diseases. With the unprecedented throughput potential, the UMRM technology will have broad impact on basic and applied biological and biomedical research and eventually on clinical practice as well. Because of the flexibility in experimental design and reagent selection and the applicability of commercially available mass spectrometers, immediate adoption of the UMRM technology for targeted protein quantitation is expected. This adoption will encompass many different levels of throughput. ACKNOWLEDGMENT This work was supported by the Cystic Fibrosis Foundation (Grant YAO07XX0) and the University of Connecticut. SUPPORTING INFORMATION AVAILABLE Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review November 16, 2009. Accepted January 5, 2010. AC9026274
Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
797
