ARTICLE pubs.acs.org/ac
Tandem Mass Tag Protein Labeling for Top-Down Identification and Quantification Chien-Wen Hung and Andreas Tholey* Institut f€ur Experimentelle Medizin—AG Systematische Proteomforschung, Christian-Albrechts-Universit€at, Niemannsweg 11, 24105 Kiel, Germany
bS Supporting Information ABSTRACT: Top-down mass spectrometry holds tremendous potential for characterization and quantification of intact proteins. So far, however, very few studies have combined top-down proteomics with protein quantification. In view of the success of isobaric mass tags in quantitative bottom-up proteomics, we applied the tandem mass tag (TMT) technology to label intact proteins and examined the feasibility to directly quantify TMT-labeled proteins. A top-down platform encompassing separation via ion-pair reversed-phase liquid chromatography using monolithic stationary phases coupled online to an LTQ-Orbitrap Velos electron-transfer dissociation (ETD) mass spectrometer (MS) was established to simultaneously identify and quantify TMT-labeled proteins. The TMT-labeled proteins were found to be readily dissociated under high-energy collision dissociation (HCD) activation. The liberated reporter ions delivered expected ratios over a wide dynamic range independent of the protein charge state. Furthermore, protein sequence tags generated either by low-energy HCD or ETD activation along with the intact protein mass information allow for confident identification of small proteins below 35 kDa. We conclude that the approach presented in this pilot study paves the way for further developments and numerous applications for straightforward, accurate, and multiplexed quantitative analysis in protein chemistry and proteomics.
D
ue to its sensitivity and specificity, mass spectrometry (MS) has emerged as the method of choice for large-scale protein identification and characterization.1 3 Two major MS-based strategies, termed “bottom-up”4 and “top-down”,5 have been developed for proteome analysis. Up to now, the majority of proteomics studies employ bottom-up approaches, where proteins are proteolyzed with specific enzymes to form small peptides prior to chromatographic separation and MS interrogation. The major advantage compared to the analysis of intact proteins is that peptides are easier to manipulate in terms of separation, quantification, and identification. However, this method bears some inherent limitations. For example, information on the total protein sequence is only partially obtained, accompanied with a potential loss of information about isoforms and posttranslational modifications (PTMs). In contrast, top-down approaches provide a complementary and promising way for characterizing intact proteins. They involve direct analysis of intact proteins without prior enzymatic proteolysis; protein primary structure is interrogated through measurement of an intact protein mass followed by direct fragmentation in the gas phase.5 The emergence of top-down MS has facilitated the identification and characterization of intact protein forms that are not readily apparent in bottom-up studies, for example, detection of protein isoforms, characterization of PTMs,6 9 and identification of alternative splicing variants and degradation products formed by proteolytic processing.10,11 Traditionally, top-down experiments are performed on Fourier transform ion cyclotron resonance (FTICR) mass spectrometers12,13 r 2011 American Chemical Society
in combination with collision-induced dissociation (CID). The high resolving power of FTICR enables the charge state of multiply charged product ions to be unambiguously assigned. However, the relatively long duty cycle hampers the online coupling to chromatographic separation which limits its application mostly to single purified proteins and simplified mixtures. The development of high mass accuracy and high-resolution hybrid instruments (e.g., LTQ-FTICR MS14 or LTQ-Orbitrap MS15 17) and the introduction of more efficient fragmentation methods [e.g., electron capture dissociation (ECD),18 electron-transfer dissociation (ETD)19,20], along with recent innovations in top-down data analysis softwares21 have made top-down experiments possible on a chromatographic time scale and enhanced the protein sequencing throughput. Furthermore, steady progresses in the development of more efficient and MS-compatible protein separation techniques14,22 28 have created the potential to extend the top-down approach for large-scale qualitative and quantitative proteome profiling. Up to now, quantitative information in top-down proteomics was generated either by label-free approaches or by in vivo labeling methods. In label-free approaches, information about the relative abundance of proteins can be obtained by direct comparison of their respective molecular ion intensities in MS spectra. This strategy has been applied to quantify relative abundances of Received: August 26, 2011 Accepted: November 21, 2011 Published: November 21, 2011 161
dx.doi.org/10.1021/ac202243r | Anal. Chem. 2012, 84, 161–170
Analytical Chemistry
ARTICLE
multiple modified histone isomers29 and for the characterization of different deamidation products of bovine ribonuclease A.30 Mazur et al. have extended this approach to perform quantitative analysis of two sets of samples by performing differential mass spectrometry.24,31 A general problem encountered with labelfree quantification methods is the need to perform a high number of replicates to generate validated data, which triggers this approach laborious and makes it not feasible for biological material only available in restricted amounts. Therefore, differential stable isotope labeling, e.g., in vivo labeling techniques like stable isotope labeling with amino acids in cell culture (SILAC) or protein metabolic labeling, has been the most widely adopted approach in the top-down quantitative studies.14,32 38 Application of these approaches for quantitative analysis has been successful; however, evaluation of the quantitative data is still a bottleneck, since laborious steps of spectrum correction and statistical computation are required. Further, in vivo labeling is not in all cases possible and the biological systems have to be adapted to the labeling conditions. The lack of suitable interpretation methods and, most severely, the limited number of available channels preventing the multiplexed analysis of different biological conditions in a single experiment are further drawbacks. Recently, isobaric stable isotope tagging, including tandem mass tags (TMT)39 and isobaric tags for relative and absolute quantification (iTRAQ),40 has been established as an efficient approach for proteomic quantification. It exploits a multiplexed set of isobaric N-hydroxy succinimide (NHS)-activated derivatives, consisting of a reporter and a mass balance group, to modify primary amino groups in proteolytic peptide mixtures. As a result, differentially labeled peptides coelute in chromatographic separation and appear as single peaks in MS scans. Relative quantitative information of isobaric-labeled peptides is obtained in MS/MS scans by dissociating the reporter groups as distinct isotopeencoded fragments. For example, each of the six available TMT tag leads to a protein/peptide mass shift of 229.1629 Da; upon CID or high-energy collision dissociation (HCD) activation, reporter ions ranging from 126.13 to 131.14 Da are released. Isobaric tag techniques have been widely combined with bottom-up approaches for accurate quantification of proteins at a global scale. Although most associated with peptide labeling, this labeling strategy was also applied to label intact proteins;41 43 notably, in these studies, protein identification as well as quantification were performed at the level of proteolytic peptides, thus in so-called “semi-top-down” approaches. In the present study, we took advantages of isobaric mass tags used in peptide quantification and established a top-down approach to simultaneously identify and quantify intact proteins by LTQ-Orbitrap Velos ETD MS. The developed method comprises labeling of intact proteins by TMT, followed by separation via ion-pair reversed-phase liquid chromatography (IP-RPLC) using polymer monolithic columns coupled online to the MS, and quantitative analysis via HCD activation as well as qualitative analysis via ETD activation. The accuracy as well as the quantitative dynamic range of the developed method were assessed, and a best practice to perform online top-down sequencing on an LTQ-Orbitrap Velos ETD MS was proposed.
erythrocytes), ovalbumin (chicken egg white), serum albumin (bovine), transferrin (bovine), acetonitrile, trifluoroacetic acid (TFA), formic acid, triethylammonium bicarbonate (TEAB), tris(2-carboxyethyl)phosphine hydrochloride (TCEP), dimethyl sulfoxide (DMSO), 50% (w/w) hydroxylamine, and iodoacetamide (IAA) were from Sigma-Aldrich (St. Louis, MO, U.S.A.). The six-plex tandem mass tags reagent kit was from Thermo Fisher Scientific (Bremen, Germany). Water used for all experiments was purified by an arium611VF System (Sartorius, G€ottingen, Germany). Labeling of Intact Proteins with TMT Six-Plex. Proteins were dissolved separately in 50 mM TEAB containing 2.5% DMSO to a final concentration of 1 μg/μL. Protein samples (except for myoglobin) were reduced with 200 mM TCEP for 1 h at 55 °C and alkylated with 375 mM IAA for 30 min in the dark at room temperature. Subsequently, 20 μg of proteins was labeled for 1 h at room temperature by adding 5 μL of the TMT reagent (0.8 mg vial redissolved in 41 μL of DMSO). The labeling reaction was then quenched by the addition of 2 μL of a 5% hydroxylamine for 15 min at room temperature. The TMTlabeled proteins were stored at 80 °C until used. For direct infusion MS experiments, TMT-labeled proteins were acidified by 2% formic acid to pH < 2. C4 ZipTip pipet tips (Millipore, Billerica, MA) for desalting and concentrating of proteins were performed following producer guidelines. Proteins were eluted with 0.1% formic acid in 50% acetonitrile. Liquid Chromatography. The monolithic 50 mm 1.0 mm i.d. ProSwift RP PS-DVB column was purchased from Dionex (Amsterdam, The Netherlands). Separations in monoliths were carried out with an integrated micro-HPLC system (model UltiMate3000, Thermo Fisher Scientific). A UV detector with a 45 nL z-shaped capillary detection cell (model UltiMate, Dionex Benelux) was utilized for UV detection at a wavelength of 214 nm. Mobile phase A was 0.05% formic acid and 0.02% TFA in water, and mobile phase B was 0.05% formic acid and 0.04% TFA in acetonitrile. The LC gradient was held at initial conditions of 20% B for 5 min followed by a ramp to 40% B over 30 min; 95% B was reached over the next 1 min and held for an additional 5 min before re-equilibrating at 5% B. The flow rate was 60 μL/min; the column temperature was 60 °C. Mass Spectrometry. MS analysis was performed on an LTQOrbitrap Velos mass spectrometer (Thermo Fisher Scientific) with ETD utility. For online measurement, the HESI-II source equipped with a 34 gauge (0.178 mm o.d., 0.076 mm i.d.) stainless steel needle (Thermo Fisher Scientific, Bremen, Germany) was used to interface the 50 mm 1.0 mm monolithic column to the MS. The instrument was operated in positive ESI mode with a sheath gas flow of 30 units, an auxiliary gas flow of 5 units, a spray voltage of 4.5 kV, and a capillary temperature of 275 °C, and was tuned and calibrated according to manufacturer’s instructions before measurements. The automatic gain control (AGC) targets of the ion trap were set to 3 104 for full MS scan, 1 104 for MSn scan, and those of the Orbitrap were set to 1 106 for full MS scan, 1 105 for MSn scan. The microscan number and maximum ion injection time of the ion trap was set to 5 and 50 ms for full MS scan, 20 and 25 ms for MSn scan, and that of the Orbitrap was set to 2 and 500 ms for full MS scan, 2 and 1000 ms for MSn scan. For offline measurements, a nanospray ion source (Upchurch Scientific, Oak Harbor, WA) equipped with an offline PicoTip (New Objective, Woburn, MA) was coupled to the mass spectrometer, no sheath gas and no auxiliary gas was applied, and the spray voltage was set to 1.0 kV.
’ EXPERIMENTAL SECTION Chemicals and Reagents. Cytochrome c (horse heart), myoglobin (horse skeletal muscle), carbonic anhydrase (bovine 162
dx.doi.org/10.1021/ac202243r |Anal. Chem. 2012, 84, 161–170
Analytical Chemistry
ARTICLE
The mass spectrometer was operated in the data-dependent mode to switch automatically between full-MS (scan 1), HCDMS2 (scan 2), and ETD-MS2 (scan 3). After a full-MS scan acquired in the ion trap MS, the most abundant protein ion (top 1) was selected for an HCD-MS2 scan and an ETD-MS2 scan. The isolation width was set to 3 m/z (setting in the instrument software; all isotopic peaks of the precursor are included in this window); the minimum ion threshold counts was set to 4500. The normalized collision energy (NCE) was set at 80% in HCDMS2. The reaction time in all MS2 experiments was set to 10 ms. RAW data files were processed manually, obtaining precursor m/z values from the MS/MS spectrum header and importing the reporter ion value into Excel table. Before quantification, the reporter ion masses were corrected for isotopic overlap using a correction matrix;44,45 for details see the Supporting Information.
an enhanced reactivity of threonine toward NHS-activated carboxyl groups in the presence of histidine residues, catalyzed by the imidazole group via hydrogen bonding with the hydroxyl oxygen at the threonine. This nonspecific labeling with TMT tags on threonine residue may be hydrolyzed upon enzymatic digestion, since signals of these nonspecifically labeled peptides were significantly weaker than those of unlabeled ones. Nonspecific labeling with TMT tags on other residues is, however, not excluded. The unexpected additional threonine-directed labeling pattern was observed repeatedly and exclusively for myoglobin but not at the other proteins investigated in this study. When TMT was replaced by iTRAQ, which shares the same NHS-activation, similar effects were observed for myoglobin (data not shown). The TMT-labeling pattern of the other proteins in this study (i.e., cytochrome c, carbonic anhydrase, superoxide dismutase, and ubiquitin) obtained under the same derivatization conditions were complete, except for carbonic anhydrase, for which one of its Lys residues (out of 18 totals) was only partially labeled with TMT. A frequently observed side product was oxidation of methionine residues, which probably occurred during the sample reduction and alkylation process; potentially, labeling in a nitrogen atmosphere can reduce this unwanted oxidation in future experiments. Overall, these results indicate that a complete TMT labeling of intact proteins is possible. Nevertheless, the occurrence of side products as nonspecific labeling and/or incomplete derivatization may not only increase the sample complexity but also lead to a dilution of the analyte, which in turn has a negative effect on the detectability of the analytes. Therefore, in further studies, an optimization of the reaction conditions should be performed. In this respect, factors like (i) the choice of buffer systems and chemicals to denature the proteins in complex mixtures, preventing the aggregation and simultaneously increasing the accessibility of protein labeling sites, (ii) the adjustment of molar ratios between labeling reagents and protein reactive sites and of reaction times in order to minimize formation of unwanted side products, including nonspecific and incomplete labeling, and (iii) treatment of crude reaction mixtures, e.g., by change of pH in order to hydrolyze nonspecific labeling products, have to be investigated. Intact Protein Quantification. In contrast to bottom-up approaches, in which protein identification and quantification are obtained from proteolytic peptides, we examined the possibility to identify and quantify TMT-labeled proteins via a topdown approach. Due to the ability to produce rich “triple-quadrupole-like” fragment patterns including fragments in the low m/z range, HCD as available in LTQ-Orbitrap XL or LTQ-Orbitrap Velos hybrid instruments has been frequently applied to fragment and quantify isobaric tagged (e.g., iTRAQ and TMT) peptides with high efficiency and accuracy. Further, high-resolution and mass accuracy HCD spectra are particularly beneficial for unambiguous assignment of protein/peptide fragments and improve the confidence of protein/peptide identification as well as de novo sequencing.47 We evaluated the applicability of HCD to quantify TMT-labeled intact proteins. As example, an [M + 21H]21+ charged protein signal (m/z at 1048.3 Da, Figure S1C, Supporting Information) from 22-fold TMT-labeled myoglobin was chosen as precursor to be fragmented by HCD in an LTQOrbitrap Velos MS. As shown in Figure 1A, the HCD spectrum shows intensive fragments in the low m/z region even at the
’ RESULTS AND DISCUSSION TMT Intact Protein Labeling. Specificity, efficiency (yield of product), and reproducibility are three major parameters to evaluate the performance of a derivatization reaction. To assess the performance of the TMT derivatization on intact proteins, myoglobin, a protein containing 20 theoretical TMT derivatization sites (i.e., the N-terminus plus 19 ε-amino groups of lysine residues), was chosen as example. Six equal amounts of myoglobin samples were labeled separately with each one of the six TMT reagents, and the reaction was stopped by the addition of hydroxylamine. The six pooled TMT-labeled myoglobin samples were desalted on C4 solid phase and analyzed by direct infusion via a PicoTip to an LTQ-Orbitrap Velos MS. Figure S1 (Supporting Information) shows the MS spectra of myoglobin in its native state (Figure S1A) and after TMT labeling (Figure S1B) acquired in the ion trap MS. In the native state, only one protein series was observed in the MS spectrum, indicating the high purity and homogeneity of myoglobin. The calculated averaged molecular weight of myoglobin (16 951.2 ( 1.7 Da) was in excellent agreement with the theoretical one (16 951.2682 Da), which means that there are no additional modifications present. In the case of complete labeling, a mass shift of 4583.3 Da, corresponding to 20-fold TMT tags (the mass shift of each tag is 229.1629 Da), on myoglobin is expected. Surprisingly, the MS spectrum of TMT-labeled myoglobin indicated the presence of a mixture of several components. The major component exhibited a mass shift of 5041.8 Da, which corresponds to a labeling with 22-fold TMT tags. Additionally, products corresponding to a 20-fold, 21-fold, and 23-fold TMT (sometimes also weak signals for 24-fold TMT) were observed with reduced intensities. On the other hand, signals corresponding to a myoglobin carrying less than 20-fold TMT tags were not observed. In order to elucidate which residues are responsible for the additional nonspecific labeling, an in-solution chymotrypsin digestion on TMT-labeled myoglobin was performed, and the resulting peptides were analyzed by matrix-assisted laser desorption ionization time-of-flight/time-of-flight (MALDI TOF/ TOF) MS on an AB SCIEX TOF/TOF 5800 mass spectrometer (experimental details, see the Supporting Information). The results demonstrated that threonine residues, particularly those located one or two positions next to a histidine residue, such as His-Xxx-Thr or His-Xxx-Xxx-Thr or Thr-Xxx-His (where Xxx = any amino acid), were additionally modified by the TMT agents (Figure S2, Supporting Information). This observation is in good agreement with earlier findings of Miller et al.,46 who described 163
dx.doi.org/10.1021/ac202243r |Anal. Chem. 2012, 84, 161–170
Analytical Chemistry
ARTICLE
Figure 1. HCD activation of myoglobin labeled with six TMT reagents, mixed in a molar ratio of 1:1:1:1:1:1 (126:127:128:129:130:131). (A) Fragment spectrum of the [M + 21H]21+ charge state of 22-fold TMT-labeled myoglobin acquired in the Orbitrap at 60 000 resolution and an NCE of 30%. (B) Zoom-in of the reporter ion region. (C) Optimization of the NCE for optimized reporter ion intensities. (D) Reporter ion ratio among 21-fold, 22-fold, and 23-fold TMT-labeled myoglobin.
relative low NCE, such as NCE of 30%. In addition to the major fragments of TMT reporter ions at m/z of 126.13 131.14 (zoom-in in Figure 1B), few frequently observed immonium ions (such as m/z at 110.07 from histidine and 159.09 from tryptophan), several protein fragments in the N-terminal region (e.g., b4+ ion at m/z 602.34 and b6+ ion at m/z 788.40), and the intact TMT tag (m/z at 230.17) were observed with an average mass deviation of 1.23 ppm. These results indicate that the majority of the protein ions were dissociated under low NCE. We optimized the NCE to generate the most intense reporter ions, as demonstrated in the Figure 1C. At NCE of 20% or less, no TMT reporter ions could be detected, but clear and interpretable protein fragmentation patterns were observed in the HCD spectrum (see Figure 3 for details). With increasing NCE, protein molecules were further fragmented to generate reporter ions (at NCE of 25% or above), and intensities of reporter ion signals increased and reached a maximum at NCE of 30%. It is worth noting that the observed reporter ion distribution exhibits a nice correspondence to the expected ratio 1:1:1:1:1:1 (126:127:128:129:130:131)
and that the ratio remained reliably conserved independent of the NCE applied (up to NCE of 90%). Furthermore, we compared reporter ion ratios among 21-, 22-, and 23-fold TMT-labeled myoglobins (shown in Figure 1D) and among different charge states (data not shown). Not surprisingly, no considerable discrepancies could be observed, since the reactivity of all six TMT tags is identical as they share the same chemical activation. An important issue for the applicability of a quantification method is its dynamic range. For this purpose, aliquots of myoglobin were differentially labeled with TMT-126 and TMT-127 tags and subsequently mixed in ratios ranging from 1:10 to 10:1 (triplicates were prepared for each ratio). The samples were then diluted with 1% formic acid to a final concentration of 4 fmol/μL followed by 1 μL of sample injection to a monolithic LC column online coupled to an LTQ-Orbitrap Velos MS. The HPLC separation and the MS detection were carried out as described in the Experimental Section, but the gradient time was shortened to 10 min, since only a single protein was applied on the column. Remarkably, myoglobin labeled with different numbers of TMT 164
dx.doi.org/10.1021/ac202243r |Anal. Chem. 2012, 84, 161–170
Analytical Chemistry
ARTICLE
As described before, the different products of TMT labeling (i.e., additional or incomplete TMT labeling) did not show significant differences on the determination of protein ratios. Therefore, a complete chromatographic separation of these related proteins is in principle not essential. However, if two or more different protein signal envelopes overlap in the mass spectrum, either the protein ratio of the less abundant ones would not be determined (because only the base peak would be chosen as precursor to be fragmented), or the determined protein ratio would be significantly biased (when the selected precursor is a mixture of multiple components). Hence, highly efficient front-end separations are prerequisites to prevent such overlapping signals and to identify and quantify proteins in complex mixtures accurately using this top-down approach. In view of several superior advantages over traditional packed columns, including high resolution, fast separation, high recoveries, and low protein carryover,26 a polymer-based monolithic reversed-phase column was employed herein to separate protein mixture. As shown in the Supporting Information, Figure S4 , more than 10 protein components can be recognized in the base peak chromatogram, indicating a very efficient separation of TMT-labeled protein mixtures on a polymer-based monolithic column. Furthermore, the results also indicated a sufficient acquisition rate of ion trap MS to record the chromatographic profile of a monolithic column properly, and an adequate data quality to determine molecular masses of proteins unambiguously, particularly those smaller than 35 kDa. It should be noted that the MS spectra have been recalibrated offline using a poly(dimethylcyclosiloxane) (PCM) background ion at m/z 684.20295 (ammonium adduct of [C2H6SiO]9) as internal calibrant, and the protein molecular masses were determined by averaging the whole protein envelope. Proteins with molecular masses larger than 35 kDa were found to be difficult to determine confidently after TMT labeling, due to the increased heterogeneity of sample resulting from incomplete labeling and/ or unwanted side products. Furthermore, the reduced sensitivity to detect high-mass species (above 25 kDa) by electrospraybased MS was reported by Compton et al.,48 resulting from massdependent increases in isotopic peaks, charge states, and chemical noise present in protein mass spectra. The detection of larger proteins (e.g., larger than 40 kDa) usually requires longer acquisition time and lots of spectra averaging, and is typically achieved with an offline analysis. In consequence, the developed method for simultaneous intact protein identification and quantification on a chromatographic scale will to date be restricted to analyze proteins smaller than 35 kDa. Nevertheless, relative quantification using HCD fragmentation can still be performed for proteins with masses above 35 kDa; in such cases, the identification and characterization has to be performed in alternative experiments, e.g., after protein digestion. The ratio of 10 protein components displayed in the Supporting Information, Figure S4, including six selected proteins and four impurities originating from carbonic anhydrase and ovalbumin, have been determined, and the results are summarized in the Table 1. The results reveal a very good correlation between the expected protein ratios and the measured ones. The accuracy of the TMT ratios averaged 10.6% relative error, and the relative variability (measure of precision) of averaged 6.1% RSD; these data were obtained from three replicates (Table 1). An elevated standard deviation is observed in proteins whose protein masses cannot be determined confidently, such as ovalbumin, ovomucoid, BSA, and apo-transferrin. This amplified measurement
Figure 2. Linear dynamic range between measured relative abundance of reporter ions (127:126) and expected ones of TMT-labeled myoglobin mixed at different molar ratios.
tags was found to be slightly separated in the monolithic column, and the retention time of TMT-labeled proteins decreases with the increased numbers of TMT labeling. As shown in Figure S3 (Supporting Information), a typical peak width (at base) of a TMT-labeled protein that eluted from the monolithic column was around 30 s. Depending on our scan settings, ca. 6 8 full-MS spectra (0.54 s per spectrum), 6 8 HCD spectra (0.48 s per spectrum), and 6 8 ETD spectra (3.18 s per spectrum) can be acquired during this time span. The full-MS spectra acquired in the ion trap MS were used to obtain the intact protein masses as well as to select a precursor ion (the most abundant ion) for further HCD and ETD fragmentation. While HCD spectra provide protein quantification information, protein sequence can be determined from the ETD spectra. For protein quantification, all HCD spectra were summed together and the reporter ion intensities were exported to Excel for calculation of the relative ratios of reporter ions. Figure 2 displays the correlation between the measured relative abundance of reporter ions (TMT-127 and TMT-126) and the expected ones of TMT-labeled myoglobin at different mix ratios. The results reveal that TMT tags provide an accurate representation of abundance ratios of proteins in the mixtures and that the reporter ions show very nice linear behavior (slope of 1.12 and R2 of 0.9995) over the entire range of protein ratios tested. These data suggest that intact protein quantification can be easily achieved by direct fragmentation of TMT-labeled proteins by HCD activation in an LTQ-Orbitrap Velos MS. In summary, this method delivers data with high sensitivity, excellent signal-to-noise ratio (S/N), and over a broad dynamic range sufficient for relative protein quantification as required in proteome analysis. In a further experiment, we applied the developed method to quantify relative abundances of proteins in a six-protein mixture. Six equal amounts of selected proteins, with protein masses ranging from 12 to 78 kDa, were labeled separately with six different TMT tags. After TMT labeling, each protein was mixed with a distinct TMT ratio (i.e., cytochrome c exhibits a ratio of 2:1:1:1:1:1, myoglobin of 1:2:1:1:1:1, carbonic anhydrase of 1:1:2:1:1:1, ovalbumin of 1:1:1:2:1:1, BSA of 1:1:1:1:2:1, and apo-transferrin of 1:1:1:1:1:2) before combining together. An amount of 5 μL of the sample (final protein concentration ranges from 0.16 to 0.5 μM) was then injected onto a monolithic column and analyzed by LTQ-Orbitrap Velos MS. The protein separation/ identification and quantification results are summarized in Figure S4 (Supporting Information) and Table 1, respectively. 165
dx.doi.org/10.1021/ac202243r |Anal. Chem. 2012, 84, 161–170
Analytical Chemistry
ARTICLE
Table 1. Results of Protein Quantification for a Six-Protein Mixturea theor MW after
final concn
ratio (av) ( SD
ID with
(kDa)
and Lys
TMT labeling (kDa)
(μM)
[R126/131:R127/131:R128/131:R129/131:R130/131]
ETD spectra
12.3
2C/19K
16.8
0.34
theor MW no. of Cys protein name cytochrome c (equine)
myoglobin (equine)
16.9
carbonic anhydrase (bovine)
29.1
carbonic anhydrase impurity 1
8.5
0C/19K
0C/18K
0C/7K
21.5
0.34
33.2
0.16
10.1
ubiquitin, partial sequence (bovine) carbonic anhydrase impurity 2 superoxide dismutase (bovine)
15.5
ovalbumin (gallus)
ovalbumin impurity 1
3C/10K
17.9
42.8
6C/20K
47.7
20.1
18C/13K
24.3
0.34
ovomucoid ovalbumin impurity 2
3.8
2C/1K
4.2
c
C-terminal sequence of ovalbumin
BSA (bovine)
apo-transferrin (bovine)
66.6
77.7
34C/60K
38C/64K
82.3
0.34
94.5
0.5
expected ratio
2.00:1.00:1.00:1.00:1.00
measd ratio (av)
2.32:1.24:1.12:1.04:1.22
(SD
0.04:0.01:0.02:0.01:0.04
expected ratio
1.00:2.00:1.00:1.00:1.00
measd ratio (av)
0.82:2.04:0.91:1.02:0.99
(SD
0.03:0.01:0.00:0.02:0.01
expected ratio
1.00:1.00:2.00:1.00:1.00
measd ratio (av)
0.99:0.96:1.99:0.90:1.08
(SD
0.04:0.04:0.09:0.04:0.03
expected ratio
1.00:1.00:2.00:1.00:1.00
measd ratio (av)
1.08:1.00:2.11:0.94:1.05
(SD
0.01:0.04:0.05:0.01:0.02
expected ratio measd ratio (av)
1.00:1.00:2.00:1.00:1.00 1.09:0.98:2.01:0.92:1.05
(SD
0.08:0.08:0.04:0.04:0.09
expected ratio measd ratio (av)
1.00:1.00:1.00:2.00:1.00 1.19:1.22:1.25:2.21:1.15
(SD
0.13:0.06:0.07:0.19:0.19
expected ratio
1.00:1.00:1.00:2.00:1.00
measd ratio (av)
1.00:0.97:1.12:2.32:0.99
(SD
0.15:0.16:0.23:0.30:0.14
expected ratio
1.00:1.00:1.00:2.00:1.00
measd ratio (av)
1.13:1.19:1.17:2.29:0.88
(SD
0.11:0.06:0.03:0.03:0.12
expected ratio
1.00:1.00:1.00:1.00:2.00
measd ratio (av)
1.17:1.21:1.24:1.14:2.27
(SD
0.04:0.09:0.06:0.07:0.05
expected ratio
0.50:0.50:0.50:0.50:0.50
measd ratio (av)
0.79:0.65:0.73:0.74:0.89
(SD
0.08:0.07:0.09:0.02:0.05
yes
yes
yes
yes
yes
nob
nob
yesd
nob
nob
a Ratios were calculated relative to the peak height at m/z 131.14. b Quantification depending on retention time. c C-terminal sequence of ovalbumin— ASVSEEFRADHPFLFCIKHIATNAVLFFGRCVSP. d ID with both CID and ETD spectra.
error of these proteins is expected due to the limited HCD spectra qualities that resulted from the weak precursor ion intensities. The identities of the recognized impurities have been confirmed by de novo sequencing depending on the protein molecular mass and the sequence tags determined from the ETD or CID spectra. Top-Down Protein Identification. Together with the determination of relative abundances of proteins at the intact protein level, the method was also tested for its potential for simultaneous top-down protein identification by (partial) sequencing at an LC time scale. Taking into account the LC peak width in relation to the mass analyzer scan speed, the MS/MS spectra quality, and the extractable protein sequence information, we propose here two measurement protocols that are feasible to perform simultaneous top-down sequencing and relative quantification on an LTQ-Orbitrap Velos MS with or without ETD. The first protocol employs HCD to dissociate intact proteins. It should be noted that, in contrast to the LTQ-Orbitrap classic,
the LTQ-Orbitrap Velos with ETD applied in this study is equipped with an HCD collision cell using an axial extraction field, which improves the ion transmission by about 3-fold, allowing a more robust and precise quantification in MS/MSbased quantification experiments. As already mentioned above, intact proteins can be readily fragmented in the HCD cell under low NCE. Figure 3 shows the single-scan MS/MS spectrum of 22-fold TMT-labeled myoglobin upon isolation of its [M + 21H]21+ charge state (m/z at 1048.3 Da) and HCD fragmentation at NCE of 20% (Figure 3A) and 25% (Figure 3B), respectively. The spectra were acquired in the Orbitrap MS at 60 000 resolution at m/z 400, and each scan consisted of four microscans. At NCE of 20%, protein molecules were dissociated primarily at the N-terminal end; the majority of fragment ions observed were complementary b- and y-ion pairs. For example, ions corresponding to y14419+, y14319+, and y14019+ were present in the HCD spectrum together with their complements, b92+, b102+, and b132+, respectively. The region below m/z 400 contained 166
dx.doi.org/10.1021/ac202243r |Anal. Chem. 2012, 84, 161–170
Analytical Chemistry
ARTICLE
Figure 3. Single-scan HCD fragmentation spectrum of the [M + 21H]21+ charge state (m/z at 1048.3 Da) of 22-fold TMT-labeled myoglobin acquired in the Orbitrap MS at resolution of 60 000 and (A) NCE of 20% or (B) 25%. (C) Expansion of the low-mass reporter ions region acquired at the NCE of 20% and 25%, respectively. The expected reporter ion ratio is 1:1:1:1:1:1 (126:127:128:129:130:131). (D) Myoglobin sequence with the observed b- and y-type ions are indicated by angled lines above and below the sequence, respectively.
numerous signals of immonium and internal fragment ions. At NCE values up to 25%, highly charged ions were significantly reduced; in consequence, more fragment ions with low m/z values derived from the C-terminal of the protein were detected. At the same time, TMT reporter ions released from the protein molecules were observed as minor signals. At NCE values of 30% and above, a dramatic loss of protein fragments (sequence information as well) occurred and the whole HCD spectrum is dominated by reporter ions. In contrast, CID is a more moderate fragmentation method compared to HCD. Fragmentation patterns of CID at NCE of 50% were found to be very similar to that of HCD at NCE of 20%, but more internal fragments were observed; the low m/z region in the CID spectra was underrepresented (data not shown). Therefore, HCD activation is superior for the simultaneous relative quantification and identification of proteins. On the other hand, CID is superior than HCD for intact protein characterization, since stepwise protein sequence mining can be accomplished by performing CID-MSn experiments on an LTQOrbitrap.49 Summing-up, HCD fragmentation at NCE of 25% provides not only the most protein sequence information for
confident protein identification but also sufficient reporter ion signal intensities for accurate protein quantification. We also found that the spectra quality (S/N ratios) can be improved by increasing the number of microscans, however, at the cost of increased total scan times. By manual inspection of the singlescan HCD fragmentation spectrum acquired with the accumulation four microscans, fragment ions exhibiting up to +19 charged state were found to be well-resolved, implying that a good quality (high mass accuracy and high resolution) fragmentation spectrum with sufficient extractable sequence information can be obtained at this relative short scanning time (ca. 2.4 s per scan), making top-down protein sequencing at a chromatographic time scale achievable. The second measurement protocol involves the use of ETD activation. ETD is a powerful fragmentation method that offers several advantages over conventional collision-activated dissociation (i.e., CID or HCD) in analysis of proteins and peptides, such as increased random cleavage along the protein/peptide backbone and preservation of otherwise labile PTMs.19 As ETD fragmentation takes place on a millisecond time scale, it is suitable for online LC MS experiments.19,50 167
dx.doi.org/10.1021/ac202243r |Anal. Chem. 2012, 84, 161–170
Analytical Chemistry Compared to peptides, where typically ETD reaction times of about 100 ms are applied, reduced reaction times were suggested for efficient protein ETD.51 However, too short or prolonged reaction times will lead to significant loss of sequence information. Too short reaction times lead to the formation of chargereduced species that compete with the production of c- and z-type ions; the recognition of the latter ions in the high-mass range (mass range above precursor ion) is hampered due to the severe peak overlay with the charge-reduced species and ions resulting from neutral losses. On the other hand, too long reaction times produce merely singly (and few doubly) charged ions, as reported by Bunger et al.;20 however, owing to the instrumental setup (normal mass scan range, 50 2000 Da), information about fragments with higher m/z values is lost in this case. This may lead to a reduction of the overall sequence coverage since only fragments smaller than 2000 (m/z) can be detected in the normal scan m/z range; note that it is also possible to apply an extended mass range with the instrumentation used here, ranging from m/z 100 4000, but the spectra quality is reduced compared to the normal mass scan range. In principle, the optimum ETD reaction time has to be adapted for each protein: the larger the proteins, the shorter the reaction time. For small proteins or large peptides ( 1000 Da). McAlister et al.52 and Mazur et al.31 have demonstrated the implementation of high mass accuracy and high-resolution Orbitrap MS to improve the recognition of highly charged species in the ETD spectra and thus the confidence of protein identification. Related topics to implement improved ETD spectra into our workflow are presently under investigation. Nevertheless, according to the known sequence of myoglobin, we are able to manually interpret 31 fragment ions, representing around 45% of protein sequence coverage. In contrast to CID/HCD, where backbone cleavage is strongly affected by side chain chemistry, ETD provides more uniform backbone cleavage, particularly at the protein terminal ends. With the use of sequence tags determined from the protein termini and the parent protein mass, intact proteins in our test mixture could be unambiguously identified. Moreover, characteristic reporter ions were observed in the ETD spectra (see insert in Figure S5A, Supporting Information). In contrast to the reporter ions generated by HCD at the m/z of 126 131 Da, ETD generates reporters at m/z of 114 119 Da.53 Resulting from a cleavage of the N Cα
ARTICLE
bond of the TMT tags, tags 126 and 127 produce an identical reporter ion at m/z of 114.1277 (12C71H1614N1), tags 128 and 129 produce an identical reporter ion at m/z of 116.1341 (12C513C21H1614N1), tag 130 produces a reporter ion at m/z of 118.1411 (12C313C41H1614N1), and tag 131 produces a reporter ion at m/z of 119.1411 (12C313C41H1615N1). This indicated that ETD can also be used for protein quantification; however, only four channels can be employed. In summary, top-down sequencing applying the two fragmentation methods based either on HCD or ETD activation are complementary, rather than competing. The complementary nature of the product ions form by HCD (b- and y-type ions) versus ETD (predominantly c- and z-type ions) allows the identities of these ions to be readily assigned. Further, we noticed that the assignment of the N-terminal fragments was facilitated in TMTlabeled proteins, as the signal intensities were elevated compared to the unlabeled forms (data not shown). Paring up HCD and ETD spectra from the same precursor should provide more protein sequence information. However, improved software tools are needed to efficiently extract the diagnostic information in the HCD and ETD spectra to improve the performance and throughput of top-down sequencing.
’ CONCLUSION The analytical approach presented here is an initial evaluation of a potential workflow for top-down qualitative and quantitative analysis of TMT-labeled proteins employing LTQ-Orbitrap Velos ETD MS. We demonstrated that direct quantification of intact proteins labeled with isobaric TMT tags can be readily achieved via HCD activation. The reporter ions liberated from labeled proteins show consistent properties over a wide dynamic range, and the charge state of TMT-labeled proteins does not affect the ability to detect the abundance ratios of protein pairs. Moreover, protein sequence tags generated either by low-energy HCD or ETD activation along with the intact protein masses information promise for confident identification of small proteins
