24-Hour Lock Mass Protection Kimberly A. Lee,* Chris Farnsworth, Wen Yu, and Leo E. Bonilla Department of Molecular Sciences, Amgen Inc., Seattle, Washington 98119, United States Received July 27, 2010
Abstract: The positive role and application of highly accurate mass measurements in proteomics is well documented. The new generation of hybrid FTMS and Q-TOF instruments, including the LTQ-Orbitrap (OT), is remarkable in their ability to routinely produce single-digit to subppm statistical mass accuracy while maintaining high analytical sensitivity. The use of mass calibrants (lock masses) to reduce the systematic error of mass-to-charge measurements has also been reported and, in some cases, incorporated in the instrument control software by the instrument manufacturers. We evaluated the use of one such calibrant in the OT (e.g., polydimethylcyclosiloxane, PCM) to study its impact on the rate of phosphopeptide annotation and found it to lack robustness under normal laboratory conditions. Therefore, we devised a strategy to improve its performance by increasing the external abundance of calibrant molecules in laboratory air. This resulted in a more robust performance of the preprogrammed lock mass recalibration feature as evidenced by improvements in both statistical mass accuracy and peptide annotation rates. Keywords: lock mass • LTQ-Orbitrap • mass accuracy • phosphoproteomics
Introduction Current mass spectrometers, particularly hybrid FTMS instruments including the LTQ-Orbitrap (OT),1 are able to measure ion mass-to-charge ratios (m/z) with exquisit accuracy and still provide the sensitivity necessary for analysis of proteomics samples (i.e., low fmol to amol). The advantages of being able to specify low tolerance values for the maximum allowable mass deviation (MMD) during and after the database retrieval of peptide sequences, including better and more confident sequence assignments, have been reviewed before both for the general proteomics case1,2 as well as for the specific cases of improved assignment of post-translational modifications (PTMs)3-5 and “middle down” mass spectrometry.6 In the case of global phosphoprotein identification and phosphorylation site analysis by MS, accurate mass measurement is indispensable to enable narrow-tolerance MMD database searches and/or filtering of search results. This results in the correct retrieval of more peptide sequences, even from the relatively low abundance phosphopeptides with their charac* To whom correspondence should be addressed. E-mail: kalee@ amgen.com.
880 Journal of Proteome Research 2011, 10, 880–885 Published on Web 12/06/2010
teristically lower quality MS/MS spectra (due to the neutral loss (NL) of the phosphate moiety), which need to be matched against combinatorially enlarged databases containing all permutations of possible phosphorylation sites. To make further improvements to the accuracy of the massto-charge (m/z) measurements obtained in the OT device, Olsen and his colleagues demonstrated the use of lock mass injection via the C-trap to obtain subparts per million (ppm) measurements;7 a turn-key implementation of their approach, it has been further facilitated by the manufacturer of the OT with the incorporation into the instrument control software of a “lock mass feature”. The lock mass feature takes advantage of a mass calibrant that is normally present in ambient air and routinely detected during electrospray ionization (ESI). The calibrant, polydimethylcyclosiloxane (PCM), produces ions that are readily detectable and useful for the internal real-time recalibration of the mass axis both in the MS (m/z ) 445.120025) and MS/MS (m/z ) 429.088735) modes.5,7 Unfortunately, in its present form, this lock mass feature also suffers from unreliable performance, especially in clean environments (such as most analytical/mass spec laboratories!) with low environmental levels of PCM. Under these circumstances, the lock mass feature often operates incorrectly and introduces a deleterious bias in the real-time recalibration process, culminating in data sets with larger mass deviations relative to similar sets acquired with the lock mass feature disabled. In this technical note, we report on a “24-hour protection” method against lock mass feature failure. It involves simply boosting the observed PCM background signal to improve the operation of the lock mass feature during data acquisition. Clever in its simplicity and ease of implementation, our method uses off-the-shelf reagents, involves a simple laboratory setup, avoids carry-over and sample cross-contamination, and is a true example of “open source” mass spectrometry!
Experimental Section Cell Culture and Peptide Preparation. Jurkat cells were cultured to a density of 106 cell/mL, washed and resuspended in PBS, and treated with 1 mM pervanadate and 50 ng/mL calyculin A at 37 °C for 20 min. Twenty-million cells were pelleted at 1000× g, 20 °C, for 5 min, and the pellets were resuspended in 2 mL lysis buffer; 9 M urea, 20 mM HEPES, pH 8.0.; sonicated to disrupt DNA; and the cell debris was removed by centrifugation at 20 000× g, 20 °C, for 30 min. The sample was then reduced with 4.5 mM DTT at 37 °C for 1 h, cooled to 20 °C, and alkylated with 10 mM iodoacetamide in the dark for 30 min. Following dilution to 2 M urea with 20 mM HEPES buffer, pH 8.0, trypsin (Princeton Separations) was added 1:50 10.1021/pr100780b
2011 American Chemical Society
technical notes
24-Hour Lock Mass Protection (wt/wt) to total protein, and the sample was incubated at 37 °C overnight. The digest was then acidified to 1% trifluoroacetic acid (TFA) and clarified by centrifugation 2000× g, 20 °C, for 5 min. The peptide fraction was enriched by solid phase extraction with a Sep-Pak C18 cartridge (Waters) as follows. After loading the digest, the cartridge was washed with 10 column volumes of 0.1% TFA, washed with 3 mL 5% CH3CN in 0.1% TFA, and eluted stepwise with 10, 15, 20, 25, 30, 35, and 40% CH3CN in 0.1% TFA. The pooled eluate was lyophilized for 48 h then frozen at -70 °C. Phosphopeptide Enrichment. For each sample, 9 mg Poros 20MC resin (Applied Biosystems, Foster City, CA) was suspended in 90 µL ethanol. Resin was washed three times with 200 µL 0.1% acetic acid, 50% CH3CN, three times with 200 µL 50 mM EDTA, and three times with 200 µL water. Resin was charged with 100 mM FeCl3, 10 mM HCl (3×, 200 µL) and washed with water followed by 1% acetic acid, 10% CH3CN (each 3× 200 µL). Resin was equilibrated with three 200 µL washes with sample buffer [50 mM glycine pH 2, 30% CH3CN], and peptides from 2 × 107 cell equivalents total were loaded in two 50 µL volumes with 30 s incubation each prior to centrifugation and removal of supernatant. Resin was washed with sample buffer; 100 mM NaCl, 1% acetic acid, 30% CH3CN; 0.1% acetic acid; and water, each 3 times with 200 µL. Phosphopeptides were eluted two times with 50 µL 7% NH4OH, with 30 s incubation prior to centrifugation and supernatant removal. Eluates were combined, dried by SpeedVac, and peptides were resuspended in 15 µL 0.1% TFA. Peptides were purified by STAGE tip (Proxeon, Odense, DK), dried by Speedvac, and resuspended in 12 µL 0.1% TFA. Lock Mass Standard Evaluation. Five potential sources of siloxane ion were evaluated: Irish Spring Deodorant Bath Bar, Original; Pert Plus Happy Medium, 2 in 1 Shampoo Plus Conditioner; Burt’s Bees Rosemary Mint Shampoo Bar with Oat Protein and Pro Vitamin B5; Flex Triple Action Moisturizing Shampoo, Balsam & Protein; and Lady Speed Stick Invisible Dry by Mennen Antiperspirant & Deodorant Stick, Powder Fresh (all from drugstore.com, Bellevue, WA). For evaluation, each candidate product was placed individually on a lab jack 15-20 cm from the LTQ-Orbitrap XL source while spraying 20% B at 400 nL/min. The Pert and Flex products were simply left uncapped; the Irish Spring and Burt’s Bees bars were removed from their packaging and placed on a Petri dish. Half of the deodorant stick was removed, divided into pieces of approximately 1 cm3, and placed in a 50 mL conical vial with the lid removed. The remainder of the stick was left in the original form, and dialed up for maximum ambient exposure. LC-MS/MS Analysis. All phosphopeptide samples were combined prior to analysis to eliminate variance due to sample preparation. Five microliters of phosphopeptide sample was loaded onto a 15 cm × 75 µm column packed in-house with Vydac C18TP 5 µm 300 Å resin. Using a two-column switching system, the loading column was washed with 2% CH3CN, 0.1% formic acid, 0.001% HFBA for 90 min at 500 nL/min, followed by elution with a four-phase linear gradient: 2-5% B over 2 min, 5-22% B over 38 min, 22-33% B over 10 min, 33-60% B over 10 min, with A ) water with 0.1% formic acid and B ) CH3CN with 0.1% formic acid delivered by an Eksigent nanoLC1D+ (Dublin, CA) at 400 nL/min. Nanospray was performed using a NewObjective (Woburn, MA) source with peptides refocused in a 5 cm packed tip, pulled and packed in-house. LC-MS analyses were performed using an LTQ-Orbitrap XL operated under Xcalibur 2.4 SP1 (ThermoFisher, San Jose, CA).
A survey scan of 400-2000 m/z was performed in the Orbitrap at 60 000 resolution with AGC target of 1 × 106 and 500 ms injection time followed by five data-dependent MS2 and MS3 scans performed in the LTQ linear ion trap with 1 microscan, 100 ms injection time, and 10 000 AGC. MS3 was specific to neutral loss of phosphate, triggered by loss of 32.7, 49, 65.4, and 98 m/z in the top three peaks, prioritized by mass. Dynamic exclusion was enabled, with repeat count of 1, exclusion duration 30 s with exclusion list of 500. Ions of charge state one were rejected for ms/ms. Database Searching and Analysis. Database searching was performed with Spectre, an in-house multiengine platform.8 Full tryptic digest was required with two missed cleavages allowed. Static modification of Cysteine (carbamidomethyl) was required and variable phosphorylation of Serine, Threonine, and Tyrsoine and oxidation of Methionine were allowed. Searches were performed against the IPI human database and a decoy database consisting of the reversed IPI human sequences. Precursor ion tolerance values were varied from 0.001-0.1 Da with fragment ion tolerance held at 0.7 Da.
Results The main motivation in our laboratory for undertaking a systematic evaluation of the OT lock mass feature was to understand its potential value for increasing the total number and quality of phosphopeptide identifications from our IMAC workflow. Therefore, we prepared a complex phosphopeptide standard mix consisting of proteolytically (trypsin) derived Jurkat peptides enriched by IMAC as described in the Experimental section above. Different IMAC preparations were mixed prior to LC-MS for homogeneity and four data-dependent LC-MS/MS runs were performed, two with the lock mass feature enabled and programmed to use the 445.120025 m/z PCM ion as the lock mass calibrant, and two with the lock mass feature not enabled. Figure 1A illustrates a comparison of the results from database searches performed repeatedly at varying Maximum allowable Mass Deviation (MMD) settings for the precursor ion mass. As can be seen, when it worked properly, using the lock mass feature resulted in data that yielded improved peptide annotation rates at every MMD setting tested (e.g., 25% increase at MMD 0.005 Da) with respect to the same data acquired with the lock mass feature not enabled (control); this is so, presumably, because enabling the lock mass feature produces higher accuracy measurements leading to better database hits. This is supported by the results shown in Figure 1B which display a clear global shift toward the zero line of the mass error distribution when going from data acquired with the lock mass feature not enabled (top left panel) to data acquired with it enabled (top right panel). However, as the bottom panel shows, we also found that often the use of the lock mass feature actually resulted in a global decrease in mass accuracy and deleterious bias of the distribution. Whenever this happened, we observed that the incorrect mass recalibration, together with a “Ufill” indication, was reported in the FT Analyzer Message for the corresponding MS scan header. Furthermore, we correlated the occurrence of this phenomenon to time periods when people were absent from the lab, particularly during analyses taking place overnight. Taken together, the previous observations led us to believe that the erroneous mass recalibration observed was due to low background levels of the lock mass calibrant ion (i.e., below the level necessary for the correct and consistent operation of the lock mass feature). As a result, we investigated a “24-hour Journal of Proteome Research • Vol. 10, No. 2, 2011 881
technical notes
Lee et al.
Figure 1. Lock mass analysis. We tested the lock mass feature, locking on the 445.120025 m/z PCM calibrant present in lab room air with our standard IMAC-enriched Jurkat phosphopeptide sample. (A) At any given MMD, and when the PCM calibrant ion is reliably detectable during acquisition (9), use of the lock mass results in improved mass accuracy and an increase in phosphopeptide data annotation rate relative to the same data acquired with the lock mass feature disabled (0). (B) When the lock mass feature is not in use (top left panel) the instrument’s native calibration offset is observed. Under optimal lock mass conditions (top right panel) this systemic offset is corrected to zero. When the PCM calibrant ion is low or absent (bottom panel), the instrument should, in theory, not apply any mass correction; however, we observe an unexplained biased and increased mass error under these conditions. Each point represents a single peptide assignment.
lock mass protection” scheme that involves simply boosting the observed PCM background signal in order to improve the reliability of operation of the lock mass feature during data acquisition. This method involves a simple laboratory setup (Figure 2) using common laboratory hardware and truly offthe-shelf reagents. Because human personal hygiene products are known to contain siloxane molecules, we acquired and tested 5 different products including a soap, a shampoo bar, two liquid shampoos, and a deodorant as sources for lock mass ions (see Experimental section for details). Lady Speed Stick (LSS) deodorant was the clear superior choice with ambient release of PCM molecules at levels approximately 50× baseline (Figure 882
Journal of Proteome Research • Vol. 10, No. 2, 2011
2b). The other products tested did not affect the baseline levels of observed PCM. To assess the PCM signal stability, the lock mass ion signal was evaluated over 24 h and over the course of the LC gradient by analyzing 20 fmol of BSA tryptic digests continuously with the LSS lock mass source present. The 445 m/z ion was found to be a stable and effective lock mass over the entire time period and throughout the gradient (data not shown). To evaluate the LSS-aided lock mass with phosphopeptide mixtures, we repeated the study described above and in Figure 1, this time using LSS as the external lock mass source of calibrant PCM ions. Once again, Jurkat peptides previously enriched by IMAC were pooled prior to LC-MS for homogene-
24-Hour Lock Mass Protection
technical notes
Figure 2. Apparatus for 24-h lock mass protection. (A) To address the problem of low PCM calibrant ion levels in room air, we devised an open source, off-the-shelf system for producing a continuous steady PCM background. After testing various personal hygiene products including shampoo, hair conditioner, and soap, we selected Lady Speed Stick deodorant (LSS) as the optimal source for PCM ions. (B) Placing the LSS on a lab jack near the instrument orifice resulted in significantly increased detectable 445 m/z PCM calibrant ion and excellent performance of the lock mass feature. Mass spectrum shown was acquired while spraying 20% B at 400 nL/min and represents signal averaged over 1 min. Journal of Proteome Research • Vol. 10, No. 2, 2011 883
technical notes
Lee et al.
Figure 3. Effect of “24-hour lock mass protection” on peptide mass error assignment. Mass error for each peptide assignment is plotted vs retention time for data collected without lock mass (red) and with lock mass 445.120025 (blue) both with (filled) and without (hollow) LSS calibrant. Squares and diamonds represent technical replicates. In the presence of the PCM calibrant, the peptide mass error distribution obtained using the lock mass is centered very close to 0 ppm.
ity and eight data-dependent LC-MS/MS analyses performed as follows: four runs in the absence of LSS (two with the lock mass feature not enabled and two using 445.120025 m/z as the lock mass ion), and four runs with the LSS present (two with the lock mass feature enabled and two without). As can be seen in Figure 3, the best results are obtained in the presence of LSS as attested by the tight distribution of mass errors around the 0 ppm line for the lock mass-enabled acquisition. In contrast, in the absence of LSS, the erroneous mass recalibration is again observed, and fewer MS2 spectra are acquired with the lock mass feature enabled. Both the reduced spectral count and the mass calibration are completely corrected by the LSS calibrant. The PCM ion does not interfere with the datadependent acquisition since it is singly charged and our acquisition methods actively exclude singly charged species from MS/MS analysis. Additionally, we evaluated the effect of the PCM ion on parent ion detection and found its addition led to a slight increase in the average signal-to-noise ratio (s/ n, calculated with QualBrowser in Xcalibur 2.1) among extracted ion chromatograms of 16 randomly selected lowintensity phosphopeptides (median 10% increased s/n, mean (calculated from log(s/n)) 36% increased s/n). Although these differences are not significant, they suggest that no degradation of MS dynamic range takes place in the presence of the LSS calibrant source.
Conclusions The implementation of the lock mass feature on the LTQOrbitrap mass spectrometer was first published in 2005 and was made available for routine use through the instrument control software by 2006. Use of this feature has the potential to bring the mass accuracy of the Orbitrap to within 1 ppm, on par with the LTQ-FTICR, with no duty cycle penalty.7 This improvement in mass accuracy is quite helpful for reliable peptide sequence annotation generally but is particularly valuable in the case of post-translationally modified peptides where searches with multiple variable modifications result in significant database amplification. Yet, most proteomics manuscripts published today using the LTQ-Orbitrap do not take advantage of the lock mass feature. 884
Journal of Proteome Research • Vol. 10, No. 2, 2011
The reason for this underutilization is clear from the data presented in Figures 1 and 3. In the absence of the LSS calibrant, more than half of the LC-MS/MS runs in our laboratory resulted in miscalibration, correlating with underfilling (Ufill) messages in the MS scan header. Each sample analyzed by LC-MS/MS in most proteomics laboratories represents days to weeks of work in preparation; some samples, particularly those of clinical origin, are indeed irreplaceable. These samples are too valuable to analyze using a feature (however useful) that is not totally reliable. We have demonstrated that introduction of a constant ambient source of the PCM calibrant eliminates Ufill messages and m/z miscalibration from the entire gradient elution time. Use of this technology (Figure 2) enables users to utilize the lock mass feature under all laboratory room-air conditions, resulting in superior instrument performance. For example, use of the 24-hour lock mass protection approach improved the average mass error in this study sufficiently (5.5 ppm (1.8 ppm to 0.1 ppm (1.7 ppm, Figure 3) to allow reliable distinction between peptide phosphorylation and sulfonation, which have a 4.8 ppm mass difference on a 2000 Da peptide. One limitation of this strategy results directly from the strong calibrant signal provided by the LSS. If unlimited data-dependent acquisition were allowed, much instrument time would be devoted to acquiring MS/MS scans from the PCM molecules themselves. Rejecting the singly charged precursor ions from fragmentation eliminates any interference from these ions, allowing the instrument to focus 100% of the duty cycle on productive MS/ MS of peptides. We expect the 24-hour Lock Mass Protection approach to be both highly beneficial and readily applied in any proteomics laboratory.
References (1) Zubarev, R.; Mann, M. On the proper use of mass accuracy in proteomics. Mol. Cell. Proteomics 2007, 6, 377–381. (2) Mann, M.; Kelleher, N. L. Precision Proteomics: The case for high resolution and high mass accuracy. Proc. Natl. Acad. Sci. U.S.A. 2008, 105 (47), 18132–18138. (3) Balarski, C. E.; Haas, W.; Dephoure, N. E.; Gygi, S. P. The effects of mass accuracy, data acquisition speed, and search algorithm choice
technical notes
24-Hour Lock Mass Protection on peptide identification rates in phosphoproteomics. Anal. Bioanal. Chem. 2007, 389, 1409–1419. (4) Olsen, J. V.; Macek, B. High Accuracy Mass Spectrometry in LargeScale Analysis of Protein Phosphorylation. In Mass Spectrometry of Proteins and Peptides, Vol. 492; Lipton, M., Pasˇa-Tolic, L., Eds.; Humana Press: Totowa, NJ, 2009; pp 131-142. (5) Tanner, S.; Shu, H.; Frank, A.; Wang, L.-C.; Zandi, E.; Mumby, M.; Pevzner, P. A.; Bafna, V. InsPecT: Identification of Posttranslationally Modified Peptides from Tandem Mass Spectra. Anal. Chem. 2005, 77, 4626–4639. (6) Boyne, M. T., II; Garcia, B. A.; Li, M.; Zamdborg, L.; Wenger, C. D.; Babai, S.; Kelleher, N. L. Tandem Mass Spectrometry with Ultrahigh Mass Accuracy Clarifies Peptide Identification by Database Retrieval. J. Proteome Res. 2009, 8, 374–379.
(7) Olsen, J. V.; de Godoy, L. M. F.; Li, G.; Macek, B.; Mortensen, P.; Pesch, R.; Makarov, A.; Lange, O.; Horning, S.; Mann, M. Parts per Million Mass Accuracy on an Orbitrap Mass Spectrometer via Lock Mass Injection into a C-trap. Mol. Cell. Proteomics 2005, 4, 2010– 2021. (8) Yu, W.; Taylor, J. A.; Davis, M. T.; Bonilla, L. E.; Lee, K. A.; Auger, P. L.; Farnsworth, C. C.; Welcher, A. A.; Patterson, S. D. Maximizing the sensitivity and reliability of peptide identification in large-scale proteomic experiments by harnessing multiple search engines. Proteomics 2010, 10 (6), 1172–1189.
PR100780B
Journal of Proteome Research • Vol. 10, No. 2, 2011 885
