Technical Note pubs.acs.org/ac
Toward Standardizing Deuterium Content Reporting in Hydrogen Exchange-MS Joey G. Sheff† and David C. Schriemer*,†,‡ †
Department of Chemistry, University of Calgary, Calgary, Alberta T2N 4N1, Canada Department of Biochemistry and Molecular Biology, University of Calgary, Calgary, Alberta T2N 4N1,Canada
‡
ABSTRACT: We introduce a method to monitor dispensing ratios during labeling reactions in hydrogen exchange (HX)-MS. The method corrects for systematic and random dispensing errors and harmonizes data incorporating variable %D2O in the experiment design. A correction factor for deuterium levels is obtained by quantifying the relative signal intensities arising from nonexchanging heavy caffeine (spiked into labeling buffer) and light caffeine (spiked into sample solutions). Dispensing variability over a wide range of % D2O composition can be detected and corrected to a common value, and although random dispensing error is usually minor, we show it can be the limiting factor in high quality signal measurements. Applying a dispensing control is therefore an effective tool for monitoring measurement precision in HX-MS. resolution labeling maps.11 And finally, even the suspicion of systematic error can lead to abandoning replicate data sets, thus reducing the statistical power of the analysis and wasting time and resources. Aside from automation, efforts to improve precision have focused primarily on controlling the rate and variability of the deuterium back exchange phenomenon: the loss of label during and after quenching of the labeling reaction.12,13 An internal standard approach has been implemented to account for runto-run variation in back exchange time, using synthetic peptide controls.14 Precision was shown to improve, and it suggests a broader standardization strategy for additional gains. In this study, we present the use of nonpeptidic standards as dispensing controls, as an additional way to improve precision by targeting errors in reagent metering during the labeling step (“dispensing error”).
A
mide hydrogen/deuterium exchange (HX) chemistry offers an exquisitely sensitive probe of a protein’s environment. Through mass spectrometry (MS), the method is experiencing a renaissance in profiling protein conformational dynamics. MS has actually eclipsed NMR as the preferred method for measuring deuterium uptake, mainly due to the expanded temporal range available with MS methods (milliseconds to days) and its suitability for large and complex protein systems.1 As in proteomics, an enzymatic digestion step is primarily responsible for success in complex systems analysis, as peptides are easier to analyze than proteins. This bottom-up HX-MS method has supported the elucidation of protein unfolding pathways,2 the identification of binding sites,3 and the conformational assessment of biotherapeutics such as monoclonal antibodies.4 Unfortunately, the strength of the HX-MS method is also its weakness. The technique is noisy.5 Deuterium incorporation rates are controlled by protein dynamics that can be sensitive to minor changes in sample formulation and in solution conditions (e.g., pH, temperature, ionic strength).6,7 The precision of deuterium measurement is influenced by slight variations in the duration of analysis, the S/N of the isotopic profile, and the multistep dispensing of reagents. Despite advancements in automation,8 interday reproducibility remains a challenge and lags significantly behind intraday repeatability.5 With very careful control over the labeling-to-analysis protocol, most peptide measurements fall within 10% of the mean deuteration values,9 although a more limited reproducibility study shows it can be better than 2.2%.10 There are good reasons to improve precision beyond the 10% level. Binding-induced perturbations of labeling can be subtle, and increased precision would translate into an improved sensitivity for mapping conformational changes and binding sites. Lack of precision also limits a successful modeling of the exchange process, which may be needed to derive higher © 2014 American Chemical Society
■
EXPERIMENTAL SECTION Caffeine Solutions and Standard Curve. Stock solutions of light caffeine and heavy caffeine were prepared in H2O and D2O, respectively, and diluted as needed. A standard curve was collected by manually injecting 50 fmol to 10 pmol of light caffeine with a fixed amount of heavy caffeine (1 pmol) into an AB Sciex TripleTOF 5600 mass spectrometer coupled to a LEAP-PAL HTX chromatography platform, outfitted with a nanoLC-ultra-2D pump. Analysis was performed in both MS and product ion modes. Separation was performed using a selfpacked precolumn (200 Å, 5 μm MagicC18, Michrom BioResources in a 25 mm × 250 μm i.d. capillary) and an analytical column (200 Å, 5 μm MagicC18, Michrom Received: September 12, 2014 Accepted: November 26, 2014 Published: November 26, 2014 11962
dx.doi.org/10.1021/ac5034424 | Anal. Chem. 2014, 86, 11962−11965
Analytical Chemistry
Technical Note
BioResources in a 70 mm × 150 μm i.d. capillary). Samples were manually loaded, and separations were performed at 4 °C. The loading mobile phase consisted of 100% H2O with 0.3% TFA. Mobile phases for separation were composed of 3% ACN with 0.1% FA for mobile phase A and 97% ACN with 0.1% FA for mobile phase B. Labeling Experiments. A sample of BSA in light caffeine was predigested with trypsin and prepared in 10 mM AMBIC (pH 7) to a final concentration of 1 μM BSA and 0.5 μM light caffeine. The labeling solution contained 10 mM AMBIC (pH 7, uncorrected) and 0.5 μM heavy caffeine in D2O. Deuteration was initiated by combining BSA and labeling solution at appropriate ratios. For example, the 50% deuteration sample was prepared by diluting 4 μL of BSA digest with 4 μL of D2O solution. The labeling reaction was quenched after 24 h at room temperature using 100 mM glycine HCl (pH 2.5), snap frozen, and stored at −80 °C until analysis. Samples were analyzed using the HX-MS platform described above. Data Analysis. The deuteration levels for all BSA peptides were measured using centroid calculations in the usual manner, corrected by the unlabeled peptide centroid mass, using MassSpec Studio.15 The intensities of both heavy and light caffeine XICs were measured in the Studio as well, from either MS or MS/MS data. A correction factor for variations in %D2O was calculated using eq 1, ⎞ ⎛ heavy Xcorr(%D2 O)target = ⎜ ⎟ ⎝ light + heavy ⎠
(1)
where Xcorr is the correction factor to be applied to the peptide deuteration data, “heavy” and “light” are the XIC intensities for light caffeine and heavy caffeine, respectively, in a given sample, and (%D2O)target is the intended composition of the labeling solution. The correction factor was applied using eq 2, Dcorr
D = meas Xcorr
Figure 1. (a) Concept for correcting dispensing error and harmonizing variable labeling conditions, using stable light and heavy isotopelabeled small molecules bearing no labile hydrogens. Light (L) and heavy (H) compounds are added to their respective solutions (protein dilution buffer in H2O and D2O labeling buffer). Labeling is initiated when combined by the dispensing system, quenched and digested, and then analyzed by LC-MS. The measured heavy fraction is used as a correction factor for the deuteration values measured for all peptides. (b) Structures of light (m/z 195.0876, left) and heavy (m/z 198.0987, right) caffeine standards used in this study.
(2)
where Dcorr is the corrected deuteration value and Dmeas is the measured peptide deuteration value. Chemicals and Reagents. Lyophilized bovine serum albumin (BSA), “light” caffeine (nonenriched, ReagentPlus grade), “heavy” caffeine (trimethyl-13C3 caffeine, 99% 13C), HPLC-grade H2O and acetonitrile (ACN), D2O (99.9% D), glycine (Gly), hydrochloric acid (HCl), trifluoroacetic acid (TFA), ammonium bicarbonate (AMBIC), and formic acid (FA) were purchased from Sigma-Aldrich (St Louis, MO). Trypsin was sequencing grade (Promega).
140 transitions were extracted from product ion scans for light and heavy caffeine, respectively. In this mode, background was very low and linearity was observed for just over 2 orders of magnitude (not shown). A good correlation is observed when the fraction of heavy caffeine is plotted as a function of %D2O (Figure 2A), showing an average RSD of 2.75%. Stock concentrations for each compound were adjusted using the slope of their respective standard curves, to ensure that peak ratios accurately reflected %D2O levels. A slight deviation from unit slope and a zero intercept likely reflects residual error in stock concentration, but the correlation is sufficient for testing the concept. Linearity of Deuteration. Raw deuteration levels for equilibrated peptide populations scale linearly with %D2O as expected, as illustrated in Figure 2B for three peptides, with RSD deuteration values of 2.7%, 3.5%, and 6.7%, comparable to the RSD value for the dispensing control. Although only shown for a limited span of D2O (25−75%), it clearly supports applying a correction factor in this range and beyond.
■
RESULTS AND DISCUSSION To control for dispensing errors (systematic and/or random), we propose using nonexchanging stable isotope labeled compounds added in both the sample solution and the labeling buffer, as shown in Figure 1, and represented here by caffeine. Linearity of Response for Dispensing Control. We first tested the linearity of response for heavy caffeine, standardized against light caffeine. Ideally, an effective reagent should generate a heavy fractional value that produces a linear dynamic range approaching 2 orders of magnitude, under conventional HX conditions, and have sufficiently high measurement precision to avoid introducing error through propagation. Spectral overlap was observed in MS mode for light caffeine (m/z 195.08); therefore, MS/MS was used for chemical noise reduction and fraction calculation. The 195 > 138 and 198 > 11963
dx.doi.org/10.1021/ac5034424 | Anal. Chem. 2014, 86, 11962−11965
Analytical Chemistry
Technical Note
Figure 2. (a) Linearity of response for fraction of heavy caffeine vs the %D2O. Data collected in MS/MS mode, for transitions m/z 195/138 (light) and m/z 198/140 (heavy). Data collected in triplicate. (b) Linearity of deuteration for three labeled BSA peptides (HPYFYAPELLYYANK in red, AEFVEVTK in blue, AFDEK in black). Data averaged from 4 replicates.
used the fit to determine a deuteration value at 50% D2O (D50,corr) and compared it to the measured deuteration value (D50,meas). D50,corr for this peptide was 3.5 ± 0.1, which is comparable to D50,meas at 3.5 ± 0.1. The example suggests that no additional error is propagated into the measurements from the manipulation, and the correction might account for a degree of random dispensing error as well. We tested this idea further. Random Pipetting Error. First, we prepared six dispensing replicates as per Figure 1, in a 1:1 ratio (50% D2O), and compared the precision achieved from dispensing replicates with the precision from technical replicates, where one preparation was sampled six times (Table 1).
Application of Correction Factor. There are two classes of correction to be considered. First, gross dispensing errors can lead to large shifts in %D2O, and at times, intentional large variations in %D2O can be useful in HX-MS experiments.16 Second, random error in dispensing systems can generate minor shifts in %D2O, depending on the technology used (manual pipettes or automated microsyringes). We investigated the ability of the correction factor to address both categories. Pipetting error was simulated by varying the D2 O composition by ±2−25%, centered on 50% D2O. For each replicate, a correction factor (Xcorr) was calculated from the heavy and light caffeine XICs (eq 1) and applied to the measured deuteration value for a BSA digest peptide in the same run (eq 2 and Figure 3). In this example peptide, the slope of the linear fit to the corrected values is very close to zero, indicating that gross measurement error can be recovered across a wide range of %D2O, and intentional variation in labeling percentages can be harmonized to a single value. We
Table 1. Precision in Six Dispensing and Technical Replicates AVG ±σ %RSD a
dispensinga
technicala
0.500 0.006 1.22
0.500 0.002 0.49
Measured as heavy/(heavy + light).
Dispensing introduces a small but significant contribution to measurement error, even under carefully controlled dispensing conditions, which might be partially recovered using the correction strategy described here. To assess the degree of correction, we applied the method to all peptides in the digest. The RSD for the D50,meas of 46 deuterated digest peptides was quantified and compared with the corresponding value for the D50,corr determinations (Figure 4). The linear fit has a slope greater than one, and an x-intercept of ∼1%. In the first place, the data indicate that peptide measurements with a high RSD do not benefit from the application of the correction, yet the precision is not worsened by the correction. It is likely that noise in the isotopic profile is the limiting factor in precision at this extreme. In the second place, at the other extreme where data quality is higher, measurement precision has a strong contribution from dispensing error and points to some value in applying the correction. Here, precision can be improved by up to 1% in the
Figure 3. Correction factors applied to the measured deuteration value of peptide LVVSTQTALA, collected over 25−75% D2O. Black squares: measured D values. Red squares: corrected D values. Data represents the average of 4 replicates (±1SD). D50%, corr calculated from a linear fit in red. 11964
dx.doi.org/10.1021/ac5034424 | Anal. Chem. 2014, 86, 11962−11965
Analytical Chemistry
Technical Note
Author Contributions
J.G.S. conducted all experiments and drafted the manuscript. D.C.S. participated in experiment design and edited the manuscript. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The work was supported by an NSERC Discovery Grant 298351-2010 (D.C.S.). D.C.S. acknowledges the additional support of the Canada Research Chair program, Alberta Ingenuity-Health Solutions, and the Canada Foundation for Innovation.
■
(1) Konermann, L.; Pan, J.; Liu, Y. H. Chem. Soc. Rev. 2011, 40, 1224−1234. (2) Englander, S. W. Annu. Rev. Biophys. 2000, 29, 213−238. (3) Mandell, J. G.; Falick, A. M.; Komives, E. A. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 14705−14710. (4) Berkowitz, S. A.; Engen, J. R.; Mazzeo, J. R.; Jones, G. B. Nat. Rev. Drug Discovery 2012, 11, 527−540. (5) Iacob, R. E.; Engen, J. R. J. Am. Soc. Mass Spectrom. 2012, 23, 1003−1010. (6) Bai, Y.; Milne, J. S.; Mayne, L.; Englander, S. W. Proteins 1993, 17, 75−86. (7) Englander, S. W.; Kallenbach, N. R. Q. Rev. Biophys. 1983, 16, 521−655. (8) Chalmers, M. J.; Busby, S. A.; Pascal, B. D.; He, Y.; Hendrickson, C. L.; Marshall, A. G.; Griffin, P. R. Anal. Chem. 2006, 78, 1005−1014. (9) Chalmers, M. J.; Pascal, B. D.; Willis, S.; Zhang, J.; Iturria, S. J.; Dodge, J. A.; Griffin, P. R. Int. J. Mass Spectrom. 2011, 302, 59−68. (10) Burkitt, W.; O’Connor, G. Rapid Commun. Mass Spectrom. 2008, 22, 3893−3901. (11) Fajer, P. G.; Bou-Assaf, G. M.; Marshall, A. G. J. Am. Soc. Mass Spectrom. 2012, 23, 1202−1208. (12) Walters, B. T.; Ricciuti, A.; Mayne, L.; Englander, S. W. J. Am. Soc. Mass Spectrom. 2012, 23, 2132−2139. (13) Zhang, H. M.; Bou-Assaf, G. M.; Emmett, M. R.; Marshall, A. G. J. Am. Soc. Mass Spectrom. 2009, 20, 520−524. (14) Zhang, Z. Q.; Zhang, A.; Xiao, G. Anal. Chem. 2012, 84, 4942− 4949. (15) Rey, M.; Sarpe, V.; Burns, K. M.; Buse, J.; Baker, C. A. H.; van Dijk, M.; Wordeman, L.; Bonvin, A. M. J. J.; Schriemer, D. C. Structure 2014, 22, 1538−1548. (16) Slysz, G. W.; Percy, A. J.; Schriemer, D. C. Anal. Chem. 2008, 80, 7004−7011.
Figure 4. Relative standard deviations for corrected deuteration vs absolute deuteration. Deuteration was measured for 46 digest peptides for 5 replicates.
dispensing configuration we used here. In other words, precision is limited by dispensing quality for peptides with low error. Selection of Dispensing Control. Caffeine is useful for the purposes of illustrating the value of a dispensing control, but other compounds may be equally suitable. A compound would ideally have a low molecular weight (e.g.,
Toward Standardizing Deuterium Content Reporting in Hydrogen Exchange-MS Joey G. Sheff† and David C. Schriemer*,†,‡ †
Department of Chemistry, University of Calgary, Calgary, Alberta T2N 4N1, Canada Department of Biochemistry and Molecular Biology, University of Calgary, Calgary, Alberta T2N 4N1,Canada
‡
ABSTRACT: We introduce a method to monitor dispensing ratios during labeling reactions in hydrogen exchange (HX)-MS. The method corrects for systematic and random dispensing errors and harmonizes data incorporating variable %D2O in the experiment design. A correction factor for deuterium levels is obtained by quantifying the relative signal intensities arising from nonexchanging heavy caffeine (spiked into labeling buffer) and light caffeine (spiked into sample solutions). Dispensing variability over a wide range of % D2O composition can be detected and corrected to a common value, and although random dispensing error is usually minor, we show it can be the limiting factor in high quality signal measurements. Applying a dispensing control is therefore an effective tool for monitoring measurement precision in HX-MS. resolution labeling maps.11 And finally, even the suspicion of systematic error can lead to abandoning replicate data sets, thus reducing the statistical power of the analysis and wasting time and resources. Aside from automation, efforts to improve precision have focused primarily on controlling the rate and variability of the deuterium back exchange phenomenon: the loss of label during and after quenching of the labeling reaction.12,13 An internal standard approach has been implemented to account for runto-run variation in back exchange time, using synthetic peptide controls.14 Precision was shown to improve, and it suggests a broader standardization strategy for additional gains. In this study, we present the use of nonpeptidic standards as dispensing controls, as an additional way to improve precision by targeting errors in reagent metering during the labeling step (“dispensing error”).
A
mide hydrogen/deuterium exchange (HX) chemistry offers an exquisitely sensitive probe of a protein’s environment. Through mass spectrometry (MS), the method is experiencing a renaissance in profiling protein conformational dynamics. MS has actually eclipsed NMR as the preferred method for measuring deuterium uptake, mainly due to the expanded temporal range available with MS methods (milliseconds to days) and its suitability for large and complex protein systems.1 As in proteomics, an enzymatic digestion step is primarily responsible for success in complex systems analysis, as peptides are easier to analyze than proteins. This bottom-up HX-MS method has supported the elucidation of protein unfolding pathways,2 the identification of binding sites,3 and the conformational assessment of biotherapeutics such as monoclonal antibodies.4 Unfortunately, the strength of the HX-MS method is also its weakness. The technique is noisy.5 Deuterium incorporation rates are controlled by protein dynamics that can be sensitive to minor changes in sample formulation and in solution conditions (e.g., pH, temperature, ionic strength).6,7 The precision of deuterium measurement is influenced by slight variations in the duration of analysis, the S/N of the isotopic profile, and the multistep dispensing of reagents. Despite advancements in automation,8 interday reproducibility remains a challenge and lags significantly behind intraday repeatability.5 With very careful control over the labeling-to-analysis protocol, most peptide measurements fall within 10% of the mean deuteration values,9 although a more limited reproducibility study shows it can be better than 2.2%.10 There are good reasons to improve precision beyond the 10% level. Binding-induced perturbations of labeling can be subtle, and increased precision would translate into an improved sensitivity for mapping conformational changes and binding sites. Lack of precision also limits a successful modeling of the exchange process, which may be needed to derive higher © 2014 American Chemical Society
■
EXPERIMENTAL SECTION Caffeine Solutions and Standard Curve. Stock solutions of light caffeine and heavy caffeine were prepared in H2O and D2O, respectively, and diluted as needed. A standard curve was collected by manually injecting 50 fmol to 10 pmol of light caffeine with a fixed amount of heavy caffeine (1 pmol) into an AB Sciex TripleTOF 5600 mass spectrometer coupled to a LEAP-PAL HTX chromatography platform, outfitted with a nanoLC-ultra-2D pump. Analysis was performed in both MS and product ion modes. Separation was performed using a selfpacked precolumn (200 Å, 5 μm MagicC18, Michrom BioResources in a 25 mm × 250 μm i.d. capillary) and an analytical column (200 Å, 5 μm MagicC18, Michrom Received: September 12, 2014 Accepted: November 26, 2014 Published: November 26, 2014 11962
dx.doi.org/10.1021/ac5034424 | Anal. Chem. 2014, 86, 11962−11965
Analytical Chemistry
Technical Note
BioResources in a 70 mm × 150 μm i.d. capillary). Samples were manually loaded, and separations were performed at 4 °C. The loading mobile phase consisted of 100% H2O with 0.3% TFA. Mobile phases for separation were composed of 3% ACN with 0.1% FA for mobile phase A and 97% ACN with 0.1% FA for mobile phase B. Labeling Experiments. A sample of BSA in light caffeine was predigested with trypsin and prepared in 10 mM AMBIC (pH 7) to a final concentration of 1 μM BSA and 0.5 μM light caffeine. The labeling solution contained 10 mM AMBIC (pH 7, uncorrected) and 0.5 μM heavy caffeine in D2O. Deuteration was initiated by combining BSA and labeling solution at appropriate ratios. For example, the 50% deuteration sample was prepared by diluting 4 μL of BSA digest with 4 μL of D2O solution. The labeling reaction was quenched after 24 h at room temperature using 100 mM glycine HCl (pH 2.5), snap frozen, and stored at −80 °C until analysis. Samples were analyzed using the HX-MS platform described above. Data Analysis. The deuteration levels for all BSA peptides were measured using centroid calculations in the usual manner, corrected by the unlabeled peptide centroid mass, using MassSpec Studio.15 The intensities of both heavy and light caffeine XICs were measured in the Studio as well, from either MS or MS/MS data. A correction factor for variations in %D2O was calculated using eq 1, ⎞ ⎛ heavy Xcorr(%D2 O)target = ⎜ ⎟ ⎝ light + heavy ⎠
(1)
where Xcorr is the correction factor to be applied to the peptide deuteration data, “heavy” and “light” are the XIC intensities for light caffeine and heavy caffeine, respectively, in a given sample, and (%D2O)target is the intended composition of the labeling solution. The correction factor was applied using eq 2, Dcorr
D = meas Xcorr
Figure 1. (a) Concept for correcting dispensing error and harmonizing variable labeling conditions, using stable light and heavy isotopelabeled small molecules bearing no labile hydrogens. Light (L) and heavy (H) compounds are added to their respective solutions (protein dilution buffer in H2O and D2O labeling buffer). Labeling is initiated when combined by the dispensing system, quenched and digested, and then analyzed by LC-MS. The measured heavy fraction is used as a correction factor for the deuteration values measured for all peptides. (b) Structures of light (m/z 195.0876, left) and heavy (m/z 198.0987, right) caffeine standards used in this study.
(2)
where Dcorr is the corrected deuteration value and Dmeas is the measured peptide deuteration value. Chemicals and Reagents. Lyophilized bovine serum albumin (BSA), “light” caffeine (nonenriched, ReagentPlus grade), “heavy” caffeine (trimethyl-13C3 caffeine, 99% 13C), HPLC-grade H2O and acetonitrile (ACN), D2O (99.9% D), glycine (Gly), hydrochloric acid (HCl), trifluoroacetic acid (TFA), ammonium bicarbonate (AMBIC), and formic acid (FA) were purchased from Sigma-Aldrich (St Louis, MO). Trypsin was sequencing grade (Promega).
140 transitions were extracted from product ion scans for light and heavy caffeine, respectively. In this mode, background was very low and linearity was observed for just over 2 orders of magnitude (not shown). A good correlation is observed when the fraction of heavy caffeine is plotted as a function of %D2O (Figure 2A), showing an average RSD of 2.75%. Stock concentrations for each compound were adjusted using the slope of their respective standard curves, to ensure that peak ratios accurately reflected %D2O levels. A slight deviation from unit slope and a zero intercept likely reflects residual error in stock concentration, but the correlation is sufficient for testing the concept. Linearity of Deuteration. Raw deuteration levels for equilibrated peptide populations scale linearly with %D2O as expected, as illustrated in Figure 2B for three peptides, with RSD deuteration values of 2.7%, 3.5%, and 6.7%, comparable to the RSD value for the dispensing control. Although only shown for a limited span of D2O (25−75%), it clearly supports applying a correction factor in this range and beyond.
■
RESULTS AND DISCUSSION To control for dispensing errors (systematic and/or random), we propose using nonexchanging stable isotope labeled compounds added in both the sample solution and the labeling buffer, as shown in Figure 1, and represented here by caffeine. Linearity of Response for Dispensing Control. We first tested the linearity of response for heavy caffeine, standardized against light caffeine. Ideally, an effective reagent should generate a heavy fractional value that produces a linear dynamic range approaching 2 orders of magnitude, under conventional HX conditions, and have sufficiently high measurement precision to avoid introducing error through propagation. Spectral overlap was observed in MS mode for light caffeine (m/z 195.08); therefore, MS/MS was used for chemical noise reduction and fraction calculation. The 195 > 138 and 198 > 11963
dx.doi.org/10.1021/ac5034424 | Anal. Chem. 2014, 86, 11962−11965
Analytical Chemistry
Technical Note
Figure 2. (a) Linearity of response for fraction of heavy caffeine vs the %D2O. Data collected in MS/MS mode, for transitions m/z 195/138 (light) and m/z 198/140 (heavy). Data collected in triplicate. (b) Linearity of deuteration for three labeled BSA peptides (HPYFYAPELLYYANK in red, AEFVEVTK in blue, AFDEK in black). Data averaged from 4 replicates.
used the fit to determine a deuteration value at 50% D2O (D50,corr) and compared it to the measured deuteration value (D50,meas). D50,corr for this peptide was 3.5 ± 0.1, which is comparable to D50,meas at 3.5 ± 0.1. The example suggests that no additional error is propagated into the measurements from the manipulation, and the correction might account for a degree of random dispensing error as well. We tested this idea further. Random Pipetting Error. First, we prepared six dispensing replicates as per Figure 1, in a 1:1 ratio (50% D2O), and compared the precision achieved from dispensing replicates with the precision from technical replicates, where one preparation was sampled six times (Table 1).
Application of Correction Factor. There are two classes of correction to be considered. First, gross dispensing errors can lead to large shifts in %D2O, and at times, intentional large variations in %D2O can be useful in HX-MS experiments.16 Second, random error in dispensing systems can generate minor shifts in %D2O, depending on the technology used (manual pipettes or automated microsyringes). We investigated the ability of the correction factor to address both categories. Pipetting error was simulated by varying the D2 O composition by ±2−25%, centered on 50% D2O. For each replicate, a correction factor (Xcorr) was calculated from the heavy and light caffeine XICs (eq 1) and applied to the measured deuteration value for a BSA digest peptide in the same run (eq 2 and Figure 3). In this example peptide, the slope of the linear fit to the corrected values is very close to zero, indicating that gross measurement error can be recovered across a wide range of %D2O, and intentional variation in labeling percentages can be harmonized to a single value. We
Table 1. Precision in Six Dispensing and Technical Replicates AVG ±σ %RSD a
dispensinga
technicala
0.500 0.006 1.22
0.500 0.002 0.49
Measured as heavy/(heavy + light).
Dispensing introduces a small but significant contribution to measurement error, even under carefully controlled dispensing conditions, which might be partially recovered using the correction strategy described here. To assess the degree of correction, we applied the method to all peptides in the digest. The RSD for the D50,meas of 46 deuterated digest peptides was quantified and compared with the corresponding value for the D50,corr determinations (Figure 4). The linear fit has a slope greater than one, and an x-intercept of ∼1%. In the first place, the data indicate that peptide measurements with a high RSD do not benefit from the application of the correction, yet the precision is not worsened by the correction. It is likely that noise in the isotopic profile is the limiting factor in precision at this extreme. In the second place, at the other extreme where data quality is higher, measurement precision has a strong contribution from dispensing error and points to some value in applying the correction. Here, precision can be improved by up to 1% in the
Figure 3. Correction factors applied to the measured deuteration value of peptide LVVSTQTALA, collected over 25−75% D2O. Black squares: measured D values. Red squares: corrected D values. Data represents the average of 4 replicates (±1SD). D50%, corr calculated from a linear fit in red. 11964
dx.doi.org/10.1021/ac5034424 | Anal. Chem. 2014, 86, 11962−11965
Analytical Chemistry
Technical Note
Author Contributions
J.G.S. conducted all experiments and drafted the manuscript. D.C.S. participated in experiment design and edited the manuscript. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The work was supported by an NSERC Discovery Grant 298351-2010 (D.C.S.). D.C.S. acknowledges the additional support of the Canada Research Chair program, Alberta Ingenuity-Health Solutions, and the Canada Foundation for Innovation.
■
(1) Konermann, L.; Pan, J.; Liu, Y. H. Chem. Soc. Rev. 2011, 40, 1224−1234. (2) Englander, S. W. Annu. Rev. Biophys. 2000, 29, 213−238. (3) Mandell, J. G.; Falick, A. M.; Komives, E. A. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 14705−14710. (4) Berkowitz, S. A.; Engen, J. R.; Mazzeo, J. R.; Jones, G. B. Nat. Rev. Drug Discovery 2012, 11, 527−540. (5) Iacob, R. E.; Engen, J. R. J. Am. Soc. Mass Spectrom. 2012, 23, 1003−1010. (6) Bai, Y.; Milne, J. S.; Mayne, L.; Englander, S. W. Proteins 1993, 17, 75−86. (7) Englander, S. W.; Kallenbach, N. R. Q. Rev. Biophys. 1983, 16, 521−655. (8) Chalmers, M. J.; Busby, S. A.; Pascal, B. D.; He, Y.; Hendrickson, C. L.; Marshall, A. G.; Griffin, P. R. Anal. Chem. 2006, 78, 1005−1014. (9) Chalmers, M. J.; Pascal, B. D.; Willis, S.; Zhang, J.; Iturria, S. J.; Dodge, J. A.; Griffin, P. R. Int. J. Mass Spectrom. 2011, 302, 59−68. (10) Burkitt, W.; O’Connor, G. Rapid Commun. Mass Spectrom. 2008, 22, 3893−3901. (11) Fajer, P. G.; Bou-Assaf, G. M.; Marshall, A. G. J. Am. Soc. Mass Spectrom. 2012, 23, 1202−1208. (12) Walters, B. T.; Ricciuti, A.; Mayne, L.; Englander, S. W. J. Am. Soc. Mass Spectrom. 2012, 23, 2132−2139. (13) Zhang, H. M.; Bou-Assaf, G. M.; Emmett, M. R.; Marshall, A. G. J. Am. Soc. Mass Spectrom. 2009, 20, 520−524. (14) Zhang, Z. Q.; Zhang, A.; Xiao, G. Anal. Chem. 2012, 84, 4942− 4949. (15) Rey, M.; Sarpe, V.; Burns, K. M.; Buse, J.; Baker, C. A. H.; van Dijk, M.; Wordeman, L.; Bonvin, A. M. J. J.; Schriemer, D. C. Structure 2014, 22, 1538−1548. (16) Slysz, G. W.; Percy, A. J.; Schriemer, D. C. Anal. Chem. 2008, 80, 7004−7011.
Figure 4. Relative standard deviations for corrected deuteration vs absolute deuteration. Deuteration was measured for 46 digest peptides for 5 replicates.
dispensing configuration we used here. In other words, precision is limited by dispensing quality for peptides with low error. Selection of Dispensing Control. Caffeine is useful for the purposes of illustrating the value of a dispensing control, but other compounds may be equally suitable. A compound would ideally have a low molecular weight (e.g.,
