Anal. Chem. 1999, 71, 4974-4980
Observation of Hydrogen-Deuterium Exchange of Ubiquitin by Direct Analysis of Electrospray Capillary-Skimmer Dissociation with Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Satoko Akashi,* Yasuhide Naito, and Koji Takio
Division of Biomolecular Characterization, The Institute of Physical and Chemical Research (RIKEN), 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
The structure of ubiquitin, a small cytoplasmic protein with an extended β-sheet and an r-helix surrounding a hydrophobic core, has been characterized by hydrogendeuterium (H/D) exchange labeling in conjunction with successive analysis by capillary-skimmer dissociation with electrospray ionization-Fourier transform ion cyclotron resonance mass spectrometry (ESI-FTICR MS). The deuterium content of each fragment ion was investigated at different times, and the results indicate that the deuterium incorporation rate into the backbone amides of ubiquitin varied depending on the environment of the amide hydrogens. Amide hydrogens of the N-terminal β-strand showed quite slow exchange while those of the 35-39 loop were exchanged within a short exposure time to deuterium oxide. It was also possible to evaluate the difference in hydrogen-bond stability. The present data are consistent with the structural features obtained by X-Ray and NMR analyses. Although some of the labeling information might be lost by the scrambling of amide protons during capillary-skimmer dissociation, the results demonstrate that the present method provides useful higherorder structural information for proteins. Since the hydrogen-deuterium (H/D) exchange rate of backbone amide hydrogens varies by several orders of magnitude depending on their environment, the measurement of the exchange rate can be used for the investigation of secondary and/ or tertiary structure of a protein. The exchange of backbone amide hydrogens that are protected from the solvent is much slower than that of those exposed to the solvent. Those that are involved in intramolecular hydrogen bonding and that are located in stable secondary structure are slower compared with nonhydrogenbonded ones. To investigate the kinetics of each amide hydrogen, high-resolution 1H NMR is the best choice, because it can determine the exchange rate individually.1-4 On the other hand, * Corresponding author and author to address reprint requests to. Tel.: +8148-467-9511. Fax: +81-48-462-4704. E-mail:
[email protected]. (1) Wagner, G.; Wu ¨ thrich, K. J. Mol. Biol. 1982, 160, 343-361. (2) Delepierre, M.; Dobson, C. M.; Karplus, M.; Poulsen, F. M.; States, D. J.; Wedin, R. E. J. Mol. Biol. 1987, 197, 111-122.
4974 Analytical Chemistry, Vol. 71, No. 21, November 1, 1999
mass spectrometry (MS) requires a smaller sample amount and is much faster than NMR. These are great advantages of MS in the structural characterization of a protein. Thus, several studies have been carried out using H/D exchange in conjunction with electrospray ionization mass spectrometry (ESI-MS)5,6 for the investigation of higher order structure,7-9 conformational changes,10-14 and interactions of proteins.15,16 Matrix-assisted laser desorption ionization (MALDI) time-of-flight mass spectrometry (TOFMS) has also been successfully used for the observation of H/D exchange rates of amide hydrogens in a protein.17 Relationships between the structure and the molecular mass changes of proteins or peptides caused by H/D exchange have been discussed in these studies. Fourier transform ion cyclotron resonance mass spectrometry (FTICR MS), together with the ESI technique, enables the resolution of isotopic peaks of molecular ions weighing as much as 110 kDa.18 In addition, the high resolution and high accuracy of FTICR MS in conjunction with 13C and 15N double depletion19 allowed the observation of ∼180 fragment ions for a small protein (3) Roder, H.; Elo ¨ve, G. A.; Englander, S. W. Nature (London) 1988, 335, 700704. (4) Englander, S. W.; Mayne, L. Annu. Rev. Biophys. Biomol. Struct. 1992, 21, 243-265. (5) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science (Washington, D.C.) 1989, 246, 64-71. (6) Smith, R. D.; Loo, J. A.; Edmonds, C. G.; Barinaga, C. J.; Udseth, H. R. Anal. Chem. 1990, 62, 882-899. (7) Miranker, A.; Robinson, C.; Radford, S. E.; Aplin, R. T.; Dobson, C. M. Science (Washington, D.C.) 1993, 262, 896-900. (8) Wagner, D. S.; Melton, L. G.; Yan, Y. B.; Erickson, B. W.; Anderegg, R. J. Protein Sci. 1994, 3, 1305-1314. (9) Kragelund, B. B.; Robinson, C. V.; Knudsen, J.; Dobson, C. M.; Poulsen, F. M. Biochemistry 1995, 34, 7217-7224. (10) Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1991, 113, 8534-8535. (11) Loo, J. A.; Ogorzalek Loo, R. R.; Udseth, H. R.; Edmonds, C. G.; Smith, R. D. Rapid Commun. Mass Spectrom. 1991, 5, 101-105. (12) Mirza, U. A.; Chait, B. T. Anal. Chem. 1994, 66, 2898-2904. (13) Feng, R.; Konishi, Y. J. Am. Soc. Mass Spectrom. 1993, 4, 638-645. (14) Smith, D. L.; Zhang, Z. Mass Spectrom. Rev. 1994, 13, 411-429. (15) Wang, F.; Blanchard, J. S.; Tang, X.-J. Biochemistry 1997, 36, 3755-3759. (16) Wang, F.; Scapin, G.; Blanchard, J. S.; Angeletti, R. H. Protein Sci. 1998, 7, 293-299. (17) Mandell, J. G.; Falick, A. M.; Komives, E. A. Anal. Chem. 1998, 70, 39873995. (18) Kelleher, N. L.; Senko, M. W.; Siegel, M. M.; McLafferty, F. W. J. Am. Soc. Mass Spectrom. 1997, 8, 380-383. 10.1021/ac990444h CCC: $18.00
© 1999 American Chemical Society Published on Web 09/24/1999
(11 kDa) subjected to capillary-skimmer collision-induced dissociation (CID) without any enzymatic digestion prior to MS analysis.20 Approaches for studying the higher order structure of a protein by H/D exchange in conjunction with FTICR MS have also been carried out. One approach was to characterize gaseous conformational changes of proteins. McLafferty and colleagues have succeeded in observing direct evidence for gas-phase conformational isomers of cytochrome c by applying several dissociation techniques to a gas-phase protein molecule in the ICR cell.21,22 Characterization of the solution structure of a protein has also been performed using FTICR MS. In this approach, peptic fragments after H/D exchange are identified with FTICR MS, thus enabling the determination of the isotopic exchange rates of amide hydrogens.23,24 The data obtained in this manner should be carefully interpreted because some back-exchange is unavoidable during the digestion step and LC/MS analysis of the peptides even though all the procedures are carried out at low pH and low temperature.25 Direct analysis of a protein immediately after H/D exchange in solution, without enzymatic digestion and LC separation, is expected to be less sensitive to the back-exchange that occurs during analysis. As a result, the techniques described here may be more effective for the investigation of the structurefunction relationship in proteins. In previous work, the course of deuterium exchange for two R-helical peptides, melittin (26 amino acid residues) and a growth hormone releasing factor analogue (32 amino acid residues), has been analyzed using ESI-MS/MS with a triple-stage quadrupole mass spectrometer.26 The deuterium content of amide hydrogens in the fragments generated by MS/MS experiments was investigated, which lead to a discussion of the localization of deuterium atoms and the stability of R-helices in these peptides. Unfortunately, larger peptides and/or proteins are difficult to examine by this strategy because they are difficult to cleave into fragments by low- or high-energy CID. No mass spectrometry study has yet been reported without peptic digestion that characterizes the secondary and/or tertiary structure of a protein in solution containing both R-helices and β-strands by solution H/D exchange. We applied capillary-skimmer CID that can generate many fragments efficiently, as previously reported,20 to the analysis of H/D exchange of a protein. ESI-FTICR MS is expected to be effective for the interpretation of deuterium incorporation into each fragment because of its high resolution which becomes important for larger peptides and/or proteins with many fragments of similar mass. The current investigation demonstrates that information on the higher order structure of a protein subjected to H/D exchange can be obtained by capillary-skimmer CID and ESI-FTICR MS (19) Marshall, A. G.; Senko, M. W.; Li, W.; Li, M.; Dillon, S.; Guan, S.; Logan, T. M. J. Am. Chem. Soc. 1997, 119, 433-434. (20) Akashi, S.; Takio, K.; Matsui, H.; Tate, S.-I.; Kainosho, M. Anal. Chem. 1998, 70, 3333-3336. (21) Wood, T. D.; Chorush, R. A.; Wampler, F. M., III; Little, D. P.; O’Connor, P. B.; McLafferty, F. W. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 2451-2454. (22) McLafferty, F. W.; Guan, Z.; Haupts, U.; Wood, T. D.; Kelleher, N. L. J. Am. Chem. Soc. 1998, 120, 4732-4740. (23) Zhang, Z.; Li, W.; Logan, T. M.; Li, M.; Marshall, A. G. Protein Sci. 1997, 6, 2203-2217. (24) Wang, F.; Li, W.; Emmett, M. R.; Hendrickson, C. L.; Marshall, A. G.; Zhang, Y. L.; Wu, L.; Zhang, Z. Y. Biochemistry 1998, 37, 15289-15299. (25) Smith, D. L.; Deng, Y.; Zhang, Z. J. Mass Spectrom. 1997, 32, 135-146. (26) Anderegg, R. J.; Wagner, D. S.; Stevenson, C. L.; Borchardt, R. T. J. Am. Soc. Mass Spectrom. 1994, 5, 425-433.
analysis. We selected ubiquitin, a small cytoplasmic protein composed of 76 amino acids as a model whose structure has already been established in detail by X-Ray27 and NMR.28-30 Capillary-skimmer dissociation was applied to H/D exchanged ubiquitin, and the percentage of deuterium incorporated into each fragment at different time points was obtained by comparing the measured value with the calculated theoretical isotope distribution of fragments with various deuterium contents. Though some H/D scrambling might have occurred during the dissociation process, it will be shown that useful secondary and/or tertiary structural information can be obtained by this method. EXPERIMENTAL SECTION Materials. Bovine ubiquitin was purchased from Sigma Chemical Co. (St. Louis, MO) and used without further purification. Acetic acid-d (98% D) and deuterium oxide (99.9% D) were obtained from Sigma Chemical Co. and Euriso-top CEA Group (France), respectively. Isotope Exchange Reaction. Ubiquitin was dissolved in D2O (99.9% D) to give a concentration of 100 pmol/µL (pD 5.5). The pD value was read directly from the pH meter. The solution was kept at room temperature for different times (t; t ) 1-240 min) for the H/D exchange reaction to occur. Aliquots of 10 µL were quenched at t min with 90 µL of cold 50 mM acetic acid-d/D2O (pD 2.9) and immediately subjected to ESI-FTICR MS. The control sample (t ) 0 min) was prepared by dissolving ubiquitin in cold 45 mM acetic acid-d/D2O to give a concentration of 10 pmol/µL and immediately subjected to ESI-FTICR MS. Each reaction was repeated three times. The mean value of the percentage of deuterium incorporated and the standard deviation for each fragment ion at t min were calculated for each time point. ESI-FTICR MS Analysis. The sample solution was introduced into the spectrometer at a flow rate of 1 µL/min using a syringe pump. The infusion syringe was kept cold by wrapping it with an ice bag to suppress any further H/D exchange after quenching. Spectra were obtained with a Bruker (Billerica, MA) BioApexII spectrometer, equipped with a 7 T magnet and an external electrospray ion source (Analytica of Branford, Branford, CT), by accumulating 32 scans for each sample. The capillary-skimmer CID spectra were obtained by increasing the capillary-skimmer potential to 200-220 V and by increasing the trapping time in an RF only hexapole ion guide to 5 s, while unfragmented spectra were obtained at 80-100 V and 0.5-1 s trapping time.20 External calibration was carried out using nondeuterated ubiquitin. Calculation of Percent Deuterium Incorporation. The isotopic distribution of every fragment ion at each H/D exchange reaction time was compared with the theoretical isotopic distribution pattern of fragments with various deuterium contents calculated by using the Bruker XMASS program (v. 4.1.0) at a resolution of 30 000 (fwhm). The six highest intensity peaks were used for the estimation of the percentage of deuterium incorporated into each fragment. (27) Vijay-Kumar, S.; Bugg, C. E.; Cook, W. J. J. Mol. Biol. 1987, 194, 531544. (28) Di Stefano, D. L.; Wand, A. J. Biochemistry 1987, 26, 7272-7281. (29) Pan, Y.; Briggs, M. S. Biochemistry 1992, 31, 11405-11412. (30) Briggs, M. S.; Roder, H. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 20172021.
Analytical Chemistry, Vol. 71, No. 21, November 1, 1999
4975
Figure 1. Capillary-skimmer dissociation FTICR MS spectra of nondeuterated ubiquitin at 7 T. Assignments of several fragment ions are indicated with the multiply charged molecular ions.
RESULTS AND DISCUSSION Figure 1 shows a capillary-skimmer CID spectrum of deuterated ubiquitin (t ) 1 min) with several assignments of the observed ions. Though the H/D exchange reaction should have caused the isotopic distribution to widen and the peak intensity to decrease, it was not difficult to interpret the deuterated spectra by comparing them with those for nondeuterated ubiquitin. Since two arginine residues are located at the third and fifth positions from the C-terminus, y-type ions were preferentially observed in the spectra. Several internal fragment ions, such as [b52y40]16+ and [b36y58]18+, were also observed. Figure 2 shows expanded capillaryskimmer CID spectra of nondeuterated and deuterated ubiquitin (t ) 0, 1, 10, 30, 60, 120, 240 min). Three ions, b162+, y243+, and y405+, were observed in this region. It can be clearly seen that the masses of these fragment ions increased as the incubation time in deuterium oxide increased. Deuterium Content of Each Fragment. Hydrogen atoms in the elemental composition of a protein are classified into three groups according to their behavior in isotopic exchange. Hydrogens attached directly to carbon atoms, including the hydrogen at the C-2 position of the imidazole group of histidine, are not labile because all the procedures were carried out under neutral or slightly acidic conditions in which the C-2 proton is difficult to exchange. Hydrogens attached to heteroatoms in the side chain and the three hydrogens on the N- and C-terminal groups (H2Nand -COOH) exchange rapidly. Backbone amide hydrogens have a variety of isotopic exchange rates depending on the environment around each backbone amide; some exchange quickly while others exchange at low rates. The deuterium exchange rate of backbone amide hydrogens reflects the higher order structure of a protein. As an example of calculating the deuterium content of a fragment, consider the fragment y243+. In this case, the elemental composition formula is C118H203N38O36, in which 148 hydrogens are attached to carbon atoms, 3 hydrogens are for retaining the 3+ charge, 26 hydrogens are attached to heteroatoms in side chains, one is a C-terminal hydrogen (-COOH), two are nascent N-terminal hydrogens (H2N-), and 23 hydrogens are backbone amide hydrogens. When ubiquitin is dissolved in deuterium oxide, hydrogens on the side-chain heteroatoms and the N- and Cterminal hydrogens are exchanged immediately up to the same 4976 Analytical Chemistry, Vol. 71, No. 21, November 1, 1999
Figure 2. Expanded CID spectra of nondeuterated, (a), and deuterated, (b)-(h), ubiquitin with different exchange times (t ) 0, 1, 10, 30, 60, 120, and 240 min). Three fragment ions, b162+, y243+, and y405+, were observed in this region. The percentage of deuterium incorporation into the backbone amide positions of each fragment is also indicated. The method for estimating the deuterium content of the backbone amides is explained in the text.
percentage occupation as the deuterium content of the solvent. Thus, their percentage of deuterium incorporation should be 99.9%, since D2O of 99.9% D content was used for the present study. The deuterium content of charge-retaining hydrogens on a protein should also be the same as that of the solvent. However, when the deuterated sample solution is sprayed in the electrospray ion source, back-exchange of deuterons with protons occurs to some extent for the fast-exchangeable deuterium atoms. When peptides, such as Leu-enkephalin (MW 555.3), renin substrate (MW 1644.9), and melittin (MW 2844.8), which had been thermally denatured in deuterium oxide (99.9% D) overnight, were subjected to ESI mass analyses, their multiply charged molecular ions indicated that the amide hydrogens of these peptides were exchanged by deuterons only to 96-97% (data not shown). This suggests that deuterium atoms are back-exchanged to some extent by hydrogens during the ESI spraying process. Such backexchange of hydrogens might have occurred because of the natural isotopic composition of the moisture in the laboratory air. An ice-cold syringe used for the sampling might have had condensation on it also contributing to back-exchange. In the present study, we used a value of 97% as the deuterium percentage of rapidly exchangeable hydrogens. The value of 0.015%, the same value of the natural abundance of deuterium, was used for the 148 hydrogens attached to the carbon atoms. Attention should be paid to the interpretation of the two hydrogen atoms on the nascent amino group in the y-series fragments. One of the nascent N-terminal hydrogens is the original backbone amide hydrogen at the cleavage site, and it should be
treated in the same manner as all other backbone amide hydrogens. The origin of the other hydrogen atom has not yet been defined clearly, especially in the case of capillary-skimmer dissociation. Several groups have investigated the formation mechanism of y-series ions in high- and low-energy CID using deuterium labeling and fast atom bombardment (FAB) ionization, and it has been shown that a hydrogen atom attached to nitrogen migrates during the cleavage of the peptide bond.31,32 The fragmentation pathway, including the formation of y-series ions, has also been discussed by others.33-36 In the case of the Asp-X peptide-bond cleavage, the hydrogen of the side-chain carboxylic acid group might be involved in the formation of y-series ions.37 However, it is not yet possible to identify where the migrant hydrogen comes from. In the present study, we have assumed that this hydrogen at the nascent N-terminus is derived from an amide, not directly from the solvent. Thus, the 203 hydrogens of y243+ are classified as follows: 148 hydrogens are unexchangeable, 30 are fast-exchangeable, 24 are amide hydrogens, and the other one is a migrant amide hydrogen. Simulations of isotope distributions were carried out by varying the deuterium content of the backbone amide hydrogens. FTICR MS permits the acquisition of high resolution ESI mass spectra in which all the isotopic peaks are fully resolved not only for fragment ions but also for multiply charged molecular ions for molecules less than 30 kDa in weight. Comparing the observed isotopic distribution with the theoretical one, it is possible to estimate the percentage of deuterium incorporated into the fragments or molecules. Since the observed peaks might have contained noise-signals, adequate attention should be paid to the interpretation, especially for peaks of low intensity. We chose the six highest peaks among the observed isotopic peaks for the interpretation. Neither the centroid mass of the isotope envelope nor the exact mass of the highest isotope peak were used for the estimation of the deuterium content. The experiments were repeated three times, and the mean value and standard deviation at each exchange time were calculated in order to increase the reliability of the data and to check the range of experimental error. Figure 3a shows the theoretical isotope distribution of y243+ at a resolution of 30 000 (fwhm) on the assumption that the percentage of deuterium incorporated into the 25 amide hydrogens was 43%. The observed isotope distribution of y243+ ion (t ) 1 min) (Figure 3b) is almost identical to the theoretical one (Figure 3a). Thus, the percentage of deuterium content in y243+ at t ) 1 min is easily estimated to be 43%. The deuterium content in each fragment ion at any isotopic exchange time (t) was deduced in the same manner. Figure 4 shows a plot of the percentage of deuterium incorporated into backbone amide hydrogen positions as a (31) Kenny, P. T.; Nomoto, K.; Orlando, R. Rapid Commun. Mass Spectrom. 1992, 6, 95-97. (32) Mueller, D. R.; Eckersley, M.; Richter, W. J. Org. Mass Spectrom. 1988, 23, 217-222. (33) Biemann, K. Methods Enzymol. 1990, 193, 455-479. (34) Cordero, M. M.; Houser, J. J.; Wesdemiotis, C. Anal. Chem. 1993, 65, 15941601. (35) Dongre´, A. R.; Jones, J. L.; Somogyi, AÄ ; Wysocki, V. H. J. Am. Chem. Soc. 1996, 118, 8365-8374. (36) Nold, M. J.; Cerda, B. A.; Wesdemiotis, C. J. Am. Soc. Mass Spectrom. 1999, 10, 1-8. (37) Jockusch, R. A.; Schnier, P. D.; Price, W. D.; Strittmatter, E. F.; Demirev, P. A.; Williams, E. R. Anal. Chem. 1997, 69, 1119-1126.
Figure 3. Theoretical (a) and observed (b) isotope distributions of y243+. The theoretical isotope envelope was calculated at a resolution of 30 000 (fwhm) using a value of 43% for the deuterium incorporation into amide hydrogens for the y243+ fragment.
Figure 4. The percentage of deuteration of amide hydrogens in CID fragments as a function of time (min) in log scale. The exchange time courses of six short C-terminal y-type fragments are indicated with broken lines and empty symbols. Those of the N-terminal b-type fragments are indicated with broken lines and filled symbols. That of y182+ is indicated with solid lines with an open circle. Those of the others (listed below) are indicated with solid lines without symbols. y243+, y253+, y374+, y394+, y404+, y424+, y434+, y444+, y454+, y474+, y484+, y494+, y504+, y524+, y535+, y575+, y585+, y595+, y605+, y615+, y625+, y635+, [b25y58]7+, [b32y58]14+, [b36y58]18+, [b52y40]16+, M7+.
function of time in log scale for the observed fragment ions in the capillary-skimmer dissociation spectra. The mean value (n ) 3) of the percentage of deuterium incorporated into each fragment was used for the plot. It can be clearly seen that there are three populations for the deuterium exchange rate of amide hydrogens in the fragments observed by the capillary-skimmer dissociation. Analytical Chemistry, Vol. 71, No. 21, November 1, 1999
4977
Figure 5. The X-Ray structure of ubiquitin27 illustrated by a ribbon diagram. The diagram was generated using the program RasMol (v. 2.6) on the basis of the coordinates registered at the Protein Data Bank (PDB).
Backbone amide hydrogens of short y-type ions such as y8+, y9 y122+, y132+, y142+, and y152+ are deuterated to 60-65% within 1 min and to more than 70% in 30 min, as shown in Figure 4. This suggests that a large number of the amide hydrogens in this region are located at very labile positions, resulting in high deuterium content in a short time. The exchange rate of y182+ seems to be different from that of the other short y-type fragments. The backbone amide hydrogens of the three residues Tyr, Asn, and Ile, located at the 18th, 17th, and 16th positions from the C-terminus (59th, 60th, and 61st from the N-terminus), should be tightly hydrogen-bonded and protected from the solvent. Backbone amide hydrogens of short N-terminal fragment ions such as b5+, b7+, b8+, b122+, b132+, and b162+ have been isotopically exchanged only up to 70% even after 4 h of exposure to D2O. This suggests that amide hydrogens of the N-terminal region are strongly hydrogen-bonded or located at hydrophobic regions. Larger fragment ions showed similar time courses of H/D exchange to those of the multiply charged molecular ions, as shown in Figure 4. Since larger fragments contain various motifs, such as R-helices, β-strands, and loops, each with different exchange rates, structural characteristics of those motifs are averaged out. It is difficult to discuss the structural features only from the H/D exchange time courses of larger fragments. Deuterium Content of Each Structural Motif. To correlate the observed deuterium incorporation to the structural features of ubiquitin, the percentage of deuterium incorporation into each of the following segments was calculated and related to the secondary structure: 1-8 (containing the N-terminal β-strand; 1-7), 9-18 (second β-strand; 10-17), 19-25 (loop), 24-34 (Rhelix; 23-34), 35-39 (loop), 40-52 (third and fourth β-strand; 40-45 and 48-50), 53-64 (loop containing 310 helix; 56-59), 6576 (fifth β-strand; 64-72).27,30 Figure 5 shows the X-Ray structure of native ubiquitin, indicating the backbone secondary structure with a ribbon diagram.27 Since all possible fragment ions generated by every peptide bond cleavage were not observed, it was difficult to divide the sequence into the segments completely according to the secondary structure of ubiquitin. Figure 6 shows the percentage of deuterium incorporated into amide hydrogen positions for each segment as a function of time (t ) 1-240 min) in log scale. The isotope exchange rate of amide hydrogens +,
4978 Analytical Chemistry, Vol. 71, No. 21, November 1, 1999
Figure 6. Deuterium incorporation into each segment of ubiquitin. Error bars indicate the sum of the standard deviations for the deuterium contents of the two fragments used for the calculation. Straight lines were plotted on the basis of the calculation using the least-squares method. The percentages of deuterium incorporated into the N- or C-terminal segments (1-8 and 65-76) are taken from the deuterium contents in the b8+ and y122+ ions, respectively, at t min. The percentage of deuterium incorporated into the 19-25 segment was calculated using the mass values of internal fragment ion [b25y58]7+ observed in the capillary-skimmer CID spectra. The percentages of deuterium incorporated into the other segments (Dx-y (%), x < y) are calculated by the following equations. (1) for the 9-18 segment D9-18 (%) ) (n18D18 (%) - n8D8 (%))/(n18 - n8) where ni is the number of the backbone amide hydrogens in the bi fragment and Di (%) is the percent incorporation of deuterium into amide hydrogens of the bi fragment, (2) for the 24-34, 35-39, 40-52, and 53-64 segments Dx-y (%) ) (nxDx (%) - n(y+1)D(y+1) (%))/(nx - n(y+1)) where nj is the number of the backbone amide hydrogens in the y(76+1-j) fragment and Dj (%) is the percent incorporation of deuterium into amide hydrogens of the y(76+1-j) fragment. For example, in the case of the segment spanning residues 24-34 (x ) 24, y ) 34), y535+ and y424+ are used for the calculation. The numbers of the backbone amide hydrogens in y535+ and y424+ (n24 and n35) are 52 and 41, respectively, since there are two Pro residues at the 37th and 38th positions with a hydrogen transferred from an amide to the nascent N-terminus, as mentioned in the text. At t ) 1 min, percent incorporation of deuterium into amide hydrogens of y535+ and y424+ is 41.3 and 43.5%, respectively (mean value, n ) 3). Then D24-34(%) (t ) 1 min) can be calculated as follows: D24-34(%) ) (52(0.413) 41(0.435))/(52 - 41) ) 33.1 (%).
represents apparent first-order kinetics. Linear regression was applied on each segment to define the best straight line of the data set and to determine the slope, the intercept, and the linearregression coefficient. The exchange reaction can be expressed by the equation Dt(%) ) kln(t) + Dt)1(%), where Dt(%) is the percentage of deuterium incorporated into amide hydrogen positions in each segment at time t min, k is the apparent firstorder rate constant, and Dt)1(%) is the calculated percentage of deuterium incorporated at t ) 1 min. One segment, 24-34, showed two linear regions of the plot while the exchange reaction of others could be represented with single straight lines. Since the 35-39 segment has two Pro residues and contains only three amide hydrogens, small mass differences might be reflected in a large change of deuterium content (%) of amide hydrogens. Though the linear regression coefficient (R2) of the hypothetical line for this segment obtained by the least-squares method is extremely small (R2 ) 0.0596), all three amide
hydrogens appeared to have been exchanged in a short time (t ) 1 min), thus suggesting that this segment is located at a flexible position. In the case of the 9-18 segment, the slope is nearly equal to that of the plot for the N-terminal 1-8 segment containing a β-strand (1-7). The 1-8 segment is sandwiched by two β-strands, 10-17 and 64-72, as shown in Figure 5. Since the β-strand 1017 is located at the edge of a β-sheet, which consists of five β-strands, amide hydrogens located on the outside of the sheet should have exchanged in a short time. Therefore, the intercept of the line for the 9-18 segment should be a larger value than that for the 1-8 segment. The small slopes observed for these two segments (1-8 and 9-18) suggest that the strength of hydrogen bonds between these two segments should be stronger those of than the others. The C-terminal segment spanning residues 65-76 contains the fifth β-strand (64-72) and a “tail” region (73-76) sticking out of the molecule (Figure 5). High flexibility of the tail region may have caused the high deuterium content observed at t ) 1 min. The slope of the plot for this segment is just a little larger than those for two other β-strand segments, 1-8 and 9-18. The segment spanning residues 40-52, containing two short β-strands, 40-45 and 48-50, showed a 1.5 times higher exchange rate than the 65-76 segment. This indicates that hydrogen bonds in which β-strand 40-45 is involved are not so strong as those between the three β-strands spanning residues 10-17, 1-7, and 64-72. The slope of the plot for the 40-52 segment is nearly equal to that for the 53-64 segment, which is a long loop containing a short 310 helix (residues 56-59). The deuterium percentage of amide hydrogens in the 53-64 segment is 75% at t ) 240 min, which indicates that ∼10 of 13 amide hydrogens have exchanged within 240 min. The difference between the exchange rates of y152+ and y182+ in Figure 4 also suggests that the structure around the three residues of Tyr59, Asn60, and Ile61 is rigid and that these amide hydrogens are difficult to exchange, as discussed earlier. X-Ray crystallography results have shown that the amide hydrogens of Tyr59, Asn60, and Ile61 are tightly hydrogen-bonded with the carbonyl oxygens of Thr55/Leu56, Ser57, and Leu56, respectively, which are located in the 310 helix (residues 56-59).27 These hydrogen bonds should contribute to a stable 310 helix, and therefore, the amides of these three residues might be difficult to exchange. A plot for the segment spanning residues 24-34, which contains R-helix 23-34, shows two linear regions along the exchange curve. The first part of this plot has a large slope (k ) 11.2), which is the largest among the values for all the segments. However, the latter part of the plot for this segment shows that no H/D exchange has occurred even at times t g 18 min. A previous NMR study indicated that all amide hydrogens except for Val25 in this segment form hydrogen bonds between oxygens of carbonyl groups.29 The results of the present study are summarized in Figure 6 and indicate that ∼60% of amide hydrogens in this segment have been exchanged within 30 min, thus suggesting that 60% of hydrogen bonds are labile. However, the remaining 40% of amide hydrogens are strong enough to resist H/D exchange for over 4 h. To localize deuterated backbone amides, the accurate calculation of deuterium incorporation into individual amide hydrogen
positions from successive fragments was performed with the equation described in the legend of Figure 6. However, some amide hydrogens showed up-and-down exchange time courses, as Anderregg et al. reported for peptides,26 when the percentage of deuterium incorporated into individual amide hydrogen positions was plotted. The experimental error is exaggeratedly reflected in the percentage of deuterium calculated using two successive fragments. For example, if the percentage of deuterium incorporated into amide hydrogen positions in two successive fragments, b8+ (7 amide hydrogens) and b9+ (8 amide hydrogens), is estimated to be 50%, the amide hydrogen of the ninth residue should be deuterated at 50%. However, if the deuterium percentage of b8+ is estimated as 52% while that of b9+ is 50%, the deuterium content of the amide hydrogen of the ninth residue is calculated to be only 36% () 8(0.5)-7(0.52)). Only 2% variance in the deuterium content of a fragment leads to a 14% difference in the deuteration of a single amide hydrogen. Such an erroneous phenomenon in H/D exchange was also observed for the short 35-39 segment that possesses only three amide hydrogens. Poor linearity of the latter part of the plot for the 24-34 segment might also be caused by a similar reason to that for the 35-39 segment, though it has 10 amide hydrogens in the segment. Ambiguity of the origin of one hydrogen on the nascent amino group in the y-series fragments should have made the potential error larger. The migrant hydrogen on the nascent amino group might have come from an amide to the N-terminal side of the cleaved bond, as previously suggested for peptides.31-36 For the cleavage of the peptide bond Asp-X, the hydrogen of the sidechain carboxylic acid group might participate in the formation of y-series ions.37 However, in the case of a protein, this hydrogen might have come from a nearby amide or side-chain hydrogen if it is located close to the cleavage site in the tertiary structure. Since hydrogens on side chains can be exchanged rapidly with deuterium in the solvent, some of the migrant hydrogens may have come from the solvent. To investigate hydrogen-transfer reactions during peptide fragmentation, Richter and co-workers carried out a low-energy MS/MS experiment for the [M + D]+ of a small peptide.32 The singly charged deuterated molecule, [M + D]+, was generated by desorption chemical ionization (DCI) using methane-d4 (CD4). A low-energy CID spectrum of [M + D]+ using triple-quadrupole MS indicated that b-series ions contain significant amounts of deuterium and that H/D exchange had occurred to some extent for every exchangeable hydrogen. This suggests that positional identity of the external proton has been lost by H/D scrambling prior to precursor-ion fragmentation.32 In a sustained off-resonance irradiation (SORI) -CID experiment of cytochrome c, McLafferty et al. reported H/D scrambling not only of the amide hydrogens but also of the exchangeable hydrogens of side chains during the dissociation process. This might have affected the deuterium content of fragments.22 On the other hand, Smith and co-workers indicated that fragment ions produced by CID MS/MS using an ion-trap instrument can be used for the ranking of amide H/D exchange rates.38 They were able to validate the accuracy of the CID MS/MS results by comparing them with NMR results. They also reported an excellent correlation between the deuterium levels in the fragment ions and those calculated using NMR data, (38) Deng, Y.; Pan, H.; Smith, D. L. J. Am. Chem. Soc. 1999, 121, 1966-1967.
Analytical Chemistry, Vol. 71, No. 21, November 1, 1999
4979
especially for b-series ions. The y-series ions often yielded deuterium levels consistent with levels derived from NMR experiments, but some discrepancies were found, suggesting that H/D transfer occurs during CID to give y-series ions. In the present study, we observed that there were, preferentially, y-series ions with several b-series ions in the capillary-skimmer CID spectra. The experimental conditions of capillary-skimmer dissociation, which forces fragmentation in the high-pressure region, might have caused H/D scrambling. This has not made it difficult, however, to correlate the deuterium content of a segment with the secondary structure, since each segment possesses more than several amide hydrogens, and a scrambling effect will be “diluted”. If this scrambling were a critical problem in analyzing the results, no relation should have been found between the H/D exchange rate and the structure. In fact, plots of the deuterium content in each segment composed of several amino acids reflect structural features, such as an existence of β-strands and an R-helix. Although some extent of the labeling information might be lost by the scrambling of amide protons during capillary-skimmer dissociation, the effect is small for the analysis of the segments derived from a protein. In the present study, precise deuterium localization failed, but H/D exchange in conjunction with capillary-skimmer CID and ESIFTICR MS provided higher order structure information for ubiquitin. Results, as a whole, were consistent with those obtained by NMR28-30 and X-Ray27 studies. The segments with an R-helix or β-strands surrounding a hydrophobic core showed relatively low exchange rates. On the other hand, amide hydrogens in a short loop, residues 35-39, were fully exchanged within 1 min. The segment spanning residues 40-52 showed a relatively high exchange rate, though it has two short β-strands. We suggest that hydrogen bonds around the 40-52 segment are not so strong. The present method enables the classification of the stability of hydrogen bonds between β-strands. A difference between an R-helix and a β-strand in the H/D exchange rate might have been observed in the plots of the segments (Figure 6). Taking into account that ubiquitin has only one R-helix, however, it is difficult to discuss the difference in H/D exchange behavior by this experiment alone. To determine the deuterium incorporation rate into backbone amide hydrogen positions in solution, peptic digestion and LC/ MS or LC/MS/MS at low pH and low temperature are often applied.14-16,23-25 Proteins are incubated in D2O as a function of time, transferred into slow-exchange conditions (pH 2-3, 0 °C, in H2O), and digested with pepsin (pH 2-3, 0 °C, in H2O). After the digestion, the sample is frozen and stored at -70 °C until LC/MS (or LC/MS/MS) analysis at low temperature. Since the
4980
Analytical Chemistry, Vol. 71, No. 21, November 1, 1999
isotopic exchange rates of the amide hydrogens are lowest at pH 2-3 and at low temperature, all the procedures except for the exchange reaction are performed at pH 2-3 and at 0 °C. However, artifactual loss of deuterium labeling to some extent is unavoidable during the digestion and analysis procedure, even under acidic conditions at low temperature. The artifactual loss of deuterium is varied depending on the environment of the amide hydrogens in each fragment peptide, and it is considered challenging to interpret the data even after compensation for deuteration loss using control data for correction. MALDI-TOFMS instead of LC/ MS or LC/MS/MS has successfully been used with H/D exchange to quickly obtain results, but it was found difficult to cover all of the sequence without any separation procedure prior to MS analysis.17 In the present method, some artifactual H/D scrambling might be also unavoidable, but the analysis after the deuteration can be carried out much more promptly. Rapid and direct analysis of the H/D exchanged protein is a great advantage of the present method, and it is possible to investigate the secondary and/or tertiary structure of a protein. CONCLUSIONS We have investigated the structure of ubiquitin, whose structure has been well studied by X-Ray and NMR, using H/D exchange in conjunction with immediate analysis by capillaryskimmer CID and ESI-FTICR MS. Since our aim was to examine a de novo strategy for higher order structural analysis of a protein, we selected ubiquitin as a model. Therefore, it was possible to interpret the results by comparison with the X-Ray structure of ubiquitin. Results were also consistent with those obtained by NMR exchange experiments. It is true that R-helices, β-sheets, and hydrogen bonds are difficult to precisely locate by the present method alone, but information regarding the stability of hydrogen bonds or structural flexibility can be obtained with a small amount of sample in a short time. This strategy should also be applicable to the structural investigation of conformational changes and to protein-protein interactions. ACKNOWLEDGMENT This research was partly supported by Kurata scholarship (to S.A.) and a Grant from the “Biodesign Research Program” of The Institute of Physical and Chemical Research (RIKEN) (to K.T.). We thank Jeffrey G. Mandell, University of California, San Diego, for his generous help with English corrections. Received for review April 27, 1999. Accepted August 11, 1999. AC990444H