Anal. Chem. 1996, 68, 2340-2344
Mass Spectrometric Determination of Isotopic Exchange Rates of Amide Hydrogens Located on the Surfaces of Proteins Kuruppu Dharmasiri and David L. Smith*
Department of Chemistry, University of NebraskasLincoln, Lincoln, Nebraska 68588-0304
The rates at which peptide amide hydrogens in folded proteins undergo isotopic exchange are reduced by factors of 100-10-8 relative to exchange rates at the same peptide linkages in unfolded proteins. To measure the isotopic exchange rates of the most rapidly exchanging peptide amide hydrogens in proteins, a flow-quench deuterium exchange-in step has been added to the protein fragmentation/mass spectrometry method (Zhang, Z.; Smith, D. L. Protein Sci. 1993, 2, 522-531). Isotopic exchange rates in eight short segments spanning the entire backbone of cytochrome c have been determined for exchangein times of 0.2-120 s. These results show that the isotopic exchange rates of 10 of the peptide amide hydrogens in cytochrome c are similar to those expected for unfolded cyt c, while the exchange rates for 33 other non-hydrogen-bonded amide hydrogens are much less than expected for unfolded cyt c. Since the isotopic exchange rates of the most rapidly exchanging amide hydrogens in folded proteins are a direct measure of their access to the aqueous solvent, the ability to determine these isotopic exchange rates points to the possibility of using quenched-flow amide hydrogen exchange and mass spectrometry as a tool for identifying protein surfaces involved with binding. Peptide linkages for all of the common amino acids, except proline, have one amide hydrogen that may be replaced with deuterium when a polypeptide is incubated in D2O. For short polypeptides with no secondary or tertiary structure, isotopic exchange of the peptide amide hydrogens is complete within seconds at pH 7. In addition, isotopic exchange rates in unstructured peptides span a 30-fold range and depend on inductive effects of the side chains flanking the amide hydrogens.1,2 This behavior contrasts with that found for large, folded polypeptides, where isotopic exchange of the peptide amide hydrogens is slowed by as much as 108. Complete exchange of some peptide amide hydrogens requires months of continuous incubation in D2O. The reduced rates of isotopic exchange in folded proteins are attributed primarily to intramolecular hydrogen bonding, as well as access of the amide hydrogens to the aqueous solvent.3,4 Amide hydrogen exchange has, therefore, become an important tool for (1) Molday, R. S.; Englander, S. W.; Kallen, R. G. Biochemistry 1972, 11, 150158. (2) Bai, Y.; Milne, J. S.; Mayne, L.; Englander, S. W. Proteins 1993, 17, 7586. (3) Woodward, C.; Simon, I.; Tu ¨ chsen, E. Mol. Cell. Biochem. 1982, 48, 135160. (4) Englander, S. W.; Kallenbach, N. R. Q. Rev. Biophys. 1984, 16, 521-655.
2340 Analytical Chemistry, Vol. 68, No. 14, July 15, 1996
probing protein structure and dynamics. The extreme sensitivity of amide hydrogen exchange rates to protein structure and dynamics can be illustrated with cytochrome c (cyt c). Although the X-ray crystallographic structures of oxidized and reduced cyt c are very similar,5,6 isotopic exchange rates of some amide hydrogens differ by as much as 30-fold.7 Results of several other studies in which amide hydrogen exchange was used as a probe of protein structure and dynamics have been reported.8-12 High-resolution nuclear magnetic resonance (NMR) has been used to measure isotopic exchange rates at specific peptide amide linkages in several small (Mr < 20 000) proteins. Most of these studies have been limited to analysis of amide hydrogens with isotopic exchange half-lives greater than 10 h because this is the time required to make the analyses. However, exchange rates for amide hydrogens with half-lives as short as minutes have been measured for 15N-labeled proteins.13,14 Gemmecker et al.15 have reported amide hydrogen exchange rates in the 50-0.05 s-1 range for a protein doubly labeled with 13C and 15N. Early studies by Tu¨chsen and Woodward, performed at low pH to slow hydrogen exchange, showed that the rapidly exchanging amide hydrogens on the surfaces of proteins are an important source of information regarding the access of specific peptide amide hydrogens to the solvent.16-18 Amide hydrogens that exchange rapidly in the folded forms of proteins have not been used to probe protein structure and dynamics, because general experimental methods suitable for measuring isotopic exchange rates in proteins on the second-to-minute time scale have not been available. The present study was directed toward developing a general method for determining isotopic exchange rates of rapidly exchanging amide hydrogens in proteins. Results of this study demonstrate that a flow-quench system can be used with the (5) Takano, T.; Dickerson, R. E. J. Mol. Biol. 1981, 153, 95-115. (6) Berghuis, A. M.; Brayer, G. D. J. Mol. Biol. 1992, 223, 959-976. (7) Marmorino, J. L.; Auld, D. S.; Betz, S. F.; Doyle, D. F.; Young, G. B.; Pielak, G. J. Protein Sci. 1993, 2, 1966-1974. (8) Kim, K. S.; Fuchs, J. A.; Woodward, C. K. Biochemistry 1993, 32, 96009608. (9) Loh, S. N.; Rohl, C. A.; Baldwin R. L. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 1982. (10) Bai Y.; Sosnick, T. R.; Mayne, L.; Englander, S. W. Science 1995, 269, 192197. (11) Loh, S. N.; Prehoda, K. E.; Wang, J.; Markley, J. L. Biochemistry 1993, 32, 11022-11028. (12) Zhang, Z.; Smith, D. L. Protein Sci., in press. (13) Schulman, B. A.; Kim, P. S. Protein Sci. 1994, 3, 2226-2232. (14) Andrec, M.; Hill, R. B.; Prestegard, J. H. Protein Sci. 1995, 4, 983-993. (15) Gemmecker, G.; Jahnke, W.; Kessler, H. J. Am. Chem. Soc. 1993, 115, 11620-11621. (16) Tu ¨ chsen, E.; Woodward, C. J. Mol. Biol. 1985, 185, 405-419. (17) Tu ¨ chsen, E.; Woodward, C. J. Mol. Biol. 1985, 185, 421-430. (18) Tu ¨ chsen, E.; Woodward, C. J. Mol. Biol. 1987, 193, 793-802. S0003-2700(96)00152-7 CCC: $12.00
© 1996 American Chemical Society
Figure 1. Diagram of the flow-quench system used to achieve deuterium exchange-in times of 0.2-120 s.
protein fragmentation/mass spectrometry method19,20 to determine isotopic exchange rates of the most rapidly exchanging peptide amide hydrogens in cyt c. Amide hydrogen exchange rate constants spanning the range from 5 to 0.005 s-1 are reported for peptide amide hydrogens located in eight segments that collectively cover the entire backbone of cyt c. EXPERIMENTAL SECTION Materials. Horse heart cyt c, pepsin, anhydrous monobasic potassium phosphate, anhydrous dibasic potassium phosphate, and trifluoroacetic acid were purchased from Sigma Chemical Co.; deuterium oxide (99.9 atom %) was purchased from Cambridge Chemical Co.; and Sephadex G-25 was purchased from Pharmacia Biotech. All chemicals were used without further purification. Static mixing tees, PEEK tubing, and various HPLC tubing were purchased from Upchurch Scientific. Cytochrome c in which all exchangeable hydrogens had been replaced with deuterium was prepared by incubating the protein in D2O buffer (pD 6.8) at 70 °C for 3 h. Flow-Quench Apparatus. The flow-quench apparatus illustrated in Figure 1 was constructed to effect isotopic exchangein times from 0.2 s to 120 s. The test protein, cyt c (10 µg/µL), was injected into a stream of H2O (5 µL/min) via a Rheodyne 8125 HPLC injector equipped with a 5 µL loop. At the first mixing tee (Model U-466), the protein was mixed with a much larger flow (100 µL/min) of D2O phosphate buffer (0.1 M, pD 6.8) at ambient temperature. Isotopic exchange was initiated at this tee and terminated at the next tee, where the protein solution was mixed with the quench buffer (0.3 M phosphate) to decrease the pH to 2.2. The deuterium exchange-in time was changed from 0.2 to 120 s by using different tubing diameters (0.005-0.30 in.) and flow rates (100-200 µL/min). The quench flow also contained an acid protease, pepsin, which fragmented the test protein into peptides (substrate/enzyme 1:1). The digestion time (4 min) was determined from the flow rate and the volume of the tubing joining the second mixing tee and the second Rheodyne 8125 (19) Zhang, Z.; Smith, D. L. Protein Sci. 1993, 2, 522-531. (20) Liu, Y.; Smith, D. L. J. Am. Soc. Mass Spectrom. 1994, 5, 19-28.
injector, which was located on a capillary HPLC system. The loop on the second injector had a volume of 50 µL. This injector was switched to the “inject” mode at the appropriate time following protein injection at the first injector to capture a portion of the proteolytically fragmented test protein. Peptides present in the digest were fractionated by reversed-phase HPLC using a 150 mm × 0.32 mm column packed with POROS 10 R2 perfusive media (PerSeptive Biosystems). Following a 3 min desalting period, peptides were eluted by a gradient of 0-60% CH3CN in 5 min. The flow rate of the mobile phases (H2O and CH3CN, both 0.05% trifluoroacetic acid) was 25 µL/min. The flow to the first HPLC injector was controlled by an Applied Biosystems syringe pump, while the other flows to the mixing tees were controlled by a Harvard Apparatus syringe pump equipped with a multiple syringe holder. The flow through the HPLC system was controlled by a Rainin pumping system. To minimize artifactual isotopic exchange during digestion, the digestion loop, the second HPLC injector, and the HPLC column were cooled to 0 °C. Mass Spectrometry. The molecular weights of partially deuterated peptides were determined by monitoring the HPLC effluent with a VG Micromass Autospec (VG Analytical) highresolution mass spectrometer (MS) equipped with a focal plane detector and a standard electrospray ionization source. The total flow from the capillary HPLC (25 µL/min) was directed to the electrospray ionization source, which was operated at 60 °C. The nebulizing and drying gas flows were 16 and 400 L/h, respectively. All mass spectra were acquired in the focal plane detection mode by scanning from 400 to 1600 u in 10 steps, where ∼18% of the central m/z was recorded in each step. Each scan required ∼3 s for completion. The resolution with the array detector was 2000 fwhm. Mass spectra were recorded in the profile mode with an OPUS data system (VG Analytical). The mass spectrometer was calibrated with a mixture of peptides and myoglobin. RESULTS Test for Rapid Mixing. Rapid and complete mixing of reagents at the tees (Figure 1) is essential for short isotope exchange times. The mixing efficiency of the tees used in this study was examined by measuring the extent to which streams of acid and base were mixed, as judged using phenolphthalein indicator.21 Base plus indicator (0.01 M NaOH, 100 µL/min) entered the tee through one arm, while acid (0.3 M, 5 µL/min) entered through the other arm. Flow rates were doubled for some measurements. The mixing tee exit was connected to a 0.5 µL cell in a UV/visible HPLC detector by various lengths of PEEK tubing. Different lengths of tubing joining the tee to the detector were used to limit the mixing time. The phenolphthalein indicator was monitored at 553 nm. For complete mixing, as judged from the longest mixing time (120 s), the absorbance decreased by 0.050 au. As the minimum mixing time was decreased to 0.18 s, the decrease in absorbance was unchanged, indicating that mixing was complete in less than 0.18 s. Protein Fragmentation/MS. Amide hydrogen exchange in small and large proteins can be followed by the protein fragmentation/MS method,19,20 which consists of five steps: (i) incubating a protein in D2O at specified pD, temperature, and time to achieve partial exchange; (ii) quenching the isotopic exchange reaction by decreasing the pD to 2-3 and decreasing the temperature to (21) Owens, G. D.; Taylor, R. W.; Ridley, T. Y.; Margerum. D. W. Anal. Chem. 1980, 52, 130-138.
Analytical Chemistry, Vol. 68, No. 14, July 15, 1996
2341
Figure 2. Amino acid sequence of horse heart cyt c indicating peptides identified following peptic digestion (enzyme/substrate 1:1, 0 °C, 4 min) in the flow-quench apparatus depicted in Figure 1.
0 °C; (iii) fragmenting the partially deuterated protein into peptides with an acid protease, such as pepsin; (iv) separating the peptides by reversed-phase HPLC; and (v) analyzing peptides as they elute from the HPLC by mass spectrometry to determine their molecular weights. Since the half-life for isotopic exchange of peptide amide hydrogens is ∼1 h under the quench conditions, all samplehandling variables, including enzyme/substrate ratio, digestion time, and HPLC conditions, were optimized for speed. This requirement demands use of low substrate/enzyme ratios (typically 1:1) and short digestion times (3-5 min). High-speed perfusion chromatography is used because it facilitates rapid fractionation of the peptides in the digest. It is important to note that the HPLC separation was performed with nondeuterated solvents, which effectively removed deuterium from the side chains and amino/carboxyl termini of the peptides.2 Isotopic exchange at these positions is much faster than at the peptide linkages. Therefore, the increase in peptide molecular weight is a direct measure of deuteration at peptide amide linkages. The protein fragmentation/MS method has been successfully implemented with fast atom bombardment12,19,20,22 and electrospray ionization mass spectrometry (ESIMS).23 For the digestion conditions used in this study, 22 peptides were detected by ESIMS. This ensemble of peptides, illustrated in Figure 2, has been used to determine the extent of isotopic exchange along the entire length of the cyt c backbone. Results for a subset of eight segments, sensing isotopic exchange at 94% of all peptide amide linkages in cyt c, are presented in Table 1 (measd kfast and measd kslow). All assignments were confirmed by MS/MS or carboxy-terminal sequencing using carboxy peptidases. Although the spatial resolution of the hydrogen exchange information is generally limited to the length of the peptides, increased resolution may be achieved for some segments when (22) Zhang, Z.; Post, C. B.; Smith, D. L. Biochemistry 1996, 35, 779-791. (23) Johnson, R. S.; Walsh, K. A. Protein Sci. 1994, 3, 2411-2418.
2342
Analytical Chemistry, Vol. 68, No. 14, July 15, 1996
overlapping peptides are found.22 For example, deuterium exchange into the segment containing peptide linkages 11-21 has been determined from the difference in the deuterium levels found in segments 1-21 and 1-10. The deuterium levels detected in two segments that include peptide amide linkages 49-64 and 96-104 following incubation of cyt c in D2O for 0.2-120 s are presented in Figure 3. Since the analysis time, which consists primarily of the 4 min digestion and 3-8 min HPLC fractionation, is a small but significant fraction of the half-life for isotopic exchange under quench conditions, adjustments for artifactual exchange during analysis have been made to these data, as described previously.19 Analysis of completely deuterated cyt c showed that 75% of deuterium located at the peptide amide linkages was recovered for the experimental conditions used in this study. Since the spatial resolution of the protein fragmentation/MS method is usually insufficient for determining isotopic exchange rates at specific amide linkages, deuterium exchange-in results, as presented in Figure 3, have been analyzed to estimate the distribution of exchange rates within short segments of proteins.22 When the deuterium concentration in the solution is large, and the pH and temperature are constant, isotopic exchange of each amide hydrogen follows first-order kinetics. Since each peptide contains several amide hydrogens, the deuterium content of a peptide can be described by the sum of N exponential terms: N
D)N-
∑exp(-k t) i
(1)
i)1
where D is the deuterium content of a peptide, ki is the exchange rate constant for each amide hydrogen, and t is the time allowed for isotopic exchange. Isotopic exchange rate constants for peptide amide hydrogens located within each segment were determined by varying ki to obtain the best fit between eq 1 and the exchange-in data. When this analysis indicates that several hydrogens have similar exchange rates, the average rate constant is given in Table 1. The number of hydrogens exhibiting the average rate constant is given in parentheses. Application of eq 1 to deuterium exchange-in data for the 49-64 segment (Figure 3a) indicates that, among the 16 amide hydrogens in the 49-64 segment, three have exchange rate constants of ∼4.6 s-1, one has an exchange rate constant of 0.18 s-1, one has an exchange rate constant of 0.07 s-1, and 11 have exchange rate constants less than 0.01 s-1. Similar analysis of the 96-104 segment (Figure 3b) shows that one amide hydrogen has an exchange rate constant of 0.02 s-1, and eight have rate constants less than 0.01 s-1. Isotopic exchange rate constants derived from deuterium exchangein data for the eight peptides used in this study are presented in Table 1. DISCUSSION Base-catalyzed amide hydrogen exchange rates in short polypeptides depend primarily on the ability of adjacent side chains to stabilize the imide intermediate.1,2 Studies of model peptides facilitate calculation of isotopic exchange rates in unfolded peptides of any amino acid sequence. This information is important because it provides a basis for comparing isotopic exchange rates in folded proteins with exchange rates expected for the same protein in an unfolded state. The effect of protein folding on
Table 1. Isotopic Exchange Rates of Peptide Amide Hydrogens Located in Eight Segments of Oxidized Cyt c segmenta 1-10 11-21
measdb kfast (s-1)
measdc kslow (s-1)
total NHd
calcde kmin (s-1)
non-hydrogen-bonded NHf
non-hydrogen-bonded NH with k < kmin
1.4 (1) >5 (1) 0.7 (1)
