Anal. Chem. 1998, 70, 1198-1202
Ion/Ion Proton-Transfer Kinetics: Implications for Analysis of Ions Derived from Electrospray of Protein Mixtures Scott A. McLuckey,* James L. Stephenson, Jr., and Keiji G. Asano
Chemical and Analytical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6365
Protein ions of different mass and charge but similar mass-to-charge ratios are shown to undergo significantly different rates of differential neutralization, defined as the rate of change of charge with time, upon initiation of reactions with oppositely charged ions in the quadrupole ion trap. Overlapping charge state distributions arising from mixtures of ions of dissimilar charge are separated on the mass-to-charge scale at short reactions times. It is also demonstrated that the time frame for near total neutralization, defined as charge reduction to the 1+ ion, is relatively insensitive to initial charge state. It is shown, for example, that the (M + 11H)11+-(M + 22H)22+ ions derived from horse skeletal muscle apomyoglobin yield the (M + H)+ ion as the major ion/ion reaction product over the same reaction period that largely converts doubly protonated bradykinin to the singly protonated species. Less than 25% of the bradykinin ions are expected to be totally neutralized when roughly 7% of the myoglobin ions are expected to be totally neutralized. The phenomenon of significantly different initial differential neutralization rates for ions of dissimilar charge, and the relative insensitivity to ion charge for total neutralization, can be used to advantage in strategies for protein ion mixture analysis. The well-known tendency for the formation of multiply charged ions in the electrospray of biopolymers1-4 has facilitated mass measurement of high-molecular-weight species using mass analyzers of relatively modest mass-to-charge range. Typically, a distribution of ion charge states is observed for a given biopolymer which allows for a straightforward determination of the charges of the ions and, as a consequence, biopolymer mass. However, multiple charging can complicate the analysis of mixtures because it can lead to extensive overlap in mass-to-charge of the charge state distributions associated with the various mixture components. The complexity of the mixture for which the masses of all
of the mixture components can be determined is limited by the extent of multiple charging, the resolving power of the mass analyzer, and/or any matrix effects in ionization. Algorithms have been described that allow for the deconvolution of overlapping charge state distributions, thereby facilitating the identification of mixture components.5-7 This approach, however, is most reliable for very simple mixtures. We recently demonstrated how the proton-transfer reactions of multiply charged protein ions to singly charged anions can be used to reduce the extent of multiple charging within a quadrupole ion trap.8-13 This capability has been shown to allow for the identification of mixture components that otherwise were not apparent in the electrospray mass spectrum.9 In this report, we discuss the implications of the kinetics of ion/ ion reactions on the separation of ions of different mass (m) and charge (z) but similar mass-to-charge (m/z) ratios. Recognition of these implications suggests different strategies for the separation of mixtures of ions, depending on the relative magnitudes of the charges associated with each component. EXPERIMENTAL SECTION The proteins chicken egg white lysozyme, horse skeletal muscle apomyoglobin, and chicken conalbumin were obtained from Sigma Chemical Co. (St. Louis, MO). Perfluoro-1,3-dimethylcyclohexane (PDCH) was purchased from Aldrich (Milwaukee, WI). Solutions for electrospray were prepared by dissolving the sample in 1 mL to give a concentration of 1-4 µM in 50:50 methanol/water (v/v). Acetic acid was added to 1% for all protein solutions. All solutions were infused at rates of 1.0-3.0 µL/min through a 120-µm-i.d. needle held at a potential of +3500-+4000 V. All experiments were carried out with a homemade electrospray source coupled with a Finnigan-MAT (San Jose, CA) ion
* To whom correspondence should be addressed. Phone: (423) 574-2848. Fax: (423) 576-8559. E-mail:
[email protected]. (1) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science 1989, 246, 64-71. (2) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Mass Spectrom. Rev. 1990, 9, 37-70. (3) Smith, R. D.; Loo, J. A.; Edmonds, C. G.; Barinaga, C. J.; Udseth, H. R. Anal. Chem. 1990, 62, 882-899. (4) Smith, R. D.; Loo, J. A.; Ogorzalek Loo, R. R.; Busman, M.; Udseth, H. R. Mass Spectrom. Rev. 1991, 10, 359-451.
(5) Mann, M.; Meng, C. K.; Fenn, J. B. Anal. Chem. 1989, 61, 1702-1708. (6) Reinhold, B. B.; Reinhold, V. N. J. Am. Soc. Mass Spectrom. 1992, 3, 207215. (7) Labowsky, M.; Whitehouse, C.; Fenn, J. B. Rapid Commun. Mass Spectrom. 1993, 7, 71-84. (8) Stephenson, J. L., Jr.; McLuckey, S. A. J. Am. Chem. Soc. 1996, 118, 73907397. (9) Stephenson, J. L., Jr.; McLuckey, S. A. Anal. Chem. 1996, 68, 4026-4032. (10) Stephenson, J. L., Jr.; McLuckey, S. A. J. Am. Chem. Soc. 1997, 119, 16881696. (11) Stephenson, J. L., Jr.; Van Berkel, G. J.; McLuckey, S. A. J. Am. Soc. Mass Spectrom. 1997, 8, 637-644. (12) Stephenson, J. L., Jr.; McLuckey, S. A. Int. J. Mass Spectrom. Ion Processes 1997, 162, 89-106. (13) Stephenson, J. L., Jr.; McLuckey, S. A. Anal. Chem. 1997, 69, 3760-3766.
1198 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
S0003-2700(97)01013-5 CCC: $15.00
© 1998 American Chemical Society Published on Web 02/14/1998
trap mass spectrometer, modified for injection of ions formed external to the ion trap through an end-cap electrode. Details of the electrospray/ion trap interface have been described.14 Further modifications have been made to this apparatus to allow anions, formed by glow discharge ionization, to be injected through a 3-mm-diameter hole drilled through the ring electrode. The discharge was gated via software control15 so that it was on only during the anion accumulation period to avoid detector noise observed when the discharge is operated continuously.16,17 Details of these modifications and others made to facilitate analysis of high mass-to-charge ions are reported elsewhere.12 Cations formed by electrospray were injected axially into the ion trap for periods ranging from 0.05 to 0.4 s. The radio frequency (rf) sine-wave amplitude applied to the ring electrode during ion injection ranged from 700 to 1200 V zero-to-peak. In all cases, helium was admitted into the vacuum system to a total pressure of 1 mTorr, with a background pressure in the instrument of 2 × 10-5 Torr, without the addition of helium. Anions were formed by sampling the headspace vapors of PDCH into the glow discharge operated at 850 mTorr and were injected after cation accumulation. For all ion/ion reactions described, anion accumulation periods ranged from 15 to 28 ms. Mutual storage times for ion/ion reactions varied from 30 to 115 ms. Residual anions were ejected prior to cation mass/charge analysis. Mass/charge analysis was effected after the completion of all ion isolation and reaction periods using resonance ejection18 to yield a mass/charge range (for singly charged apomyoglobin) as high as 18 000, using resonance ejection amplitudes of 2-3 V peak-to-peak. The mass/charge scale for the ion/ion reaction spectra was calibrated initially using the electrospray mass spectra of apomyoglobin. In this work, the mass/charge ratios of the various charge states of the parent compound were known and could be used to determine a correction for the mass scale provided by the ion trap data system. As charges were removed and the mass scale was extended, calibrations were made in a stair-step fashion.9 The mass/charge ratios of ions which were measured at a lower mass range extension factor were used to determine the new mass range extension factor when the mass range was extended further. In this way, the mass scale associated with the largest mass range extension factor could be related back to the original mass scale calibration. The spectra shown here were typically the result of an average of 50-100 individual scans. RESULTS AND DISCUSSION We have noted that the rates for ion/ion proton-transfer reactions involving multiply charged proteins follow a chargesquared dependence under typical quadrupole ion trap operating conditions (viz., using a helium bath gas pressure of roughly 1 mTorr) and when there is a large excess of anions.8 We have also noted that the predicted rate for an ion/ion capture collision, (14) Van Berkel, G. J.; McLuckey, S. A.; Glish, G. L. Anal. Chem. 1990, 62, 1284-1295. (15) ICMS software provided by N. Yates and the University of Florida. (16) McLuckey, S. A.; Goeringer, D. E.; Asano, K. G.; Vaidyanathan, G.; Stephenson, J. L., Jr. Rapid Commun. Mass Spectrom. 1996, 10, 287-298. (17) Duckworth, D. C.; Smith, D. H.; McLuckey, S. A. J. Anal. At. Spectrom. 1997, 12, 43-48. (18) Kaiser, R. E., Jr.; Cooks, R. G.; Moss, J.; Hemberger, P. H. Rapid Commun. Mass Spectrom. 1989, 3, 50-53.
kc, in which the collision partners form a stable orbiting complex, is also expected to follow a charge-squared relationship. That is,
kc ) νπ[z1z2e2/(µν2)]2
(1)
where ν is the relative velocity of the oppositely charged ions, µ is the reduced mass of the collision partners, and z1 and z2 are the charges of the cations and anions, respectively. While it is not firmly established that eq 1 provides for a quantitative prediction of the rate constant, due to uncertainty in the measurement of the anion density, the predicted z12 dependence has been noted consistently in our laboratory. Given the highly exothermic nature of mutual neutralization, these reactions can be regarded as irreversible. It is, therefore, straightforward to model the time evolution of the product cations via differential equations describing a set of consecutive, irreversible, pseudo-first-order reactions.11 We have noted that the time evolution of the cation reactants and products can be fit with a single rate, R, corresponding to that of the singly charged cation reacting with the singly charged anion. The reaction rates for the higher cation charge states are simply taken as n2R, where n is the cation charge.11 Modeling the cationic product ion evolution in this way is useful because it allows for the prediction of reaction times necessary to reach a desired product ion charge distribution under typical reaction conditions. As a result of the charge-squared dependence of the reaction rates, separation of ions of different m and z but similar m/z ratio can occur quickly or slowly, depending on the relative magnitudes of the charges. For example, in a scenario in which there are many mixture components yielding similar charge state distributions, it might be necessary to reduce the charges to 2+ and 1+ in order to provide sufficient separation in the m/z dimension to allow for identification of all mixture components. This scenario typically requires relatively long reaction times. On the other hand, the separation of components of widely different z can occur very quickly and may not require reducing charge states of all mixture components to 2+ and 1+. In fact, it may be desirable to reduce charge states only to the extent necessary to allow for the individual components to be identified. In discussing ion/ion proton-transfer reactions for the purpose of mixture analysis, it is useful to establish definitions for total neutralization, near total neutralization, and differential neutralization. The definition for total neutralization, as it applies to the multiply charged cation MHnn+, is the removal of all excess protons to yield the neutral molecule. Near total neutralization for the mixture analysis context is defined as reduction to the 1+ charge state. As shown below, there may also be significant numbers of 3+ and 2+ ions and neutral molecules at reaction times leading predominantly to the 1+ charge state. Differential neutralization is simply the rate of change in cation charge as a function of reaction time. The rapid separation in mass-to-charge of mixture components with significantly different differential neutralization rates is illustrated in Figure 1. Figure 1a shows the electrospray mass spectrum of chicken egg white lysozyme (14.3 kDa) and chicken conalbumin (76.2 kDa), wherein the 9+, 8+, and 7+ charge states of lysozyme and the 50+-37+ charge states of chicken conalbumin show significant overlap in mass-to-charge. In particular, the Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
1199
Figure 2. (a) Electrospray mass spectrum of a mixture containing bradykinin (charge states italicized) and horse skeletal muscle apomyoglobin. (b) Spectrum obtained after subjecting the ions of (a) to reaction with anions derived from PDCH for 30 ms.
Figure 1. (a) Electrospray mass spectrum (resonance ejection at 45 kHz) of a mixture containing chicken egg white lysozyme and chicken conalbumin (charge states italicized). (b) Spectrum obtained after subjecting the ions of (a) to reaction with anions derived from PDCH for 30 ms (resonance ejection at 45 kHz). (c) Spectrum obtained after an ion/ion reaction period of 50 ms (resonance ejection at 35 kHz).
7+ lysozyme ion and 38+ conalbumin ions are not resolved. In this case, it is clear that at least two components are present in this spectrum, due to the high abundance of the 9+ and 8+ lysozyme ions. It would be far less clear if the abundances of these ions were comparable to or less than those of conalbumin, such that they would essentially be buried within the conalbumin charge state envelope. Figure 1b shows the spectrum resulting from the reaction of the protein cations with anions derived from PDCH for a reaction period of 30 ms. Measurably different 1200 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
differential neutralization rates are observed for the lysozyme and conalbumin ions. For example, the most abundant lysozyme cations in Figure 1a,b are the 8+ and 7+ ions, respectively, whereas for conalbumin they are the 43+ and 18+-22+ ions, respectively. The lysozyme cations tend to lose two or three protons over this time period, while the conalbumin ions tend to lose over 20 protons. If the lysozyme and conalbumin ions were to lose the same charge fraction with time, they would shift along the mass-to-charge scale at the same rate and continue to tend to show overlap in mass-to-charge. As is clear in Figure 1b, very little overlap between the two components remains after 30 ms of reaction time. The separation of mixture components is even greater after 50 ms of reaction time, as shown in Figure 1c. The 7+-3+ lysozyme ions appear at m/z values less than 5000, whereas only the most highly charged conalbumin ions, 15+17+, appear in the plotted m/z range. Most of the signal arising from conalbumin is shifted to m/z values greater than 6000 (data not shown). It is clear upon inspection of eq 1 that it predicts protein charge to be the most important parameter in determining the differences in reaction rates for the lysozyme and conalbumin ions (the anion charge, z2, is constant in these studies at 1-). The differences in relative velocity and reduced mass, the other variables in eq 1, are very small for lysozyme and conalbumin cations reacting with anions derived from PDCH. The conalbumin cations are, there-
Figure 3. (a) Spectrum acquired, using a resonance ejection frequency of 12 100 Hz, after the bradykinin/apomyoglobin mixture was subjected to ion/ion reactions for 115 ms. (b) Same as (a) except that a resonance ejection frequency of 89 202 Hz was used.
fore, expected to react much more rapidly than the lysozyme cations, as long as they are more highly charged. That is, the differential reactivity of the initial conalbumin cations is significantly greater than that of the lysozyme cations. The relative time frame for near total neutralization, however, is a related but different issue. If the time frame for reducing highly charged ions to 1+ is insensitive to the initial charge, all mixture components can be expected to yield 1+ ions using the same reaction period. However, if the time frame for near total neutralization is highly sensitive to initial reactant cation charge, different mixture components would reach the 1+ charge state over different reaction periods, thereby complicating a mixture analysis strategy based on ion/ion reactions designed to form singly charged ions. We have found the time frames for near total neutralization to be remarkably insensitive to initial cation charge, which is consistent with predictions based on eq 1 (see below). This finding is illustrated here for a rather extreme case in which the reactant cation charge states of the mixture components differ by as much as a factor of 10. Figure 2a shows the electrospray mass spectrum of a mixture containing bradykinin and horse skeletal muscle apomyoglobin. The major cation formed from bradykinin via electrospray is the doubly protonated molecule. If any singly protonated ions are formed, they are not resolved from the more abundant 16+ ions of apomyoglobin. (Electrospray of bradykinin in the absence of apomyoglobin yields a singly charged ion signal less than 5% of the doubly charged
Figure 4. (a) Simulated ion abundance versus time curves for the (M + 16H)16+ ion of apomyoglobin and its products, using a singly charged cation/anion rate of 4.5 s-1. (b) Simulated ion abundance versus time curves for the bradykinin (M + 2H)2+ ion and its products, using a singly charged cation/anion rate of 4.5 s-1.
ion.) Figure 2b shows the spectrum obtained after the cations were allowed to react with anions derived from PDCH for 30 ms. All of the apomyoglobin ions have moved to mass-to-charge values greater than m/z 1061, the mass-to-charge of protonated bradykinin. Essentially all of the apomyoglobin ions with more than 14 charges have disappeared within 30 ms, whereas most of the doubly charged bradykinin ions remain, illustrating once again the significant difference in differential neutralization for ions of similar mass-to-charge but very different numbers of charges. Figure 3 compares spectra of the bradykinin/apomyoglobin mixture acquired after an ion/ion reaction period of 115 ms. A different resonance ejection frequency was used to collect each spectrum, 12 100 Hz for Figure 3a and 89 202 Hz for Figure 3b, because the entire mass-to-charge range necessary to observe all of the ions from doubly protonated bradykinin (m/z 531.5) to singly protonated apomyoglobin (m/z 16,950) could not be accessed with a single resonance ejection frequency. This comparison shows that the reaction time necessary to make singly charged apomyoglobin the most abundant myoglobin ion also results in the 1+ species of bradykinin being the most abundant bradykinin ion. Furthermore, a small, doubly protonated bradykinin ion signal remains while essentially all of the apomyoglobin ions of charge 4+ and greater are depleted. The spectra of Figure 3, however, do not directly show the fraction of ions from each component that are totally neutralized. Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
1201
Figure 4 compares simulated relative abundance curves, including those of the totally neutralized species, for the 16+ apomyoglobin ion and its products (Figure 4a) and the bradykinin dication (Figure 4b) and its products as a function of reaction time. In each case, consecutive irreversible reactions are assumed, with the reaction rates for each cation taken as n2R, where R ) 4.5 s-1. This rate was chosen to give the approximate product ion abundances observed in the data of Figure 3. These curves illustrate the relatively high rate of differential neutralization at high charge states and the similar time frame for near total neutralization. Under the conditions of this simulation, roughly 7% of the 16+ ions are totally neutralized after 115 ms of reaction. Although bradykinin starts with an initial charge state of 2+, only about 24% of the initial ion population is totally neutralized after 115 ms of reaction time. As a result of the charge-squared reaction rate dependence, the more highly charged apomyoglobin ions can nearly catch up to the bradykinin ions, in terms of charge state distribution, prior to total neutralization. CONCLUSIONS Ion/ion proton-transfer reaction rates are highly sensitive to the reactant ion charge states. Highly charged ions react dramatically faster than ions of low charge. This situation leads to significantly different initial differential neutralization rates for
1202 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
ions of different charge, such that ions of similar mass-to-charge but different charge separate on the mass-to-charge scale very quickly. In terms of protein mixture analysis, this situation is most likely to prevail in mixtures comprised of small proteins or polypeptides of moderate to low charge present with larger proteins of higher average charge. While initial differential neutralization rates can vary dramatically with protein ion charge, the time frame for near total neutralization is relatively insensitive to initial charge state. Therefore, significant fractions of each mixture component, regardless of initial charge state distribution, can be observed as singly charged ions over an appropriately chosen range of reaction times. This situation allows for a mixture analysis strategy designed to reduce charge states to low values, primarily 2+ and 1+, using a single reaction period. ACKNOWLEDGMENT This work was supported by the National Institutes of Health under Grant R01GM45372. Oak Ridge National Laboratory is managed for the U.S. Department of Energy under Contract DEAC05-96OR22464 by Lockheed Martin Energy Research Corp. Received for review September 15, 1997. January 15, 1998. AC9710137
Accepted
