Stabilization of Gas-Phase Noncovalent Macromolecular Complexes

Mar 10, 2001 - (for example, with ammonium bicarbonate or acetate solution). The analytical interest of TEAB for the analysis of macromolecular specie...
0 downloads 0 Views 154KB Size
Anal. Chem. 2001, 73, 1699-1706

Stabilization of Gas-Phase Noncovalent Macromolecular Complexes in Electrospray Mass Spectrometry Using Aqueous Triethylammonium Bicarbonate Buffer David Lemaire, Ge´rald Marie, Laurent Serani, and Olivier Lapre´vote*

Laboratoire de Spectrome´ trie de Masse, Institut de Chimie des Substances Naturelles, CNRS, 91198 Gif sur Yvette Cedex, France

The use of triethylammonium bicarbonate (TEAB) solution in electrospray mass spectrometry proved to be a very efficient way for studying proteins or noncovalent protein complexes under “nondenaturing” conditions. The low charge states observed in the mass spectra improve the separation of ions arising from macromolecular species of close masses. Moreover, the multiply charged ions generated in a TEAB solution are significantly more stable than those formed under more conventional conditions (for example, with ammonium bicarbonate or acetate solution). The analytical interest of TEAB for the analysis of macromolecular species that can easily dissociate in the gas phase, such as hemoglobin or other macromolecular noncovalent complexes, is demonstrated. Electrospray ionization (ESI) has emerged as a method of choice for the characterization of biological macromolecules by mass spectrometry.1-3 The mass spectra obtained by ESI show a distribution of peaks corresponding to different charge states of multiply charged ions. The intimate processes of formation of such ions are related to the behavior of molecular species initially solvated, then submitted to a high electric field, and finally transferred into the gas phase. The charge state distributions and hence the underlying mechanisms of protein ion formations depends on factors intrinsic to the protein itself (sequence, conformation, presence or absence of disulfide bridges, number of acidic and basic sites)3-6 and to the solution properties (solvent,7 * Corresponding author: (fax) (33) 169 077 247; (e-mail) [email protected]. (1) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M.Science 1989, 64, 246. (2) Loo, J. A.; Udseth, H. R.; Smith, R. D. Anal. Biochem. 1989, 179, 404. (3) Covey, T. R.; Bonner, R. F.; Shushan, B. I.; Henion, J. Rapid Commun. Mass Spectrom. 1991, 5, 249. (4) Smith, R. D.; Loo, J. A.; Ogorzalek Loo, R. R.; Busman, M.; Udseth, H. R. Mass Spectrom. Rev. 1991, 10, 359. (5) Loo, J. A.; Edmonds, C. G.; Udseth, H. R.; Smith, R. D.Anal. Chem. 1990, 62, 693. (6) Loo, J. A.; Ogorzalek Loo, R. R.; Udseth, H. R.; Edmonds, C. G.; Smith, R. D. Rapid Commun. Mass Spectrom. 1991, 5, 101. (7) Edmonds, C. G.; Loo, J. A.; Barinaga, C. J.; Udseth, H. R.; Smith, R. D. J. Chromatogr. 1989, 474, 21. 10.1021/ac001276s CCC: $20.00 Published on Web 03/10/2001

© 2001 American Chemical Society

pH,8 ionic strength,9 counterions10). Moreover, the distribution of charge states is sensitive to some instrumental parameters such as the voltages applied on the capillary exit, the counter electrode and the sampling cone, the ion source pressure, and the direction and temperature of the bath gas.11-16 Last, gas-phase reactions can occur and affect the number of charges present in protein ions.17-24 In the particular case of noncovalent macromolecular complexes, it is necessary to use experimental conditions that would preserve their native state in the solution phase, as far as possible, during their transfer into the gas phase. For this purpose, the pH is usually maintained at a near-neutral value by using a buffer solution. The volatility of buffer salts is of primary importance for limiting the addition of metal ions or solvent molecules on protein ions. Protein solutions are thus usually buffered by ammonium acetate or ammonium bicarbonate at concentrations lower than 0.1 mol‚L-1. In the present study, we propose to compare these two widely used ammonium salts toward their effect on the mass spectra of various proteins ionized under nondenaturing solution conditions. We will also try to demonstrate the interest of triethylammonium bicarbonate as a novel buffer for investigating protein supramolecular associations. (8) Chowdury, S. K.; Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1990, 112, 9012. (9) Wang, G.; Cole, R. B. Anal. Chem. 1994, 66, 3702. (10) Mirza, U. A.; Chait, B. T. Anal. Chem. 1994, 66, 2898. (11) Ashton, D. S.; Beddell, C. R.; Cooper, D. J.; Green, B. N.; Oliver, R. W. A. Org. Mass Spectrom. 1993, 28, 721. (12) Ikonomou, M. G.; Blades, A. T.; Kebarle, P. Anal. Chem. 1990, 62, 957. (13) Fenn, J. B. J. Am. Soc. Mass Spectrom. 1993, 4, 524. (14) Kebarle, P.; Tang, L. Anal. Chem. 1993, 65, 972A. (15) Gatlin, C. L.; Turecˇek, F. Anal. Chem. 1994, 66, 712. (16) Kelly, M. A.; Vestling, M. M.; Fenselau, C. C.; Smith, P. B. Org. Mass Spectrom. 1992, 27, 1143. (17) Stephenson, J. L., Jr.; McLuckey, S. A. J. Am. Chem. Soc. 1997, 119, 1688. (18) Stephenson, J. L., Jr.; McLuckey, S. A. J. Am. Chem. Soc. 1996, 118, 7390. (19) Ogorzalek Loo, R. R.; Smith, R. D. J. Mass Spectrom. 1995, 30, 339. (20) Ogorzalek Loo, R. R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 1994, 5, 207. (21) Winger, B. E.; Light-Wahl, K. J.; Smith, R. D. J. Am. Soc. Mass Spectrom. 1992, 3, 624. (22) Ogorzalek Loo, R. R.; Loo, J. A.; Udseth, H. R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 1992, 3, 695. (23) McLuckey, S. A.; Van Berkel, G. J.; Glish, G. L. J. Am. Chem. Soc. 1990, 112, 5668. (24) LeBlanc, J. C. Y.; Siu, K. W. M.; Guevremont, R. Anal. Chem. 1994, 66, 3289.

Analytical Chemistry, Vol. 73, No. 8, April 15, 2001 1699

Figure 1. Electrospray mass spectra of 20 µM myoglobin in buffered solution (pH 6.9) of (a) 50 mM NH4HCO3 and (b) 50 mM NH4OAc, recorded at a Vsc - Vs value of 48 V.

Figure 2. Electrospray mass spectra of 20 µM myoglobin in buffered solution (pH 6.9) of (a) 50 mM NH4HCO3 and (b) 50 mM NH4OAc, recorded at a Vsc - Vs value of 144 V.

EXPERIMENTAL SECTION Materials. Bovine pancreatic insulin, hen egg lysozyme, horse heart myoglobin, human hemoglobin A0 and S, bovine serum albumin (monomer and dimer), horse liver alcohol dehydrogenase, and nicotinamide adenine dinucleotide (NADH) were purchased from Sigma (Saint-Quentin Fallavier, France). Ammonium bicarbonate (99%), ammonium acetate (99%), and 1 M aqueous triethylammonium bicarbonate were purchased from Sigma and used without any further purification. Mass Spectrometry. Experiments were performed by using a Zabspec/T mass spectrometer (Micromass, Manchester, U.K.) 1700 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

equipped with an electrospray ionization source. They were carried out at a skimmer voltage (Vs) of 4 kV and the sampling cone potential (Vsc), controlled by the operator, was adjusted to attain Vsc - Vs values between 0 and 240 V. The temperature of the source was held at 100 °C. The mass spectrometer was scanned over the m/z range of interest, and the mass scale was calibrated by injecting a solution of cesium iodide (Aldrich, Saint-Quentin Fallavier, France). Samples were delivered into the electrospray ion source by means of a syringe pump PHD2000 (Harvard Apparatus, Les Ulis, France) at a flow rate of 5 µL‚min-1.

Figure 3. Electrospray mass spectra of 20 µM bovine albumin in buffered solution (pH 7.9) of (a) 20 mM NH4HCO3 and (b) 20 mM TEAB.

Sample Preparation. Protein solutions were desalted by using 10 kDa cutoff ultrafiltration cartridges (YM 10 membrane, Amicon-Millipore) with a Jouan BR4i centrifuge (6000g, 4 °C) prior to mass spectrometric analyses. The 20 µM protein samples were washed with several volumes of 50 mM ammonium or triethylammonium bicarbonate buffers. RESULTS AND DISCUSSION Anion Effect: Acetate vs Bicarbonate. Figure 1 displays two mass spectra of myoglobin dissolved in 50 mM ammonium bicarbonate or acetate aqueous solutions at pH 6.9. The voltage difference between the sampling cone and the skimmer (Vsc Vs) was set at 48 V. In both cases, two ion peaks attributed to the holoprotein (myoglobin and heme) were observed at m/z 1952.9 and 2197.0 corresponding to the 9+ and 8+ charge states, respectively. By increasing the interface voltage difference at 144 V (Figure 2), two new ion peaks appeared at m/z 1884.4 and 2119.9, indicating the presence of apomyoglobin with identical charge numbers (9+ and 8+, respectively). The gas-phase dissociation efficiency of holomyoglobin into apoprotein and heme was distinctly different according to the buffer salt present in the solution phase. The dissociation threshold of the complex protein/ heme formed in the ammonium acetate solution was clearly lower than that of the same species generated in the presence of ammonium bicarbonate. This difference could be related to the hydration energies of the two anions, bicarbonate (-710 kJ‚mol-1) and acetate (-695 kJ‚mol-1).25 Under given experimental conditions, the energy transferred collisionally to the ion species entering the interface region between the atmospheric pressure ion source and the low-pressure area following the skimmer achieves the desolvation of ions and is further deposited as internal energy on the fully desolvated ions. Thus, if a greater amount of (25) Handbook of Chemistry and Physics, 47th ed.; CRC Press: Cleveland, OH, 1966.

energy is needed for the desolvation of proteins interacting with HCO3-, the available energy for the collisional activation of ions is expected to be reduced. Hence, to attain the same dissociation rate as that of ions arising from the ammonium acetate solution, it is necessary to enhance the collision energy by a significant increase of the sampling cone voltage. Moreover, the charge states of the protein ions being identical irrespective of the anion used and the voltage applied to the sampling cone, the proton affinities of acetate and bicarbonate are probably very similar.10 Effect of Cation on the Charge-State Distribution: Ammonium vs Triethylammonium. Triethylammonium bicarbonate (TEAB) is often employed for the purification and preparation of oligonucleotides in view of their analysis by electrospray mass spectrometry in negative ion mode.26 The very different gas-phase basicities of triethylamine (221 kcal‚mol-1) and ammonia (193 kcal‚mol-1) prompted us to undertake a comparative study of the behavior of various protein samples in 20 mM ammonium and triethylammonium bicarbonate solutions. The positive ion mass spectrum of the covalent dimer of bovine albumin (form D, 133 kDa) dissolved in ammonium bicarbonate showed an envelope of six ion peaks, corresponding to the charge states 23+ to 28+ of the protein, the 26+ ion giving rise to the most abundant signal (Figure 3a). For m/z values between 3500 and 4500, some peaks were attributed to the monomeric form of albumin (M). A different behavior, with regard to the extent of protonation, was observed in the case of a protein solution containing 20 mM TEAB (Figure 3b). The most striking feature of the spectrum was indeed a spectacular decrease of the average charge state, the peaks envelope being centered around the 14+ species. To generalize this observation, six other proteins of various molecular weights were analyzed. The results concerning the charge state of the most abundant ion species shown by the mass spectra are summarized in Table 1. It appears clearly from these (26) Huber, C. G.; Krajete, A. Anal. Chem. 1999, 71, 3730.

Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

1701

Table 1. Most Abundant Charge States Observed by ESI-MS Studies of Various Proteins in Ammonium Bicarbonate and Triethylammonium Bicarbonate Solutions most abundant charge state protein insulin lysozyme myoglobin hemoglobin A0 and S bovine serum albumin (monomer) alcohol dehydrogenase (noncovalent dimer) bovine serum albumin (covalent dimer)

ammonium triethylammonium MW (103) bicarbonate bicarbonate 5.8 14.3 17.5 64.5 66.5

3 6 9 17 17

2 4 5 10 9

79.6

20

11

133.0

26

14

data that for every protein studied the use of TEAB leads to very reduced charge states in comparison with the ions formed in the ammonium bicarbonate solution. This decrease could be correlated to the gas-phase basicity (GB) of the protein basic sites. Cassady et al. showed, by submitting multiprotonated ubiquitin to a reaction with various gaseous amine, that the observed decrease of the average charge state was related to the difference of the gas-phase basicities of the reagent gas and of the protein basic sites (GBArg, 236 kcal‚mol-1; GBLys, 219 kcal‚mol-1; GBHis, 213 kcal‚mol-1).27 The gas-phase basicity of ammonia being equal to 193 kcal‚mol-1, it could be postulated that most of the arginine, lysine, and histidine residues are protonated when the experiments are carried out in the presence of ammonium bicarbonate. The significantly higher gas-phase basicity of triethylamine (221 kcal‚mol-1) suggests that exothermic proton-transfer reactions between the multicharged protein and the reagent gas could preferentially involve histidines and to a lesser extent lysines. In the case of such a “pure” gasphase protonation, arginine should always remain in its protonated form. In fact, the reactivity of a basic residue is strongly influenced by the neighboring functional groups, whose spatial distribution is also dependent on the tertiary structure of the protein. It is thus impossible to predict the average number of protonated basic sites within a multiprotonated protein from considerations of gasphase basicity along with the composition in amino acids of the protein under investigation. For instance, in the case of the proteins listed in Table 1, the overall effect of the base used in the buffer solution manifested itself in the form of a linear relationship between the logarithm of the predominant charge state and the logarithm of the protein mass (Figure 4). Such a correlation was already carried out by Standing et al. in a study concerning 30 proteins of molecular masses between 10 kDa and 4.65 MDa.28 The empirical relation was also in the form Ln z ) A Ln m + B (z, most abundant charge state; m, molecular weight) with A being equal to 0.55, a value that should be compared with our own values (0.67 in ammonium bicarbonate, 0.60 in TEAB). This slope difference could be attributed to the mass spectrom(27) Cassady, C. J.; Wronka, J.; Kruppa, G. H.; Laukien, F. H. Rapid Commun. Mass Spectrom. 1994, 8, 394. (28) Chernushevich, I. V.; Krutchinsky, A. N.; Ens, W.; Standing, K. G. Proceedings of the 44th ASMS Conference on Mass Spectrometry and Allied Topics, Portland, OR, 1996; p 751.

1702 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

Figure 4. Most abundant charge-state values in function of the molecular mass of each protein in ammonium bicarbonate and TEAB buffers, and corresponding linear regression (Y ) AX + B; r, correlation coefficient).

eters used, an electrospray ion source coupled with a magnetic analyzer in our study and a nanoelectrospray source fitted on a time-of-flight instrument for the experiment of Standing et al. The buffer salt used by these authors was ammonium acetate and not ammonium bicarbonate; however, from our own experience, the counteranion is of no significant effect on the charge-state distribution of the multiply charged ions (see above). Effect of Cation on the Gas-Phase Stability of Protein Ions: Ammonium vs Triethylammonium. The heme/myoglobin complex was used for investigating the effect of ammonium vs triethylammonium on the gas-phase stability of protein ions in a way similar to the comparison study between acetate and bicarbonate anions. The results showed a different behavior of holomyoglobin ions according to the buffer salts used for these experiments. Half of the heme/myoglobin complexes were thus dissociated at 84 V (Vsc - Vs) by using ammonium bicarbonate whereas 144 V was needed for the same result in the case of a TEAB solution (data not shown). In other words, the ions formed with TEAB are significantly more stable than those obtained with ammonium bicarbonate. The first hypothesis for explaining this difference is the loss of kinetic energy of multiply charged ions, which is proportional to the loss of charges. In the case of the ammonium bicarbonate solution, the most abundant ion corresponds to the 8+ charge state, for which 50% dissociation occurs at a Vsc - Vs value of 84 V. The kinetic energy Ek of this ion could be calculated as 672 eV ()84 × 8). In the case of the TEAB solution, the most abundant 5+ species displays 50% dissociation at 144 V, corresponding to a 720 eV kinetic energy. These two Ek values are sufficiently close to induce a similar dissociation rate, all other experimental parameters being identical. However, in the absence of pressure measurements at the ion source interface, it is difficult to confirm this assumption by collision cross section calculations. Another factor influencing the gas-phase stability of the noncovalent complex is the occurrence of intramolecular Coulombic repulsions. Smith et al. showed that the highly charged ions were destabilized by comparison with those of reduced charge state, the unimolecular dissociation barriers being lowered when the number of charges was enhanced.29 Consequently, the

Figure 5. Electrospray mass spectra of 20 µM alcohol dehydrogenase (dimer) with NADH (1:1) in buffered solution of (a) 20 mM NH4HCO3 and (b) 20 mM TEAB.

Figure 6. Electrospray mass spectrum of 20 µM human hemoglobin A0 in buffered solution of 25 mM NH4HCO3. R2β2H4 tetramers and RβH2 dimers are labeled by the letters T and D, respectively.

gas-phase structure of a protein ion bearing a limited number of charges should be more easily maintained in the gas phase than species possessing a large excess of charges. Indeed, it has been shown that a high number of protons on a protein molecule induced an important scrambling effect, leading to the loss of a number of electrostatic bonds and, hence, to the denaturation of the protein under investigation.30 (29) Busman, M.; Rockwood, A. L.; Smith, R. D. J. Phys. Chem. 1992, 96, 2397.

Further, if we consider that the proton transfer can occur in the gas phase, the different exothermicities of the protonation reaction according to the gas-phase basicity of the reagents should also be taken into account. The difference of gas-phase basicity between ammonia and triethylamine is 28 kcal‚mol-1 (1.2 eV). Thus, for a given number of charges z, the protonation reaction (30) Schnier, P. D.; Gross, D. S.; Williams, E. R. J. Am. Chem. Soc. 1995, 117, 6747.

Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

1703

Figure 7. Electrospray mass spectrum of 20 µM human hemoglobin A0 in buffered solution of 25 mM TEAB. R2β2H4 tetramers and RβH2 dimers are labeled by the letters T and D, respectively.

of a protein with the ammonium cation is more exothermic than the similar reaction carried out with the triethylammonium cation, this difference corresponding to 1.2z eV. As in the case of chemical ionization, the internal energy of the protonated (or multiprotonated) species is expected to be higher when the reagent gas is of lower proton affinity. Such an effect could contribute to the highest gas-phase stability of the ions generated in a TEAB buffer. Use of Triethylammonium Bicarbonate for Analytical Purposes. The possibility to obtain more stable noncovalent protein complexes with TEAB than with other salts is obviously of great interest from an analytical point of view. Another advantage of this buffer salt is the decrease of the charge states of the multiply charged ions formed by electrospray. It is particularly useful for examining mixtures of proteins or protein complexes of close molecular weights, for instance in the case of a protein in its free form in admixture with its complex formed with a small molecule (cofactor, substrate, inhibitor, etc..). The mass difference ∆m between the two species is calculated from the ∆m/z value measured on the mass spectrum: the higher the charge number z, the closer the diagnostic peaks. Overlapping peaks can thus be better resolved by reducing the charge states of the proteins. Figure 5 illustrates the separation of peaks corresponding to the enzyme alcohol dehydrogenase with and without its cofactor NADH (665.4 Da). One equivalent of cofactor was added to 1 equiv of protein dimer (D). The solution of protein and NADH was desalted by ultrafiltration prior to mass spectrometric analysis. In the ammonium bicarbonate solution, the spectrum displayed ions corresponding, on one hand, to the dimer species of alcohol dehydrogenase bearing 16-22 charges and, on the other hand, to the monomer characterized by its 12+ to 14+ charge states. The molecular weights calculated from the m/z values of the multiply charged ions were characteristic of the free protein, the protein/NADH complexes probably merging 1704

Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

at the right side of peaks (see peak labeled D19+ in Figure 5a). By contrast, the spectrum obtained with TEAB allowed the discrimination between the dimer and its NADH adduct (Figure 5b). As mentioned before, the increased stability of the noncovalent interactions under such conditions reinforced the relative intensity of the dimer/NADH complex, the dissociation of this species in the gas phase being responsible for its much weaker abundance in the ammonium bicarbonate solution. Further, the mass difference between the diagnostic species was 666 Da, thus characterizing unambiguously the NADH adduct. Another example of the usefulness of TEAB for studying noncovalent macromolecular complexes is the result obtained in the case of human hemoglobin A0. This protein is a heterotetramer comprising two R-globin chains (15 126 Da) and two β-chains (15 867 Da), each of them bearing a heme molecule (616 Da). The whole protein thus corresponds to a species of R2β2H4 stoichiometry possessing a mass of 64 450 Da. The first challenge with such a sample is to maintain the whole complex in the gas phase. The ESI mass spectrum recorded in an ammonium bicarbonate solution reveals the presence of the R2β2H4 species associated with the charge states 15+ to 18+ (Figure 6). Ions belonging to the minor heterodimeric species RβH2 appear at charge states 11+ and 12+, and the low-mass region of the spectrum is dominated by abundant ions corresponding to the individual globins R and β (with 5+ to 11+ charges). Such a spectrum does not allow confirmation of the presence of a heterotetrameric structure in the solution phase even though the absence of any R2 or β2 dimers and the relatively higher abundance of the tetramer with regard to the dimer suggests a dissociation in the gas phase of intact hemoglobin rather than nonspecific associations of isolated globin chains. In contrast with the previous spectrum, analysis of hemoglobin in a triethylammonium bicarbonate solution shows the unambiguous presence of the whole

Figure 8. Electrospray mass spectra of 20 µM human hemoglobin A0 in buffered solution of 25 mM TEAB, recorded at Vsc - Vs values of (a) 108, (b) 132, (c) 156, (d) 180, (e) 204, and (f) 228 V. R2β2H4 tetramers and RβH2 dimers are labeled by the letters T and D, respectively.

tetramer with intense ion peaks between m/z 5800 and 8100 (Figure 7). Two other groups of peaks are attributed to the heterodimeric species RβH2 from which two consecutive losses of heme can occur. On the lower mass region of spectrum, the single globin chains deprived of their heme give minor signals. The unexpected observation of the heme loss from the dimer RβH2 and its total absence from the tetramer R2β2H4 prompted us to examine carefully the collisional activation effects on the behavior of the tetrameric hemoglobin in the gas phase. For this purpose, six spectra were recorded at various sampling cone

voltages leading to Vsc - Vs values between 108 and 228 V (Figure 8). For the lower values of Vsc - Vs (108 and 132 V), the RβH2 dimer bearing six or seven positive charges leads to the major ion signals. Increasing the sampling cone voltage significantly enhances, in terms of relative intensity, the tetrameric ions observed for the charge states 9+ and 10+. The highest relative abundance detected for the intact tetramer is attained at Vsc - Vs ) 180 V. For the highest voltages applied to the sampling cone, the dimeric species seem to increase again at the expense of the Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

1705

whereas the tetramer remains stable. All steps preceding the transfer into the gas phase of the protein ions could contribute to the partial denaturation of hemoglobin, including both the initial ultrafiltration step and the electrospraying process. Further efforts are needed for a better understanding of denaturation in the solution phase, to maximize the mass spectrometric response of such noncovalent protein complexes.

Figure 9. Breakdown curves of the human hemoglobin A0 complexes: tetramer T, dimer D, dimer-heme D-H, dimer-2hemes D-2H, and monomers M as a function of Vsc - Vs. Relative intensities correspond to Ii/(IT + ID + ID-H + ID-2H + IM) ratios.

tetrameric species. Under such conditions, the dissociation of the RβH2 ions by heme losses leads to the major ions. This fragmentation pathway is more pronounced for the ion [RβH2 + 7H]7+ than for the [RβH2 + 6H]6+ species, as seen especially in Figure 8d. This is in good agreement with the observations mentioned above on the effect of the charge state on the gas-phase stability of multiply charged ions. Summing the relative intensities of the ion peaks attributable to the same ion species and reporting these data on a graph provides further insight into the gas-phase stabilities of ions arising from hemoglobin by electrospray ionization (Figure 9). The tetrameric species R2β2H4 (labeled T in Figure 9) corresponds to a constant fraction of the total ion current (between 30 and 33%). The monomeric R- and β-chains also show a constant relative intensity except for the lowest value of Vsc - Vs (108 V). The absence of monomers under such conditions is probably due to the weak ionization efficiency (broad peaks and noisy spectrum). By contrast, the ion signal of the dimer RβH2 (D) decreases steadily in favor of the RβH (D - H) and Rβ (D - 2H) species: the collisional activation due to the enhancement of the sampling cone voltage leads to the gas-phase dissociation of the dimer RβH2

1706 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

CONCLUSION The use of triethylammonium bicarbonate solution in electrospray mass spectrometry proved to be very efficient for studying proteins or noncovalent protein complexes under “nondenaturing” conditions. The low charge states observed in the mass spectra help to improve the separation of ions arising from macromolecular species of close masses. Moreover, the multiply charged ions generated in a TEAB solution are significantly more stable than those formed under more conventional solution conditions (for example, with ammonium bicarbonate or acetate). This is of great interest for the analysis of macromolecular species that can easily dissociate in the gas phase, such as hemoglobin or other macromolecular noncovalent complexes. One of the disadvantages of TEAB, related to the charge-state reduction, lies in the need of mass spectrometers presenting wide mass ranges. However, coupling of electrospray ion sources with magnetic mass analyzers or, more recently, with time-of-flight or extended mass range quadrupoles makes TEAB a compound well-adapted to modern mass spectrometric techniques for studying noncovalent associations. ACKNOWLEDGMENT The authors gratefully acknowledge Dr. Bhupesh C. Das for his assistance. We also thank the Association pour la Recherche contre le Cancer, France, for financial support. D. L. and G. M. are indebted to the Institut de Chimie des Substances Naturelles (CNRS) and to the Ministe`re de l′Education Nationale, de la Recherche et de la Technologie, respectively, for Ph.D. research fellowships.

Received for review October 30, 2000. Accepted February 5, 2001. AC001276S