Nanoelectrospray Ionization of Protein Mixtures: Solution pH and

Jan 9, 2004 - Several reports have presented evidence that gaseous native protein ions are likely to be formed in ESI via the so-called charged residu...
0 downloads 8 Views 265KB Size
Download PDF
Anal. Chem. 2004, 76, 1165-1174

Nanoelectrospray Ionization of Protein Mixtures: Solution pH and Protein pI Peng Pan, Harsha P. Gunawardena, Yu Xia, and Scott A. McLuckey*

Department of Chemistry, Purdue University, West Lafayette, Indiana 47907-1393

Solutions consisting of single proteins and mixtures of proteins at different pH values have been subjected to both positive ion and negative ion nanoelectrospray ionization to study the influence of solvent pH and protein pI on the ionization responses of proteins. As has been noted previously, it is possible to form protein ions of one polarity despite the fact that the proteins are present as the opposite polarity in solution. However, total response under this condition tends to be at least an order of magnitude less than the condition in which the nanoelectrospray ionization polarity is the same as the net charge of the proteins in solution. Furthermore, maximum signals in positive ion mode were noted when the pH value of the solution was 4-5 units lower than the protein pI. In the negative ion mode, maximum protein anion signals were observed when the pH was roughly 5 units higher than the protein pI. While only small changes in the abundance-weighted average charge were noted as a function of solution conditions, the extent of sodium ion incorporation was seen to depend strongly on the relationship between net protein charge in solution and gasphase ion polarity. Sodium ion incorporation was minimized under conditions of maximum signal (i.e., low pH positive ion mode and high pH negative ion mode). Sodium ion incorporation was highest when the protein ion polarities in solution and the gas phase were opposite. These observations are consistent with the charged residue model for electrospray ionization and suggest that a degree of selectivity for electrospray ionization applied to protein mixtures can be realized via judicious selection of solution pH and ionization polarity. Furthermore, the relative extent of sodium ion incorporation under a given set of conditions appears to correlate, at least qualitatively, with protein pI. Electrospray ionization (ESI)1-4 plays a central role in many applications of mass spectrometry to bioanalysis problems due in large part to its ability to form gaseous ions from biomolecules * Phone: (765) 494-5270. Fax: (765) 494-0239. Email: [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.; Baringa, 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. 10.1021/ac035209k CCC: $27.50 Published on Web 01/09/2004

© 2004 American Chemical Society

and the facility with which it enables the coupling of condensedphase separations with mass spectrometry. Most applications of ESI to complex mixtures of biomolecules have involved either on-line or off-line separations to minimize the number of components subjected to ionization simultaneously. This is particularly desirable for analyte species that tend to form multiply charged ions, such as proteins. The multiple charging phenomenon associated with electrospray complicates mass measurement when the spectrum is sufficiently complex that the peaks arising from each component cannot be associated unambiguously with one another. Extensive overlap of analyte charge states has tended to limit the development of applications of ESI to relatively complex mixtures of proteins. The complexity of the mixture of multiply charged ions amenable to analysis is limited by the resolving power of the mass analyzer. Hence, high magnetic field strength Fourier transform ion cyclotron resonance mass spectrometry5,6 currently provides the greatest capability for resolving mixture components and establishing ion charge and, as a result, analyte mass. Another approach to dealing with the multiple-charging complication associated with the ESI of protein mixtures is to manipulate the charge states of the ions to enable the determination of protein mass. Ion/ion proton-transfer reactions have proved to be an effective means for manipulating protein ion charge states7-12 and have been demonstrated to enable protein mixture analysis using ESI. With the development of so-called “top down” protein identification/characterization methodologies,13-17 (5) Kelleher, N. L.; Lin, H. Y.; Valaskovic, G. A.; Aaserud, D. J.; Fridriksson, E. K.; McLafferty, F. W. J. Am. Chem. Soc. 1999, 121, 806-812. (6) Li, W.; Hendrickson, C. L.; Emmett, M. R.; Marshall, A. G. Anal. Chem. 1999, 71, 4397-4402. (7) McLuckey, S. A.; Stephenson, J. L., Jr.; Asano, K. G. Anal. Chem. 1998, 70, 1198-1202. (8) Stephenson, J. L., Jr.; McLuckey, S. A. J. Am. Soc. Mass Spectrom. 1998, 9, 585-596. (9) Stephenson, J. L., Jr.; McLuckey, S. A. J. Am. Chem. Soc. 1996, 118, 73907397. (10) Stephenson, J. L., Jr.; McLuckey, S. A. Anal. Chem. 1998, 70, 3533-3544. (11) Scalf, M.; Westphall, M. S.; Krause, J.; Kaufman, S. L.; Smith, L. M. Science 1999, 283, 194-197. (12) Scalf, M.; Westphall, M. S.; Smith, L. M. Anal. Chem. 2000, 72, 52-60. (13) Mortz, E.; O’Connor, P. B.; Roepstorff, P.; Kelleher, N. L.; Wood, T. D.; McLafferty, F. W.; Mann, M. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 82648267. (14) Meng, F.; Cargile, B. J.; Miller, L. M.; Forbes, A. J.; Johnson, J. R.; Kelleher, N. L. Nat. Biotechnol. 2001, 19, 952-957. (15) Reid, G. E.; Shang, H.; Hogan, J.; Lee, G. U.; McLuckey, S. A. J. Am. Chem. Soc. 2002, 124, 7353-7362. (16) Reid, G. E.; McLuckey, S. A. J. Mass Spectrom. 2002, 37, 663-675. (17) VerBerkmoes, N. C.; Bundy, J. L.; Hauser, L.; Asano, K. G.; Razumovskaya, J.; Larimer, F.; Hettich, R. L.; Stephenson, J. L., Jr. J. Proteome Res. 2002, 1, 239-252.

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004 1165

it is of interest to explore the strengths and limitations of ESI for application to protein mixtures. In the application of any ionization method to a mixture analysis scenario, the degree of ionization selectivity is particularly important. In some cases, it is desirable that the ionization method be more or less universal, whereas a high degree of selectivity can be valuable when targeted species are of interest. In any case, it is important to understand and recognize the factors that lead to discrimination in ionization of complex mixtures. A number of studies have been reported that relate the effects of various condensed-phase factors on the nature of ions formed by ESI and signal responses. These studies include the effects of solution pH on charge-state distribution,18,19 effects of solvent composition,20,21 effects of analyte concentration,22-26 effects of analyte structures/conformations,27,28 and effects of the presence of other analytes.24-26,29,30 Relatively little attention, however, has been directed to factors that affect relative signal responses for proteins present in complex mixtures. We recently related a study describing ESI responses to proteins under conditions of low pH and high protein solubility.24 In this scenario, for the set of proteins studied, ESI response was found to be similar for each protein on a charge-normalized basis. Such a condition is desirable for protein mixture analysis when more or less universal ionization is appropriate. Under these conditions, the effect of proteins in mixtures on the observed signal responses of each other can be accounted for largely on the basis of competition for the limited excess charge. On a charge-normalized basis, each protein competes more or less equally for the excess charge. A subsequent study focused on the effect of small cations and small highly basic molecules on the electrospray responses of proteins at low pH.25 In this work, it was found that the effect of proteins on the ESI response of other proteins is as high as that of the most surface-active small cations/molecules studied, on a charge-normalized basis. Proteins therefore have a greater effect on the ionization responses of other proteins on a molar basis, because each protein molecule consumes more charge than does a singly charged surface-active molecule. This work described herein is motivated by the possibility for ionization selectivity associated with electrospray of protein mixtures. To this end, it is useful to relate the effects of solution variables, such as pH, on ionization response to properties of the proteins, such as, for example, isoelectric point (pI). These relationships can facilitate the development of strategies for the

selective electrospray of complex protein mixtures. While many studies have focused on the factors that affect observed chargestate distributions of protein ions formed via ESI,18,19,31-38 this work is primarily focused on electrospray response, which has received somewhat less attention.

(18) Chowdhury, S. K.; Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1990, 112, 9012-9013. (19) Guevremont, R.; Siu, K. W. M.; Le Blanc, J. C. Y.; Berman, S. S. J. Am. Soc. Mass Spectrom. 1992, 3, 216-224. (20) Cole, R. B.; Harrata, A. K. Rapid Commun. Mass Spectrom. 1992, 6, 536539. (21) Cole, R. B.; Harrata, A. K. J. Am. Soc. Mass Spectrom. 1993, 4, 546-556. (22) Sunner, J.; Nicol, G.; Kebarle, P. Anal. Chem. 1988, 60, 1300-1307. (23) Kostiainen, R.; Bruins, A. P.; Kebarle, P. Rapid Commun. Mass Spectrom. 1994, 8, 549-558. (24) Pan, P.; McLuckey, S. A. Anal. Chem. 2003, 75, 1491-1499. (25) Pan, P.; McLuckey, S. A. Anal. Chem. 2003, 75, 5468-5474. (26) Pan, P.; McLuckey, S. A. Proceedings of the 51st ASMS Conference on Mass Spectrometry and Allied Topics, Montreal, QC, 2003. (27) Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1993, 115, 6317-6321. (28) Loo, J. A.; Edmonds, C. G.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1990, 62, 693-698. (29) Tang, L.; Kebarle, P. Anal. Chem. 1991, 63, 2709-2715. (30) Enke, C. G. Anal. Chem. 1997, 69, 4885-4893.

(31) Peschke, M.; Blades, A.; Kebarle, P. J. Am. Chem. Soc. 2002, 124, 1151911530. (32) Chowdhury, S. K.; Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1990, 112, 9012-9013. (33) Schnier, P. D.; Gross, D. S.; Williams, E. R. J. Am. Soc. Mass Spectrom. 1995, 6, 1086-1097. (34) Carbeck, J. D.; Severs, J. C.; Gao, J.; Wu, Q.; Smith, R. D.; Whitesides J. Phys. Chem. B 1998, 102, 10596-10601. (35) Grandori, R. J. Mass Spectrom. 2003, 38, 11-15. (36) Wang, G.; Cole, R. B. Org. Mass Spectrom. 1994, 29, 417-427. (37) de la Mora, J. F. Anal. Chim. Acta 2000, 406, 93-104. (38) Dobo, A.; Kaltashov. I. A. Anal. Chem. 2001, 73, 4763-4773. (39) Reid, G. E.; Wells, J. M.; Badman, E. R.; McLuckey, S. A. Int. J. Mass Spectrom. 2003, 222, 243-258. (40) Kelleher, N. L.; Senko, M. W.; Siegel, M. M.; McLafferty, F. W. J. Am. Soc. Mass Spectrom. 1997, 8, 380-383. (41) Van Berkel, G. J.; Asano, K. G.; Schnier, P. D. J. Am. Soc. Mass Spectrom. 2001, 12, 853-862. (42) Hager, J. W. Rapid Commun. Mass Spectrom. 2002, 16, 512-526.

1166

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

EXPERIMENTAL SECTION Materials. Protein and peptide samples were purchased from Sigma (St. Louis, MO). Acetic acid, ammonium hydroxide, and ammonium acetate were purchased from Aldrich (Milwaukee, WI). All samples were used without further purification. Working solutions were prepared daily by dilution of 1-5 mg/mL stock solutions of the proteins or peptides (prepared in pure water solution) to the final molar concentrations. Water was obtained from an 18 MΩ NANOpure ultrapure water system from Barnstead/Thermolyne Corp. (Dubuque, IA). pH buffers were made by titration of acetic acid and ammonium hydroxide. The reported concentrations are accurate to within (10%. Mass Spectrometry. The experimental data presented in this paper were acquired using a Hitachi (San Jose, CA) model M-8000 quadrupole ion trap mass spectrometer modified for ion introduction through the ring electrode by atmospheric sampling glow discharge ionization (ASGDI), as discussed in detail elsewhere.39 Ionization was accomplished via nanoelectrospray. Nanoelectrospray emitters were pulled from borosilicate glass capillaries with a 1.5-mm o.d. and a 0.86-mm i.d. using a Sutter Instruments micropipet puller, model P-87 (Sutter Instruments, Novato, CA). The nanoelectrospray assembly consists of a microelectrode holder (Warner Instruments, P/N ESW-MISP, Hamden, CT) with a metal (stainless steel or platinum) wire that is inserted into the capillary.40,41 The voltage applied to the wire for nanoelectrospray was typically (1.2 kV for positive mode and negative mode, respectively. Many of the experiments described herein were also carried out using a Q TRAP hybrid tandem mass spectrometer (Applied Biosystems/MDS Sciex, Concord, Ontario, Canada) using the nano-ESI source, as described in detail elsewhere.42 The Hitachi system employs a heated capillary for ion sampling, whereas the Sciex system employs a plate nozzle ion sampling arrangement. Although differences in charge-state distributions and overall signal responses were often noted between instruments, very similar behaviors were noted in comparing data collected with a given instrument. That is, observations noted with changes in nano-ESI conditions (i.e., polarity and solution com-

position) were similar for both instruments. Hence, the results reported herein are not expected to be particularly sensitive to the atmosphere/vacuum arrangement. Data acquired at “constant” pH were conducted using a stainless steel wire and pH-buffered solutions and were collected at relatively short times following initiation of the spray (typically less than 5 min). Experiments intended to maximize changes in solution pH employed a platinum wire, unbuffered solutions (nominally pure water as solvent), and relatively long spraying times (typically in excess of 30 min). Eight or more nanoelectrospray emitters were pulled and used for each experiment. Average responses were calculated from these identical experiments and recorded as a data point. RESULTS AND DISCUSSION Results. ESI is known to be most effective for species that are either already ions in solution or can readily form ions via acid-base chemistry or, in some cases, via electrochemical reactions. For ionized species, the usual practice is to employ ESI using the same polarity as that of the ions of interest. In the case of multiamphiphilic species, such as proteins, that can form net positive, net negative, or net neutral species, gaseous ions have been formed under a variety of ESI conditions. Of particular interest in the context of this work is the formation of gaseous ions of polarity opposite to that of the proteins in bulk solution. It has been noted, for example, that positively charged protein ions can be formed in positive ESI from solutions of sufficiently high pH that the proteins themselves carry a net negative charge in solution.43,44 This situation, whereby analyte ion polarity in bulk solution is opposite to the ESI polarity, has been referred to as “wrong-way-round” electrospray.45 Initial reports demonstrating wrong-way-round electrospray of proteins indicated that overall ESI response for a given protein tended to be significantly poorer in the wrong-way-round condition relative to “right-way-round” ESI, at least for positive ions.44 Given the wide range of pI values associated with proteins, manipulation of solution pH might enable a degree of ionization selectivity to be effected. Within the context of establishing conditions for ionization selectivity, therefore, it is of interest to determine the role that net protein charge in solution can play in the ESI responses of proteins in mixtures. In this work, we emphasize relative responses within a given ionization mode (i.e. positive or negative) because signal levels for a given ion polarity are relatively straightforward to make. Comparisons between absolute signal responses in negative ion ESI versus positive ion ESI are more problematic because of possible differences in spray conditions, detection efficiencies, etc. Figure 1 shows both positive and negative ESI mass spectra of 5 µM aqueous solutions of bovine ubiquitin (MW ) 8565 Da, pI ) 5.2) buffered at pH ) 2.3 and pH ) 8.0. Positive ion mass spectra are shown in Figure 1a (pH ) 2.3) and 1b (pH ) 8.0) and negative ion mass spectra are shown in Figure 1c (pH ) 8.0) and 1d (pH ) 2.3). ESI polarity and solution pH determine whether the protein is subjected to either right-way- or wrong-way-round (43) Loo, J. A.; Udseth, H. R.; Smith, R. D. Rapid Commun. Mass Spectrom. 1988, 2, 207-210. (44) Kelly, M. A.; Vestling, M. M.; Fenselau, C. C. Org. Mass Spectrom. 1992, 27, 1143-1147. (45) Mansoori, B. A.; Volmer, D. A.; Boyd, R. K. Rapid Commun. Mass Spectrom. 1997, 11, 1120-1130.

Figure 1. Nanoelectrospray mass spectra of bovine ubiquitin (pI ) 5.2) obtained under right-way-round and wrong-way-round conditions in both negative and positive ionization modes: (a) positive ion, pH ) 2.3; (b) positive ion, pH ) 8.0; (c) negative ion, pH ) 8.0; and (d) negative ion, pH ) 2.3.

conditions. The following conditions apply to the spectra of Figure 1: (a) positive ion, right-way-round; (b) positive ion, wrong-wayround; (c) negative ion, right-way-round; and d) negative ion, wrong-way-round. Two particularly striking observations can be made from the positive ion wrong-way-round versus right-wayround comparison. First, the signal response in the wrong-wayround ESI condition is significantly less than obtained from the right-way-round ESI condition. Figure 1b shows the mass spectrum of bovine ubiquitin from pH ) 8.0 buffer in which abundance was normalized relative to the highest signal obtained from the pH ) 2.3 buffer solution (Figure 1a, right-way-around ESI). The total positive ion abundance in Figure 1b is only a few percent of that in Figure 1a. This result is typical for the single protein solutions we have examined in this way and is consistent with the original report devoted to wrong-way-round ESI.44 The second observation is the extensive sodium ion adduction observed in the wrong-way-round condition, particularly for the 5+ charge state. No sodium is intentionally added to the solution, and its bulk concentration is unknown. There are several possible sources of sodium salt contamination, and no stringent measures, aside from the use of 10-18 Ω water, were taken to avoid it. In general, we noted significantly greater sodium and, in some cases, potassium ion incorporation into positive protein ions formed under wrong-way-round conditions than under right-way-round conditions. However, such data were usually collected at pH values greater than those associated with the use of 1-2% acetic acid solutions. Therefore, the concentration of adventitious sodium ions is larger relative to the hydronium ion concentration as pH increases such that greater incorporation of contaminating metal Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1167

ions might be expected. In any case, sodium ion incorporation was often seen to be much more extensive in the lower charge states observed than in the higher charge states. A third observation, which is somewhat more subtle, concerns the charge-state distributions observed in right-way versus wrong-way ESI. As noted in Figure 1a and 1b, the weighted average charge in the wrong-way-round mode is lower than in the right-way-round condition, in this case by 1-2 charge states. In general, the weighted average charge state in the wrong-way-round positive ion mode was usually slightly less than in the right-way-round mode for the proteins examined in this study. However, we note that the weighted average charge state for a given set of conditions was found to vary by as much as 1-2 charge states, depending upon the interface conditions, as well as from one nanoelectrospray tip to the next. This may well be due to variations in the nanoelectrospray tip diameter, as has been discussed recently.62 At pH ) 8.0, ubiquitin is in the right-way-round condition in negative ESI (Figure 1c), and at pH ) 2.3, ubiquitin is in the wrong-way-round condition in negative ESI (Figure 1d). The comparison of negative ESI right-way versus wrong-way-round data shows similarities with the positive ESI comparison. For example, the absolute signal levels are higher in the right-wayround condition, in this case by about an order of magnitude. Furthermore, there is relatively greater sodium ion incorporation into the protein ions under the wrong-way-round condition, and the weighted average charge state of the protein is slightly higher in the right-way-round condition (i.e., a roughly one-half chargestate difference). In this particular case, a relatively small shift in charge is noted in the change in the relative abundances of the 5- and 4- charge states. In both negative ion spectra, the presence of sodium ions is much greater than in the right-wayround positive ion case (Figure 1b), but there is a greater relative abundance of sodium-containing ions in the wrong-way-round negative ESI spectrum (Figure 1d). (46) Van Berkel, G. J.; McLuckey, S. A.; Glish, G. L. Anal. Chem. 1992, 64, 1586-1593. (47) Van Berkel, G. J.; Zhou, F.; Aronson, J. T. Int. J. Mass. Spectrom. Ion Process 1997, 162, 55-62. (48) Van Berkel, G. J. In Electrospray Ionization Mass Spectrometry: Fundamentals, Instrumentation and Applications; Cole, R. Ed.; Wiley: New York, 1997; Chapter 2, pp 65-105. (49) Van Berkel, G. J.; Giles, G. E.; Bullock J. S., IV.; Gray, L. J. Anal. Chem. 1999, 71, 5288-5296. (50) Van Berkel, G. J.; Kertesz, V. J. Mass Spectrom. 2001, 36, 204-210. (51) Karas, M.; Bahr, U.; Du ¨ lcks, T. Fresnius’ J. Anal. Chem. 2000, 366, 669676. (52) Kebarle, P. In Electrospray Ionization Mass Spectrometry; Cole, R. B., Ed.; John Wiley and Sons: New York, 1997; Chapter 1. (53) Verkerk, U. H.; Peschke, M.; Kebarle, P. J. Mass Spectrom. 2003, 38, 618631. (54) Juraschek, R.; Du ¨ lcks, T.; Karas, M. J. Am. Soc. Mass Spectrom. 1999, 10, 300-308. (55) Felitsyn, N.; Peschke, M.; Kebarle, P. Int. J. Mass Spectrom. 2002, 219, 39-62. (56) Dole, M.; Mack, L. L.; Hines, R. L.; Mobley, R. C.; Ferguson, L. D.; Alice, M. B. J. Chem. Phys. 1968, 49, 2240-2249. (57) Iribarne, J. V.; Thomson, B. A. J. Chem. Phys. 1976, 64, 2287-2294. (58) Thomson, B. A.; Iribarne, J. V. J. Chem. Phys. 1979, 71, 4451-4463. (59) Smith, J. N.; Flagan, R. C.; Beauchamp, J. L. J. Phys. Chem A 2002, 106, 9957-9967. (60) Gomez, A.; Tang, K. Q. Phys. Fluids 1994, 6, 404-414. (61) Grewal, R. N.; Aribi, H. E.; Smith, J. C.; Rodriquez, C. F.; Hopkinson, A. C.; Siu, K. W. M. Int. J. Mass Spectrom. 2002, 219, 89-99. (62) Li, Y.; Cole, R. B. Anal. Chem. 2003, 75, 5739-5746.

1168 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

Figure 2. Total response/maximum total response (R/Rmax) versus pH derived from (a) 5 µM single protein aqueous solutions of bovine cytochrome c, lysozyme, and myoglobin subjected to positive ion nano-ESI and (b) 5 µM single protein aqueous solutions of bovine ubiquitin subjected to negative ion nano-ESI. The error bars represent (1σ of the measurements of R/Rmax.

It is apparent from Figure 1 and from previous observations44 that signal response for a protein tends to be higher when it is subjected to right-way-round conditions. It is of interest to determine the relationship between signal response and pH for proteins of different pI. Figure 2a summarizes positive ion nanoESI responses for myoglobin (pI ) 7.0), cytochrome c (pI ) 10.6), and lysozyme (pI ) 11.0) from single-protein solutions acquired as a function of pH. The plot is displayed as R/Rmax versus solution pH, where R and Rmax are total integrated response (signal) and maximum total integrated response, respectively. Each data point represents the average of results for 10 nanospray tips. All three proteins show that R/Rmax approaches unity at pH values several units lower (4-5) than the isoelectric point of the protein. In the cases of lysozyme and cytochrome c, which have similar pI values, comparable normalized response versus pH behaviors are observed. Myoglobin, which is more acidic than either of the other proteins, shows similar behavior in a qualitative sense but does not approach R/Rmax until a pH is reached that is lower than that required for the more basic proteins. These results suggest that the pH value must be sufficiently low that the protein charge in solution is more or less maximized. From these data alone, however, it is not clear if other factors, such as protein conformation or concentrations of buffer ions, might play roles in the shapes of the response versus pH curves. Figure 2b shows a plot for ubiquitin acquired in the negative ion mode. Behavior in the negative ion mode mirrors that in the positive ion mode insofar as signal is maximized when the pH of the solution is several units greater than the pI of the protein. This behavior is expected on the basis of the positive ion data if it is assumed that maximum signals are expected when the analyte is fully charged and at the same polarity as the analyte. Figure 3 provides spectra of an equimolar mixture of bovine ubiquitin (pI ) 5.2) and bovine cytochrome c (pI ) 10.6), each present at roughly 5 µM, under various operating conditions.

Figure 3. Mass spectra of an equimolar bovine cytochrome c (C, pI ) 10.6, 12 234 Da) and bovine ubiquitin (U, pI ) 5.2, 8565 Da) mixture acquired via nano-ESI using different pH buffers and electrospray ion modes: (a) positive mode, pH ) 2.3; (b) positive mode, pH ) 8.0; (c) negative mode, pH ) 8.0; and (d) negative mode, pH ) 11.

Figure 3a shows a positive ion spectrum collected at pH ) 2.3. In this case, both proteins are net positively charged in solution, and both give abundant signals with relatively little sodium ion incorporation for any of the observed charge states. Figure 3b shows the positive ion electrospray mass spectrum of the mixture at pH ) 8.0. In this case, cytochrome c is expected to be positively charged in solution (i.e., the right-way-round condition), whereas ubiquitin is expected to be negatively charged (i.e., the wrongway-round condition). Ions derived from cytochrome c clearly dominate this spectrum. The relative contribution from ubiquitin ions is markedly smaller than in the case of Figure 3a, and much more sodium ion incorporation is apparent in the ubiquitin ions than in the cytochrome c ions. Furthermore, both proteins show a shift to lower charge states when the pH is changed from 2.3 (Figure 3a) to 8.0 (Figure 3b). Figure 3c shows the negative ion spectrum acquired from the same solution used to acquire Figure 3b (i.e., equimolar cytochrome c and ubiquitin buffered at pH ) 8.0). In this case, ubiquitin is in the right-way-round condition, and cytochrome c is in the wrong-way-round condition. Only ubiquitin ions are clearly apparent, and they are formed with very little incorporation of metal ions. Figure 3d shows the negative ion electrospray mass spectrum of the two-component mixture from a solution containing 2% NH4OH. In this case, the pH of the solution was roughly 11. This condition places cytochrome c near its isoelectric point. In this case, it is difficult to classify cytochrome c as being in either the right-way- or wrong-way-round condition. In any case, the net charge of cytochrome c is expected to be

significantly less positive in this solution than in the solution that gave rise to Figure 3c. The data of Figure 3d show ubiquitin to continue to dominate the spectrum, but there is detectable signal from cytochrome c that corresponds to the 5- ion. This ion shows extensive incorporation of sodium ions. As is the general trend, the ubiquitin ions show increasing sodium ion incorporation as the number of charges decreases. Furthermore, there are also relatively small signals associated with ions of each protein with an adduct species of roughly 215 Da. There appears to be more sodium ion incorporation in the ubiquitin ions of this spectrum than in those of the spectrum of Figure 3c. The additional sodium and the species adducted to some of the protein ions may well have been introduced into the solution with the addition of ammonium hydroxide. The most important observation arising from Figure 3d is the appearance of anion signal arising from cytochrome c, which is observed only at the highest pH values. Of particular interest to this study is the influence of the presence of one or more proteins in the right-way-round condition on the response of a protein in the wrong-way-round condition. In this regard, the spectra of Figures 1b and 3b are relevant in that 1b shows a single protein component subjected to ESI under the wrong-way-round condition, whereas 3b shows the case in which a second protein component is present that experiences the right-way-round condition. These spectra, however, were not collected under identical or nearly identical conditions. To address the issue of the effect of a protein present under the right-wayround condition on the signal response of a protein under the wrong-way-round condition, spectra of binary mixtures of ubiquitin, present at a fixed concentration of 5 µM, and bovine cytochrome c, present in concentrations ranging from 0 to 10 µM, in aqueous solutions buffered at pH ) 8.0 were subjected to the same ESI and mass spectrometry conditions (see Figure 4). Neither the charge-state distribution nor the absolute signal response for ubiquitin (subjected to wrong-way-round conditions) were observed to change significantly as the cytrochrome c concentration (subjected to right-way-round conditions) increased up to 10 µM (data not shown). The cytochrome c signal dominated all spectra in which it was present at 2 µM and greater and increased linearly with concentration over this range, with an invariant charge-state distribution, indicating that the total protein concentration had not reached a point whereby protein signals were limited by the excess charge associated with the droplets.24 Therefore, this set of experiments indicates that, at least when there is an excess of charge, proteins present in the right-way-round condition do not have a significant influence on the ionization of proteins exposed to wrong-way-round conditions. The preceding results show that protein signal responses in nano-ESI are strongly dependent upon solution pH, protein pI, and ESI polarity. This suggests that pH and ionization polarity can be adjusted to provide either a more or less universal ionization condition or a selective condition wherein the ionization method can serve as either a high-pass or low-pass pI filter. A simple four-component mixture consisting of lysozyme (pI ) 11.0), cytochrome c (pI ) 10.6), R-lactalbumin (pI ) 5.4), and β-lactoglobulin A (pI ) 5.2) is used here to illustrate the possibilities. Assuming all other variables that affect protein signal are fixed, the means to maximize the possibility for forming ions of every Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1169

Figure 4. Positive ion nano-ESI mass spectra of a 5 µM ubiquitin solution at pH 8.0 containing (a) no cytochrome c, (b) 2 µM cytochrome c, (c) 5 µM cytochrome c, and (d) 10 µM cytochrome c. All spectra are normalized against U5+ abundance as 100%.

mixture component is to use, where possible, conditions such that all ions are either positively charged in solution, for positive ESI, or all ions are negatively charged in solution, for negative ESI. Figure 5a shows the positive nano-ESI mass spectrum of the four component mixture in a pH ) 2.5 solution that clearly shows signals from all four components. In this case, the pH of the solution is significantly lower than the pI values of each of the mixture components. The mirror image experiment is to subject the mixture to negative ESI under conditions of high pH in which all components are expected to be negatively charged in solution. This is illustrated in Figure 5b, which shows the negative ESI mass spectrum from a pH ) 14.0 solution. Signals from all four components are clearly apparent. To observe strong signals from the most basic proteins, it was necessary to operate at very high pH levels. Therefore, sodium hydroxide was used to give such a high pH. The high levels of sodium present in solution account for the extensive sodium incorporation observed with all of the mixture components in Figure 5b. The use of conditions to give a degree of selectivity is illustrated with the spectra of Figure 5c and d, which were both acquired at a solution pH of 8.0. Figure 5c shows the positive ion nano-ESI mass spectrum in which only the basic proteins are observed, whereas Figure 5d shows the negative ion ESI mass spectrum in which only the acidic proteins give prominent signals. All of the data presented thus far were collected under conditions in which only small changes in pH from the solution introduced into the nano-ESI capillary are expected (i.e., buffered solutions, stainless steel wire for electrical contact to solution, and data collected within a few minutes of the initiation of a spray). 1170 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

Figure 5. Nano-ESI mass spectra of solutions of a roughly equimolar lysozyme (pI ) 11.0, L), cytochrome c (pI ) 10.6, C), R-lactalbumin (pI ) 5.4, R), and β-lactoglobulin A (pI ) 5.2, β) via nano-ESI using of different pH buffers and electrospray ion modes: (a) positive mode, pH ) 2.5; (b) negative mode, pH ) 14; (c) positive mode, pH ) 8.0; and (d) negative mode, pH ) 8 .0.

When unbuffered solutions were used, significant changes in spectra were often noted as a function of nanoelectrospray time. These changes were presumed to result from the well-known electrochemical processes that take place in ESI. For example, previous studies44-51 have demonstrated that under certain ESI conditions, solution pH can change significantly (e.g., by up to 5 pH units41,47) as a result of the electrolytic oxidation or reduction of water or other solvents. By taking advantage of this phenomenon, it is possible to follow pH related changes in nano-ESI mass spectra using a single emitter by monitoring mass spectra as a function of time. During this time, the hydronium ion concentration either increases or decreases, depending on the ESI polarity. Figure 6 shows three nano-ESI mass spectra of a three-component mixture consisting of ubiquitin (pI ) 5.2), myoglobin (pI ) 7.0), and cytochrome c (pI ) 10.6) in the negative ion mode. The mixture was prepared in water with no added acid, base, or buffers and was initially at near neutral pH. A platinum wire was used to make electrical contact with the solution so that the reduction of water would take place to maintain the electrospray. Figure 6a shows the negative ion mass spectrum acquired upon initiation of the nanoelectrospray. This particular spectrum persisted for 5-10 min with little change in total ion signal or in the identities of the ions. As seen in the spectrum, only anions derived from ubiquitin are observed. At near neutral pH, it is the only protein of the three that would be expected to be net negatively charged (i.e., it is the only protein in the right-way-round condition). Relatively extensive incorporation of sodium ions is noted in the

Figure 6. Nano-ESI mass spectra of solutions of a roughly equimolar myoglobin (pI ) 7.0, M), cytochrome c (pI ) 10.6, C), and ubiquitin (pI ) 5.2, U) via nano-ESI out of pure water after a spray times of (a) 5, b) 20, and c) 40 min.

ubiquitin ions, and the 5- charge state is most abundant. This spectrum contrasts with that of Figure 3c with respect to the degree of sodium incorporation. Although the pH values of the solutions used to acquire Figures 3c and 6a are similar, the buffered solution associated with Figure 3a contains 5-10 mM ammonium acetate buffer, which may have a major influence on the identities of the counterions associated with the ubiquitin ion. After roughly 10 min of continuous spraying time, the total anion signal began to increase and changes in the spectra were noted. Figure 6b shows the negative ion mass spectrum acquired after a spray time of 20 min. In this case, the ubiquitin ions are seen to shift slightly to higher charge states, and anions arising from both myoglobin and cytochrome c appear. Figure 6c shows the negative ion mass spectrum acquired after 40 min of spray time. This spectrum represents data collected when no further changes in the spectra are observed. In this case, the 7- ubiquitin ion is the most abundant ion formed from this protein, and much of the sodium ion incorporation noted at earlier times is no longer apparent. The changes noted here are consistent with those expected if the pH of the solution increased with time on the basis of data shown from buffered solution and on previous results showing time-dependent mass spectra with this type of nanoelectrospray arrangement.41 These results, however, have the advantage that the same nanospray tip was used throughout and that no external addition of reagents/buffers was required that might change the concentrations of contaminants, such as sodium. Analogous data have also been collected in positive ion mode for unbuffered protein mixtures introduced into the emitter at near neutral pH (data not shown) whereby some of the proteins were not initially in the right-way-round condition. Hence, the “wire in capillary” mode for nanoelectrospray is useful in qualitatively

following changes in mass spectra as the solution pH either increases or decreases. However, it should be noted that this approach does not lend itself well to precise measurements. For example, the location of the metal wire tip relative to the capillary tip and flow rate are factors in determining how the spectra evolve with time. We found, for example, that it is possible to establish a pH gradient near the capillary tip that can give rise to a degree of isoelectric separation when mixtures of proteins are present. Although the overall trends apply, depending upon how the solution is sampled from the capillary tip, scan-to-scan irreproducibilities can be introduced. Discussion. It is apparent from the results described above that the isoelectric point of the protein, the solution pH, and the nano-ESI ionization mode (positive versus negative) are important parameters in determining protein response. In this section, we comment on the observations described above within the context of current notions of electrospray mechanisms and the implications of these results for the analysis of mixtures of proteins subjected to nano-ESI. Of particular interest are the following experimental observations: 1. protein response in the right-way-round versus wrong-wayround conditions, 2. sodium ion incorporation both as a function of right-wayround versus wrong-way-round and as a function of charge state, and 3. R/Rmax versus pH Several reports have presented evidence that gaseous native protein ions are likely to be formed in ESI via the so-called charged residue model (CRM),31,37,52-55 originally described by Dole.56 In this process, gaseous ions are formed via small highly charged offspring droplets produced in the charge-driven disintegration of the initially formed ESI droplet. A distinction is made between the relatively small droplets of relatively high charge that are emitted during Rayleigh fission and the relatively massive residue droplet of lower relative charge that accounts for the remaining mass. Gas-phase ions are believed to arise from the small highly charged droplets, whereas some fraction of the mass of involatile species in the initial droplet are condensed into a particle of low charge that does not lead to mass spectrometrically observed ions. Ionization selectivity, therefore, is expected to depend on differential partitioning of nonvolatile species in the initial ESI droplet between the small highly charged droplets that tend to yield the majority of gaseous ions and the relatively low charged residues. We believe that the results described above can be rationalized within the context of the CRM56 and recent further elaborations of the mechanism.53,54 Our experience in comparing signal responses under rightway-round versus wrong-way-round conditions has indicated that there tends to be at least 1 order of magnitude greater signal in the right-way-round condition than in the wrong-way-round condition for both ion polarity modes. In our previous report concerning the effects of proteins on the positive ion electrospray responses of other proteins from solutions of low pH (in which all of the proteins were subjected to right-way-round conditions), we presented a simple expression for protein electrospray response, Rn, when no other strongly basic species are present as

Rn ) kproteinCnqav,nE Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

(1) 1171

where Cn represents protein concentration, qav,n represents the abundance weighted average charge, kprotein represents a phenomenological rate constant for protein ionization, and E represents an efficiency factor. For the restricted, but commonly used, conditions that applied to that study, relative responses could be accounted for on the basis of Cnqav,n, suggesting that the kprotein and E factors were fairly constant for the proteins studied. However, given that the differences in weighted charge-state distributions for right-way versus wrong-way nanoelectrospray alone cannot account for differences in responses, it is clear that the product kprotein × E differs for the two spray conditions. The E factor is broken down further, consistent with the terminology originally introduced by Tang and Kebarle,29 into the product f × P × T × D where P is the efficiency of the gas-phase ion transfer into the mass spectrometer, T is the ion transmission efficiency within the mass spectrometer, D is the detection efficiency, and f is the efficiency of converting the excess surface charge into gas phase ions amenable to mass spectrometry. For a given protein subjected to right-way versus wrong-way conditions, it is likely that the P, T, and D terms of the overall efficiency E are very similar, given that the charge-state distributions are not dramatically different. It is therefore of interest to examine the kprotein × f product to determine if differences in either of these terms can rationalize the differences in right-way versus wrongway responses. The kprotein term was originally introduced to give expression 1 the units of current and was not intended to imply that ionization takes place via the so-called ion evaporation mechanism (IEM) introduced by Iribarne and Thomson.57,58 Within the context of the charged residue mechanism, if a single protein molecule is present in a small multiply charged droplet, a gaseous protein ion is formed when the solvent and weakly bound adducts are evaporated. Provided there are available sites on the protein for excess charge to condense and survive the desolvation/ion transmission process, there is likely to be little difference in the likelihood for protein ionization based on its initial polarity in the droplet. Rather, the rate of protein ion formation is more likely to be determined by the rate at which small highly charged droplets containing the proteins are produced. Hence, a term to account for partitioning of proteins between the small highly charged droplets that eventually yield ions and the rest of the initially formed droplet that does not should be included. In some ESI models that assume the IEM, a surface activity term is included.30 Such a term could also be added to 1. However, the f term mentioned above can also include this effect if it is assumed that this term is analyte-dependent (i.e., f becomes fprotein). The key point is that the differences in signal responses associated with right-way versus wrong-way-round ESI can probably be accounted for simply on the basis of the likelihood that the proteins are at or near the surface. All other factors being equal, it is expected that protein ions with the same net polarity in solution as the charge on the droplet are more likely to occupy sites at the droplet surface. Clearly, a protein with a net charge in solution that is the same polarity as the droplet is more likely to participate in providing the droplet surface with its excess charge than another protein that is either neutral or of opposite net polarity in solution. The latter species are more likely to be ion-paired and in the bulk of the solution than are the protein ions of the same polarity as 1172

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

the droplet. Furthermore, it has been argued that distortions on the droplet surface caused by the presence of protein ions can stimulate droplet disintegration at those sites, thereby leading to the formation of small highly charged droplets containing proteins.54 Hence, for these reasons, responses in the right-way-round mode are expected to be generally greater than in the wrongway-round mode, provided all other factors are constant. The extent of sodium ion incorporation into the ions under the various conditions used in this study provides possible insights into mechanistic aspects related to gas-phase ion formation. It was not the original intent of this study to examine the extent of sodium ion incorporation as a function of ESI conditions. Only in the instance that NaOH was used to give a solution of high pH was sodium intentionally added to an ESI solution. However, sodium ion contamination is very difficult to eliminate because it can arise from reagents added to the solution, the emitter glass, and contamination of the metal wire used to make electrical contact to the solution. Therefore, due to variability in the degree of sodium contamination, the extent of sodium ion incorporation for a given set of ESI conditions tended to vary from one set of experiments to another. However, the general trends noted in the data presented herein were observed consistently. The increased degree of sodium ion incorporation in ions of relatively low charge states in the positive ion mode would not be expected on the basis of the number of cationizing species needed to provide the excess charge associated with the ion. If this were the case, a greater absolute number of sodium ions would be expected to be present in the higher charge states. It has recently been suggested that ions of different charge states can arise from different droplet disintegration pathways.54 Ions that arise early in the fission process (e.g., from the first generation of offspring droplets that give rise to ions) are produced from droplets with the lowest relative concentrations of salts, whereas those that are formed late in the process are formed from droplets that have experienced an enrichment in nonvolatile salt concentration due to exposure to longer solvent evaporation times. Electrolyte species without a particular preference for the surface, such as sodium ions, tend to be enriched in droplets formed relatively late in the disintegration process. If the low-charge-state ions arise from droplets later in the disintegration process than the highcharge-state ions, the likelihood for sodium ion incorporation is higher for the low-charge-state species. It is useful to consider some crude approximations associated with the droplet disintegration process to estimate the extent of salt concentration that can take place. For the nth generation droplet, the charge at which Rayleigh fission is predicted is

qn ) 8π(0γRn3)1/2

(2)

where qn is Rayleigh charge, 0 is permittivity of vacuum, γ is the surface tension of the solution, and Rn is Rayleigh radius. On the basis of the data of Gomez and Tang,59 if it is assumed that there is a loss of 20-40% of the charge59 and 2% of the mass of the large droplet60 upon asymmetric fission, the charge of the ensuing droplet that undergoes fission, qn+1, is 0.85 that of the previous drop, that is,

qn+1/qn ) (Rn+1/Rn)3/2 ) 0.8 - 0.6

(3)

and the ratio of the cubes of the radii and, hence, volumes is 0.640.36 from

Rn+13/Rn3 ) Vn+1/Vn ) 0.64 - 0.36

(4)

The mass (or number of moles) of sodium, assuming it is equally distributed within and on the surface of the droplet, shows a small decrease with each fission, that is,

massn+1/massn ≈ 0.98

(5)

Hence, the ratio of concentrations of sodium in the parent droplet from one fission to the next is approximated by

Cn+1/Cn ≈ (0.98/0.64) - (0.98/.36) ) 1.5 - 2.7

(6)

The cumulative concentration effect over n fission events, assuming that the loss of 20-40% charge and 2% mass per fission is constant with n, is estimated as

Cn/C0 ≈ (1.5)n - (2.7)n

(7)

It is not clear how accurate the assumptions are for submicrometer-sized droplets, because the values were established from studies of micrometer-sized droplets.59,60 Furthermore, the constancy of the fractional charge and mass losses for multiple events might not hold, particularly as the droplet size decreases and as the nonvolatile components of the solution become more highly concentrated. Nevertheless, it is of interest to evaluate the expected concentration increase with this very simplistic picture for the initial fission events, in which changes in the characteristics of the droplets are less pronounced. This picture predicts an order of magnitude increase in sodium ion concentration in the large “parent” droplet for roughly every 2-7 fission events. The relative change for the smaller offspring droplets might be expected to be even greater if surface active species occupy the surface such that the sodium ion concentration at the surface is less than that in the bulk. Hence, the very early fission events might yield offspring droplets that are depleted in sodium. This situation would be expected to prevail until the surface-active species were removed sufficiently to allow sodium ions to occupy the surface at concentration levels comparable to those in the bulk. Differences in disintegration pathways for right-way-round versus wrong-way-round ESI might also account for the observation that the extent of sodium ion incorporation tends to be higher for proteins subjected to wrong-way-round conditions than under right-way-round conditions. In the case of wrong-way-round ESI, the net charge of the protein must be inverted from its polarity in the bulk phase to that applied to the ESI solution. This process is believed to proceed via neutralization of the nascent charges of the protein by counterions present in solution followed by, or in concert with, ionization of acidic or basic sites, as the case may be, via interactions with ions associated with the excess charge.52 When the counterions lead to neutralization products that are readily liberated in the mass spectrometer interface, adducts are not observed. For example, in the case of ammonium or hydronium ions, neutralization of a carboxylate site on a protein does

not show ammonium or hydronium ion condensation on the protein because either ammonia or water is readily driven off in the desolvation process. The neutralization of a carboxylate group by a sodium ion, however, gives rise to a relatively stable species. From this perspective, it is not surprising that an acidic protein subjected to positive ion electrospray under wrong-way-round conditions would show a greater degree of sodium ion incorporation than a basic protein subjected to the same conditions. In the case of the negatively charged protein, it could be argued that there are more sites for sodium ions to attach. Although this may play a role in the positive ion case, it does not readily account for the observation that sodium ion incorporation is also seen to be greater in the negative ion mode for wrong-way-round conditions (see, for example, Figure 1c and d). In this case, positive charges on basic sites must be neutralized and cations removed from carboxylates to yield a net negatively charged gaseous ion. Aside from the neutralization of anionic sites, assuming the net positively charged protein to be a zwitterion,35 the incorporation of a metal ion into the protein is not relevant to this process. Nevertheless, extensive sodium ion incorporation can be observed, even at low pH. For example, in the comparison of Figure 1c and d, slightly greater sodium ion incorporation is observed in the ubiquitin ions formed under wrong-way-round conditions (Figure 1d) than under right-way-round conditions, despite the fact that the hydronium ion concentration in the experiment leading to Figure 1d was over 5 orders of magnitude greater in the bulk solution than in the solution giving rise to Figure 1c. The results of Figure 6 are also particularly noteworthy in this regard. At pH ) 8.0 (Figure 6a), ubiquitin is nominally in the right-way-round condition for negative ESI, but it is not yet at the pH where total anion signal is maximized. As the pH is raised, the total signal level increases, and there is markedly less sodium ion incorporation, despite the fact that the hydronium ion concentration decreases. This result is more significant than the observation of lower sodium ion incorporation in positive ESI as the pH is decreased (which is also observed), because this result could be interpreted as being due simply to an increase in the hydronium ion concentration relative to the sodium ion concentration. The tendency for sodium ion incorporation to be greater under wrong-way-round ESI conditions is consistent with the ions being formed later in the ESI process than ions formed under rightway-round conditions. Proteins present in droplets that evolve later in the disintegration process experience greater electrolyte concentrations that can lead to extensive metal ion incorporation.61 Perhaps, for this reason, no clear matrix effect upon the ionization of ubiquitin (wrong-way-round condition) by the addition of cytochrome c up to 10 µM (right-way-round condition) was noted. For the cases in which no sodium ions were intentionally introduced into the solution, total ion signals were greatest under right-way-round conditions when the pH value was several pH units lower than the pI value of the protein in positive ion mode (see Figure 2). An analogous situation appeared to be the case for negative ions in the right-way-round condition in that the strongest signals with least sodium ion incorporation were noted at pH values significantly higher than the pI values of the proteins. It is not clear why such a significant excess of acid or base above the isoelectric point is needed for maximum ion yield. However, on the basis of both ion yields and the degree of sodium ion Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1173

incorporation, it is likely to be directly related to the partitioning of the protein ions into the early generation small droplets that lead to ions. CONCLUSIONS A degree of ionization selectivity in the nanoelectrospray ionization of proteins can be affected by the adjustment of solution pH and selection of ionization polarity. Although proteins with a net charge in solution opposite to that of the ionization polarity can yield ions, the signal levels are significantly lower than the instance in which the protein ions are of the same net charge as the ionization polarity. The most universal means for ionization are at the extremes of the pH scale. Low pH values are optimal for positive ions, and high pH values are optimal for negative ions. Maximum signal responses in the positive ionization mode are observed at pH values three or more units lower than the protein isoelectric point. Negative ion signals for a given protein tend to be maximized at pH values several units above the isoelectric point of the protein. At near neutral pH, therefore, only very acidic proteins provide maximum yields in negative ion mode and only very basic proteins yield maximum signals under positive ion mode. Under conditions in which a protein in solution is of charge opposite that of the ionization polarity, signal responses are relatively low, and the likelihood for the incorporation of alkali metal ions, such as sodium, is highest. The signals obtained in

1174

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

this way do not appear to be sensitive to the presence of other proteins with the same net charge in solution as the electrospray droplets, at least when there is sufficient excess charge. The results reported here are consistent with the charged residue mechanism for protein ionization and partitioning of protein analyte species between the small highly charged droplets that lead to gaseous ions and the residual material that does not. The extent of metal ion incorporation into the ions appears to be a marker for nonvolatile electrolyte concentration in droplets that lead to ions. If so, the lower signals and greater sodium ion incorporation into ions whose polarity in the bulk is opposite to that of the electrospray ionization mode suggests that these ions are formed later in the electrospray droplets than those subjected to electrospray of the same polarity as the ions in solution. ACKNOWLEDGMENT This research was sponsored by the National Institutes of Health, Institute of General Medical Sciences under Grant GM 45372.

Received for review December 5, 2003. AC035209K

October

13,

2003.

Accepted