Method for Identifying Nonspecific Protein−Protein Interactions in

Oct 5, 2007 - Method for Identifying Nonspecific Protein−Protein Interactions in ... are investigated using the protein ubiquitin (Ubq) as a model sys...
0 downloads 0 Views 487KB Size
Download PDF
Anal. Chem. 2007, 79, 8301-8311

Method for Identifying Nonspecific Protein-Protein Interactions in Nanoelectrospray Ionization Mass Spectrometry Jiangxiao Sun, Elena N. Kitova, Nian Sun, and John S. Klassen*

Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2

The nonspecific self-association of proteins in nanoflow electrospray ionization mass spectrometry (nanoES-MS), and the influence of experimental conditions thereon, are investigated using the protein ubiquitin (Ubq) as a model system. Extents of nonspecific protein association generally increase with protein concentration and, interestingly, with decreasing ES spray potential. The extent of selfassociation is also sensitive to the duration of the accumulation event in an external rf hexapole. Notably, the relative abundance of metal (Na+ and K+) adducts generally increases with the size of nonspecific Ubq multimer. This result suggests that the gaseous ions of monomeric and nonspecific multimeric Ubq have, on average, different ES droplet histories, with monomer ions originating earlier in the ES process than the nonspecific multimeric complexes. This finding forms the basis for a new method for distinguishing between specific and nonspecific protein complexes in ES-MS. A reporter molecule (Mrep), which does not bind specifically to the proteins and protein complexes of interest, is added to the ES solution at high concentration. The distribution of Mrep bound nonspecifically to gaseous ions of the proteins and protein complexes, as determined from the ES mass spectrum, is used to determine whether a given protein complex originates in solution or whether it forms from nonspecific binding during the ES process. The method is demonstrated in cases where the ions of protein complexes detected by nanoES-MS originate exclusively from nonspecific association, exclusively from specific interactions in solution, or from both specific and nonspecific interactions. Mass spectrometry (MS), combined with electrospray ionization (ES) or its low-flow variant nanoflow ES (nanoES), has emerged as a powerful and direct method for studying the structure of multiprotein complexes in solution,1 monitoring their assembly/disassembly2 and subunit exchange reactions3,4 and identifying and quantifying their interactions with other biopoly* To whom correspondence should be addressed. E-mail: john.klassen@ ualberta.ca. (1) Loo, J. A. Mass Spectrom. Rev. 1997, 16, 1-23. (2) Fandrich, M.; Tito, M. A.; Leroux, M. R.; Rostom, A. A.; Hartl, F. U.; Dobson, C. M.; Robinson, C. V. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 14151-14155. (3) Sobott, F.; Benesch, J. L. P.; Vierling, E.; Robinson, C. V. J. Biol. Chem. 2002, 277, 38921-38929. 10.1021/ac0709347 CCC: $37.00 Published on Web 10/05/2007

© 2007 American Chemical Society

mers, ligands, and cofactors.5 Combined with gas-phase ion dissociation techniques, i.e., tandem mass spectrometry (MS/MS), which can be used to disassemble multiprotein complexes in the gas phase, ES-MS/MS can also provide unique insights into the subunit composition and binding topology of heterocomplexes.8,9 While ES-MS represents an important addition to the arsenal of tools currently available to probe the structure and function of protein assemblies in vitro, the full potential of the technique has not been fully realized. One of the challenges to using the direct ES-MS assay to study proteins that self-assemble into different multimeric forms or assemble with other proteins into heterocomplexes is the difficulty in reliably quantifying the relative abundance of the different protein forms present in solution. To a first approximation, the relative abundance of the gaseous ions detected for proteins and their noncovalent assemblies by ESMS will reflect their relative abundance in solution. However, differences in ES response factors, which reflect the transfer efficiency of analyte from solution to the gas phase (ionization efficiency), and the sampling, transmission, and detection efficiencies (referred to collectively as detection efficiency) can dramatically alter the relative abundance of the protein/complex ions.10 Additionally, the formation of nonspecific complexes during the ES process (false positives) and in-source gas-phase dissociation of weak complexes (false negatives) can influence the measured ion abundance. Gaseous ions of macromolecules, such as proteins and their noncovalent complexes, produced by ES are generally thought to form by the charge residue model (CRM).11 According to this model, the initial droplets generated from the ES tip undergo solvent evaporation until the Rayleigh limit is reached at which point they undergo Coulombically driven fission, releasing multiple small, highly charged droplets (called offspring or progeny droplets), which may contain zero, one, or several analyte (4) Keetch, C. A.; Bromley, E. H. C.; McCammon, M. G.; Robinson, C. V. J. Biol. Chem. 2005, 280, 41667-41674. (5) Daniel, J. M.; Friess, S. D.; Rajagopalan, S.; Wendt, S.; Zenobi, R. Int. J. Mass Spectrom. 2002, 216, 1-27. (6) Winston, R. L.; Fitzgerald, M. C. Mass Spectrom. Rev. 1997, 16, 165-179. (7) Veenstra, T. D. Biochem. Biophys. Res. Commun. 1999, 257, 1-5. (8) Benesch, J. L. P.; Robinson, C. V. Curr. Opin. Struct. Biol. 2006, 16, 245251. (9) Heck, A. J. R.; van den Heuvel, R. H. H. Mass Spectrom. Rev. 2004, 23, 368-389. (10) Kuprowski, M. C.; Konermann, L. Anal. Chem. 2007, 79, 2499-2506. (11) Dole, M.; Mack, L. L.; Hines, R. L.; Mobley, R. C.; Ferguson, L. D.; Alice, M. B. J. Chem. Phys. 1968, 49, 2240-2249.

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007 8301

molecules. The parent droplets may repeat the evaporation/fission process multiple times, producing multiple generations of offspring droplets. Solvent evaporation from the offspring droplets ultimately yields gaseous, multiply charged analyte ions. While there may be many factors that influence ionization efficiency, surface activity has a dominant role, with hydrophobic and charged analytes exhibiting higher ionization efficiencies.12 Quantitative ES-MS binding studies performed on complexes of proteins and small molecules or small biopolymers, such as oligosaccharides, have shown that differences in ionization/detection efficiency are generally small in cases where the ligand is small, compared to the protein, such that protein and protein-ligand complex have similar surface activities.13,14 However, the ionization efficiencies are unlikely to be the same in the case of protein self-association or association of proteins with other proteins or macromolecules. Additionally, differences in detection efficiency may also be significant in cases where the protein/complex ions span a large range of mass-to-charge (m/z) ratios. Several methods for estimating relative ES response factors in conventional (pump-driven) ES for analyte molecules/complexes have been reported.15,16 To our knowledge, the applicability of these correction methods has not been demonstrated for nanoES, which is more commonly used in MS analysis of protein complexes because of its small volume (sample) requirements, high efficiency, and tolerance to buffered aqueous solutions. ES-MS analysis of protein complexes can also be influenced by the dissociation of gaseous ions, promoted by collisional heating, within the ES ion source.13,17 The extent of collisional heating and concomitant gas-phase dissociation can often be minimized with the judicious choice of source parameters, in particular lens voltages. However, in some instances, it has been shown that, even with very gentle conditions, noncovalent protein complexes undergo dissociation in the ES source.18,19 Recently, our laboratory described a method to stabilize weakly interacting noncovalent biological complexes in nanoES-MS.18 The method, which involves the use of a solution additive such as imidazole, is based on the enhancement of evaporative cooling achieved by the dissociation of additives, bound nonspecifically to the complex, in the ion source.20 The reduction in the average internal energy during the desolvation process can minimize or even eliminate in-source dissociation of the complex.18 False positives can arise from the formation of nonspecific complexes, i.e., complexes that are not originally present in solution and that form through nonspecific interactions during the ES process. Such nonspecific interactions can occur whenever the ES offspring droplets, which ultimately lead to gaseous ions, contain multiple analyte molecules. The resulting complexes may be sufficiently stable in the gas phase that they survive to the (12) Cech, N. B.; Enke, C. G. Anal. Chem. 2000, 72, 2717-2723. (13) Wang, W.; Kitova, E. N.; Klassen, J. S. Anal. Chem. 2003, 75, 4945-4955. (14) Peschke, M.; Verkerk, U. H.; Kebarle, P. J. Am. Soc. Mass Spectrom. 2004, 15, 1424-1434. (15) Gabelica, V.; Galic, N.; Rosu, F.; Houssier, C.; De Pauw, E. J. Mass Spectrom. 2003, 38, 491-501. (16) Chitta, R. K.; Rempel, D. L.; Gross, M. L. J. Am. Soc. Mass Spectrom. 2005, 16, 1031-1038. (17) Håkansson, K.; Axelsson, J.; Palmblad, M.; Håkansson, P. J. Am. Soc. Mass Spectrom. 2000, 11, 210-217. (18) Sun, J.; Kitova, E. N.; Klassen, J. S. Anal. Chem. 2007, 79, 416-425. (19) Clark, S. M.; Konermann, L. Anal. Chem. 2004, 76, 7077-7083. (20) Wang, W.; Kitova, E. N.; Klassen, J. S. Anal. Chem. 2005, 77, 3060-3071.

8302

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

detector.13,20 Recently, our laboratory developed a straightforward and quantitative method to correct nanoES mass spectra for nonspecific protein-ligand interactions.21 The method involves the addition of a reference protein (Pref), which does not bind specifically to the protein and ligand of interest in solution, to the ES solution. The occurrence of nonspecific protein-ligand binding is monitored by the appearance of ions corresponding to nonspecific complexes of Pref and ligand in the mass spectrum. Furthermore, the fraction of Pref undergoing nonspecific ligand binding provides a quantitative measure of the contribution of nonspecific binding to the measured intensities of protein and specific protein-ligand complexes. While this method is easily implemented in the case of protein-small molecule binding, it is not convenient for monitoring the occurrence of nonspecific interactions formed between proteins during the ES process. Houssain and Konermann recently reported an elegant method to distinguish specific from nonspecific multiprotein complexes detected by ES-MS.22 The method combines on-line pulsed hydrogen/ deuterium (H/D) exchange and tandem mass spectrometry. The authors demonstrated that the protein complexes formed in the solution (specific) and during the ES process (nonspecific) exhibit different H/D exchange patterns, which were revealed by ESMSn. However, one of the limitations with this approach is that it is not readily implemented in the case of nanoES. The present work describes the first detailed study into the phenomenon of nonspecific self-association of proteins in nanoESMS and the influence of experimental conditions thereon. Drawing on the new insights gained from this study, our laboratory has developed a novel method for distinguishing specific from nonspecific protein complexes detected in nanoES-MS. Briefly the method is based on the observation that, on average, specific and nonspecific protein complexes have different droplet histories. While the specific complexes are formed in solution, the nonspecific complexes are produced in the ES process, specifically the evaporating offspring droplets. To distinguish specific from nonspecific protein complexes, a reporter molecule (Mrep), which does not interact specifically with the proteins and protein complexes of interest, is added to the ES solution. Differences in the distribution of Mrep bound nonspecifically to the gaseous ions of the proteins and protein complexes serves to distinguish between protein complexes originating in solution and nonspecific complexes formed during the ES process. The application of this method is demonstrated for cases where the protein complexes originate exclusively from nonspecific binding, exclusively from specific binding in solution, or from both specific and nonspecific binding. MATERIALS AND METHODS Proteins and Carbohydrates. β-Lactoglobulin isoform B (Lg; MW 18 281), and bovine ubiquitin (Ubq; MW 8565) were purchased from Sigma-Aldrich Canada (Oakville, OnN, Canada). The Ubq was used without any further purification. The Lg was dissolved in deionized water, dialyzed using an Amicon ultracentrifugation filter (Millipore Corp., Bedford, MA) with a molecular mass cutoff of 10 kDa, and lyophilized. The Lg was weighed immediately after removing it from the lyophilizer, (21) Sun, J.; Kitova, E. N.; Wang, W.; Klassen, J. S. Anal. Chem. 2006, 78, 30103018. (22) Houssain, B. M.; Konermann, L. Anal. Chem. 2006, 78, 1613-1619.

dissolved in a known volume of deionized water, and stored at -20 °C, if not used immediately. The synthetic trisaccharide, RTal[RAbe]RMan (1) was provided by D.R. Bundle (University of Alberta). The nanoES solutions were prepared from stock solutions of protein and carbohydrate with known concentrations. A 50 mM aqueous solution of ammonium acetate was added to yield a final concentration of 5 mM. Mass Spectrometry. All experiments were performed on an Apex II 9.4-T Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometer (Bruker, Billerica, MA) equipped with an external nanoES ion source. NanoES was performed using an aluminosilicate capillary (1.0-mm o.d., 0.68-mm i.d.), pulled to ∼47-µm o.d. at one end using a P-2000 micropipet puller (Sutter Instruments, Novato, CA). The electric field required to spray the solution was established by applying a voltage of ∼(800 V to a platinum wire inserted inside the glass tip. The solution flow rate was estimated to be between 20 and 50 nL/min. The droplets and gaseous ions emitted by the nanoES tip were introduced into the mass spectrometer through a stainless steel capillary (i.d. 0.43 mm) maintained at an external temperature of 66 °C. The ion/gas jet sampled by the capillary ((50 V) was transmitted through a skimmer (0 to (2 V) and stored electrodynamically in an rf hexapole. Unless otherwise specified, a hexapole accumulation time of 1 s was used for all experiments. Ions were ejected from the hexapole and accelerated to ∼(2700 V into a 9.4-T superconducting magnet, decelerated, and introduced into the ion cell. The trapping plates of the cell were maintained at a constant potential of (1.4-1.8 V throughout the experiment. The typical base pressure for the instrument was ∼5 × 10-10 mbar. Isolation of the reactant ions for the BIRD and SORI-CID experiments was achieved using a combination of single rf and broadband rf sweep excitation. The temperature of the ion cell for the BIRD experiments was controlled with two external flexible heating blankets placed around the vacuum tube in the vicinity of the cell.23 The isolated ions were stored inside the heated cell for varying reaction times before excitation and detection. SORICID was performed using a single-frequency excitation applied at 400-500 Hz below the cyclotron frequency of the reactant ion. An excitation time of 500 ms was used for all experiments. The collision gas, argon, was introduced into the vacuum chamber using a pulse valve to give a pressure of ∼1 × 10-6 Torr in the ion cell. For all SORI-CID experiments the gas pulse started 10 ms before the beginning of the SORI excitation pulse. A 5-s delay following the introduction of collision gas and prior to detection was used to pump away the collision gas. Data acquisition was controlled by an SGI R5000 computer running the Bruker Daltonics XMASS software, version 5.0. Mass spectra were obtained using standard experimental sequences with chirp broadband excitation. The time domain signals, consisting of the sum of 40-100 transients containing 128 or 512K data points per transient, were subjected to one zero-fill prior to Fourier transformation. RESULTS AND DISCUSSION Influence of Experimental Conditions on Nonspecific Association of Proteins in NanoES-MS. The nonspecific as(23) Felitsyn, N.; Kitova, E. N.; Klassen, J. S. Anal. Chem. 2001, 73, 46474661.

Figure 1. NanoES mass spectra obtained in positive ion mode for aqueous solutions of 5 mM ammonium acetate and Ubq at (a) 10, (b) 100, and (c) 400 µM.

sociation of proteins with metal ions,24 salts,25 and carbohydrates20 during the ES process has been the subject of extensive investigation. However, a detailed understanding of the phenomenon of nonspecific association of macromolecules, such as proteins and their noncovalent complexes, in ES-MS is lacking. To address this deficiency, our laboratory undertook a systematic investigation of the influence of experimental and instrumental parameters on the nonspecific self-association of Ubq, a small 8.5-kDa protein that exists as a monomer in solution, in nanoES-MS. a. Protein Concentration. To investigate the influence of protein concentration on the extent of nonspecific protein-protein binding, nanoES mass spectra were acquired for aqueous solutions of Ubq at concentrations ranging from 10 to 400 µM in both positive (Figure 1) and negative (Figure 2) ion modes. Shown in Figure 1 are illustrative mass spectra acquired in positive ion mode at three different concentrations (10, 100, and 400 µM). At a concentration of 10 µM, the only protein ions detected corresponded to the monomeric form of Ubq, i.e., Ubqn+ at n ) 4-6 (Figure 1a). However, at higher concentrations of Ubq, typically >50 µM, ions corresponding to Ubq multimers were also detected. For example, at a Ubq concentration of 100 µM, the Ubqn+ ions dominated the spectrum, but Ubq2n+ and Ubq3n+ ions were also observed (Figure 1b). At 400 µM, Ubqxn+ ions, where x ) 1-4, were detected (Figure 1c). The Ubqxn+ ions were almost entirely in their protonated form, i.e., (Ubqx + nH)n+; metal ion adducts, i.e., (Ubq + (n - 1)H + M)n+, where M ) Na+ or K+, were also (24) Felitsyn, N.; Peschke, M.; Kebarle, P. Int. J. Mass Spectrom. 2002, 219, 39-62. (25) Verkerk, U. H.; Kebarle, P. J. Am. Soc. Mass Spectrom. 2005, 16, 13251341.

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8303

Figure 2. NanoES mass spectra obtained in negative ion mode for aqueous solutions of 5 mM ammonium acetate and Ubq at (a) 10, (b) 100, and (c) 400 µM.

detected but at low abundance. To the best of our knowledge, there is no experimental evidence that Ubq associates into multimers or aggregates in aqueous solutions at the concentrations used in the present study. Consequently, the multimeric Ubqxn+ ions identified in the nanoES mass spectra must originate from nonspecific binding during the ES process. To quantify the extent of nonspecific Ubq association during the ES process, the fraction (f x) of Ubq detected as monomer or as one of the Ubqx multimers was calculated at each concentration investigated using the following expression:

x

∑

I(Pn+ x ) n

n

fx)

∑x∑ x

n

I(Pn+ x )

(1)

n

n+ where I(Pn+ x ) represents the measured intensity of the Ubqx ions (including all metal adducts). Because the FT-ICR MS ion signal is proportional to ion abundance and charge state (n), the measured Ubqxn+ ion intensities were divided by charge to obtain the ion abundance. The f x values determined at four different Ubq concentrations (10, 100, 200, and 400 µM) are shown in Figure 3a. It can be seen that, with increasing concentration, the relative abundance of the Ubqn+ ions decreases from 1.0 (at 10 µM) to ∼0.5 (at 400 µM). It can also be seen that higher Ubq concentrations lead not only to more nonspecific association, i.e., a greater fraction of Ubq engaged in nonspecific binding, but also to the formation of larger nonspecific multimers. It should be noted that the measured distribution of Ubqxn+ ions was found to exhibit considerable variability, being sensitive not only to protein

8304

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

Figure 3. Fraction (f x) of Ubq present as monomer (Ubq), dimer (Ubq2), trimer (Ubq3), and tetramer (Ubq4) determined from nanoES mass spectra acquired in (a) positive ion mode (Figure 1) and (b) negative ion mode (Figure 2) versus the concentration of Ubq in solution. The errors correspond to one standard deviation.

concentration but also to the ES and instrumental conditions. Nevertheless, for a given set of experimental conditions, the fraction of protein undergoing nonspecific binding and the size of the nonspecific multimers generally increase with concentration. This result is consistent with the findings of previous studies.26 The other experimental factors that were found to have the most significant influence on the measured distributions are described in the following sections. NanoES mass spectra for solutions of Ubq at a range of concentrations were also measured in negative ion mode. Shown in Figure 2 are illustrative mass spectra acquired at three different Ubq concentrations (10, 100, and 400 µM). Consistent with the results obtained in positive ion mode, higher protein concentrations led to an increase in the extent of nonspecific association and the appearance of larger nonspecific multimers in negative ion mode. Interestingly, a dependence of the relative abundance of metal ion (Na+ and K+) adducts of the detected Ubqxn- ions with the size of the multimer was also evident. Specifically, little or no metal ion adducts were detected for the Ubq monomer ions, while such adducts were clearly present for the nonspecific multimers and their relative abundance increased with increasing size of the multimer. For example, the major protein ions detected in Figure 2c correspond to monomer and nonspecific dimer and trimer. The metal ion adducts correspond to ∼8% of the signal for the Ubqn- ions, ∼12% for Ubq2n- ions, and ∼32% for Ubq3nions. It should be noted that metal ion adducts were also detected in positive ion mode. However, the trend toward more abundant metal ion adducts with the size of the multimer was less evident. (26) Winger, B. E.; Light-Wahl, K. J.; Loo, R. R. O.; Udseth, H. R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 1993, 4, 536-545.

The influence of concentration on the extent of nonspecific self-association of Ubq (and macromolecules in general) is consistent with the formation of gaseous protein ions via CRM. Although the exact nature of the droplet histories involved in gaseous protein ion formation cannot be ascertained, the number of protein molecules in the offspring droplets is expected to increase with increasing concentration of protein in the ES solution. Consequently, the probability of nonspecific association is expected to increase with increasing protein concentration. In addition to a greater fraction of protein undergoing nonspecific association, higher protein concentration will also lead to larger nonspecific multimers, in agreement with the present experimental results. The dependence of the relative abundance of metal ion adducts on the size of the Ubq multimer can also be explained within the framework of CRM provided that the monomer and nonspecific Ubq multimer ions have, on average, different droplet histories. Specifically, the ions of the nonspecific multimers must be produced later in the ES process, from “old” parent droplets that have experienced many previous fission events, compared to monomer ions, which are produced earlier in the ES process. Longer-lived ES droplets, which give rise to late generation offspring droplets, will contain higher analyte concentrations because of solvent evaporation. Consequently, their offspring droplets will have a higher probability of containing multiple proteins, as well larger numbers of other species including metal ions, compared to early generation offspring droplets. It is important to note that the lower relative abundance of metal ion adducts detected for the Ubqn+ and nonspecific Ubqxn+ ions in positive ion mode is not inconsistent with the specific and nonspecific complexes have different droplet histories. Rather, the absence of significant ion signal corresponding to salt adducts in positive ion mode can be explained by the competition between Na+ or K+ and NH4+, a component of buffer that is present at high concentration, for the offspring droplets. It is generally accepted that NH4+ can form nonspecific adducts with protein ions during the ES process. However, as described by Verkeck and Kebarle, these adducts, unlike the alkali metal ion adducts, are relatively unstable and readily dissociate in the ion source.25 Similar arguments have been made to explain the absence of acetate adducts in mass spectra acquired from protein solutions containing ammonium acetate buffer.25 b. ES Parameters. In the present study, the distribution of nonspecific Ubqx species was found to be exquisitely sensitive to the ES conditions. The variability in distributions measured with different nanoES tips was considerable. Additionally, a striking influence of the ES potential on the extent of nonspecific association of Ubq was observed. Specifically, in both positive and negative ion modes, lower ES potentials were found to enhance the detection of nonspecific multimer ions. Shown in Figures S1 and S2 (Supporting Information) are illustrative mass spectra acquired for solutions of Ubq in positive and negative ion modes, respectively, using the same nanoES tip but different ES potentials. The distributions of Ubqx species are shown in Figure 4. In positive ion mode at +1300 V (the highest spray voltage used), ∼80% of Ubq was detected as monomer ions (Ubqn+) with the remaining ∼20% as dimer (Ubq2n+) and trimer (Ubq3n+) ions (Figure 4a). Reduction of the spray voltage results in an increase in relative abundance of the nonspecific multimer ions, as well as an increase

Figure 4. Fraction (f x) of Ubq present as monomer (Ubq), dimer (Ubq2), trimer (Ubq3), tetramer (Ubq4), and pentamer (Ubq5) determined from nanoES mass spectra acquired in (a) positive ion mode (Figure S1) and (b) negative ion mode (Figure S2) versus the ES voltage.

in the size of the multimers. For example, at +700 V, ∼40% of Ubq was detected as Ubqn+ ions, ∼40% as Ubq2n+ ions, with the remaining ∼20% present as Ubq3n+ and Ubq4n+ ions (Figure 4a). Similar results were obtained in negative ion mode; i.e., lower negative ES spray potentials resulted in a greater extent nonspecific association and larger nonspecific multimers (Figures 4b and S2, Supporting Information). Additionally, the relative abundance of metal ion adducts was found to increase with the size of the Ubqxn- ions, consistent with the results described in the preceding section. The influence of ES spray potential on the extent of nonspecific association is intriguing and the origin of this effect is unclear. Given that a reduction of spray potential is accompanied by a decrease in ES spray current (data not shown), this phenomenon may be related to differences in droplet sampling, with lower potentials promoting the sampling of late-generation offspring droplets. According to available models,27,28 the radius (r) of the initial droplets emitted from the ES tip scale with flow rate (Vf l), r ∝ (Vfl)c, where the exponent c ranges from 2/7 to 2/3. In turn, Vf l scales with spray current (IES), IES ∝ (Vf l)c, where c ranges from 1/2 to 4/7. Assuming the above equations hold for ES in the nanoflow (nL/min) regime, a reduction in spray potential is expected to lead to a decrease in the size of the initial droplets emitted from the nanoES tip. It is possible that the smaller ES droplets produced at the lower potentials are more effectively sampled into the mass spectrometer where they contribute to the (27) de la Mora, J. F.; Loscertales, I. G. J. Fluid Mech. 1994, 260, 155-184. (28) Pfeifer, R. J.; Hendricks, C. D., Jr. AIAA J. 1968, 6, 496-502.

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8305

formation of late-generation offspring droplets. In the extreme, the initial ES droplets may be sufficiently small that desolvation leads directly to gaseous protein ions, bypassing the fission process altogether. In this case, every droplet emitted from the nanoES tip that contains two or more protein molecules will contribute to the formation of nonspecific protein complexes. Regardless of the exact nature of the spray potential effect, the present measurements reveal that, in terms of nonspecific selfassociation of Ubq, decreasing the ES spray potential has qualitatively similar effects as increasing protein concentration. c. External Accumulation. Collisional heating of gaseous ions within the ES ion source can result in their dissociation.13,17 Insource dissociation has been shown to be particularly problematic for the analysis of noncovalent complexes that have relatively low gas-phase stability.18,19 The extent of collisional heating depends on the configuration of the ion source and the experimental and instrumental parameters employed. In the instrument used for the present study, collisional heating is most significant while the ions are trapped electrodynamically within the external rf hexapole. The pressure in this region has been estimated to be in the 10-3-10-5 Torr range. At high space charge, the trapped ions spread out radially and are accelerated to high kinetic energies. Collisions between the translationally excited ions and background gas molecules convert the kinetic energy to internal energy.17,29,30 It has been shown that collisional heating in the rf hexapole is sufficient to induce cleavage of covalent bonds within biopolymers, including protein ions.30,31 However, there have been few attempts to characterize the internal energy distribution of the trapped ions. From a comparison on the measured rate of dissociation of protonated ions of a 27-kDa protein-trisaccharide complex in the hexapole and rate data acquired under thermal conditions, it was estimated that the effective temperature of the gaseous ions within the hexapole could exceed 140 °C.13 Given that in-source dissociation can have a significant effect on the detection of noncovalent complexes, the influence of hexapole storage time on the detection of nonspecific Ubqxn( ions was investigated in the present study. Shown in Figures S3 and S4 (Supporting Information) are mass spectra acquired for solutions of Ubq in positive and negative ion modes, respectively, with hexapole accumulation times ranging from 1 to 20 s. The corresponding distributions of Ubqx species are shown in Figure 5. It can be seen that, in positive ion mode, the distribution of the Ubqxn+ ions varies significantly with accumulation time, with longer storage times increasing the relative abundance and the size of the nonspecific multimers. For example, the fraction of monomer decreased from 90 to 52%, and the dimer increased from 10 to 45% upon increasing the accumulation time from 2.5 to 20 s (Figure 5a). In negative ion mode, increasing accumulation/ storage times initially increases the relative abundance of the nonspecific complexes, similar to what was observed in positive ion mode. However, at longer times, the opposite effect is observed with a systematic increase in the relative abundance of the monomer ions (Figure 5b). (29) Sannes-Lowery, K. A.; Hofstadler, S. A. J. Am. Soc. Mass Spectrom. 2000, 11, 1-9. (30) Belov, M. E.; Gorshkov, M. V.; Udseth, H. R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2001, 12, 1312-1319. (31) Pan, C.; Hettich, R. L. Anal. Chem. 2005, 77, 3072-3082.

8306 Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

Figure 5. Fraction (f x) of Ubq present as monomer (Ubq), dimer (Ubq2), and trimer (Ubq3) determined from nanoES mass spectra acquired in (a) positive ion mode (Figure S3) and (b) negative ion mode (Figure S4) versus the hexapole accumulation time.

The increase in the relative abundance of the nonspecific multimer ions observed in positive ion mode cannot be explained by in-source dissociation, since dissociation of the Ubqxn+ ions would be expected to increase the relative abundance of the monomeric Ubqn+ ions. Instead, the shift to larger multimers is likely due to differential ion transmission. It has been shown previously that increasing the accumulation time in an rf hexapole leads to a shift in mass distribution toward higher m/z ions.32 This phenomenon was attributed to changes in the transmission characteristics of the multipole due to space charge effects. Specifically, the effective offset potential of the hexapole increases at high space charge, resulting in the preferential loss of lower m/z ions. The influence of accumulation time observed in negative ion mode, wherein the nonspecific Ubqxn- ions initially increase with time, but then decrease in favor of the Ubqn- ions, may result from the combined effects of space charge and in-source dissociation. Mass discrimination resulting from space charge effects will favor the detection of larger, Ubqxn- ions, while in-source dissociation will promote the detection of monomeric Ubqn- ions. The underlying assumption in this explanation is that the Ubqxnions are more prone to in-source dissociation than are the Ubqxn+ ions. To demonstrate that the Ubqxn- ions are thermally less stable than the corresponding Ubqxn+ ions, time-resolved blackbody infrared radiative dissociation (BIRD) experiments were performed on the Ubq310( and Ubq27( ions. Shown in Figure 6a,b are the illustrative BIRD mass spectra acquired for Ubq310( ions at a cell temperature of 166 °C and a reaction time of 2.5 s. Both (32) Miyabayashi, K.; Naito, Y.; Miyake, M.; Tsujimoto, K. Eur. J. Mass Spectrom. 2000, 6, 251-258.

Figure 6. BIRD mass spectra acquired at a reaction temperature of 166 °C and reaction time of 2.5 s for (a) Ubq310- and (b) Ubq310+. SORI-CID mass spectra acquired for (c) Ubq310- with offset frequency of 475 Hz, (d) Ubq310+ with offset frequency of 475 Hz, (e) Ubq27- with offset frequency of 400 Hz, and (f) Ubq27+ with offset frequency of 400 Hz.

ions were found to dissociate into monomer and dimer ions at this temperature. However, the Ubq310- ion was found to ∼50% more reactive than the Ubq310+ ion under these conditions, with a dissociation rate constant of 1.1 s-1 (Ubq310-) compared to 0.7 s-1 (Ubq310+). The Ubq27( ions were found to be kinetically stable at reaction temperatures of up to 170 °C, which is the highest temperature accessible with current apparatus. Therefore, sustained off-resonance irradiation collision-induced dissociation (SORI-CID) was used to compare the stability of the ions. Shown in panels c-f in Figure 6 are illustrative SORI-CID mass spectra measured for the Ubq27( and Ubq310( ions using identical experimental conditions. The results obtained by SORI-CID for the Ubq310( ions indicate that the Ubq310- ions are kinetically less stable than the Ubq310+ ions, consistent with the BIRD data. Similarly, the Ubq27- ions are kinetically less stable than the Ubq27+ ions. Based on the results of the dissociation experiments, it is concluded that the nonspecific Ubqxn- ions are generally less stable (thermally) than the Ubqxn+ ions, at least under the experimental conditions investigated here. The lower stability of the Ubqxn- ions explains the more pronounced contribution of in-source dissociation to the distributions measured in negative ion mode.

Method for Distinguishing Specific from Nonspecific Protein Complexes. In addition to providing new insights into the influence of experimental and instrumental conditions on nonspecific protein association in nanoES-MS, the results of the present study strongly suggest that the gaseous ions of nonspecific Ubq complexes are produced from offspring droplets that, on average, are formed later in the ES process compared to the droplets that produce the Ubq monomer ions. This finding forms the basis for a new method for distinguishing between specific and nonspecific multiprotein complexes in nanoES-MS. The proposed method is outlined in Figure 7 for the special case where only ions corresponding to protein (P) monomer (Pn+) and homodimer (P2m+) are present in the mass spectrum. To establish whether nonspecific self-association of P occurs during the ES process and contributes to the P2m+ ion signal, a reporter molecule (Mrep), which does not interact specifically with the P or P2 (if present) in solution, is added to the ES solution. The Mrep is added at high concentration (>100 µM) to promote its nonspecific binding with P and P2. Nonspecific binding leads to the appearance of peaks in the mass spectrum corresponding to the P(Mrep)in+ and P2(Mrep)im+ ions, where i ) 0, 1, 2, .... The fraction (f x,i) of P and of P2 bound nonspecifically Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8307

Figure 7. Hypothetical mass spectra illustrating the proposed method for distinguishing between specific and nonspecific protein complexes in nanoES-MS. (a) Mass spectrum in case where only ions of a protein monomer (Pn+) and dimer (P2m+) are present. Distribution of Mrep molecules bound nonspecifically to Pn+ and P2m+ in the case where (b) P2m+ originates exclusively from a specific P2 complex in solution and (c) P2m+ originates wholly or in part from nonspecific self-association of P during the ES process. The f x,i values represent the fraction of Pn+ (x ) 1) and P2m+ (x ) 2) bound nonspecifically to i molecules of Mrep.

to i molecules of Mrep (relative to all possible numbers of Mrep) is given by the following expression:

∑

I(Px,i)n+

n

f x,i )

∑∑ i

n

n I(Px,i)n+

(2)

n

where I(Px,i) is the measured intensity for the Pn+ (x ) 1) and P2m+ (x ) 2) ions (including all metal adducts) bound nonspecifically to i molecules of Mrep. In the absence of in-source dissociation, the distribution of Mrep bound nonspecifically to P and P2 will be identical (i.e., f1,i ) f2,i) if the Pn+ and P2m+ ions have identical droplet histories.21 If, on the other hand, the droplet histories of Pn+ and P2m+ ions differ, then the distributions of Mrep bound nonspecifically to P and P2 will not be equivalent (i.e., f1,i * f2,i). If, as suggested from the results described above, the nonspecific protein complexes are formed preferentially from lategeneration droplets, which are enriched in protein, then it follows that late-generation offspring droplets will also be enriched in Mrep. Consequently, the P2m+ ions that result from nonspecific association are expected to undergo more extensive nonspecific binding to Mrep than the Pn+ ions. 8308

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

In principle, any molecule or ion can be used to play the role of Mrep. However, the proposed methodology is most easily implemented when a relatively small, neutral molecule is used. In this case, the nonspecific binding of Mrep to the proteins and protein complexes will not alter the charge state of the protein ions and will not spread the protein ion signal over a broad range of m/z values (more than a few hundred), where differences in detection efficiency may complicate the comparison of f x,i values. To test the utility of the proposed method for identifying the occurrence of nonspecific protein binding in nanoES-MS, it was applied under the following conditions: (i) where the protein complex originates exclusively from nonspecific binding, (ii) where the protein complex originates exclusively from specific binding in solution, and (iii) where the protein complex originates from both specific and nonspecific binding. A synthetic trisaccharide, 1, was used as the reporter molecule for these experiments. Small carbohydrates such as 1 are well suited for the role of Mrep since they are neutral molecules and they form relatively strong nonspecific interactions with proteins in the gas phase.20,33 i. Protein Complex Originating from Nonspecific Binding. Ubq was used to demonstrate the efficacy of the new method in cases where the protein complexes originate exclusively from nonspecific binding. NanoES-MS measurements were performed on solutions of Ubq, at a range of concentrations, and 1. Shown in Figure 8a is an illustrative mass spectrum acquired from a solution containing Ubq 210 µM and 1 420 µM. The dominant Ubq ions, as revealed by the mass spectrum, correspond to monomer and dimer bound to between 0 and 4 molecules of 1, i.e., Ubqxn+ (1)i ions, where x ) 1, 2 and i ) 0-4. Shown in Figure 8b are the corresponding normalized distributions of Ubq(1)i and Ubq2(1)i species determined from the mass spectrum. It can be seen that the Ubq(1)i and Ubq2(1)i distributions are statistically different, with the Ubq2n+ ions participating in greater nonspecific binding to 1 compared to the Ubqn+ ions. The enhanced nonspecific binding of 1 to the Ubq2n+ ions is consistent with the formation of these ions from offspring droplets that are produced later in the ES process, compared to droplets that give rise to the Ubqn+ ions. Listed in Table 1 are the ratios of the f x,i values, for a given i, determined for the Ubq monomer and dimer at three different Ubq concentrations (120, 210, and 310 µM). Notably, at each concentration. the ratios of f x,i values are not equal to unity (i.e., f2,i/f1,i * 1) indicating that Ubq(1)i and Ubq2(1)i distributions are nonequivalent. ii. Protein Complex Originating from Specific Binding. Bovine β-lactoglobulin is a relatively small protein (18.4 kDa), which, under physiological conditions, exists as both monomer (Lg) and dimer (Lg2).34,35 An association constant, Kassoc, of 9.5 ( 0.9 × 104 M-1 has been determined for the monomer-dimer equilibrium in aqueous solution at 25 °C and neutral pH.36 Shown in panels a and b in Figure 9 are representative mass spectra acquired for a solution of β-lactoglobulin (100 µM) in the absence and presence of 1 (210 µM), respectively. In the absence of 1, the major ions detected correspond to Lg and Lg2. The equilibrium (33) Wang, W.; Kitova, E. N.; Sun, J.; Klassen, J. S. J. Am. Soc. Mass Spectrom. 2005, 16, 1583-1594. (34) Qin, B. Y.; Bewley, M. C.; Creamer, L. K.; Baker, H. M.; Baker, E. N.; Jameson, G. B. Biochemistry 1998, 37, 14014-14023. (35) Sakurai, K.; Goto, Y. J. Biol. Chem. 2002, 277, 25735-25740. (36) Apenten, R. K. O.; Galani, D. Thermochim. Acta 2000, 359, 181-188.

Figure 8. (a) NanoES mass spectrum acquired for an aqueous solution of 5 mM ammonium acetate, 210 µM Ubq, and 420 µM 1 in positive ion mode. (b) Fraction (f x,i) of Ubq and Ubq2 bound nonspecifically to i molecules of 1. The errors correspond to one standard deviation. Table 1. Comparison of the f x,i Terms for the Nonspecific Binding of Mrep (≡ 1) to Ubqn+ (x ) 1) and Nonspecific Ubq2n+ (x ) 2) Ions Determined by NanoES-MSa,b [Ubq] (µM)

[1] (µM)

f2,0/f1,0

f2,1/f1,1

f2,2/f1,2

120 210 310

300 210 310

0.73 ( 0.02 0.79 ( 0.06 0.69 ( 0.05

1.01 ( 0.08 1.39 ( 0.19 1.69 ( 0.09

1.79 ( 0.53 1.62 ( 0.53 1.50 ( 0.38

a Ratios calculated from average f x,i values taken from 5 measurements. b Errors correspond to one standard deviation.

constant (Kassoc) for the Lg association reaction, 2Lg H Lg2, can be calculated directly from the mass spectrum using eq 3:

Kassoc )

[Lg2]eq [Lg]eq2

∑ )

n

I(Lg2)n+ n

(∑ ) I(Lg)n+

n

2

(3)

n

where [Lg]eq and [Lg2]eq are the equilibrium concentrations of Lg and Lg2, respectively, and I(Lgn+) and I(Lg2n+) are the measured intensities of the Lgn+ and Lg2n+ ions, respectively.

Figure 9. NanoES mass spectra acquired for an aqueous solution of 5 mM ammonium acetate and (a) 100 µM Lg or (b) 100 µM Lg and 210 µM 1. (c) Fraction (f x,i) of Lg and Lg2 bound nonspecifically to i molecules of 1. The errors correspond to one standard deviation.

Because the Lgn+ and Lg2n+ ions are detected over a range of charge states, the signal intensities corresponding to ions of different charge states were normalized for charge and the ion abundances were summed together. Importantly, the average value of Kassoc (6 ( 3 × 104 M-1) calculated from the measured intensities of the Lgn+ and Lg2n+ ions, and assuming equivalent response factors, is in agreement with the reported value. Although not definitive, this result does suggest that nonspecific association of Lg during the nanoES process does not contribute appreciably to the detected Lg2n+ ions. The addition of 1 to the ES solution leads to the appearance of Lgx(1)in+ ions, where i ) 0-3. Shown in Figure 9c are the normalized distributions of Lgx(1)i species for Lg and Lg2. Notably the two distributions are identical, within experimental error, indicating that the Lgn+ and Lg2n+ ions have similar droplet histories and, consequently, that the detected Lg2n+ ions originate entirely from the specific Lg2 complex in solution. Measurements were performed on solutions Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8309

Table 2. Comparison of the f x,i Terms for the Nonspecific Binding of Mrep (≡ 1) to Lgn+ (x ) 1) and Lg2n+ (x ) 2) Ions Determined by NanoES-MSa,b [β-lactoglobulin] (µM)

[1] (µM)

f2,0/f1,0

f2,1/f1,1

30 100 180

210 210 210

0.95 ( 0.04 0.99 ( 0.03 0.98 ( 0.04 (0.97 ( 0.04)c

0.96 (( 0.07 0.96 ( 0.05 0.97 ( 0.03 (0.96 ( 0.05)c

a Ratios calculated from average f x,i values taken from 5 measurements. b Errors correspond to one standard deviation. c Average values of f2,i/f1,i based on all measurements.

of Lg at two other concentrations (30 and 180 µM) and the ratio of f x,i values, for i ) 0 and 1, determined for Lg and Lg2 are listed in Table 2. In each case, the ratios are, within error, equal to 1.0, indicating that the Lgn+ and Lg2n+ ions have similar droplet histories and that Lg2n+ ions originate exclusively from the specific Lg2 complex. iii. Protein Complex Originating from Both Specific and Nonspecific Binding. It was also of interest to demonstrate the utility of this method in cases where both specific and nonspecific binding contribute to the ion signal for protein complexes. To accomplish this, nanoES-MS was performed on solutions of Lg at elevated concentration (>200 µM) and basic pH (8.5). The Kassoc for the monomer-dimer equilibrium is 1 order of magnitude smaller at pH 8.5 compared to pH 7.37 This, combined with the higher concentration of Lg is expected to increase the probability of nonspecific self-association of Lg during the ES process. Shown in panels a and b in Figure 10a are mass spectra acquired for solutions of Lg (290 µM) in the absence and presence of 1, respectively, at pH 8.5. In the absence of 1, only protein ions corresponding to Lgn+ and Lg2n+ ions are detected. Importantly, the absence of Lg3n+ ions in the mass spectrum suggests that nonspecific association is absent under these conditions. Shown in Figure 10c are the distributions calculated for the Lg(1)i and Lg2(1)i species. Notably, the distributions are nonequivalent, which indicates that nonspecific association of Lg does in fact contribute to the signal for the Lg2n+ ions. However, the fractional contribution of nonspecific association of Lg to the Lg2n+ ion signal cannot be established from these data. Taken together, the results described above support the hypothesis that the gaseous ions of multiprotein complexes originating from specific binding in solution and from nonspecific binding during the ES process have, on average, different ES droplet histories. More importantly, the results confirm that the proposed Mrep method, which involves the addition of a reporter molecule to the ES solution, can be used to identify the contribution of nonspecific binding to the ion signal of multiprotein complexes in nanoES-MS. It should also be noted that the Mrep method is not limited to the study of nonspecific protein-protein binding in nanoES-MS. Rather, this method can serve as a general approach for identifying nonspecific complexes involving macromolecules. The application of this method to the study of nonspecific protein-ligand binding in nanoES-MS will be described in a forthcoming publication.38 (37) Invernizzi, G.; Sˇ amalikova, M.; Brocca, S.; Lotti, M.; Molinari, H.; Grandori, R. J. Mass Spectrom. 2006, 41, 717-727.

8310 Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

Figure 10. NanoES mass spectra acquired for an aqueous solution (at pH 8.5) of (a) 290 µM Lg or (b) 290 µM Lg and 310 µM 1. (c) Fraction (f x,i) of Lg and Lg2 bound nonspecifically to i molecules of 1. The errors correspond to one standard deviation.

CONCLUSIONS Using Ubq as a model protein, the first systematic study of the nonspecific self-association of proteins in nanoES-MS, and the influence of experimental conditions thereon, was carried out. Importantly, it was shown that extent of nonspecific binding is strongly sensitive to protein concentration, spray potential, and accumulation time in the external rf hexapole of the ion source. The influence of concentration on the extent of nonspecific Ubq binding is consistent with formation of the gaseous ions via the charge residue model of ES. In negative ion mode, it was noted that relative abundance of the metal ion (Na+ and K+) adducts exhibited a dependence on the size of the multimeric state of the Ubq ions, with more abundant adducts detected for the larger multimeric ions. This observation strongly suggests that the gaseous ions of nonspecific Ubqx complexes are produced, on average, from offspring droplets that are formed later in the ES process compared to the droplets that produce the monomeric Ubqn( ions. The greater extent of nonspecific binding observed at low spray potentials is intriguing and suggests that the lower (38) Sun, J.; Sun, N.; Kitova, E. N.; Klassen, J. S. Manuscript in preparation.

spray potentials promote the sampling of ions originating from late-generation droplets. The origin of this effect is not fully understood. However, it is speculated that it is related to a decrease in the size of the initial ES droplets formed at lower spray potentials. The extent of nonspecific protein self-association was also found to increase with increasing accumulation time in the rf hexapole in positive ion mode. The effect was attributed to the preferential trapping of high m/z ions in the hexapole at high space charge. In negative ion mode, the extent of nonspecific selfassociation was found to initially increase and then decrease with accumulation time. It was suggested that this behavior results from the combined effects of high space charge and collisioninduced dissociation of the deprotonated Ubqxn- ions, which were shown to be kinetically less stable than the corresponding Ubqxn+ ions. The finding that the monomeric and nonspecific multimeric Ubq ions have, on average, different droplet histories forms the basis for a new method for distinguishing between specific and nonspecific multiprotein complexes in nanoES-MS. The method involves the addition of a reporter molecule, which does not bind specifically to the proteins/complexes of interest in solution, to the ES solution at relatively high concentration. The distribution of Mrep bound nonspecifically to gaseous ions of the proteins and

protein complexes, as determined from the mass spectrum, can be used to establish whether a given protein complex originates exclusively in solution or whether nonspecific binding during the ES process contributes to its formation. The method was successfully implemented in cases where the gaseous ions of a protein complex detected by nanoES-MS originated exclusively from nonspecific association, exclusively from specific interactions in solution, and from both specific and nonspecific interactions. ACKNOWLEDGMENT The authors acknowledge the Natural Sciences and Engineering Research Council of Canada and the Alberta Ingenuity Centre for Carbohydrate Science for generous funding and P. Kebarle for helpful comments during the preparation of the manuscript. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review May 8, 2007. Accepted August 3, 2007. AC0709347

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8311