Anal. Chem. 1997, 69, 3188-3192
Analysis of Macromolecules Using Nanoelectrospray Ionization Mass Spectrometry and Low-Energy Collision Activation Philippe A. Guy and Robert J. Anderegg*
Department of Analytical Chemistry, Glaxo Wellcome Inc., Five Moore Drive, Research Triangle Park, North Carolina 27709
The mass spectrometric analysis of several proteins using nanoelectrospray (nanoES) with elastic collisions showed an improvement in the sensitivity over nanoES without collisional activation. We believe this effect is due to a better declusterization/ionization process. Optimization of the collision parameters can be easily performed during the long experiment time allowed using the nanoES source. Moreover, an apparent shift in the charge-state distribution is observed, with lower charged ions becoming relatively more abundant with increasing either target gas pressure or kinetic energy of the precursor ions. Higher charge-state ions might be expected to have higher collision frequencies and correspondingly lose more kinetic energy than lower charge-state ions. Mass spectrometry (MS) has proven to be a powerful method for the characterization of macrobiomolecules such as proteins or oligonucleotides. This is especially true since the late 1980s, with the development of two gentle ionization methods: electrospray (ESI)1 and matrix-assisted laser desorption (MALDI).2 These methods have opened new opportunities for the MS of macromolecules by greatly extending the molecular size and type of samples amenable to MS.3 A standard ESI source operates at a flow rate of 1-20 µL/min; higher, if nebulization is used to assist in droplet formation. However, when the sample amount is limited, it is advantageous to use a minimal volume of sample solution introduced at the lowest possible flow rate. This has recently been accomplished with the development of capillary electrophoresis coupled to ESIMS without liquid sheath flow4 or using microelectrospray5 and nanoelectrospray (nanoES)6,7 ionization sources. The nanoES source produces a stable ion signal at a flow rate of 20-40 nL/min, leading to smaller droplet diameter compared to the standard electrospray process (180 nm predicted for nanoES versus 3 µm for standard ESI).7,8 Using as little as 1 µL of sample solution, it is possible to perform analysis for a period of up to 50 min, which allows extensive MS/MS investigation of several peptides in a complex mixture.7,9-11 Thus, in combination with a micropurification procedure, it has been shown that (1) Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984, 88, 4451-4459. (2) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301. (3) Siuzdak, G. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 11290-11297. (4) Wahl, J. H.; Goodlett, D. R.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1992, 64, 3194-3196. (5) Emmett, M. R.; Caprioli, R. M. J. Am. Soc. Mass Spectrom. 1994, 5, 605613. (6) Wilm, M. S.; Mann, M. Int. J. Mass Spectrom. Ion Processes 1994, 136, 167180. (7) Wilm, M. S.; Mann, M. Anal. Chem. 1996, 68, 1-8. (8) Tang, L.; Kebarle, P. Anal. Chem. 1993, 65, 3654-3668.
3188 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997
femtomole amounts of digested proteins either from a gel9 or from a microcolumn packed with immobilized trypsin10 provide sufficient information for protein identification. Because of both the low voltage applied at the microspray capillary and the absence of nebulizer gas, the microelectrospray ionization source may play an interesting role in the study of noncovalent interactions due to a gentler environment compared to classical electrospray.12 Nevertheless, the nanoES source has been almost exclusively focused on peptide studies. So far, the only protein spectra reported are in the analysis of a glycoprotein (ovalbumin) and tissue plasminogen activator (after carboxypeptidase digestion) using the nanoES source7 and a protein-ligand complex using microspray.12 However, protein analysis could be of great interest, especially in cases when the amount of protein is limiting or when better mass accuracy, obtained by averaging the signal over a long time period, is desired. Thus, we have investigated the analysis of proteins using nanoES and compared the protein mass spectra obtained by either scanning the first quadrupole or the third quadrupole with collision activation. The latter approach has already been used in electrospray studies either for probing the modification of charge-state distribution (CSD) in proteins13-16 or with ion trap MSn experiments for obtaining sequence information.17 At low collision energies, rather than dissociating into sequence-specific fragments, multiply charged ions have been reported to undergo ionmolecule reactions with appropriate targets, resulting in a shift of the CSD to lower charge (higher m/z ratio) values.13-15,18,19 We describe the utilization of nanoES for the analysis of proteins using elastic collision carried out in the radio frequency(9) Wilm, M. S.; Shevchenko, A.; Houthaeva, T.; Breit, S., Schweigerer, L.; Fotsis, T.; Mann, M. Nature 1996, 379, 1, 466-469. (10) Blackburn, K. R.; Anderegg, R. J. J. Am. Soc. Mass Spectrom. 1997, 8, 483494. (11) Carr, S. A.; Huddleston, M. J.; Annan, R. S. Anal. Biochem. 1996, 239, 180192. (12) Robinson, C.; Jarvis, S.; Chung, E.; Hugues, C.; Green, B. Proceedings of the 44nd ASMS Conference on Mass Spectrometry and Allied Topics; 1996; p 1408. (13) Ikonomou, M. G.; Kebarle, P. Int. J. Mass Spectrom. Ion Processes 1992, 117, 283-298. (14) Feng, R.; Konishi, Y. Proceedings of the 39th ASMS Conference on Mass Spectrometry and Allied Topics, 1991; pp 1432-1433. (15) Cox, K. A.; Julian, R. K.; Cooks, R. G.; Kaiser, R. E., Jr. J. Am. Soc. Mass Spectrom. 1994, 5, 127-136. (16) Smith, R. D.; Barinaga, C. J. Rapid Commun. Mass Spectrom. 1990, 4, 5457. (17) McLuckey, S. A.; Glish, G. L.; Van Berkel, G. L. Anal. Chem. 1991, 63, 1971-1978. (18) Loo, R. R.; Loo, J. A.; Udseth, H. R.; Fulton, J. L.; Smith, R. D. Rapid Commun. Mass Spectrom. 1992, 6, 159-165. (19) Ogorzalek Loo, R. R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 1994, 5, 207-220. S0003-2700(96)01293-0 CCC: $14.00
© 1997 American Chemical Society
only collision cell of a triple quadrupole mass spectrometer. Because of the long sample analysis time of nanoES, the same sample can be analyzed under a variety of Q1 and Q3 scanning conditions. Q3 scanning with collision activation has shown a sensitivity improvement of 3-5-fold, better peak shape, and a reduction in chemical noise in the lower m/z ratios. The effect of several parameters involved in the signal response, such as the voltage applied to the nanoES needle (ISV), the interface temperature, the interface plate (IN), and the orifice skimmer (OR) potentials, were studied. We have also investigated the influence of collision gas thickness (CGT) and the effects of collision energy (R0-R2) on the mass spectra of proteins. EXPERIMENTAL SECTION Samples and Reagents. Poly(propylene glycol) and methanol were purchased from Aldrich Chemical Co. (Milwaukee, WI). Formic acid was purchased from J. T. Baker Inc. (Phillipsburg, NJ). Chicken egg white lysozyme and bovine hemoglobin R chain were purchased from Sigma Chemical Co. (St. Louis, MO) and used without further purification. The SH2 domain of pp60c-src tyrosine kinase (residues 144-259) was expressed and purified in-house at Glaxo Wellcome. Instrumentation. Nanoelectrospray ionization mass spectrometry (nanoESMS) was carried out using an API III+ triple quadrupole mass spectrometer (Perkin-Elmer Sciex Instruments, Thornhill, ON, Canada) equipped with a nanoES source obtained from European Molecular Biology Laboratory (Heidelberg, Germany). NanoES capillaries were obtained precoated from the Protein Analysis Co. (Odense, Denmark). The instrument was calibrated using poly(propylene glycol) with the Sciex Ionspray source. Samples were dissolved at a concentration of 5 pmol/µL (SH2 and hemoglobin proteins) and 10 pmol/µL (lysozyme) in 60% methanol, 5% formic acid (pH 3.1). One microliter of the sample solution was loaded into the capillary as already described.7 When the flow rate stopped, a soft touch of the needle tip on the valve gate plate (voltage on) re-initiated flow. However, comparisons performed in this study were achieved without repositioning the needle tip or modifying the flow rate by rebreaking the needle tip. A new capillary was used for each sample. Several parameters involved in ion signal such as the voltage applied to the nanoES needle (ISV 600-1020 V), the interface plate (IN 80-180 V), and orifice skimmer (OR 10-100 V) potentials and the interface temperature (50-80 °C) were studied. The effect of the collision energy (R0-R2) was also investigated. Nitrogen curtain gas flow was set at 0.6 L/min. During the acquisition using Q3 with collision activation, a mixture of argon/nitrogen (9:1) was used. Collision gas thickness was used in the range of (0.2-3.0) × 1014 molecules/cm2 (CGT 20-300). Spectra were typically recorded in the 300-2300 range of mass-to-charge (m/z) ratios in step of 0.3 m/z, with a 0.7 ms dwell time. Our ISV, IN, and R2 potentials and interface temperature standard analysis parameters were typically set at 650, 100, and 0 V and 50 °C, respectively while OR potential was set at 60 (for SH2 and hemoglobin) and 80 V (for lysozyme). The signal was accumulated over 10 (for SH2 and hemoglobin) and 20 scans (for lysozyme) in the MCA data acquisition mode. RESULTS AND DISCUSSION Analysis of Proteins Using Q1. In the first part of our study, we analyzed proteins using nanoESMS by scanning Q1. On our
Figure 1. NanoES spectra of native chicken egg white lysozyme by scanning Q1 with OR set at (a) 40, (b) 60, (c) 80, and (d) 100 V. The counts per second observed by the detector has been reported at the right top of each spectrum. The inset spectra correspond to the expanded region of the +9 charged ion.
triple quadrupole instrument, the energy of the incoming ions is controlled by the OR. Figure 1 shows the effect on the mass spectra of chicken egg white lysozyme of changing the OR in a 40-100 V range. The CSD ranges from +8 to +13 and maximizes at the +9 charged ion. Increasing the OR potential results in a better desolvation/ionization process leading to higher signal-tonoise ratio; with an optimum at OR ) 80 V (Figure 1c). Standard ESIMS analysis performed on the same sample has shown an optimum ion detection at OR ) 100 V with a lower CSD, ranging from +7 to +11, but also maximizing at the +9 charged ion. The number in the upper right of each panel refers to the counts per second observed by the detector and is a measure of the magnitude of the mass spectrometric response. However, no differences in the mass assignments were observed at any orifice potential tested using either standard or nanoelectrospray ionization techniques. One can notice a better declusterization efficiency, particularly for the +8 to +10 charged ions when increasing the OR potential, as evidenced by the sharpening of the mass spectral signals and the diminution of the apparent adduct peaks to the right of each protein ion (see the expanded region of the +9 charged ion). This improvement in ion abundance is in agreement with previous reports using a standard electrospray source.19 Some “tailing” on the high-mass side remains for lower charged ions (i.e., the right side of the +8 to +10 predicted ions), perhaps because of incomplete desolvation or adduct formation. The difference in ions behavior between the low and high CSD (i.e., +8 to +10 versus +11 to +13) might be explained by the presence of two conformations of the protein in which the +8, +9, and +10 charged ions correspond to a more folded structure, preserving either solvent or other adducts. Analytical Chemistry, Vol. 69, No. 16, August 15, 1997
3189
Figure 2. Comparison of lysozyme nanoES spectra by scanning (a) Q1 and (b) Q3 with collision gas thickness set at 1.8 × 104 molecules/cm2. The counts per second observed by the detector has been reported at the right top of each spectrum. The inset spectra correspond to the expanded region of the +8 charged ion.
The overall ionization efficiency in electrospray involves a combination of desolvation, ionization, and transfer efficiencies into the vaccum system. Compared to standard electrospray ionization, the desolvation efficiency using the nanoES source is reported to be increased because the droplets are smaller and monodisperse.6 Using a synthetic peptide, Wilm and Mann have demonstrated almost 2 orders magnitude higher efficiency of converting analyte molecules into ion current in the mass spectrometer using the nanoES source.7 However, this effect is balanced by the lower potential applied to the nanoES capillary and the absence of nebulizer gas, resulting in a gentler environment, allowing the preservation of weakly associated proteinligand12 and, presumably, protein-solvent complexes. Our results with the orifice potential (see above) suggest that enhanced desolvation/declustering help in the protein mass spectra analyzed using nanoES, so we explored the use of collisional activation of the proteins. Comparison of Spectra Using NanoESMS by Scanning Q1 or Q3 Quadrupoles with Collision Activation. Figure 2 shows the mass spectra of lysozyme using nanoESMS by scanning Q1 (a) or Q3 with collisional activation (b). The counts per second observed at the detector indicate an improvement in sensitivity when scanning Q3 with collision gas by a factor of 5 compared to the Q1 results. The CSD was found almost identical, except for the appearance of a small +7 charge ion when scanning Q3. Furthermore, the observation of the expanded region of the +8 charged ion (Figure 2, inset spectra) shows a greater degree of decluterization with collision activation, as evidenced by the loss of the small signals to the right of each protein ion. No change 3190 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997
Figure 3. NanoES spectra of hemoglobin by scanning (a) Q1 and (b) Q3 with a collision gas thickness set at 1.0 × 104 molecules/cm2. The mass of the protein was calculated at 15 053.2 Da. The counts per second observed by the detector has been reported at the right top of each spectrum.
in protein mass has been observed during these experiments, which might suggest the loss (or gain) of protons or other charged species. In Figure 3, we show the mass spectra derived from a 5 pmol/ µL solution of bovine hemoglobin chain R. In this example, the improvement of sensitivity was a factor of 3. After peak centroid was assigned, the mass of the protein using Q1 and Q3 with collision activation was measured to be M ) 15 058.8 ( 1.5 and 15 054.9 ( 1.4 Da, respectively. Compared to the 15 053.2 Da calculated mass, it appears in this example that using Q3 with elastic collisions can lead to better accuracy in protein mass assignment. Under our denaturing conditions, we also can note the presence of the heme (m/z 616.1). Moreover, using these analysis conditions, one can note the improvement of the signalto-noise ratio. The SH2 domain of the tyrosine kinase pp60c-src was analyzed with and without collision activation (Figure 4) and provided an even more dramatic example. The poor relative intensity of the protein-derived ions in Figure 4a (scanning Q1) is due to the high chemical noise observed in the lower m/z range. This noise can be strongly reduced by scanning Q3 with collision activation, which results in a higher signal-to-noise ratio for the protein ions, even though the overall ion current for the spectrum has decreased by 90% (Figure 4b). Thus, the mass of the protein can be easily measured at 12 287.5 ( 1.3 Da, which is in agreement with the 12 286.9 Da calculated mass. Similar sample solution and instrument parameters were used during both analyses, so that the only difference observed resulted from the addition of argon/nitrogen gas (9:1) in the collision cell (Q2). On the basis
Figure 4. NanoES spectra of SH2 domain of the tyrosine kinase pp60c-src by scanning (a) Q1 and (b) Q3 with a collision gas thickness set at 1.5 × 104 molecules/cm2. The counts per second observed by the detector has been reported at the right top of each spectrum.
of these results, we then studied the influence of several other instrument parameters on the protein signal. Study of Several Parameters Involved in NanoES Ionization by Scanning Q3 with Collision Activation. First, we studied the modification of collision gas pressure in the collision cell. In Figure 5, we show the mass spectra of lysozyme, varying the collision gas thickness in a range of (1.5-2.5) × 1014 molecules/cm2 while the kinetic energy of the precursor ions was set at 30 eV (R0-R2). Higher gas pressure (beyond CGT 250) leads to decreased sensitivity, presumably due to ion scattering, until eventually the signal dissapears (CGT 300). It is clear that collision gas pressure is important in the sensitivity improvement. Similar experiments were done on several proteins. However, no standard optimum pressure has been found during our experiments, rather the best pressure seems to depend on size and conformation of the protein. We noticed a reduction of peak adducts at each charge state when the gas pressure was increased. An apparent change in ion abundance is observed, with higher charge states (+13 to +10) becoming less abundant with increasing target gas pressure. However, the abundance of the +7, +8, and +9 charged ions remains more or less constant. This behavior can be explained based on collision cross section of the ions. A larger ion will undergo many more collisions than a smaller one traversing the same path. Thus, the larger ion will lose more kinetic energy than the smaller ion and, with sufficient energy loss, will reach a point at which it is unable to exit the second quadrupole. Moreover, the ion scattering resulting from the high gas pressure can lead to the same effect. These results are in agreement with other authors who have used ESIMS to study collisional activation
Figure 5. NanoES spectra of lysozyme by scanning Q3 with collision gas thickness set at (a) CGT 150, (b) CGT 180, and (c) CGT 250. The counts per second observed by the detector has been reported at the right top of each spectrum.
when collisions take place either in the collision cell13-15 or in the mass spectrometer source after installation of a small reactor.16,19 Compared to the 19 basic sites found in the lysozyme sequence (calculated by summation of N-terminal and arginine, lysine, and histidine amino acids), only 13 charges were observed in the mass spectrum. The presence of the four disulfide bridges in lysozyme still maintains a certain conformation of the protein even under denaturing conditions. After reduction of the disulfide bridges, the CSD ranged from +11 to +22, indicating that other amino acids become protonated (results not shown). Using ESIMS, Loo et al. have studied the protein structural effects in gas phase ion/ molecule reactions with several amines before and after disulfidebridges reduction.18,19 They have observed that the reduced protein ions are less reactive than the native protein ions of the same charge state because in the latter case the charges would be spaced more closely (more constrained conformation). Another important parameter involved in collisional activation is the kinetic energy (R0-R2) of the precursor ions. This energy is modified by varying the R2 potential on the Sciex mass spectrometer. Collision gas thickness was set at 1.8 × 104 molecules/cm2 (CGT 180) because of the high sensitivity observed (see above). In Figure 6, the kinetic energy of the precursor ions derived from lysozyme was set at +10, +40, and +60 eV (i.e., R2: +20, -10, and -30 V, respectively). Other instrument voltages were adjusted correspondingly. The recorded spectra exhibit a CSD shift to lower charges, with the appearance of a small +7 ion when the kinetic energy was set above 40 eV (Figure 6b and c). Indeed, as has previously observed with gas target pressure, peak adducts tend to be lower as kinetic energy increases. Analytical Chemistry, Vol. 69, No. 16, August 15, 1997
3191
temperature, which is in agreement with a previous report.20 Other parameters such as nanoelectrospray voltage and interface potential modified in the range of 600-1000 and 80-180 V, respectively, were studied. We noticed only a small improvement in sensitivity by increasing IN potential, but peak broadening occurred (results not shown). The increasing of nanoES (ISV) voltage resulted in broad peaks without any gain in sensitivity. Indeed, in certain cases, lowering the ISV voltage increased the signal intensity. Using collision activation, Cox et al. observed the presence of two conformational forms for several proteins.15 It seems plausible that larger cross sections of ions in higher charge state results from a more open, extended structure of the proteins, possibly from Coulombic repulsion of the charges. However, different protein conformations need not necessarily give different cross sections. Indeed, the average over all orientations may be most important. For example, a rod-shaped protein ion can give either a larger or smaller cross section than a spherical protein ion, depending on its orientation.
Figure 6. NanoES spectra of lysozyme by scanning Q3 with kinetic energy of precursor ions set at (a) +10, (b) +40, and (c) +60 eV. Collision gas thickness was fixed at 1.8 × 104 molecules/cm2. The counts per second observed by the detector has been reported at the right top of each spectrum.
However ion signals broadened with increasing collision energy, and a M-17 Da fragment ion was observed when the kinetic energy was set above 40 eV. In order to complete our study, we investigated the importance of other instrumental parameters during the analysis of proteins using collision activation, but none had as a significant effect as those described above. When the interface temperature was increased, the relative abundance of the higher charged ions increased slightly (results not shown). This variation can be explained by denaturation of the protein occurring at high (20) Mirza, U. A.; Cohen, S. L.; Chait, B. T. Anal. Chem. 1993, 65, 1-6.
3192 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997
CONCLUSIONS The results of our studies show that protein analysis using nanoES with collision activation leads to an improvement in sensitivity. Examination of the phenomenon reveals that the relative abundance of ions in the CSD can be altered by changing the collision conditions. Those optimizations can be easily performed during the time of analysis allowed using nanoES source. Both collision gas thickness and kinetic energy of the precursor ions were shown to play an important role in the CSD shift. As reported by several authors, these energy loss studies for multiply charged ions could be used for demonstrating the presence of several conformations of gas-phase proteins. Additional nanoES investigations exploring these issues are being pursued. Received for review December 30, 1996. Accepted June 2, 1997.X ACKNOWLEDGMENT Kevin Blackburn is acknowledged for fruitful discussions. AC961293A X
Abstract published in Advance ACS Abstracts, July 15, 1997.
