Functional Probing of Arginine Residues in Proteins Using Mass

Anal. Chem. 2005, 77, 4481-4488

Functional Probing of Arginine Residues in Proteins Using Mass Spectrometry and an Arginine-Specific Covalent Tagging Concept Alexander Leitner* and Wolfgang Lindner*

Christian Doppler Laboratory for Molecular Recognition Materials, Institute of Analytical Chemistry and Food Chemistry, University of Vienna, 1090 Vienna, Austria

The reactivity of arginine residues in model proteins (ubiquitin, cytochrome c, myoglobin, ribonuclease A, lysozyme) was examined using a selective tagging reaction in combination with on-line monitoring of the reaction progress by electrospray ionization mass spectrometry (ESI-MS). The kinetics of this reaction, based on the cyclization of the guanidine group of arginine with 2,3butanedione and phenylboronic acid at pH 8-10, allow the grouping of arginines in “exposed” or “partially buried” residues, because they differ substantially in their reaction rate constants for the conversion of the guanidine groups. The method allows one to differentiate between different protein conformations as shown for myoglobin and its apo form and native and reduced ribonuclease A: Removal of the heme group in myoglobin resulted in an increased reactivity for the two partially buried arginines. For RNAse A, quantitative reduction of the disulfide bonds lead to the exposure of an additional arginine residue and two different conformations of the reduced protein were observed by ESI-MS that could be distinguished according to their charge-state distribution. Experimentally obtained accessibilities were compared with solvent-accessibility data calculated from 3D structures and substantial agreement between both techniques was observed. In recent years, mass spectrometry (MS) has increasingly become a complementary tool for investigating structural properties of proteins, especially in combination with electrospray ionization (ESI). For example, MS is being used in protein crosslinking1,2 and hydrogen/deuterium exchange (HDX) studies.3-7 A unique feature of ESI-MS spectra is also the possibility to differentiate between different protein conformations according to their charge-state distributions.5,7-8 * To whom correspondence should be addressed. Fax: +43 1 4277 9523. E-mail: [email protected]. [email protected]. (1) Sinz, A. J. Mass Spectrom. 2003, 38, 1225-1237. (2) Back, J. W.; De Jong, L.; Moijsers, A. O.; De Koster, C. G. J. Mol. Biol. 2003, 331, 303-313. (3) Englander, S. W. Annu. Rev. Biophys. Biomol. Struct. 2000, 29, 213-238. (4) Kaltashov, I. A.; Eyles, S. J. Mass Spectrom. Rev. 2002, 21, 37-71. (5) Konermann, L.; Simmons, D. A. Mass Spectrom. Rev. 2003, 22, 1-26. (6) Hoofnagle, A. N.; Resing, K. A.; Ahn, N. G. Annu. Rev. Biophys. Biomol. Struct. 2003, 32, 1-25. (7) Yan, X.; Watson, J.; Ho, P. S.; Deinzer, M. L. Mol. Cell. Proteomics 2004, 3, 10-23. 10.1021/ac050217h CCC: $30.25 Published on Web 06/03/2005

© 2005 American Chemical Society

Perhaps one of the most significant advantages of using mass spectrometric detection over other techniques is the minute amount of sample that is required, typically orders of magnitude less than, for example, for nuclear magnetic resonance (NMR) spectroscopy. In combination with enzymatic digestion and hyphenation to high-performance liquid chromatography, structural features can be assigned on the level of individual amino acid residues.9-12 In addition, even large proteins or protein complexes (>100 000 Da) can be probed by ESI-MS, something that is quite difficult to achieve with NMR spectroscopy. A most fundamental parameter defining the location of individual amino acid side chains in a protein is the solvent-accessible surface area (ASA).13 It is defined by the area covered by rolling a spherical probe over the surface of a protein, typically using the van der Waals radius of water (1.4 Å) as the probe size. There are a number of programs publicly available that allow the calculation of the ASA from protein 3D structures, e.g., GETAREA14 or NACCESS.15 Another, more refined concept to determine the location of residues within a protein is the determination of residue/atom depth.16-17 Such basic structural parameters, however, will not always directly reflect reactivities of specific functional groups of amino acid residues toward larger molecules, such as modifying reagents, even if the reaction conditions do not dramatically alter the protein structure. These kinds of reagents are frequently used for structural probing of proteins in combination with mass spectrometry. To name just two examples, bifunctional tags that target the -amino group of lysine are commonly used to crosslink proteins to obtain spatial constraints,1-2 and different reactivities of cysteine residues can be investigated using alkylating (8) Grandori, R. Curr. Org. Chem. 2003, 7, 1589-1603. (9) Pan, H.; Raza, A. S.; Smith, D. L. J. Mol. Biol. 2004, 336, 1251-1263. (10) Hamuro, Y.; Anand, G. S.; Kim, J. S.; Juliano, C.; Stranz, D. D.; Taylor, S. S.; Woods, V. L., Jr. J. Mol. Biol. 2004, 340, 1185-1196. (11) Yan, X.; Broderick, D.; Leid, M. E.; Schmierlik, M. I.; Deinzer, M. L. Biochemistry 2004, 43, 909-917. (12) Black, B. E.; Foltz, D. R.; Chakravarthy, S.; Luger, K.; Woods, V. L., Jr.; Cleveland, D. W. Nature 2004, 430, 578-582. (13) Lee, B.; Richards, F. M. J. Mol. Biol. 1971, 55, 379-400. (14) Fraczkiewicz, R.; Braun, W. J. Comput. Chem. 1998, 19, 319-333. (15) Hubbard, S. J.; Thornton, J. M. ‘NACCESS’, Computer Program, Department of Biochemistry and Molecular Biology, University College London, 1993. (16) Chakravarty, S.; Varadarajan, R. Structure 1999, 7, 723-732. (17) Pintar, A.; Carugo, O.; Pongor, S. Trends Biochem. Sci. 2003, 28, 593597.

Analytical Chemistry, Vol. 77, No. 14, July 15, 2005 4481

tags.18-19 MS has also been used to probe protein modifications from photochemical20,21 or radiolytic22,23 oxidation processes, although these technqiues are less residue-specific. Arginine is another amino acid that is of significant importance for protein function. Owing to the outstanding basicity of the guanidino group, arginine residues frequently serve as recognition sites for other proteins or RNA, for example, by interacting with phosph[on]ate or sulf[on]ate moieties.24 Targeting the guanidino group for covalent modification is, however, not easily feasible, and there is only a relatively small number of reagents that is employed for this task. Typically, R-dicarbonyl compounds such as 2,3-butanedione and 1,2-cyclohexanedione are used.25-26 Zenobi’s group has also shown that arginine residues exposed on the surface of proteins can be probed by noncovalent attachment of appropriate selectors such as, for example, arylsulfonic acids.27-29 Recently, we have presented an alternative covalent labeling technique that specifically targets arginine residues in peptides and proteins.30,31 This tagging reaction is also based on the modification with R-dicarbonyls but employs an additional cyclization step with an arylboronic acidstypically phenylboronic acids and allows the direct monitoring of Arg tagging by ESI-MS at basic pH. In the present work, we have applied the concept to the investigation of the reactivity of arginine residues in model proteins. In contrast to largely unstructured small peptides, the reactivity of arginines in these proteins should be related to their structural features such as the above-mentioned solvent accessibilities. The data presented herein confirm this assumption and demonstrate that tagging of the guanidino groups in proteins can differentiate between more and less exposed arginine residues. EXPERIMENTAL SECTION Materials. All proteins used in this study were obtained from Sigma and of the highest purity available. Proteins were generally used without further purification, with the exception of ribonuclease A, which was desalted prior to use, using Amicon UltrafreeMC devices (5-kDa cutoff, Millipore, Bedford, MA) according to the manufacturer’s protocol. Phenylboronic acid (PBA) was purchased from Aldrich, while all other chemicals were from Fluka. Ultrapure water, also from Fluka, was used throughout. For hydrogen/deuterium exchange experiments, glass-distilled D2O from Aldrich was used. Reduction and Alkylation of Ribonuclease A. A 1-mg sample of RNAse A was dissolved in 250 µL of water containing (18) Hubale´k, F.; Pohl, J.; Edmondson, D. E. J. Biol. Chem. 2003, 278, 2861228618. (19) Kim, Y. J.; Pannell, L. K.; Sackett, D. L. Anal. Biochem. 2004, 332, 376383. (20) Sharp, J. S.; Becker, J. M.; Hettich, R. L. Anal. Biochem. 2003, 313, 216225. (21) Sharp, J. S.; Becker, J. M.; Hettich, R. L. Anal. Chem. 2004, 76, 672-683. (22) Xu, G. Z.; Takamoto, K.; Chance, M. R. Anal. Chem. 2003, 75, 6995-7007. (23) Xu, G. Z.; Chance, M. R. Anal. Chem. 2004, 76, 1213-1221. (24) Schug, K.; Lindner, W. Chem. Rev., 2005, 105, 67-113. (25) Riordan, J. F.; McElvany, K. D.; Borders, C. L., Jr. Science 1977, 195, 884886. (26) Riordan, J. F. Mol. Cell. Biochem. 1979, 26, 71-92. (27) Friess, S. D.; Zenobi, R. J. Am. Soc. Mass Spectrom. 2001, 12, 810-818. (28) Friess, S. D.; Daniel, J. M.; Hartmann, R.; Zenobi, R. Int. J. Mass Spectrom. 2002, 219, 269-281. (29) Friess, S. D.; Daniel, J. M.; Zenobi, R. Phys. Chem. Chem. Phys. 2004, 6, 2664-2675. (30) Leitner, A.; Lindner, W. J. Mass Spectrom. 2003, 38, 891-899. (31) Leitner, A.; Lindner, W. Anal. Chim. Acta, 2005, 528, 165-173.

4482

Analytical Chemistry, Vol. 77, No. 14, July 15, 2005

2 mg of tris(2-carboxyethyl)phosphine hydrochloride (TCEP, ∼100 molar equiv). After reduction for 60 min at room temperature, the sample was purified using self-packed MicroSpin columns packed with Sephadex G-10 material (Amersham Biosciences, Uppsala, Sweden) and diluted as appropriate. Samples were freshly prepared before the experiments to avoid reoxidation that could lead to the formation of nonnatural disulfide bonds. Labeling of Arginine Residues in Proteins. Unless otherwise noted, protein stock solutions were prepared in ultrapure water at a concentration of 1 mg mL-1. A 100-µL aliquot of this solution was diluted into 900 µL of a 10 mM aqueous ammonium acetate solution, adjusted to the appropriate pH (9.1 or 9.5) with ammonium hydroxide solution (25%). Immediately before the measurements, 20 µL of butanedione solution (100 mM in water) and 80 µL of phenylboronic acid solution (50 mM in water) were added and 500 µL of the mixed reaction solution was transferred to a gastight syringe. Data acquisition was started 1 min after the addition of PBA. During data acquisition, the solutions were kept at room temperature (23 ( 1 °C). Mass Spectrometry. Mass spectra were acquired on an Agilent 1100 MSD ion trap SL instrument (Agilent Technologies, Waldbronn, Germany) equipped with a standard electrospray source. Immediately after addition of the reagents, the protein solutions were infused via a syringe pump (kd Scientific, New Hope, PA) at a flow rate of 5 µL min-1. Instrumental parameters were optimized for ubiquitin as a model protein and were kept constant for all other experiments. Specifically, the following settings were used: nebulizer gas pressure 15 psi, drying gas flow 5 L min-1, drying gas temperature 225 °C, capillary voltage 3800 V, skimmer voltage 40 V, capillary exit voltage 180 V, and trap drive voltage 145 V. Spectra were acquired in “enhanced resolution” mode, continuously over a reaction time of 90 min and then after several hours to obtain the maximum labeling state. Data Analysis. From the arginine labeling experiments, averaged mass spectra were obtained from 2-min intervals ((1 min from the time shown in the figures). No smoothing or other spectral postprocessing was performed. Relative signal intensities were obtained (for each charge state individually) by dividing the absolute signal intensity of a labeling state by the sum of the intensities of all unlabeled and labeled forms. Comparable ionization efficiency for labeled and unlabeled forms was assumed, as the reduction in the proton affinity of labeled Arg residues is relatively small.30-31 The average number of Arg tags, ht, incorporated was calculated according to the following equation:

∑ t ‚ ri )/100

ht ) (

t

t

where t is the number of arginine tags and rit the relative signal intensity (in %) for the protein carrying t tags. The average charge state was calculated in an identical manner, with the number of tags replaced by the charge state. No shift in the charge-state distributions due to the incorporation of the tags was observed. Calculations of Solvent-Accessible Surface Areas. ASAs were calculated from Protein Data Bank (PDB) data files using GETAREA14 using the default probe radius of 1.4 Å and the “area per residue” output option. Assignment of residues as “solvent exposed” and “buried” was made according to the default levels

Figure 1. Reaction scheme for the modification of arginine residues with excess 2,3-butanedione (BD) and phenylboronic acid (PBA).

of GETAREA (>50 and 8.5), makes it necessary to perform the ionization in the so-called “wrong way round” mode, meaning the formation of positively charged ions from alkaline solutions. This may not be considered optimal, especially in the case of proteins. In this context, Pan et al. systematically examined nanoelectrospray ionization of proteins in dependence of the solution pH39 and observed that sensitivity in the “wrong way round” mode is generally at least 1 order of magnitude lower than at low pH. This may be considered a limitation of our method, yet we have additional factors that strongly influence the sensitivity of our method: First, the number of charge states that are formed from ESI distributes the total intensity over multiple signals in the spectra. An example is apomyoglobin, for which charge states from +8 to around +20 (albeit with very different signal intensities) were observed. Second, arginine-rich proteins are present in a number of different labeling states at any time point during the experiment. This was the case for lysozyme, containing eleven arginines: Up to eight labeling states were observed simultaneously, distributed over three charge states (+7-+9). Still, even for proteins such as the two mentioned, sensitivity was good enough to obtain accurate results at the concentration used (∼100 µg mL-1 corresponding to roughly 10 µM) and the given flow rate (5 µL min-1). For most of the proteins in this study, it would easily be possible to reduce the concentration by 1 order of magnitude, although this was not studied in detail. In particular, instrumental conditions can be optimized for a particular protein, which was not done in this study to avoid potential unfolding in the gas phase, especially for apomyoglobin or reduced RNAse. (Such unfolding would not have allowed us to study possible charge-state dependencies of the tagging reaction.) Therefore, the total amount of sample required could be as low as several micrograms compared to 50 µg for one experiment with our setup. In addition, sensitivity can be further increased by using nanoelectrospray ionization, which would also have the additional benefit that nanoESI is less

(37) Hoerner, J. K.; Xiao, H.; Dobo, A.; Kaltashov, I. A. J. Am. Chem. Soc. 2004, 126, 7709-7717. (38) Kaltashov, I. A. Int. J. Mass Spectrom. 2005, 240, 249-259.

(39) Pan, P.; Gunawardena, H. P.; Xia, Y.; McLuckey, S. A. Anal. Chem. 2004, 76, 1165-1174.

Analytical Chemistry, Vol. 77, No. 14, July 15, 2005

4487

susceptible to suppression effects.40-42 Higher sensitivity might also be achieved by a more thorough purification (desalting) of the proteins or the PBA reagent. Another important aspect to be considered in “wrong way round” ESI is the influence on the charge-state distribution observed. Generally, a slight shift to lower charge states is observed when ESI-MS spectra are obtained from high-pH solutions compared to acidic pH. Other factors that influence the CSD include instrumental parameters and the salt content of the electrosprayed solution.43-46 Therefore, both the high pH and the presence of excess reagents could lead to changes in the CSDs in our case. For this reason, we compared the charge-state distributions for all model peptides for neutral pH (10 mM ammonium acetate, pH ∼6.7) and alkaline pH (10 mM ammonium acetate adjusted to pH 9.5). The results are given in Table 2. It can be observed that there are only minor differences, typically 0.2 charges or less, that are within the range of experimental uncertainty. For apomyoglobin, the difference is larger (+1.3 charges) and may be caused by small conformational differences. However, the presence of more than 10 different charge states at a time may also lead to more variation between mass spectra. The presence of the reagents did not further influence the CSD and did not lead to noticeable decrease in the signal-to-noise ratio. No adverse effect of the slightly alkaline milieu on the structural integrity of the proteins examined was observed. However, larger proteins or protein complexes are beyond the (40) Juraschek, R.; Du ¨ lcks, T.; Karas, M. J. Am. Soc. Mass Spectrom. 1999, 10, 300-308. (41) Schmidt, A.; Karas, M.; Du ¨ lcks, T. J. Am. Soc. Mass Spectrom. 2003, 14, 492-500. (42) Smith, R. D.; Shen, Y.; Tang, K. Acc. Chem. Res. 2004, 37, 269-278. (43) Gumerov, D. R.; Dobo, A.; Kaltashov, I. A. Eur. J. Mass Spectrom. 2002, 8, 123-129. (44) Verkerk, U. H.; Peschke, M.; Kebarle, P. J. Mass Spectrom. 2003, 38, 618631. (45) Pan, P.; McLuckey, S. Anal. Chem. 2003, 75, 5468-5474. (46) Samalikova, M.; Matecko, I.; Mu ¨ ller, N.; Grandori, R. Anal. Bioanal. Chem. 2004, 378, 1112-1123.

4488

Analytical Chemistry, Vol. 77, No. 14, July 15, 2005

mass range of our instrument (up to m/z 2200) and were not examined at this stage. CONCLUSION We have shown that our novel covalent arginine tagging strategy can be successfully applied to the functional (solvent accessibility) probing of arginine residues in native proteins. The accessibility of the guanidino functionalities correlated well with solvent accessibilities obtained from 3D structural data, and the experimental setup provided highly reproducible results. Most importantly, significant structural changesse.g., the removal of the heme in the case of myoglobin and the reduction of disulfide bonds in ribonuclease Aswere directly reflected in different reactivities of the arginine residues. In the future, we plan to expand our studies to other proteins and we think that our method complements other MS-based assays for the investigation of protein structure, in particular with regard to the biological importance of arginine. ACKNOWLEDGMENT Financial support for the Christian Doppler Laboratory for Molecular Recognition Materials by the Chrisian Doppler Society (Vienna, Austria), Merck KgaA (Darmstadt, Germany), AstraZeneca (Mo¨lndal, Sweden), and piChem (Graz, Austria) is gratefully acknowledged. Funds were also provided by a grant from the Austrian Science Fund, project P15482. SUPPORTING INFORMATION AVAILABLE A detailed discussion on the determination of the frequency of Xxx-Arg-Xxx motifs in proteins. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review February 4, 2005. Accepted May 3, 2005. AC050217H