Distinguishing between Phosphorylated and Nonphosphorylated

Texas 77843, and Chemistry and Drug Metabolism, Intramural Research Program, NIDA, NIH,. Baltimore, Maryland 21224. Received March 4, 2002. Abstract: ...
6 downloads 0 Views 186KB Size
Distinguishing between Phosphorylated and Nonphosphorylated Peptides with Ion Mobility-Mass Spectrometry Brandon T. Ruotolo,† Guido F. Verbeck IV,† Lisa M. Thomson,‡ Amina S. Woods,§ Kent J. Gillig,† and David H. Russell*,† Laboratory for Biological Mass Spectrometry, Department of Chemistry, Texas A&M University, College Station, Texas 77843, Laboratory for Molecular Simulation, Texas A&M University, College Station, Texas 77843, and Chemistry and Drug Metabolism, Intramural Research Program, NIDA, NIH, Baltimore, Maryland 21224 Received March 4, 2002

Abstract: Mass spectrometry has become an indispensable tool in identifying post-translationally modified proteins, but multiple peptide mass-mapping/peptide-sequencing experiments are required to answer questions involving the site and type of modification present. Here, we apply ion mobility-mass spectrometry (IM-MS), a high-throughput analysis method having high selectivity and sensitivity, to the challenge of identifying phosphorylated peptides. Ion mobility separation is based on the collision cross-section of the ion. Phosphorylation can result in a conformational change in gas-phase peptide ions, which can be detected by IM. To demonstrate this point, a peptide mixture containing a variety of peptide sequences is examined with IM-MS and molecular dynamics calculations. During the course of these studies, two classes of phosphopeptide were identified: (i) phosphorylated peptide ions that have conformers that differ from the nonphosphorylated ion and (ii) phosphorylated peptide ions that have conformations that are very similar to the nonphosphorylated peptide. The utility of IM-MS peptide mass mapping for identifying both types of phosphorylated peptides is discussed. Keywords: phosphopeptides • high-throughput • MALDI • conformation

quencing to determine the site(s) of phosphorylation.4 In principle, the separation step for modified peptides can be circumvented in favor of automated approaches such as “datadependant” MS/MS sequencing.5 Although these techniques are more sensitive to the presence of phosphorylated sequences,6 they are inherently inefficient both in terms of the chromatography utilized and the time required for extensive interrogation of multiple ion signals for potential posttranslational modification. Several laboratories have demonstrated the utility of ion mobility spectrometry (IMS) for studies of protein conformation7,8 and are developing IM-MS techniques for protein identification and characterization.9,10 This work investigates the utility of matrix-assisted laser desorption-ionization (MALDI)-ion mobility (IM)-time-offlight (TOF) mass spectrometry (MS)11 for distinguishing between phosphorylated and nonphosphorylated peptides. For example, a mass-mobility map (plot of IM drift time versus m/z) for a mixture of peptides (MW range of 500-3000) is approximately linear and varies with charge state;10 however, some peptide ions that have well-defined secondary/tertiary structures have been found to deviate from the linear relationship.12 Charge-carrying residues can effect the shapes of gasphase peptide ions because polar functional groups, e.g., amide carbonyl group, C-terminal carboxylic acid group, and polar side chains, “solvate” the charge site. This investigation examines the effects of phosphorylation on the conformation and mobilities of gas-phase peptide ions by employing molecular dynamics (MD) to assign gas-phase conformations to the observed ion signals.

Introduction

Experimental Section

The detection of phosphopeptides presents an analytical challenge for current mass spectrometry techniques that are designed to meet the needs of proteomics. Although phosphorylation is one of the most common types of post-translation modifications found in bioactive peptides and proteins,1 the concentration of modified proteins is often low, making the detection of modified analytes a significant challenge.2 Current methodologies rely upon labor-intensive, time-consuming separation techniques that isolate specific proteins for further analysis, for example, ion-metal affinity chromatography in the case of phosphorylated proteins.3 The separated proteins are then subjected to proteolytic digestion and MS/MS se-

Spectra were obtained using a MALDI-IM-TOF MS instrument that has been described elsewhere.11 Ions are formed at the operating pressure of the drift cell (1 Torr He) using a nitrogen laser (337 nm) and standard MALDI sample preparation methods. Ions travel through a periodic focusing drift cell13 (millisecond transit times) and are separated by mobility, with a typical resolution of 50 (t/∆t). After mobility separation, the ions are extracted and mass analyzed with a small linear timeof-flight mass spectrometer (m/z resolution of 200). Linear regression analysis was performed with Curve Expert (Microsoft, Seattle WA), and deviations noted in the text are reported in percent difference relative to the drift time predicted by a strict linear model. Model peptides were obtained from AnaSpec, Inc. (San Diego, CA.) and prepared, from 100 pmol/µL stock solutions, by diluting the peptide mixture with



Laboratory for Biological Mass Spectrometry, Texas A&M University. Laboratory for Molecular Simulation, Texas A&M University. § Intramural Research Program, NIDA, NIH. ‡

10.1021/pr025516r CCC: $22.00

 2002 American Chemical Society

Journal of Proteome Research 2002, 1, 303-306

303

Published on Web 05/16/2002

communications

Figure 1. Mass-mobility plot of a tryptic digest of β-casein (bovine). The trend line (slope ∼0.44) has been added to guide the eye. Identified on the plot are the phosphorylated sequence (m/z 2063, FQpSEEQQQTEDELQDK) and several sequences that exhibit broadening (observed by Clemmer et al. in ref 17): m/z 1383, LLYQEPVLGPVR and m/z 2187, DMPIQAFLLYQEPVLGPVR.

R-cyano-4-hydroxycinnamic acid to a matrix-to-analyte ratio of ca. 1000:1. Enzymatic digestion was performed using sequencing-grade modified trypsin (Promega, Madison, WI) as described previously.14 Molecular dynamics (MD) calculations were performed using simulated annealing15 with the CFF 1.01 class II force field (Accelerys, Inc., San Diego, CA). In all modeling studies peptides were protonated on the most basic side chains (arginine or lysine) of a given peptide. Theoretical collision cross-section calculations were performed according to methods previously described by Jarrold.16

Results and Discussion The plot shown in Figure 1 is a mass-mobility map of a tryptic digest of bovine β-casein, and these data are representative of mass-mobility plots for tryptic digests. In a previous paper, we discussed the mass-mobility trends observed for a larger dataset of peptide ions, the results of which show that the majority of ion signals exhibit high correlation (average r ) 0.991) to a line having a slope of ∼0.44 (shown in Figure 1).17 Several features are apparent upon first inspection of this figure. For example, several ion signals, such as m/z 1383 (A) and m/z 2187 (C), are broadened in drift time relative to other peaks in the spectrum. Peptides A and C both contain multiple proline residues, and Clemmer and co-workers have proposed that this type of broadening is due to the presence of multiple cis or trans proline isomers of the gas-phase peptides.18 Another prominent feature of the data is the ion signal at m/z 2063; which corresponds to FQpSEEQQQTEDELQDK. Note that the mass-mobility value for this phosphopeptide deviates from the expected value by approximately 10%. This observation is consistent with data that we have obtained on a number of phosphoprotein digests, in which phosphopeptides often exhibit negative deviations from the trend exhibited for nonmodified peptides. That is, the mobility observed for phosphopeptides is higher (corresponds to a smaller collision crosssection) than would be predicted based strictly on the massto-charge ratio for the peptide ion,19 and we propose that this 304

Journal of Proteome Research • Vol. 1, No. 4, 2002

deviation is a result of changes in the conformation (a more compact secondary/tertiary structure) of the gas-phase phosphopeptide relative to its nonphosphorylated analogue. In this paper, we report mass-mobility data for several model peptides, see Figure 2, to clarify the trends observed for phosphopeptides in tryptic digests. Figure 2 contains a massmobility plot of a series of eight model signaling peptides20 observed as [M + H]+ ions: (1) RRApSPVA (835.9 amu), (2) KRpTIRR (909.0 amu), (3) DRVYIHPF (angiotensin II) (1046.2 amu), (4) DRVpYIHPF (1126.2 amu), (5) VRKRTLRRL (1197.5 amu), (6) RRREEETEEE (1362.4 amu), (7) RRREEEpSEEEAA (1570.5 amu), and (8) RRREEEpTEEEAA (1584.5 amu). Figure 2 also contains signals for two in-source decay products ([1 PO3H]+ and 7 and 8 - PO3H]+). The peptides RRREEEpTEEEAA (m/z 1585.4) and RRREEEpSEEEAA (m/z 1571.5) do not appear to be resolved in Figure 2, but two distinct maxima are observed in the mass spectrum, suggesting that the peptide ions have very similar mobilities. On the other hand, the ion signals corresponding to the fragment ions formed by loss of phosphate from RRREEEpTEEEAA and RRREEEpSEEEAA exhibit a single broad maximum that cannot be unambiguously assigned to 7 or 8. The mass-mobility results for phosphorylated peptides shown in Figure 2 fit into two categories: (i) ions that are shifted in m/z relative to their nonphosphorylated analogues and (ii) ions that exhibit both a shift in m/z and mobility. Also shown in Figure 1 is the average trend line observed for peptide ions (shown in black). The average trend line, corresponding to a slope of 0.44 (( 1.5% total drift time for the majority of peptide ions), was obtained by pooling data from 13 separate protein digest datasets along with the data shown in Figure 2.17 The mass-mobility values for four phosphopeptides in this mixture deviate from the linear relationship by varying degrees. Those in the first class of phosphopeptides (for example peptides 1 and 4) exhibit moderate negative deviations (approximately 5% for peptides 1 and 4), whereas the nonphosphorylated analogues display higher correlation to the peptide trend line.

communications

Figure 2. Mass-mobility plot of eight model peptides: (1) RRApSPVA, (2) KRpTIRR, (3) DRVYIHPF, (4) DRVpYIHPF, (5) VRKRTLRRL, (6) RRREEETEEE, (7) RRREEEpSEEEAA, (8) RRREEEpTEEEAA. The superimposed trend line has a slope of ∼0.44, indicative of the average trend for peptide [M + H]+ ions (from ref 21) and a trend line for RRREEEXEEE motif peptides with the slope ∼0.35.

Peptide 2 may also fall into the first class of phosphopeptides, as it deviates from the expected mobility value by 2%; however, the in-source decay product ion signal corresponding to the nonphosphorylated analogue of the sequence was not observed. Those peptides placed in the second class exhibit much larger deviations from other peptides in the mass-mobility plot (approaching 10% in most cases), but exhibit a high correlation coefficient when fit to a separate linear trend (as shown for peptides 6, 7, and 8 in Figure 2). MD calculations were utilized to obtain low energy structures for peptides 3, 4, 8, and 8 - PO3H.15 Figure 3 contains the lowest energy structures for RRREEEpTEEEAA (8) and RRREEETEEEAA (8 - PO3H).21 Note the hairpin orientation (indicated by a dashed box in Figure 3) in the nonphosphorylated sequence, consisting of a sharp turn in residues 4-7. Presumably, this conformation is preferred due to extensive hydrogen bonding within the molecule. The hairpin turn is not appreciably disrupted in the phosphorylated version of this peptide, as shown in Figure 3 (bottom). The mobility data for 8 and 7 and 8 - PO3H are consistent with the structures shown in Figure 2. That is, RRREEETEEEAA exhibits both a mass and a mobility shift upon phosphorylation, indicating that the some portion of the peptide conformation is conserved. Theoretical and experimental collision cross-sections both indicate that the two molecules are unresolved in the mobility dimension (matching the experiment), and the calculated collision crosssection matched the experimental data within the error of the theoretical methods ((2% total error).22 The second set of low energy structures (Figure 4) were obtained for angiotensin II (DRVYIHPF) 3 and its phosphorylated analogue (DRVpYIHPF) 4. The gas-phase structure of angiotensin II appears to be less constrained and elongated, i.e., exhibits more random coil character. The experimentally measured mobility for DRVpYIHPF (phosphorylated angiotensin II) suggests that the ion is more compact; e.g., the [M + H]+ ion is shifted on the mass axis but appears at roughly the same drift time as 3 (centroids differ by less than 10 µs). MD calculations indicate that the most probable structure for 4 is

Figure 3. Lowest energy structures from MD calculations for RRREEETEEEAA (top) and RRREEEpTEEEAA (bottom). The N and C termini are labeled, and the purple sphere denotes the position of the phosphate group. The conserved portion of the structure (noted in text) is indicated with a dashed box.

one where the protonated arginine side chain solvates the phosphate group, thus producing a small change in the overall Journal of Proteome Research • Vol. 1, No. 4, 2002 305

communications This suggests that the two class model presented here does not account for the detailed behavior of every phosphopeptide, and future experiments will be directed toward further evaluating the gas-phase conformations of phosphopeptides in the context of complex mixtures. In all cases, it is apparent that the effect of phosphorylation on gas-phase peptide ion conformation is dependent upon the overall amino acid sequence of the peptide. However, there is enough information present in Figures 1 and 2 to suggest that in most cases, where peptides exhibit a mass shift but a negligible shift in mobility, phosphopeptides are discernible from their nonphosphorylated analogues through mass-mobility analysis alone. Conceivably, the fact that the majority of phosphopeptides exhibit a negative deviation relative to the peptide population as a whole can be used as a screening parameter, allowing investigators to concentrate data dependent sequencing and characterization efforts on a smaller population of peptides (