A Modified Database Search Strategy Leads to Improved Identification

Technical Note pubs.acs.org/jpr

A Modified Database Search Strategy Leads to Improved Identification of in Vitro Brominated Peptides Spiked into a Complex Proteomic Sample Huiling Liu,†,‡ Cheryl F. Lichti,†,‡ Barsam Mirfattah,† Jennifer Frahm,§ and Carol L. Nilsson*,† †

University of Texas Medical Branch, Department of Pharmacology and Toxicology, 301 University Boulevard, Galveston, Texas 77555-0617, United States § Thermo Fisher Scientific, 1400 Northpoint Parkway, West Palm Beach, Florida 33407, United States S Supporting Information *

ABSTRACT: Inflammation leads to activation of immune cells, resulting in production of hypobromous acid. Few investigations have been performed on protein bromination on a proteomic scale, even though bromination is a relatively abundant protein modification in endogenously brominated proteomes. Such studies have been hampered by the lack of an optimized database search strategy. In order to address this issue, we performed nanoLC−MS/MS analysis of an in vitro generated, trypsin-digested brominated human serum albumin standard, spiked into a complex trypsin-digested proteomic background, in an LTQ-Orbitrap instrument. We found that brominated peptides spiked in at a 1−10% ratio (mass:mass) were easily identified by manual inspection when higher-energy collisional dissociation (HCD) and collision induced dissociation (CID) were employed as the dissociation mode; however, confident assignment of brominated peptides from protein database searches required a novel approach. By addition of a custom modification, corresponding to the substitution of a single bromine with 81Br rather than 79Br for dibromotyrosine (79Br81BrY), the number of validated assignments for peptides containing dibromotyrosine increased significantly when analyzing both high resolution and low resolution MS/MS data. This new approach will facilitate the identification of proteins derived from endogenously brominated proteomes, providing further knowledge about the role of protein bromination in various pathological states. KEYWORDS: protein bromination, mass spectrometry, HCD, CID, post-translational modification, proteomics, bioinformatics

■

Information).15 The aromatic side chains of Phe, His, and Trp can also be monobrominated. Bromine has an unusual isotopic abundance (79Br, 50.69%, 81 Br, 49.31%)16 compared to the elements most often encountered in proteins, C, H, O, N, P, and S. Brominated peptide mass spectra thus display a characteristic isotopic pattern which simplifies identification of this peptide modification through visual examination of MS data.17 The bromine-specific isotopic pattern can also be observed in the b and y fragment ions18 of HCD spectra for brominated peptides. The isotope pattern is markedly different from the pattern produced by nonbrominated peptides (Figure 2 in the Supporting Information). Brominated peptides may elude identification when conventional database searching strategies are used because peptide assignments are universally made by comparisons to “average” peptides, not halogenated ones. We optimized the data analysis strategy for identification of

INTRODUCTION Protein halogenation by activated eosinophils1 and other granulocytic cells may occur in inflammation in skin following photoaging,2,3 in asthma,4,5 parasitic infections, and certain gliomas.6 Inflammation thus results in endogenous production of hypobromous acid and brominated biomolecules, including nucleotides and proteins. Correlations between the production of brominated nucleotides and mutagenesis has been well studied,7−11 but few investigations of protein bromination have been performed on a proteomic scale. The characterization of endogenously brominated proteomes is essential to define alterations in protein−protein and protein−DNA interactions and induction of autoimmunity.12 The reaction of H2O2 with bromide ions in vivo is catalyzed by eosinophil peroxidase or myeloperoxidase (MPO) to produce hypobromous acid (HOBr).10,13,14 HOBr, released by activated macrophages, monocytes, and eosinophils, reacts with certain amino acid side chains more readily than peptide backbone amide groups. HOBr undergoes electrophilic attack on the aromatic side chain of Tyr to form either mono- or dibrominated side chains (Figure 1 in the Supporting © 2013 American Chemical Society

Received: May 20, 2013 Published: July 31, 2013 4248

dx.doi.org/10.1021/pr400472c | J. Proteome Res. 2013, 12, 4248−4254

Journal of Proteome Research

Technical Note

times were identical to those described above. Bound peptides were eluted by gradient elution (250 μL/min) as follows: 5− 35% (v/v) ACN, 0.1% (v/v) formic acid for 35 min (total LC run time of 60 min). All LC−MS/MS data were acquired using XCalibur, version 2.0.7 (Thermo Fisher Scientific). For low resolution data driven analyses, the survey scans (m/z 350−1600) (MS1) were acquired in the Orbitrap at 120,000 resolution (at m/z 400) in profile mode, followed by ten MS/MS events in the linear ion trap in centroid mode with collision activation in the ion trap. The automatic gain control target for the LTQ was 1 × 104. Monoisotopic precursor selection was enabled, and dynamic exclusion was used to remove selected precursor ions (±10 ppm) for 15 s after MS/MS acquisitions. For the isotopic data dependent acquisition, the mass difference and expected ratio between the monoisotopic and A + 2 peaks for the 2+ and 3+ charge states of the brominated peptides were entered [monobrominated, 0.9990 Da, 1 for the 2+, and 0.6660, 1 for the 3+; dibrominated, 0.9990 Da, 0.5 for the 2+, and 0.6660 Da, 0.5 for the 3+; tribrominated, 0.9990 Da, 0.33 for 2+, and 0.6660 Da, 0.33 for 3+; and tetrabrominated, 0.9990 Da, 0.25 for 2+, and 0.6660 Da, 0.25 for 3+]. The match tolerance was set to 20%. An isotopic data dependent method was also developed with “use m/z values as masses” enabled. In this method, the only difference was the uncharged mass difference was entered in the isotopic data dependence table. When isotopic data dependence is enabled, the instrument will trigger MS2 if the monoisotopic and A + 2 peaks match the specified mass difference and ratio within the specified tolerance. All other method settings were the same as the high resolution data-driven analyses. For high resolution data-driven analyses (DDA), the survey scans (m/z 350−1600) (MS) were acquired in the Orbitrap at 60,000 resolution (at m/z = 400) in profile mode, followed by five HCD fragmentation MS/MS spectra, acquired at 15,000 resolution in the Orbitrap in centroid mode. The automatic gain control targets for the Orbitrap were 1 × 106 for the MS scans and 1 × 105 for MS/MS scans. A full description of MS2 parameters may be found in the Supporting Information.

brominated peptide spectra to enable their study on a proteomic scale.

■

METHODS

Preparation of an in Vitro Brominated Albumin Standard

Three 100 μL volumes of human serum albumin (SigmaAldrich) containing 200 pmol each were reacted with fresh hypobromous acid (2.0 mM). HOBr was prepared immediately prior to the reaction with HSA and was formed from the reaction of HOCl (40 mM in H2O, pH 13) with NaBr (45 mM in H2O). After 5, 15, and 25 min time points the reacted samples were buffer-exchanged using 0.5 mL 10K Centrifugal Filters (Amicon Ultra) with 50 mM NH4HCO3 buffer according to the manufacturer’s instructions. Preparation of Cell Lysates and Protein Digestion

U373 glioblastoma cell pellets (2 × 106) were lysed with RIPA buffer (25 mM Tris·HCl pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS), containing protease and phosphatase inhibitors and deoxyribonuclease (all products Thermo Fisher Scientific, Rockford, IL). Volumes containing 100 μg of total protein were aliquoted, and brominated albumin was spiked into the protein mixture (0.1%, 1%, and 10% by weight). Proteins were reduced, alkylated, and then precipitated in four volumes (440 μL) of ice cold acetone for 2 h at −20 °C. Pellets were air-dried and resuspended in 12.5 μL of 8 M urea. Trypsin (10 μg in 87.5 μL of TEAB buffer) was added, and the samples were incubated for 24 h at 37 °C. For a full description of these procedures, please see the Supporting Information. Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry

HSA bromination was verified by screening of the HOBrreacted protein by MALDI-TOF/TOF mass spectrometry (UltrafleXtreme, Bruker Daltonics) equipped with a 2 kHz solid state laser (Bruker Smartbeam-II). Saturated sinapinic acid matrix (Sigma-Aldrich) was prepared in 0.1% TFA:ACN (7:3 v/v). One microliter of sample was mixed with 1 μL of matrix solution; 0.5 μL was applied on a stainless steel target by the dried-droplet method.19 Samples were analyzed by MALDITOF/TOF MS in linear positive mode. A full description of the MALDI-MS method is found in the Supporting Information.

Data Analysis

Raw data files were processed and analyzed using Proteome Discoverer 1.3 (Thermo Scientific), which combines SEQUEST21 and Mascot 2.3 (Matrix Science) search engines, against the Uniprot-Human database (Feb 25, 2012 version, 81,213 protein sequences) using trypsin as the enzyme with two missed cleavages allowed. Carbamidomethylation of cysteine (Cys-CAM) was set as fixed modification, and methionine oxidation (Met-ox), monobromination (YHWF), and dibromination (Y) were set as variable modifications. Precursor ion tolerance was set to 10 ppm, and fragment mass tolerance was set to 0.8 Da for low resolution and 0.1 Da for high resolution spectra. We filtered for high-confidence peptides (defined according to the Percolator algorithm,23 with a q-value threshold of 0.01) and for proteins with at least two unique peptides per protein. For PEAKS22,23 (version 6, Bioinformatics Solutions Inc., Waterloo, ON, Canada) searches, data refinement was performed with no merged scans, with precursor charge correction (charge states 1−8), and with no filtering. De novo sequencing was performed with 20 ppm parent tolerance and 0.1 Da fragment tolerance using carbamidomethyl Cys as a fixed modification, and variable modifications of monobromi-

Nano-LC−MS/MS Analysis

Chromatographic separation and mass spectrometric analysis was performed with a nano-LC chromatography system (EasynLC 1000, Thermo Scientific), coupled online to a hybrid linear ion trap-Orbitrap mass spectrometer (Orbitrap Elite, Thermo Scientific) through a Nano-Flex II nanospray ion source (Thermo Scientific).20 Mobile phases were 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (ACN, B). After equilibrating the column in 95% solvent A and 5% solvent B, the samples (5 μL in 5% v/v ACN/0.1% (v/v) formic acid in water, corresponding to 1 pmol of HSA or 1 μg of Br-HSA spiked into U373 protein digest) were injected (5 μL/min, 4 min) onto a trap column (C18, 100 μm ID × 2 cm) and subsequently eluted (250 nL/min) by gradient elution onto a C18 column (10 cm × 75 μm ID,15 μm tip, ProteoPep II, 5 μm, 300 Å, New Objective). For Br-HSA spiked into U373, the gradient was as follows: isocratic at 5% B, 0−5 min; 5% to 35% B, 5−75 min; 35% to 95% B, 75−80 min; and isocratic at 95% B, 80−90 min. Total run time, including column equilibration, sample loading, and analysis, was 104 min. For Br-HSA, sample loading and column equilibration 4249



Technical Note

the Supporting Information and Tables 3 and 4 in the Supporting Information. The isotopic data dependent trigger in Xcalibur will trigger MS2 if the monoisotopic and A + 2 peaks match the specified mass difference and relative ratio within the specified tolerance. The setting was designed for singly charged small molecule analysis. With peptides the multiple charge states and brominations increase the heterogeneity, which adds to the number of possible mass differences and expected isotopic ratios. When we simulated the isotopic distribution for several peptides, we found that increasing the number of Br modifications and peptide charge reduced the characteristic 1:1 relative abundance of the monoisotopic and A + 2 isotopes, which the instrument uses to trigger for MS2. The heterogeneity would explain the large number of nonspecific triggers observed. Also, we entered information for mono- to tetrabrominated peptides for the 2+ and 3+ charge states based on preliminary data. In our samples we detected several 4+ peptides, so the isotopic trigger would have missed many of the 4+ peptides and, therefore, would not detect as many brominated peptides as a nonspecific data dependent method. Further, originally we selected a match tolerance of 20%, which would allow some deviation in the isotopic ratios between the monoisotopic and A + 2 peaks. Both increasing and decreasing the match tolerance criteria did not increase the number of brominated peptides identified compared to the standard DDA. Therefore, this strategy was abandoned for the current study. Figures 1 and 2 show representative examples of singly (Figure 1) and doubly (Figure 2) brominated HSA peptides identified in our initial database searches. As with MS spectra, the b and y fragment ions for brominated peptides in MS/MS spectra display a characteristic isotope pattern that can be useful in determining the site of bromination. For Figure 1B, one notes the difference in abundance between the modification site-determining y6 and y7 ions (see insets); the approximate 1:1 intensity of the monoisotopic (A) and A + 2 peaks for y7 supports the presence of a single bromine atom. The MS/MS of a dibromotyrosine-containing peptide (Figure 2B) illustrates the characteristic 1:2:1 abundance pattern for the A, A + 2, and A + 4 ions in the y ion series starting with the site-determining y4 ion, particularly noticeable for y7 (see insets). This pattern is helpful both in confirming the presence of dibromotyrosine and in localizing the site of modification when multiple tyrosine residues are present within the same peptide.

nation (FHWY), dibromination (Y), oxidation (M), bromoisotope (FHWY), and dibromo-isotope (Y), and trypsin as the enzyme (1 missed cleavage). A maximum of 3 variable modifications were allowed per peptide, and the top 5 peptides were reported. The resulting peptide sequences were searched against the same Uniprot-Human database. The initial search was performed with the previously specified modifications, and a final search for unexpected modifications was performed with the entire Unimod database. Finally, homology searching was performed using the SPIDER algorithm24 to identify peptides resulting from nonspecific cleavages or amino acid substitutions. Theoretical isotopic distributions for identified brominated peptides were created by use of IsoPro software, and each correct assignment was manually verified by examination of .raw files (XCalibur) against theoretical fragment lists generated in MS Product (prospector.ucsf.edu). MS/MS fragment ion tables for each verified spectrum can be found in the Supporting Information, along with theoretical and observed MS1 and annotated MS/MS spectra. The experiments described above were performed twice, with comparable results.

■

RESULTS AND DISCUSSION

Development of a Brominated HSA Standard, Spike-In Method

To investigate the sensitivity of detection of brominated peptides derived from in vitro-brominated HSA,4 we first generated the protein standard. The reaction was monitored by MALDI-TOF/TOF MS at various time points (5 min, 15 min, and 25 min, Figure 3 in the Supporting Information). Because full bromination was achieved in 5 min, we selected a 5 min reaction time for further study. The brominated HSA was digested with trypsin, and the resulting proteolytic mixture was analyzed by nano-LC−MS/MS. We identified brominated Tyr and His-containing peptides by this approach.10 No brominated Trp or Phe-containing peptides were detected. A list of identified brominated peptides can be found in Table 1 in the Supporting Information. LC−MS/MS and Data Analysis

To test our ability to detect brominated peptides in a complex proteomic sample, we spiked 0.1%, 1%, and 10% brominated HSA into U373 cell lines, digested the mixtures with trypsin, and analyzed each sample by nano-LC−MS/MS. After database searching, we identified approximately 1000 proteins derived from U373 (Table 2 in the Supporting Information). Within this complex sample, we were able to identify Br-HSA peptides in the 10% and 1% spike-in samples. Spike-in levels lower than 1% (150 fmol) did not yield any reliable assignments of brominated peptides, although HSA was identified. In the 10% spike-in sample, we identified 26 brominated HSA peptides by HCD and 14 additional peptides by CID, for a total of 40 peptides. These peptides represent 19 amino acid sequences with varying charge states, number of bromine atoms, and methionine oxidation status. Of these 40 peptides, nine peptides contained bromohistidine, nine contained monobromotyrosine, and 21 contained dibromotyrosine. The 1% spike-in sample gave four peptides by HCD and seven peptides by CID, with ten total peptide identifications. One monobromotryosine and two bromohistidine-containing peptides were identified; the remainder were dibromotyrosine-containing peptides. Complete lists of identified brominated peptides, their MS/MS spectra, and ion tables are available in Figure 4 in

New Data Analysis Strategy

For doubly brominated peptides with very low abundance, there are instances where the monoisotopic peak is either of such low abundance that it is not detected experimentally, or is not recognized by data analysis software. As a result, the parent mass for the corresponding MS/MS spectrum is incorrectly assigned. This can lead to erroneous or missed peptide assignments. This is also true for MS/MS spectra of peptides containing dibromotyrosine. In this case, isotope peaks for doubly brominated b and y fragment ions are present in a 1:2:1 ratio, with the 79Br81Br-containing ion (A + 2) being the most abundant. Like MS spectra, this can result in the absence of the true monoisotopic peakand often the 13C isotope peak as wellin MS/MS spectra of doubly brominated peptides, leading to unassigned spectra or incorrectly assigned peptide sequences. These related issues became clear when analyzing the initial de novo sequencing and search results from PEAKS. Very few 4250



Technical Note

Figure 1. A: Experimental (left) and theoretical (right) peptide isotopic distributions for [M + 2H]2+ ion of SLH$TLFGDK (monobrominated His). B: MS/MS spectrum of the brominated peptide acquired in a proteomic background. H$ = brominated His. Insets: The y6 ion shows a typical unmodified isotopic pattern while the mono- and diprotonated y7 ions display a typical monobrominated isotopic pattern. * indicates signals in the isotope cluster.

Figure 2. A: Experimental (left) and theoretical (right) peptide isotopic distributions for [M + 2H]2+ ion of RPcFSALEVDETY#VPK (dibrominated Tyr). B: MS/MS spectrum of the [M + 2H]2+ brominated peptide ion acquired in a proteomic background. Y# = dibrominated Tyr, c = carbamidomethylated Cys. Insets: The y3 ion shows a typical unmodified isotopic pattern while the y4 and y7 ions display a typical monobrominated isotopic pattern. * indicates signals in the isotope cluster.

brominated peptides were identified, far less than seen by Mascot and SEQUEST. In examining the results to determine the reason for the discrepancy, we noted that many HSA peptides containing bromotyrosine or dibromotyrosine residues, previously confirmed as being brominated, were incorrectly assigned as containing a phosphate group. The change in mass associated with phosphorylation (79.966331 Da, www.unimod.org) is approximately 2 Da higher than that for bromination (77.91051 Da, www.unimod.org), and the value for phosphorylation (79.9084) is close to the value that would be associated with substitution of H by 81Br (Δm = 79.966331 − 79.9084, 724 μDa). Closer examination of the corresponding parent ion for the incorrectly assigned peptides revealed that the A + 2 isotope peak had been incorrectly identified as the monoisotopic m/z by PEAKS in both the MS and MS/MS spectra, and similar misassignments were made in the MS/MS spectra. In order to resolve this problem, we added custom modifications to PEAKS, corresponding to replacement of H by 81Br. In the case of dibromotyrosine, the new custom modification was 79Br81BrY. This change dramatically increased the number of brominated peptides identified (see Table 2 in the Supporting Information and Figure 3). In the case of HCD acquisition for the 10% spike-in sample, over twice as many peptides (27) were identified by PEAKS as by SEQUEST or Mascot (12). For singly brominated peptides, PEAKS and SEQUEST/Mascot identified approximately the same number of peptides (13 and 9, respectively). The most significant

improvement was in the number of identifications for dibromotyrosine-containing peptides: 14 were identified by PEAKS, while only 3 were identified by SEQUEST and Mascot. A similar discrepancy was observed for the CID search results: PEAKS identified 18 monobromohistidine/tyrosine and 20 dibromotyrosine peptides, while SEQUEST and Mascot identified 16 monobromohistidine/tyrosine and 10 dibromotyrosine peptides. Based upon these results, we added a custom 79Br81BrY modification to both SEQUEST and Mascot and repeated the database searches. A significant improvement was seen in the number of identifications of peptides containing dibromotyrosine, shown in Figures 3C and 3F. (See Table 5 in the Supporting Information for a complete list of dibromotyrosinecontaining peptides identified using the new search strategy.) For HCD analysis of the 10% Br-HSA spike-in sample, 11/14 peptides previously identified only by PEAKS were added to the identifications in SEQUEST and Mascot. (Two of the 14 peptides identified by PEAKS were not included in the following comparison since they would not be identified based upon the search parameters used for Mascot and SEQUEST.) Of the remaining 12 peptides, all peptides were identified by both SEQUEST and Mascot. One peptide was unique to PEAKS. A similar improvement was seen in the number of assigned dibromotyrosine-containing peptides for the 10% spike-in samples analyzed by CID (Figures 3E and 3F). In this case, 4251



Technical Note

Figure 3. Venn diagrams illustrating identification of brominated peptides by PEAKS (orange circle), Mascot (blue striped circle), and SEQUEST (clear circle) for brominated HSA spiked into a complex unmodified background proteome at 10%. Diagrams indicate number of peptides as follows: A, all brominated peptides, HCD; B: dibromotyrosine-containing peptides before custom modification added to Mascot and SEQUEST; C: dibromotyrosine-containing peptides identified after addition of custom modification to Mascot and SEQUEST, HCD; D, all brominated peptides, CID; E: dibromotyrosine-containing peptides before custom modification added to Mascot and SEQUEST, CID; F: dibromotyrosine-containing peptides identified after addition of custom modification to Mascot and SEQUEST, CID.

with HCD spectra. In addition to the greater confidence in assignment due to higher mass accuracy, HCD spectra for highly charged peptides can be deconvoluted, greatly facilitating database searches and increasing the confidence for peptide assignments in search engines such as Mascot. It is often difficult to distinguish fragment charge states for CID spectra if the charge is greater than 1; this is not the case for high resolution HCD MS/MS spectra. Furthermore, the change in mass induced by 81Br bromination is very close to that of phosphorylation; the ability to distinguish between the two modifications benefits from high resolution data.25 Complex proteomic samples are likely to contain many phosphopeptides, and the fragment mass tolerances used for database searching for low resolution MS/MS spectra would likely lead to many false positive matches. For example, in a study of the phosphorylation of topoisomerase IIβ for which LC−MS/MS analysis was performed on a LCQ-Deca, two phosphopepeptides identified by Mascot and SEQUEST were found upon manual verification to be brominated.26 This finding highlights the importance of high resolution MS and MS/MS in correctly identifying and localizing post-translational modifications. An additional benefit of HCD spectra over CID spectra acquired in the ion trap is the improved ability to detect characteristic immonium ions for brominated amino acid residues.27 These ions are diagnostic for the presence of a particular brominated amino acid residue, so we initially attempted to filter .mgf files by the mass of each immonium ion. Unfortunately, not all brominated peptides produced a significant amount of the corresponding immonium ion. However, when the ion was present, it did provide additional confidence in localization of bromination site to a particular amino acid residue. Figure 1 shows an example of a peptide for which the brominated immonium ion (indicated by H$) is

20 dibromotyrosine-containing peptides were identified by PEAKS in the initial search, while only 10 were identified by SEQUEST and Mascot. Of the 20, 2 were semitryptic and 1 contained an N-terminal pyroglutamate, leaving 17 peptides that should be identified by SEQUEST and Mascot. When the database search was repeated with the custom dibromotyrosine modification, the number of identified dibromotyrosinecontaining peptides increased to 14 for both SEQUEST and Mascot. Five peptides were identified only with the new custom modification. For the remaining peptides identified by more than one search engine, all were identified using both the regular and custom modifications. However, in comparing all identifications for the same peptide sequence, both the Mascot and SEQUEST scores tended to be better with identifications coming from the custom modification. A detailed list of dibromotyrosine-containing peptides identified by CID can be found in Table 5 in the Supporting Information. For the 1% spike-in sample, in the CID analysis, four dibromotyrosine-containing peptides were identified. One peptide was unique to PEAKS; one was unique to SEQUEST and Mascot. The revised database searching strategy did not identify additional dibromotyrosine-containing peptides. It did, however, increase the number of MS/MS spectra assigned to each one of these peptides. For the samples analyzed by HCD, however, the same three dibromotyrosine containing peptides were identified in all three search engines, and all three were identified only with the custom modification. HCD vs CID for Brominated Peptides

CID-based fragmentation gave a higher number of identifications for brominated peptides (Table 3 in the Supporting Information), likely due to the inherent difference in spectral acquisition time between CID and HCD. However, we believe that the identification of brominated peptides is best performed 4252



Technical Note

for confirming the charge state as +2 rather than +1, is missing. HCD data is therefore of great importance to making valid conclusions about the type and site of modification by Br.

present, while Figures 2 and 4 illustrate spectra for which the brominated immonium ion is absent.

■

CONCLUSIONS We developed a method to reliably assign peptide bromination based on a new database search strategy applied to nano-LC− MS/MS data sets, identifying brominated peptides present in substoichiometric (1%) concentrations in the background of a complex proteomic mixture of tryptic peptides. We found that the number of assignments for dibromotyrosine-containing peptides could be significantly increased by adding the 79 81 Br BrY custom modification to the list of amino acid modifications within the search engine. This increase was seen for both low resolution (CID) data and high resolution (HCD) MS/MS data using multiple search engines, supporting the broad applicability of our modified database search strategy. Due to the possibility of false positive matches, it is advisable to perform manual validation for all identified peptides containing bromine. The levels at which bromination of proteins occurs in inflammatory disease are in the picomolar range;10,13 those levels are amenable to analysis with the novel analytical workflow described in this report. The new methodology would be applicable to the study of human tissues exposed to endogenous bromination in disease5,15 and to structural proteomic experiments that involve chemical modification by exogenous bromination.28,29

■

ASSOCIATED CONTENT

S Supporting Information *

Figure 4. A: Experimental (left) and theoretical (right) peptide isotopic distributions for the [M + 3H]3+ ion of (ALVLIAFAQY#LQQcPFEDHVK). B: CID MS/MS spectrum of the [M + 3H]3+ brominated peptide ion acquired in a proteomic background. The corresponding HCD MS/MS spectrum is found in Figure 4 in the Supporting Information.

Additional figures and tables as discussed in the text. This material is available free of charge via the Internet at http:// pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

Figure 4 illustrates the difference between the low resolution CID spectrum and high resolution HCD spectrum for a triply protonated peptide assigned as a dibrominated species (ALVLIAFAQY#LQQcPFEDHVK). The high resolution MS spectrum (Figure 4A) is consistent with the corresponding theoretical spectrum (Figure 4A) in the case of the CID spectrum (Figure 4B), but interpretation of the spectrum is complicated due to the characteristic isotopic distribution for doubly brominated species. As previously discussed, the A + 2 isotopic peak for doubly brominated peptides is two times more abundant than the monoisotopic m/z peak, sometimes leading to a complete absence of the monoisotopic m/z peak or the 13C isotope (A + 1) peak. The CID spectrum (Figure 4B) is characterized by an abundance of doubly protonated y ions which appear to be singly protonated due to the absence of 13C isotope peaks (see Figure 4B inset for an example). The spectrum shows a ladder of b ions and a couple of y ions up to the site of bromination and intense, seemingly singly protonated peaks, the masses of which match the theoretical values for the doubly protonated y ion series, supporting the proposed sequence assignment. The HCD spectrum for this peptide (Figure 4 in the Supporting Information) also supports the assignment. Even though the custom dibromo-Tyr modification facilitated the identification of this peptide, the experimental CID-MS/MS data does not back up the assignment because the 13C isotopic peak, which is necessary

*E-mail: [email protected]. Phone: (409) 747-1804. Fax: (409) 772-9648. Author Contributions ‡

These authors contributed equally to this work.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Funding provided by the Cancer Prevention Research Institute in Texas (CPRIT) and the University of Texas Medical Branch is gratefully acknowledged. The helpful comments offered by Dr. George Yeh were helpful in the preparation of this study. The authors also appreciate the assistance of Mark Sowers in writing a custom script for filtering .mgf files and Dan Maloney, Brian Munro, and Bin Ma from BSI for helpful discussions.

■

REFERENCES

(1) Jacobsen, E. A.; Helmers, R. A.; Lee, J. J.; Lee, N. A. The expanding role(s) of eosinophils in health and disease. Blood 2012, DOI: 10.1182/blood-2012-06-330845. (2) Ishitsuka, Y.; Maniwa, F.; Koide, C.; Douzaki, N.; Kato, Y.; Nakamura, Y.; Osawa, T. Detection of Modified Tyrosines as an Inflammation Marker in a Photo-aged Skin Model. Photochem. Photobiol. 2007, 83, 698−705, DOI: 10.1562/2006-07-24-ra-978.

4253



Technical Note

(3) Ishitsuka, Y.; Maniwa, F.; Koide, C.; Kato, Y.; Nakamura, Y.; Osawa, T.; Tanioka, M.; Miyachi, Y. Increased halogenated tyrosine levels are useful markers of human skin ageing, reflecting proteins denatured by past skin inflammation. Clin. Exp. Dermatol. 2012, 37, 252−258, DOI: 10.1111/j.1365-2230.2011.04215.x. (4) Hawkins, C. L.; Davies, M. J. The role of reactive N-bromo species and radical intermediates in hypobromous acid-induced protein oxidation. Free Radical Biol. Med. 2005, 39, 900−912, DOI: 10.1016/j.freeradbiomed.2005.05.011. (5) Wu, W.; Samoszuk, M. K.; Comhair, S. A. A.; Thomassen, M. J.; Farver, C. F.; Dweik, R. A.; Kavuru, M. S.; Erzurum, S. C.; Hazen, S. L. Eosinophils generate brominating oxidants in allergen-induced asthma. J. Clin. Invest. 2000, 105, 1455−1463, DOI: 10.1172/jci9702. (6) Curran, C. S.; Bertics, P. J. Eosinophils in glioblastoma biology. J. Neuroinflammation 2012, 9 (11), DOI: 10.1186/1742-2094-9-11. (7) Henderson, J. P.; Byun, J.; Mueller, D. M.; Heinecke, J. W. The Eosinophil Peroxidase-Hydrogen Peroxide-Bromide System of Human Eosinophils Generates 5-Bromouracil, a Mutagenic Thymine Analogue. Biochemistry 2001, 40, 2052−2059, DOI: 10.1021/bi002015f. (8) Henderson, J. P.; Byun, J.; Takeshita, J.; Heinecke, J. W. Phagocytes Produce 5-Chlorouracil and 5-Bromouracil, Two Mutagenic Products of Myeloperoxidase, in Human Inflammatory Tissue. J. Biol. Chem. 2003, 278, 23522−23528, DOI: 10.1074/ jbc.M303928200. (9) Henderson, J. P.; Byun, J.; Williams, M. V.; McCormick, M. L.; Parks, W. C.; Ridnour, L. A.; Heinecke, J. W. Bromination of deoxycytidine by eosinophil peroxidase: A mechanism for mutagenesis by oxidative damage of nucleotide precursors. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 1631−1636. (10) Henderson, J. P.; Byun, J.; Williams, M. V.; Mueller, D. M.; McCormick, M. l. L.; Heinecke, J. W. Production of Brominating Intermediates by Myeloperoxidase. J. Biol. Chem. 2001, 276, 7867− 7875, DOI: 10.1074/jbc.M005379200. (11) Valinluck, V.; Sowers, L. C. Inflammation-Mediated Cytosine Damage: A Mechanistic Link between Inflammation and the Epigenetic Alterations in Human Cancers. Cancer Res. 2007, 67, 5583−5586, DOI: 10.1158/0008-5472.can-07-0846. (12) Westman, E.; Harris, H. E. Alteration of an Autoantigen by Chlorination, a Process Occurring During Inflammation, Can Overcome Adaptive Immune Tolerance. Scand. J. Immunol. 2004, 59, 458−463, DOI: 10.1111/j.0300-9475.2004.01428.x. (13) Gaut, J. P.; Byun, J.; Tran, H. D.; Heinecke, J. W. Artifact-Free Quantification of Free 3-Chlorotyrosine, 3-Bromotyrosine, and 3Nitrotyrosine in Human Plasma by Electron Capture−Negative Chemical Ionization Gas Chromatography Mass Spectrometry and Liquid Chromatography−Electrospray Ionization Tandem Mass Spectrometry. Anal. Biochem. 2002, 300, 252−259, DOI: 10.1006/ abio.2001.5469. (14) Weiss, S. J.; Test, S. T.; Eckmann, C. M.; Roos, D.; Regiani, S. Brominating oxidants generated by human eosinophils . Science 1986, 234, 200−203. (15) Wu, W.; Chen, Y.; d’Avignon, A.; Hazen, S. L. 3-Bromotyrosine and 3,5-Dibromotyrosine Are Major Products of Protein Oxidation by Eosinophil Peroxidase: Potential Markers for Eosinophil-Dependent Tissue Injury in Vivo. Biochemistry 1999, 38, 3538−3548. (16) Williams, D.; Yuster, P. Isotopic Constitution of Tellurium, Silicon, Tungsten, Molybdenum, and Bromine. Phys. Rev. 1946, 69, 556−557. (17) Nair, S. S.; Nilsson, C. L.; Emmett, M. R.; Schaub, T. M.; Gowd, K. H.; Thakur, S. S.; Krishnan, K. S.; Balaram, P.; Marshall, A. G. De Novo Sequencing and Disulfide Mapping of a BromotryptophanContaining Conotoxin by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Anal. Chem. 2006, 78, 8082−8088, DOI: 10.1021/ac0607764. (18) Roepstorff, P.; Fohlman, J. Proposal for a common nomenclature for sequence ions in mass spectra of peptides. Biomed. Mass Spectrom. 1984, 11, 601.

(19) Karas, M.; Hillenkamp, F. Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons. Anal. Chem. 1988, 60, 2299−2301. (20) Michalski, A.; Damoc, E.; Lange, O.; Denisov, E.; Nolting, D.;Müller, M.; Viner, R.; Schwartz, J.; Remes, P.; Belford, M.; Dunyach, J.-J.; Cox, J.; Horning, S.; Mann, M.; Makarov, A. Ultra High Resolution Linear Ion Trap Orbitrap Mass Spectrometer (Orbitrap Elite) Facilitates Top Down LC MS/MS and Versatile Peptide Fragmentation Modes. Mol. Cell. Proteomics 2012, 11, DOI: 10.1074/ mcp.O111.013698. (21) Eng, J. K.; McCormack, M.; Yates, J. R. An approach to correlate tandem mass spectral data of peptides with amino acid sequences in a protein database. J. Am. Soc. Mass Spectrom. 1994, 5, 976−989. (22) Zhang, J.; Xin, L.; Shan, B.; Chen, W.; Xie, M.; Yuen, D.; Zhang, W.; Zhang, Z.; Lajoie, G. A.; Ma, B. PEAKS DB: De Novo sequencing assisted database search for sensitive and accurate peptide identification. Mol. Cell. Proteomics 2012, 11, DOI: 10.1074/ mcp.M111.010587. (23) Han, X.; He, L.; Xin, L.; Shan, B.; Ma, B. PeaksPTM: Mass Spectrometry-Based Identification of Peptides with Unspecified Modifications. J. Proteome Res. 2011, 10, 2930−2936, DOI: 10.1021/ pr200153k. (24) Han, Y.; Ma, B.; Zhang, K. SPIDER: software for protein identification from sequence tags with de novo sequencing error. J. Bioinform. Comput. Biol. 2005, 3, 697−716. (25) Scherl, A.; Shaffer, S.; Taylor, G.; Hernandez, P.; Appel, R.; Binz, P.-A.; Goodlett, D. On the benefits of acquiring peptide fragment ions at high measured mass accuracy. J. Am. Soc. Mass Spectrom. 2008, 19, 891−901, DOI: 10.1016/j.jasms.2008.02.005. (26) Grozav, A. G.; Willard, B. B.; Kozuki, T.; Chikamori, K.; Micluta, M. A.; Petrescu, A.-J.; Kinter, M.; Ganapathi, R.; Ganapahti, M. K. Tyrosine 656 in topoisomerase IIβ is important for the catalytic activity of the enzyme: Identification based on artifactual +80-Da modification at this site. Proteomics 2011, 11, 829−842. (27) Hung, C.-W.; Schlosser, A.; Wei, J.; Lehmann, W. D. Collisioninduced reporter fragmentations for identification of covalently modified peptides. Anal. Bioanal. Chem. 2007, 389, 1003−1016. (28) Roeser, J.; Bischoff, R.; Bruins, A. P.; Permentier, H. P. Oxidative protein labeling in mass-spectrometry-based proteomics. Anal. Bioanal. Chem. 2010, 397, 3441−3455. (29) Palaniappan, K. K.; Pitcher, A. A.; Smart, B. P.; Spiciarich, D. R.; Iavarone, A. T.; Bertozzi, C. R. Isotopic Signature Transfer and Mass Pattern Prediction (IsoStamp): An Enabling Technique for Chemically-Directed Proteomics. ACS Chem. Biol. 2011, 6, 829−836.

4254


A Modified Database Search Strategy Leads to Improved Identification

Recommend Documents