Technical Note pubs.acs.org/jpr
Optimization of LTQ-Orbitrap Mass Spectrometer Parameters for the Identification of ADP-Ribosylation Sites Florian Rosenthal,†,§,⊥ Paolo Nanni,‡,⊥ Simon Barkow-Oesterreicher,‡ and Michael O. Hottiger*,† †
Institute of Veterinary Biochemistry and Molecular Biology, ‡Functional Genomics Center Zurich, University of Zurich/ETH Zurich, §Life Science Zurich Graduate School, University of Zurich, CH-8057 Zurich, Switzerland
Downloaded via UNIV OF NEW ENGLAND on October 10, 2018 at 14:01:34 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: ADP-ribosylation of proteins alters their function or provides a scaffold for the recruitment of other proteins, thereby regulating several important cellular processes. Mono- or poly-ADP-ribosylation is catalyzed by different ADP-ribosyltransferases (ARTs) that have different subcellular localizations and modify different amino acid acceptor sites. However, our knowledge of ADP-ribosylated proteins and their acceptor amino acids is still limited due to the lack of suitable mass spectrometry (MS) tools. Here, we describe an MS approach for the detection of ADP-ribosylated peptides and identification of the ADP-ribose acceptor sites, combining higher-energy collisional dissociation (HCD) and electron-transfer dissociation (ETD) on an LTQ-Orbitrap mass spectrometer. The presence of diagnostic ions of ADP-ribose in the HCD spectra allowed us to detect putative ADP-ribosylated peptides to target in a second LC−MS/MS analysis. The combination of HCD with ETD fragmentation gave a more comprehensive coverage of ADP-ribosylation sites than that with HCD alone. We successfully identified different ADP-ribose acceptor sites on several in vitro modified proteins. The combination of optimized HCD and ETD methods may be applied to complex samples, allowing comprehensive identification of ADPribosylation acceptor sites. KEYWORDS: ADP-ribosylation, histone, PARP, LTQ-Orbitrap, mass spectrometry, electron-transfer dissociation, ETD, higher-energy collisional dissociation, HCD
■
INTRODUCTION ADP-ribosylation is an ancient post-translational protein modification (PTM) catalyzed by ADP-ribosyltransferases (ARTs).1 ARTs use nicotinamide adenine dinucleotide (NAD+) as substrate to transfer the ADP-ribose moiety onto specific acceptor sites of target proteins or existing ADP-ribose units, thereby generating mono- or poly-ADP-ribosylated proteins.1 The acceptor sites for mono-ADP-ribose are side chains of certain amino acid residues. The following amino acids have been reported to be covalent acceptor sites for ADPribose in eukaryotic cells: Lys, Arg, Gln, Cys, Asp, Glu, diphthamide, phospho-Ser, and Asn.2−6 However, these sites have been primarily identified by cumbersome chemical or mutational analyses that do not allow a comprehensive analysis of all proteins. On the basis of their sequence similarity to bacterial diphtheria toxin and cholera toxin, ARTs are subdivided into ARTDs and ARTCs,2 respectively. The ARTD subfamily comprises the intracellular ARTs, which are implicated in many regulatory processes and linked to various diseases and pathological conditions.7−9 In cells, ADP-ribose modification is highly transient because of the hydrolytic action of enzymes that degrade poly-ADP-ribose (PAR) to mono-ADP-ribose (MAR) as well as enzymes that remove MAR from proteins.10−12 © 2015 American Chemical Society
ADP-ribosylation is considered to be an element of the histone code and coregulates gene transcription, protein function, differentiation processes, and stress responses as well as the onset and progression of diseases.10,13−16 Although research on ADP-ribosylation has a long history, the inherent technological difficulties for analyzing this complex PTM are a major reason for the large gaps in understanding its cellular and molecular functions. Among the challenges, ADP-ribosylation can occur during sample preparation because ARTD1 (PARP1), the main intracellular ARTD family member, is activated by fragmented DNA. On the other hand, the modification is easily lost during extraction or sample preparation. The heterogeneity of the PAR chain further complicates analysis due to the unpredictable mass shift of a complex and potentially branched polymer. This problem can be addressed by enzymatically degrading PAR to MAR with poly(ADP-ribose) glycohydrolase (PARG) prior to analysis.10 Another limiting factor for the functional analysis of ADPribosylation is the lack of suitable methods to identify and quantify the actual amino acid acceptor sites on ADPribosylated proteins. Without such tools, it is impossible to link physiological and cellular readouts to the modification Received: May 19, 2015 Published: July 27, 2015 4072
DOI: 10.1021/acs.jproteome.5b00432 J. Proteome Res. 2015, 14, 4072−4079
Technical Note
Journal of Proteome Research Table 1. Overview of the Evaluated Tandem Mass Spectrometry Methodsa MS2 events MS method a. b. c. d. e. f. a
HCD ETD (IT) ETD (FT) HCD-ETD (IT) HCD-PD1-ETD (IT) two-stage HCD-ETD (IT)
HCD
ETD
10
10 10 10
data collection FT
IT
FT
IT
1 × 10
1 × 104
c p p p p
reagent
5
p 10 8 10 10 10
MS2 max injection time (ms)
MS2 AGC target
c c c
1 1 1 1
× × × ×
105 105 105 105
1 × 104 1 × 104 1 × 104
1 1 1 1 1
× × × × ×
105 105 105 105 105
FT
IT
200
50 50
200 200 200 200
50 50
reagent 100 100 100 100 100
product ions
136.0623
FT, Orbitrap; IT, ion trap; p, profile; c, centroid.
Velos system for the identification of ADP-ribosylated peptides and their modification sites.
status of a specific amino acid and the activity of a certain ARTD. Liquid chromatography−mass spectrometry strategies (LC− MS/MS) are frequently used to identify specific PTMs, and the most commonly employed fragmentation method is collisioninduced dissociation (CID). However, labile PTMs such as ADP-ribosylation are unstable during CID fragmentation. Indeed, a typical CID MS/MS spectrum of an ADP-ribosylated peptide is dominated by fragment ions that originate from the ADP-ribose backbone, but it lacks sufficient peptide backbone fragment ions to identify the amino acid sequence.17,18 In particular, fragment ions corresponding to adenine, adenosine− H2O, adenosine monophosphate (AMP), adenosine diphosphate (ADP), and ADP-ribose are commonly observed. Recently, precursor ion scanning-triggered MS/MS strategies that monitor CID-specific marker ions of the ADP-ribose moiety have been proposed, allowing the selective detection of ADP-ribosylated peptides in relatively low complexity protein digests.19,20 In contrast to CID, the ADP-ribose modification is maintained on the peptide during electron-transfer dissociation (ETD) or electron capture dissociation (ECD), allowing the peptide sequence to be identified and the modification site to be localized. For this reason, these fragmentation techniques have been considered to be the best choice for the identification of ADP-ribosylation sites.17,21 Hengel et al. proposed a marker ion approach that circumvents the difficulties in peptide identification and takes advantage of the altered fragmentation pattern of ADPribosylated peptides.21 In a first LC−MS/MS run, putative ADP-ribosylated peptides are detected (based on the presence of marker ions in the CID spectra), which are then targeted for ETD fragmentation and subsequent peptide identification in a second LC−MS/MS analysis.17,21 Although this strategy requires twice as much sample and instrument time and could be affected by the loss of low-mass ions during ion trap CID fragmentation (with consequent loss of marker ion information), it allows for better characterization of ADPribosylated peptides. Surprisingly, the use of higher-energy collisional dissociation (HCD) has been applied only very rarely to ADP-ribosylated proteins, although two recent studies have led to the identification of hundreds of ADP-ribosylated peptides.22,23 In both of these studies, a Q-Exactive MS instrument (Thermo Scientific) was used, which allows only HCD fragmentation to be used as the dissociation technique. Since a comparison between HCD and ETD has not been performed before, we set out to systematically evaluate the use of HCD and ETD fragmentations on an ETD-enabled hybrid LTQ-Orbitrap
■
EXPERIMENTAL SECTION
In Vitro Modification of Standard Peptide and Histone Mix
The peptide biotin-KAARKSAPATGGVKKPHRYR (H3) or a mixture containing the four core histones as well as the H1 linker histone as full-length proteins were ADP-ribosylated as described earlier.10 Briefly, 1 μg of peptide or 3 μg of histone mix was incubated with 10 pmol of GST-ARTD10 (818−1025) for 15 min at 30 °C in reaction buffer (50 mM Tris-HCL, pH 8.0, 4 mM MgCl2, 250 μM DTT) in the presence of 160 μM NAD+. The reactions were stopped by the addition of PJ34 (ARTD inhibitor) and frozen until desalting, or further processed by reducing (5 mM DTT for 60 min), alkylating (15 mM IAM for 45 min), and digesting with trypsin (1:50 (w/ w) for 2 h). Nanoliquid Chromatography−Tandem Mass Spectrometry
Mass spectrometry analysis was performed on an LTQOrbitrap Velos ETD mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) coupled to an Eksigent-NanoHPLC system (Eksigent Technologies, Dublin, CA, USA). Solvent composition at the two channels was 0.1% formic acid for channel A and 0.1% formic acid, 99.9% acetonitrile for channel B. Peptides were loaded on a self-made tip column (75 μm × 150 mm) packed with reverse-phase C18 material (ReproSil-Pur 120 C18-AQ, 1.9 μm, Dr. Maisch GmbH) and eluted at a flow rate of 300 nL/min by a gradient from 2 to 35% B in 30 min (for standard peptides) or 60 min (for complex histone samples). Full-scan MS spectra (300−1700 m/z) were acquired at a resolution of 30 000 at 400 m/z, an accumulation gain control (AGC) of 1 × 106, and a maximum injection time of 250 ms. The AGC values for MS/MS analysis were set to 1 × 104 for ETD experiments with ion trap detection (IT, 100 ms injection time) and to 1 × 105 for HCD and ETD experiments with detection in the Orbitrap (FT, 200 ms injection time). The HCD-normalized collision energy was set to 40%, enabling the collision energy to be stepped (width 15%, three steps) and detecting the ions at a resolution of 7500 at 400 m/z. In all experiments, only one microscan was used for detection. The isolation width was set to 2 and 4 amu for HCD and ETD experiments, respectively. Full scans and Orbitrap MS/MS scans were acquired in profile mode, whereas ion trap mass spectra were acquired in centroid mode. All experiments were recorded in data-dependent mode from signals above a threshold of 2000. Charge state screening was enabled, and singly charge states were rejected. For ETD experiments, the ETD anion target value was set to 1 × 105, and the activation time to 100 ms. Charge state-dependent ETD reaction times 4073
DOI: 10.1021/acs.jproteome.5b00432 J. Proteome Res. 2015, 14, 4072−4079
Technical Note
Journal of Proteome Research
Figure 1. Optimal HCD fragmentation of the ADP-ribosylated standard peptide using ramped CE. (A) Overview of the sample preparation, liquid chromatography mass spectrometry, and data analysis workflow employed for the evaluation of the HCD fragmentation behavior of the ADPribosylated biotin-KAARKSAPATGGVKKPHRYR (H3) peptide. (B) Nomenclature of ADP-ribose fragments as described by Hengel et al.17 The ADP-ribose fragment ions with strong signals in the HCD MS/MS spectra are shown (m1, m3, m6, m8, m10). (C) HCD spectrum of the ADPribosylated biotin-KAARKSAPATGGVKKPHRYR (H3) peptide. The loss of adenine, adenosine + water, adenosine monophosphate (AMP), adenosine diphosphate (ADP), and ADP-ribose can be observed in the zoomed panel. The arrows mark the four most abundant ADP-ribose fragment ions.
Orbitrap full MS scan followed by top 10 data-dependent HCD and an additional ETD ion trap event when a peak at 136.0623 (ADP-ribose fragment ion) was observed among the top 10 most abundant peaks of the HCD scan (HCD-PD1-ETD); (f) two-stage, consisting of HCD and an additional ETD ion trap MS/MS acquisition on the top 10 precursors where the marker ions 136.0623, 250.0940, and 348.07091 were observed in a previous HCD run. A complete description of all tandem MS experiments employed in this study can be found in Table 1.
were enabled, setting a reference value of 100 ms for doubly charged peptides. A supplemental activation with 25% normalized collision energy was always enabled. For the analysis of standard peptides, precursor masses already selected three times for MS/MS were excluded for further selection for 45s. For the analysis of complex histone samples, the dynamic exclusion was set to 1 occurrence. The exclusion window was set to 10 ppm, and the exclusion size was limited to a maximum of 500 entries. The following LC−MS/MS acquisition methods were performed: (a) Orbitrap full MS scan followed by top 10 data-dependent Orbitrap HCD MS/MS; (b) Orbitrap full MS scan followed by top 10 data-dependent ion trap ETD MS/MS or (c) top 8 data-dependent Orbitrap ETD MS/MS; (d) Orbitrap full MS scan followed by top 10 double-play datadependent HCD MS/MS and ion trap ETD MS/MS; (e)
Database Analysis and Configuration of Mascot Modifications
MS and MS/MS spectra were converted into Mascot generic format (mgf) using Proteome Discoverer, v1.4 (Thermo Fisher Scientific, Bremen, Germany). When both HCD and ETD 4074
DOI: 10.1021/acs.jproteome.5b00432 J. Proteome Res. 2015, 14, 4072−4079
Technical Note
Journal of Proteome Research
Figure 2. Analysis of ADP-ribosylated peptides using a combination of HCD and ETD fragmentation. Schematic overview of the mass spectrometry methods employed in the present study. The histone mixture is ADP-ribosylated and analyzed using various combination of HCD and ETD fragmentation techniques. ETD spectra can be acquired in the ion trap (IT) or in the Orbitrap (FT). The red circle represents the ADP-ribose marker ion present in the HCD spectrum that triggers an ETD fragmentation event. PTM MarkerFinder is software that screens for ADP-ribose marker ions and creates an inclusion list containing putative ADP-ribosylated peptides to be targeted in a second LC−MS/MS analysis.
scans were present in the same raw files, separate mgfs for the two fragmentation techniques were created. All high-resolution HCD and ETD MS/MS spectra were deconvoluted using MS Spectrum Processor, v0.9.24 In the ETD spectra, the precursor, the charge-reduced precursor(s), and the neutral losses were removed. The mgfs were searched against the UniProtKB human database (taxonomy 9606, version 20140422), including 35 787 Swiss-Prot, 37 802 TrEMBL entries, 73 589 decoy hits, and 260 common contaminants. Mascot 2.4.1 (Matrix Science) was used for the identification with the following search settings: the combined HCD and Orbitrap ETD mgfs were searched with a peptide tolerance of 10 ppm and MS/MS tolerance of 0.05 Da, whereas the MS/MS tolerance for the ion trap ETD spectra was set to 0.6 Da. For the HCD runs, singly charged b and y ion series, immonium ions, and water and ammonia loss ion series were searched. For ETD, multiply charged c, y, z, z + 1, z + 2 series were considered. Enzyme specificity was set to trypsin, allowing up to 4 missed cleavages. Decoy hits were used to control the false discovery rate at the peptide and protein levels. The ADP-ribose variable modification was set up differently for HCD or ETD spectra searches. For HCD mgfs, the modification was set to a mass shift of 541.0611, with scoring of the neutral losses equal to 347.0631 and 249.0862. The marker ions at m/z 428.0372, 348.0709, 250.0940, 136.0623 were ignored for scoring. Nine additional neutral losses were set as satellite (Figure S1). For the identification of ADP-ribosylation from ETD spectra, only the mass shift of 541.0611 was used. An ADP-ribosylated peptide was considered to be correctly identified when a Mascot score higher than 20 and an expectation value lower than 0.05 were obtained. To assess the location of the ADPribosylation sites, we used the site localization analysis provided by Mascot, which is based on the work by Savitski et al.25 and was developed especially for phosphorylation. This method is not optimized for ADP-ribosyl modification due to the lack of standard peptides with known modification sites. For this reason, even if Mascot states a correctness of 95% for the site localization, this value is arbitrary and cannot be experimentally validated. Due to the lack of a better estimate, we define
correctness as having a confidence of at least 95% in the Mascot site localization analysis. PTM MarkerFinder
For LC−MS/MS experiments where HCD fragmentation has been used, the Mascot output.dat files have been further analyzed using PTM MarkerFinder.26 Briefly, PTM MarkerFinder screens the Mascot outputs for HCD spectra containing the ADP-ribose marker ions at m/z 136.0623, 250.0940, 348.0709, and 428.0372. The spectra containing at least two marker ions and where the sum of the marker ion intensities covers at least 5% of the total ion intensities are considered to be spectra from putative ADP-ribosylated peptides. Finally, PTM MarkerFinder summarizes information about the presence and intensity of marker ions and annotates the spectra with the corresponding peptide sequence.
■
RESULTS AND DISCUSSION
Optimal HCD Fragmentation of the ADP-Ribosylated Standard Peptide Using Ramped CE
The currently available MS/MS protocols are not optimized for the accurate and reliable detection of ADP-ribosylated peptides and their acceptor sites in complex samples. To improve MSbased detection of complex ADP-ribosylated peptides, two model samples with different complexities were prepared as described in the Experimental Section. To better comprehend the behavior of ADP-ribosylated peptides during HCD fragmentation and to detect the diagnostic ions characteristic for the ADP-ribose moiety of modified peptides, an ADPribosylated H3 standard peptide was first fragmented using HCD with NCE (normalized collision energy) values ranging from 23 to 60% (Figure 1A,B and as described by Hengel et al.17). This analysis revealed that the MARylated standard peptide showed high-intensity signals for the expected ADP-ribose marker ions but low-intensity signals for the peptide fragment at higher NCE values (Figure S2). A good balance between the intensity of ADP-ribose marker ions and peptide backbone fragments was achieved by using ramped NCE (width 15%, 4075
DOI: 10.1021/acs.jproteome.5b00432 J. Proteome Res. 2015, 14, 4072−4079
Technical Note
Journal of Proteome Research
Figure 3. Analysis of histones ADP-ribosylated by ARTD10 in vitro using combinations of HCD and ETD fragmentation. Histones ADP-ribosylated by ARTD10 in vitro were analyzed by MS/MS using six different combinations of HCD and/or ETD fragmentation techniques. ADP-ribosylated peptides where the site localization confidence score is higher than 95% (blue bars) and lower than 95% (red bars) are plotted. (A) Number of ADPribosylated peptide spectrum matches (PSMs). (B) Number of unique ADP-ribosylated peptides (different modification sites on the same peptide are considered to be unique peptides). (C) Venn diagrams showing the differences and overlaps between HCD and ETD identifications.
acquisition of the ETD spectra in the IT allowed a higher throughput to be achieved compared to that with FT detection (due to the shorter length of the scan cycles), resulting in a slightly higher number of identifications (Figure 3). When ADP-ribosylated histone samples were analyzed by either HCD or ETD fragmentation alone, HCD resulted in nearly 2 times more ADP-ribosylated peptide spectrum matches (PSMs) than those with ETD (Figure 3A). In principle, combining ETD and HCD may improve the identification of ADP-ribosylated peptides and modification sites, as it takes advantage of the information generated by both approaches. However, subjecting the sample to HCD-ETD (IT) results in increased cycle times and in a lower number of identifications and assigned spectra (Figure 3). This strategy is particularly inefficient for complex samples because all peptides, including the ones that are not modified, are fragmented and analyzed twice. To overcome these shortcomings, HCD was followed by ETD on the same precursors only when one or more marker ions was present in the HCD spectra (productdependent approach, HCD-PD-ETD17). Although this approach has been successfully used for glycopeptide studies,27−29 this is the first description, to the best of our knowledge, of it being used for the identification of ADP-ribosylated peptides using an LTQ-Orbitrap machine. The ETD (IT) was triggered upon detection of the marker ion at m/z 136.0623 (HCD-PD1ETD (IT)). In general, marker ion 136.0623 proved to be specific enough to trigger ETD fragmentation, whereas the inclusion of three marker ions (136.0623, 250.0940, 348.0709 (HCD-PD3-ETD (IT)) increased the number of missed PDETD activations (data not shown). The HCD-PD1-ETD (IT) approach identified approximately the same number of unique peptides as that from the two single HCD and ETD (IT) runs together, indicating that all modified peptides with a clear ion
three steps centered around 40%). The HCD spectra acquired with such parameters were dominated by the marker ions originating from the internal fragmentation of ADP-ribose (with the ion at m/z 136 being the most abundant one) as well as by additional fragments corresponding to the peptide plus a residual mass from ADP-ribose fragmentation (Figure 1C). These ADP-ribose fragments are often not caused by neutral losses and thus may exhibit charge states different from that of the precursor. On the basis of these results, ramped NCE was defined as the optimal setting and used for all following analyses. Moreover, the Mascot settings for the ADP-ribose modification were improved by (i) exclusion of the marker ion for scoring and (ii) use of m6 and m3 neutral losses for scoring, (iii) use of m1, m1 + H2O, m1 + 2H2O, m3 + H2O, m6 + H2O, m8, m8 + H2O, m8 + 2H2O, and m10 as satellite neutral losses (Figure 1C and Figure S1). The Combination of HCD and ETD Fragmentation Increases the Number of Identified ADP-Ribosylation Sites from a Mixture of in Vitro Modified Full-Length Histone Proteins
Until recently, ETD fragmentation was considered to be the method of choice for detecting and identifying ADP-ribosylated peptides and was therefore compared to HCD fragmentation (Table 1 and Figure 2). In a next step, a mixture of all four core histones and linker histone H1 was ADP-ribosylated by ARTD10 (818−1025) in vitro, digested with trypsin, and all of the resulting peptides were analyzed by HCD (as described above) or by ETD fragmentation acquired in the ion trap (IT) or Orbitrap (FT). Since the average charge states of the identified peptides were relatively low (less than 4, Figure S3), the acquisition of high-resolution ETD (FT) MS/MS spectra did not lead to identification of a higher number of unique ADP-ribosylated peptides (Figure 3B). In contrast, the 4076
DOI: 10.1021/acs.jproteome.5b00432 J. Proteome Res. 2015, 14, 4072−4079
Technical Note
Journal of Proteome Research Table 2. Identification of ADP-Ribosylated Histone Peptides Having Unambiguous Site Localizationa
a
Newly identified ADP-ribosylation sites from histones in vitro modified by ARTD10. The modification sites are presented in red. All ADPribosylated peptides with a Mascot localization score over 95% are listed. PSMs, peptide spectrum matches.
Taken together, the identification and mapping of the ADPribosylation sites in a more complex sample was best achieved by the application of HCD-PD1-ETD (IT) or the two-stage HCD-ETD (IT) approach. Both methods identified a large number of acceptor sites and recorded the additional ETD information. However, in most cases, a single HCD run might be sufficient to capture the majority of ADP-ribosylated peptides and their acceptor sites. Two things considered if complex cellular extracts are to be analyzed. As complex mixtures also contain poly-ADPribosylated peptides, an additional sample handling step that reduces the complexity of the PAR structure can significantly increase the number of successful identifications. Poly-ADPribose can be reduced to mono-ADP-ribose by treatment with PAR-degrading enzymes such as PARG or ARH3, thereby creating a defined mono-ADP-ribose moiety that can be reliably detected with our optimized method. In addition, ADPribosylation is not abundant in cellular extracts, and most attempts to identify the ADP-ribosylome without prior enrichment have failed or have generated only poor results. Therefore, recent studies have employed different methods to enrich ADP-ribosylated peptides from complex mixtures to successfully identify ADP-ribosylated proteins.22,23,30 Yet, unambiguous and unbiased site localization still remains a technical challenge, and the combination of the herein described new site identification methods, together with PARG pretreatment and the enrichment of ADP-ribosylated peptides, is therefore a powerful tool to identify the cellular ADP-ribosylome. The different MS/MS protocols described above led to the identification of 167 ADP-ribosylated peptides (854 PSMs) of in vitro modified ARTD10 and histones (Table 2 and Table S1). Several sites or stretches of ADP-ribosylation were covered
pattern were already detected and identified by the two single runs (Figure 3B). The previously suggested two-stage (marker ion) approach,17 which has not yet been tested with HCD and ETD, is an alternative method. To evaluate this method, samples were first analyzed using HCD fragmentation. Subsequently, the Mascot output was screened for spectra containing ADP-ribose marker ions using PTM MarkerFinder26 to obtain a list of precursor ions corresponding to putative ADP-ribosylated peptides. Only these precursor ions were targeted for HCD and ETD (IT) fragmentation in a second LC−MS/MS experiment. Interestingly, when compared to the HCD-PD1-ETD (IT) approach, the two-stage HCD-ETD (IT) experiment identified almost twice as many ADP-ribosylated PSMs but fewer unique ADP-ribosylated peptides (Figure 3B). However, to increase the chances of acquiring good quality MS/MS spectra, we used different dynamic exclusion settings between single HCD and marker ion-dependent runs (after 1 and 3 occurrences, as described in the Experimental Section). Despite the different repeat counts, chosen for optimal performance, the difference in output may indicate that the more abundant modified peptides were repeatedly selected for MS/MS and identified with a high score, thus limiting the detection of less abundant ones. To validate the potential benefits of the ETD fragmentation in these approaches, the contributions from HCD and ETD spectra to the identified peptides from the combined HCDETD (IT) runs were compared. A significant proportion of the unique ADP-ribosylated peptides was identified uniquely using ETD, suggesting that the ETD fragmentation provided additional information compared to that with HCD alone (Figure 3C). 4077
DOI: 10.1021/acs.jproteome.5b00432 J. Proteome Res. 2015, 14, 4072−4079
Technical Note
Journal of Proteome Research Author Contributions
by overlapping peptides, underlining the robustness of this approach. Interestingly, ADP-ribosylation of histones by ARTD10 in vitro was not limited to a single type of amino acid residue; modifications were found on Glu and Asp as well as on Lys and Arg. Representative HCD (FT) and ETD (IT) annotated spectra of ADP-ribosylated peptides modified on Glu, Asp, Lys, and Arg are shown in Supporting Information, Figures S4 and S5, respectively. Previously, Glu and Asp residues were mainly identified as automodification sites for ARTD10, an enzyme that, due to substrate-assisted catalysis, is predicted to modify acidic amino acids of selected modified target proteins in vitro.31 However, our analyses imply that ARTD10 is able to modify more than one type of amino acid in vitro. Whether the identified basic acceptor amino acids are modified also by a substrate-assisted mechanism remains to be determined.
⊥
F.R. and P.N. contributed equally to this work. Samples were prepared and processed by F.R. MS/MS analysis, data evaluation, and method optimization were done by P.N. and F.R. Tools for data analysis were programmed by S.B. The manuscript was written through contributions of all authors. M.O.H. supervised the study and edited the manuscript.
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported in part by the Swiss National Science Foundation, grant 310030B_138667, and the Kanton of Zurich (to M.O.H.). Florian Freimoser and Stephan Christen (Institute of Veterinary Biochemistry and Molecular Biology, University of Zurich, Switzerland) provided editorial assistance and critical input during the writing of the paper. We thank Peter Gehrig and Christian Panse from the FGCZ for critical input and help with the PTM MarkerFinder, respectively.
■
CONCLUSIONS The approach for the reliable identification of ADP-ribosylated proteins and their specific modification sites described in the present study combines methods for the preparation of ADPribosylated samples, data acquisition on an LTQ-Orbitrap system, and data analysis (Figures 1A and 2A). The combination of HCD and ETD fragmentations, preferably by HCD-PD1-ETD (IT) or two-stage HCD-ETD (IT), enables the identification of ADP-ribosylation sites with high reliability and confidence for in vitro modified proteins. With these approaches, we were able to identify several ADP-ribose acceptor sites (Glu, Asp, Lys, and Arg) on various different proteins with high confidence. We thus present a robust method for the exact identification of ADP-ribosylation sites using an ETD-enabled LTQ-Orbitrap mass spectrometer that can be applied to various proteins of interest and that expands the possibility of reliably identifying ADP-ribose acceptor sites in complex samples such as cellular extracts.
■
■
ABBREVIATIONS AGC, accumulation gain control; ADP, adenosine diphosphate; AMP, adenosine monophosphate; Arg, arginine; ART, ADPribosyl transferase; ARTC, ART cholera toxin-like; ARTD, ART diphtheria toxin-like; Asn, asparagine; Asp, aspartic acid; CID, collision-induced dissociation; Cys, cysteine; DTT, dithiothreitol; ECD, electron capture dissociation; ETD, electron-transfer dissociation; Gln, glutamine; Glu, glutamic acid; H3, histone 3; HCD, higher-energy collisional dissociation; IAM, iodoacetamide; IT, ion trap; L, leupeptin; Lys, lysine; LC, liquid chromatography; MAR, mono-ADP-ribose; MS, mass spectrometry; NAD + , nicotinamide adenine dinucleotide; NCE, normalized collision energy; NL, neutral loss; FT, Orbitrap; P, pepstatin; 32P, phosphorus-32; PAR, poly-ADP-ribose; PARG, poly-ADP-ribosyl glycohydrolase; PTM, post-translational modification; PD, product-dependent; Ser, serine
ASSOCIATED CONTENT
■
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jproteome.5b00432. Mascot settings used for the configuration of ADP-ribose modification (Figure S1); HCD spectra of peptide biotin-KAARKSAPATGGVKKPHRYR + 1 ADP-ribose (Figure S2); average charge state of the identified ADPribosylated peptides (Figure S3); representative HCD (FT) annotated spectra of ADP-ribosylated histone peptides modified on Glu, Asp, Lys, and Arg (Figure S4); and representative ETD (IT) annotated spectra of ADP-ribosylated histone peptides modified on Glu, Asp, Lys, and Arg (Figure S5) (PDF). List of ADP-ribosylated peptides identified from in vitro modified full-length histone proteins with ambiguous site localization (
