B American Society for Mass Spectrometry, 2014

J. Am. Soc. Mass Spectrom. (2014) DOI: 10.1007/s13361-014-0835-x

RESEARCH ARTICLE

Mass Spectral Enhanced Detection of Ubls Using SWATH Acquisition: MEDUSA—Simultaneous Quantification of SUMO and Ubiquitin-Derived Isopeptides John R. Griffiths,1 Navin Chicooree,1,2 Yvonne Connolly,1 Milla Neffling,3 Catherine S. Lane,3 Thomas Knapman,3 Duncan L. Smith1 1

CRUK Manchester Institute, University of Manchester, Manchester, M20 4BX, UK School of Chemistry, University of Manchester, Manchester, M13 9SU, UK 3 AB SCIEX, Phoenix House, Warrington, WA1 1RX, UK 2

Abstract. Protein modification by ubiquitination and SUMOylation occur throughout the cell and are responsible for numerous cellular functions such as apoptosis, DNA replication and repair, and gene transcription. Current methods for the identification of such modifications using mass spectrometry predominantly rely upon tryptic isopeptide tag generation followed by database searching with in vitro genetic mutation of SUMO routinely required. We have recently described a novel approach to ubiquitin and SUMO modification detection based upon the diagnostic a′ and b′ ions released from the isopeptide tags upon collision-induced dissociation of reductively methylated Ubl isopeptides (RUbI) using formaldehyde. Here, we significantly extend those studies by combining data-independent acquisition (DIA) with alternative labeling reagents to improve diagnostic ion coverage and enable relative quantification of modified peptides from both MS and MS/MS signals. Model synthetic ubiquitin and SUMOderived isopeptides were labeled with mTRAQ reagents (Δ0, Δ4, and Δ8) and subjected to LC-MS/MS with SWATH acquisition. Novel diagnostic ions were generated upon CID, which facilitated the selective detection of these modified peptides. Simultaneous MS-based and MS/MS-based relative quantification was demonstrated for both Ub and SUMO-derived isopeptides across three channels in a background of mTRAQ-labeled Escherichia coli digest. Key words: SUMOylation, Ubiquitination, SWATH, MEDUSA, Quantitation, Isopeptides Received: 22 November 2013/Revised: 10 January 2014/Accepted: 11 January 2014

P

rotein ubiquitination [1] and SUMOylation [2] are important reversible post-translational modifications involved in many critical cellular processes [3–5]. Both modifications occur via the covalent addition of a small protein (Ub or SUMO) to a target protein through a triple enzyme cascade reaction. Modification of the target protein takes place predominantly at acceptor lysine residues between a truncated C-terminal diglycine on the Ub/SUMO and the epsilon (ε)-NH2 group of the acceptor lysine. These modifications result in a mass increase of approximately 8.5 kDa per ubiquitin addition and around 12 kDa for every SUMO.

Electronic supplementary material The online version of this article (doi:10.1007/s13361-014-0835-x) contains supplementary material, which is available to authorized users. Correspondence to: John R. Griffiths; e-mail: [email protected]

Disregulation of these two modifications has important consequences in human pathogenesis, and both have been implicated in a number of diseases such as the cancers [6, 7]. In addition to the individual impact of aberrations in either post-translational modification (PTM), there is growing evidence that the two modifications may interact and participate in crosstalk in a number of ways [8, 9]. There is, therefore, increasing interest within the biological community in understanding more deeply the roles of these modifications and the interplay that exists between them and with other modifications such as phosphorylation and acetylation [10]. Most mass spectrometry-based methods for the analysis of ubiquitination rely upon the generation of a GG remnant remaining on the acceptor lysine upon digestion with trypsin, which may subsequently be used as a variable modification (+114.0429 Da) parameter in database searching [11]. Unfortunately, this approach is prone to

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

inaccurate assignments (false positives) because (1) the mass addition for GG is identical to that of an asparagine residue, (2) cysteine alkylation using iodoacetamide has been shown to potentially interfere [12], and (3) no known modification-specific ions are released upon CID fragmentation of these isopeptides [13]. Recently, Chicooree et al. demonstrated that upon reductive methylation of ubiquitin-derived isopeptides, several GG-related ions become stabilized and diagnostic ions are subsequently generated by collision cell CID [14]. This method, termed reductive methylation of ubiquitin isopeptides (RUbI), improves confidence in ubiquitination identifications as a result of the selectivity of the a′ and b′ diagnostic ions along with the previously reported increase in backbone b ion coverage as a consequence of dimethyl labeling [15]. Although the analysis of sites of ubiquitination by proteomics is made more tractable because of the advantageous tryptic cleavage site (Arg) N-terminal to the diglycine residues, no such analogous site exists in SUMO. This has, until recently, rendered the detection of SUMOylation extremely difficult with common proteomic approaches requiring either the generation of mutant forms of SUMO with a suitable tryptic cleavage site engineered into their sequence [16] or employment of specialized fragmentation techniques on ultra high resolution instrumentation [17]. However, it was recently demonstrated that atypical cleavages did occur when SUMOylated proteins were digested with trypsin [18], allowing the same approach of dimethyl labeling (RUbI) to be applied to the detection of SUMOderived isopeptides [19]. Upon proteolysis, proteins modified by SUMOylation were found to generate appreciable amounts of GG- for Human SUMO-1-ylation and TGG- or QTGG- for Human SUMO-2/3-ylation [18]. Although the approach of dimethyl labeling offers a clear improvement in the detection of Ub and SUMO-modified peptides, the ability to quantify these isopeptides is less favorable. In theory at least, it should be possible to relatively quantify three samples simultaneously using light, intermediate, and heavy labeling reagents [20]; however, in practice this is not the case. Experiments using the light version of dimethyl labeling demonstrate that the b2′ ion at m/z 143.0815 offers poor selectivity, making its use impractical for ubiquitination analysis via RUbI method (data not shown). In addition, the a1′ ions of all three versions of the reagents, which are significantly more abundant than equivalent b2′ ions, are in a low mass region of the mass spectrum, rendering their transmission potentially problematic on certain instruments. Consequently, we decided to assess the performance of alternative labeling reagents for the detection and quantification of Ub and SUMO-derived isopeptides. We hypothesized that the mTRAQ reagents would potentially be suitable for such a workflow [21]. mTRAQ reagents are non-isobaric variations of iTRAQ [22], (indeed the mTRAQ Δ4 reagent is chemically identical to iTRAQ-117) and are available in

three forms, Δ0, Δ4, and Δ8, which differ in mass from one another by 4.0071 Da, enabling quantification based on the isotope dilution principle [23]. In addition, since labeling occurs at the N-termini of peptides (as well as on lysine residues), any product ions resulting from Nterminal fragmentation (e.g., a, b, or c ions [24]) will be characteristic of each mTRAQ label and may also be used for quantitation. The detection of specific diagnostic product ions for posttranslational modification analysis may occur in real-time or post-acquisition. Real-time detection, such as the detection of the phosphite anion, m/z 78.96, utilized to locate phosphorylated peptides, has typically been performed on triple quadrupole instruments using precursor ion scanning [25]. Although this has proven to be effective for numerous modifications, triple quadrupole instruments are low resolution devices thus reducing the selectivity of diagnostic ion detection. Also, only one product ion can be monitored at any one time, resulting in either lower selectivity or multiple acquisitions of the same sample. Data may also be interrogated post-acquisition in either a data/information-dependent acquisition (D/IDA) [26, 27] or a data-independent acquisition (DIA) [28–30] experiment. A DDA workflow consists of one (or more) precursor ion survey scan(s) (MS acquisition) followed by a set of product ion scans (MS/MS acquisition) of the highest intensity precursor ions per cycle, typically of around 3 s. Modern QTOF and Orbitrap-based instruments are capable of selecting in excess of 20 precursors for tandem MS in such a cycle time. Data may then be interrogated by extracting product ions from the MS/MS acquisitions in order to pinpoint the desired modification(s). Although this approach may prove successful for the higher abundant modified peptides, the inherent under-sampling biases the analysis to those of higher abundance (or strictly speaking those peptides that give rise to a higher MS ion current). Preenrichment strategies may help to mitigate this effect; however, there will always be a proportion of peptides that go undetected. The concept of DIA was introduced in 2006 by Plumb et al., who used a Waters (Milford, MA, USA) Q-Tof Premier to acquire all fragment ion information from all potential precursor ions in a complex mixture simultaneously for the identification of endogenous metabolites in rat urine [28]. The approach, termed MSE, acquired data with the collision cell alternating between low energy and elevated (E) energy in rapid cycles. This MSE strategy was also successfully extended to the analysis of peptides [31]. Since all MS/MS data are acquired in a single shot, the elevated collision energy is ramped over the acquisition time in order to ensure all precursors are fragmented. This compromises the fragmentation efficiency for all ions since any precursor ion is only subjected to its optimal collision energy for a small proportion of the time. Recently, a new method of DIA was described by Gillet et al. in which a cycle consists of one full scan MS

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

acquisition followed by 32 consecutive 25 Da precursor isolation windows (swaths) on an AB SCIEX TripleTOF 5600 quadrupole time-of-flight mass spectrometer [30]. This SWATH approach offers the advantages of customizable window sizes, more focused/optimal collision energy settings, and far less complex product ion scans compared with the entire MS mass range of MSE. A chemical derivatization approach has been used to modify the phosphate groups of phosphoserine peptides, enabling improved detection by mass spectrometry [32]. Upon CID, diagnostic ions specific to the derivatized phosphoserine residues were generated. When the data were acquired under DIA conditions, a significant reduction in data complexity was observed. Here, we present a combination of mTRAQ derivatization with SWATH (DIA) acquisition on a TripleTOF 5600 QTOF for the selective detection of synthetic Ub and SUMO-derived isopeptides. The approach termed Mass spectral enhanced detection of Ubls using SWATH acquisition (MEDUSA) is also shown to be capable of providing relative quantification across three channels for both modifications simultaneously. Since the mTRAQ labels are non-isobaric, and labeling takes place at the N-termini of peptides, quantification is shown to be achievable in both precursor and product ion acquisitions.

Experimental Chemicals Synthetic isopeptides were obtained from Peptide Protein Research Ltd. (Fareham, UK) at 998 % purity and used without further purification (see Supplementary Table S1 for isopeptide sequences and precursor masses predicted, post-labeling). mTRAQ reagents were obtained from AB SCIEX (Foster City, CA, USA). Water (HPLC grade), trifluoroacetic acid (TFA, 99 %), formic acid (mass spectrometry grade), isopropanol, and tetra-ethyl ammonium bicarbonate (1 M, TEAB) were all obtained from Sigma Aldrich Ltd. (Poole, Dorset, UK). HPLC grade acetonitrile was purchased from Fisher Scientific (Loughborough, Leicestershire, UK). Lyophilized E. coli digest was purchased from Waters Corporation (Milford, MA, USA).

Sample Preparation Preparation of mTRAQ-Labeled Background Matrix Material A background matrix of mTRAQ-labeled E. coli-derived tryptic peptides was prepared by reconstituting 100 μg of lyophilized E. coli digest in 100 μL of 500 mM TEAB and dividing equally into 3 × 30 μL aliquots (30 μg per aliquot). Each tube was derivatized with mTRAQ reagent Δ0, Δ4, or Δ8 (AB SCIEX, Foster City, CA, USA) as described by the manufacturer. After allowing the reactions to proceed to completion for 1 h at room temperature, the samples were dried on a SpeedVac evaporator and

reconstituted in 10 μL 0.1 % TFA (giving a concentration of 1000 ngμL–1 of peptide for each label). A mixed sample for the assessment of background ion interference was prepared by mixing all three labeled E. coli digest samples in the ratio 1:1:1 (diluted in 0.1 % TFA), such that the final concentration of each component was 100 ngμL–1 (i.e., a 2 μL injection volume on column resulted in 600 ng total peptide loading (200 ng of each label). Preparation of mTRAQ-Labeled Synthetic Isopeptide Standards Each of the lyophilized synthetic isopeptide standards (in addition to the linear equivalent peptide) was reconstituted in a suitable volume of 50:50 acetonitrile:water to give a nominal concentration of 1 nmolμL–1. Standards were stored at –20 °C in 10 μL aliquots (10 nmol) prior to derivatization. Ten nmol (one aliquot) of each of the synthetic standards were combined and dried on a SpeedVac. The lyophilized isopeptide mix was next reconstituted in 100 μL of 500 mM TEAB and dividing equally into 3 × 30 μL aliquots (100 μg per aliquot). Each tube was derivatized with mTRAQ reagent Δ0, Δ4, or Δ8 (AB SCIEX, Foster City, CA USA) as described by the manufacturer. After allowing the reactions to proceed to completion for one hour at room temperature, the samples were dried on a SpeedVac evaporator and reconstituted in 200 μL 0.1 % TFA (giving a concentration of 10 pmol μL–1 for each labeled isopeptide). Two working mixtures were prepared in 0.1 % TFA in the ratio of 1:2:10 and 1:5:10 (Δ0, Δ4, and Δ8 respectively) in the presence of the background E. coli digest mix described above (200 ng of each label). The working mixture at ratio 1:2:10 is equivalent to an approximate on-column loading of 100 fmol, 200 fmol, and 1 pmol of each peptide and the ratio 1:5:10 is equivalent to an approximate on-column loading of 100 fmol, 500 fmol, and 1 pmol of each peptide for the labels Δ0, Δ4, and Δ8, respectively, based upon a 2 μL injection volume.

LC-MS/MS Analysis For all experiments, 2 μL injection volume was used. Samples were loaded onto an Eksigent cHiPLC C18 trap column (0.5 mm × 200 μm i.d., 3 μm ChromXP (Eksigent Technologies, Dublin, CA, USA) at a flow rate of 2.5 μLmin–1 for 20 min in loading buffer, 97.9 % water + 2 % acetonitrile + 0.1 % formic acid. The peptides were separated over a 55-min linear gradient from 5 % to 30 % Buffer B (acetonitrile + 0.1 % formic acid) at a flow rate of 300 nLmin–1 on an Eksigent Technologies cHiPLC C18 reversed phase analytical column (15 cm × 75 μm i.d., 3 μm). Buffer A consisted of 99.9 % water + 0.1 % formic acid. The column was maintained at a temperature of 60 °C. Mass analysis was carried out on an AB SCIEX TripleTOF 5600

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

Figure 1. Schematic representation of the MEDUSA workflow. Ub and SUMO diagnostic ions are generated when mTRAQlabeled isopeptides are subjected to collision-induced dissociation. SWATH DIA data are then interrogated for the presence of these ions, which are used for quantitation

quadrupole time-of-flight mass spectrometer (AB SCIEX, Foster City, CA, USA). The mass spectrometer was operated in a data independent acquisition (DIA) mode utilising SWATH acquisition. The SWATH acquisition cycle consisted of an initial TOF MS full scan for 100 ms followed by 80 × 40.4 ms MS/MS acquisitions with 6 Da precursor windows (1 Da overlap between adjacent windows). The resolution in MS/MS for these acquisitions was approximately 20,000. Total time for each full SWATH cycle was 3.4 s. Post-acquisition MS/MS data was manually interrogated for the presence of diagnostic a′ and b′ ions pertaining to the presence of a GG-, TGG-, or QTGGmodification tag using PEAKVIEW software ver. 1.1.0.1 (AB SCIEX, Foster City, CA, USA). A schematic representation of the workflow is presented in Figure 1.

Results and Discussion Generation of Diagnostic Isopeptide Ions Previously, it was demonstrated that upon collision induced dissociation, dimethyl labeled isopeptides release diagnostic a′ and b′ ions from the modification tags of GG- for ubiquitination and Human SUMO-1-ylation and TGG- or QTGG- for Human SUMO-2/3-ylation [14, 19], greatly aiding their identification. In order to determine if similar diagnostic ions result from mTRAQ labeling, we analyzed a set of labeled synthetic isopeptides using SWATH acquisition on a TripleTOF 5600 mass spectrometer, and interrogated the product ion spectra for the predicted a′ and b′ ions given in Table 1. An overlay of the extracted a′ and b′ ions for the mTRAQ Δ8-labeled isopeptides with the backbone sequence of

NSSYVLLKTGK, with K having GG-, TGG- and QTGGmodification tags covalently attached is presented in Figure 2. It can be seen from Figure 2a that for the isopeptide, NSSYVLLKGGTGK labeled with mTRAQ Δ8, three diagnostic ions overlay perfectly at an elution time of approximately 37.1 min. These traces correspond to the extracted ion chromatograms for a1′ (m/z 178.1430), b1′ (m/z 206.1379), and b2′ (m/z 263.1594) ions. Under the experimental conditions employed, the b2′ ion is the most abundant and, therefore, the ion of choice for further MS/MS-based quantitation (see later), although it may also be possible to quantify on other diagnostic ions. Figure 2b shows the extracted diagnostic ion chromatograms corresponding to the isopeptide, NSSYVLLKTGGTGK, eluting at approximately 37.4 min. The major ion at m/z 364.2070 corresponds to the b3′ ion and is used for subsequent MS/MS-based quantitation (see later), with the a1′ (m/z 222.1692) and b1′ (m/z 250.1641) showing perfect co-elution. Figure 2c depicts a similar overlay of the extracted ion chromatograms for the diagnostic ions for the isopeptide, NSSYVLLKQTGGTGK. In this case, the major ion at m/z 492.2656 corresponds to the b4′ ion with the a1′ (m/z 249.1801), b1′ (m/z 277.1750), and b3′ (m/z 435.2442) again showing perfect co-elution at approximately 37.1 min. Similar results were obtained for Δ0 and Δ4 mTRAQ-labeled versions of the same isopeptides (data not shown). In addition, all other isopeptides were examined and found to release the same diagnostic ions as determined a priori, suggesting that the use of mTRAQ (all three labels) would be a suitable choice as a labeling reagent for isopeptides. It should be noted that compared with reductive methylation, more diagnostic ions are generated with mTRAQ

Table 1. Predicted a′ and b′ Diagnostic Product ions for mTRAQ-Labeled Isopeptides

GG-tag; Δ0 GG-tag; Δ4 GG-tag; Δ8 TGG-tag; Δ0 TGG-tag; Δ4 TGG-tag; Δ8 QTGG-tag; Δ0 QTGG-tag; Δ4 QTGG-tag; Δ8

a1′

b1′

b2′

b3′

b4′

170.1288 174.1359 178.1430 214.1550 218.1621 222.1692 241.1659 245.1730 249.1801

198.1237 202.1308 206.1379 242.1499 246.1570 250.1641 269.1608 273.1679 277.1750

255.1452 259.1523 263.1594 299.1714 303.1785 307.1856 370.2085 374.2156 378.2227

356.1928 360.1999 364.2070 427.2300 431.2371 435.2442

484.2514 488.2585 492.2656

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

spectrometers. In addition, all three mTRAQ variants show excellent diagnostic ions enabling three-plex analysis.

Specificity of Diagnostic Ions in a Complex Sample of E. coli-Derived Linear Peptides

Figure 2. Post-acquisition extracted diagnostic a′ and b′ ions generated during SWATH acquisition of the synthetic isopeptides with the backbone sequence; NSSYVLLKTGK, after labeling with mTRAQ Δ8. (a) GG-tag with a1′ = 178.1430 (green), b1′ = 206.1379 (red), and b2′ = 263.1594 (orange); (b) TGG-tag with a1′ = 222.1692 (green), b1′ = 277.1750 (red), and b3′ = 364.2070 (blue); and (c) QTGG-tag with a1′ = 249.1801 (green), b1′ = 277.1750 (red), b3′ = 435.2442 (blue), and b4′ = 492.2656 (gray)

labeling, which results in a greater level of selectivity (see below). The masses of the diagnostic ions are at higher m/z values compared with those derived from dimethyl labeling, which would prove advantageous on instrumentation with poor low mass ion transmission or, potentially, on ion trap mass

In order for MEDUSA to have utility in successfully identifying modified peptides, the diagnostic ions generated must be sufficiently selective for the analytes of interest (i.e., the ions must not be present in non-modified peptides to any appreciable level. We chose to test this by taking a complex sample consisting of a mixture of 200 ng of each mTRAQlabeled E. coli digest (containing no known SUMOylated or ubiquitinated isopeptides) and subjecting it to the MEDUSA approach. Each SWATH window was systematically manually interrogated for all diagnostic ions. The results of this are summarized in Supplementary Table S2. For a match to be considered as positive, the following selection criteria were applied. First, the major diagnostic ion has to be the b2′ for GG-, the b3′ for TGG-, and the b4′ for QTGG-. These ions must have an intensity of at least 1000 counts without smoothing or baseline subtraction. A GG-tag isopeptide is only deemed to be detected if both the a1′ and b1′ ions are also present and perfectly co-elute with the b2′. Similarly, a putative positive identification of a TGG-tag isopeptide requires the presence of both a1′ and b1′ ions and perfect coelution with the b3′. Finally, a QTGG-tag isopeptide is considered to be potentially modified if the a1′, b1′, and b3′ ions are all shown to perfectly co-elute with the b4′. Figure 3a shows the diagnostic ion extracted ion chromatogram overlay (a1′- green, b1′- red, and b2′- orange, for GG-tag Δ8) for the SWATH precursor window; 569– 575 Da. At a retention time of 51.37 min, a putative isopeptide containing a GG-tag has clearly been detected. Upon inspection of the corresponding MS/MS spectrum (Figure 3b) the diagnostic ions for a GG-tag are clearly seen (inset) indicating the presence of an isopeptide or a GG-Nterminating linear peptide. The MS/MS spectrum generated is too complex for de novo sequencing in this instance; however, a subsequent analysis of 200 ng of just the Δ8labeled E. coli allowed successful identification of the peptide with the sequence being GGVIPGEYIPAVDK. Since this peptide contains an N-terminal diglycine, the a1, b1, and b2 ions are identical to the GG-tag isopeptide diagnostic ions. However, as only around 0.5 % of tryptic peptides from both human and E. coli proteomes Nterminate in GG [14], this does not represent a significant problem. Indeed, the ability of MEDUSA to data reduce a complex dataset some 200-fold for putative GG isotag bearing isopeptides is one of its key strengths. Examination of all SWATH data for the mixed E. coli sample resulted in the identification of only 37 precursor ions, which are deemed to be putative Ub/SUMO positives (23 GG-, 11 TGG-, and 3 QTGG-); however, several of these are the same peptide with different mTRAQ labels and present in more than one charge state (Supplementary Table S2).

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

Figure 3. (a) Overlay of extracted ion chromatograms of an E. coli digest for the three GG-tag diagnostic ions of mTRAQ Δ8, in SWATH window 569–575 Da. Inset shows perfect chromatographic overlay at a retention time of 51.37 min (a1′-green, b1′-red, and b2′-orange). (b) MS/MS spectrum at 51.37 min for precursor ions in SWATH window 569–575 Da with inset highlighting the region containing diagnostic ions

Examination of a data-dependent analysis of m-TRAQ-labeled E. coli digest at the same time detected approximately 5200 multiply charged (+2, +3, and +4) precursors, which equates to a false positive rate for GG-tag of 0.44 %. This huge data reduction from over 5000 unscreened E. coli peptides to less than 23 is entirely consistent with the theoretical presence of a small percentage carrying an N-terminal diglycine.

Detection of mTRAQ-Labeled Isopeptides in the Presence of a Background of E. coli-Derived Linear Peptides After demonstrating that mTRAQ-labeled isopeptides (1) release diagnostic a′ and b′ ions upon collision-induced dissociation, and (2) these ions are not subject to appreciable

interference from non-modified peptides at the levels used in these model experiments, we next determined if we could detect a set of synthetic isopeptides, which were spiked into a complex background. In addition, an isomeric linear peptide (GGLIFAGKQLEDGR) was added to the isopeptides (containing the isomeric isopeptide; LIFAGKGGQLEDGR) to assess a worst-case scenario in terms of misleading GGtag detection. Figure 4a shows the overlaid extracted diagnostic ion chromatograms for SWATH window 649–655 Da, [which contains the isopeptide VFAGKGGQLEDFLK (+3 charge state, m/z = 651.7176)] for the diagnostic ions of a1′ (m/z 178.1430), b1′ (m/z 206.1379), and b2′ (m/z 263.1594). Several peaks are shown throughout the chromatographic profile; however, it is clearly evident that only one peak contains

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

other isopeptides with GG-, TGG-, and QTGG-tags upon extraction of the diagnostic ions. When the TOF-MS spectrum for this SWATH window at 56.92 min is interrogated, only one potential precursor is present with m/z = 651.7191 and +3 charge state (Figure 4b). The MS/MS spectrum associated with this precursor is amenable to de novo sequencing, and it was found to result from the predicted isopeptide, VFAGK GG QLEDFLK (Figure 4c). Next, we determined whether or not it was possible to distinguish two isomers, one of which was an isopeptide, using the MEDUSA approach. The isopeptide, LIFAGKGGQLEDGR and its linear isomer, GGLIFAGKQLEDGR, were present in the test mixture at nominally equal amounts. The precursor ions for all three labels (Δ0-red, Δ4-green, Δ8-blue) were first extracted from the TOF-MS, and two closely eluting peaks were observed at retention times of 45.06 and 45.92 min (Figure 5a). Examination of the two MS/MS spectra generated at these two retention times (from the precursor SWATH window 584–590 Da corresponding to the Δ8-label) showed that both species contained the diagnostic ions for a GG-tag (Figure 5b and c). Figure 5b represents the MS/MS spectrum acquired at 45.06 min over the m/z range 150–400 Da. This spectrum clearly contains both a b2 ion (375.2834 Da) from the peptide backbone and a b2′ ion (263.1596) from the GG-tag. The corresponding a and a′ ions are also observed. The spectrum acquired at 45.92 min does not contain these pairs of a/a′ and b/b′ ions (Figure 5c). In this case, the ions at the same m/z values as the GG-tag diagnostic ions correspond to the backbone ions as confirmed by the b3 ion at m/z 376.2444. Even in this worst-case scenario of a GG-terminating peptide, a false positive identification is not produced because of the high quality b-ion generation post-mTRAQ labeling.

MS and MS/MS-Based Quantitation of Isopeptides

Figure 4. (a) Overlay of extracted diagnostic ion (GG-tag, Δ8) chromatograms in SWATH window 649–655 Da. Inset shows perfect chromatographic overlay of the three GG-tag diagnostic ions corresponding to isopeptide; VFAGKGGQLEDFLK (+3); RT = 56.92 min; m/z 651.7176. (b) MS spectrum at 56.90 min in the range 649–655 Da. (c) MS/MS spectrum for SWATH window 649–655 Da at RT = 56.92 min

all diagnostic ions which co-elute perfectly, the major ion being the b2′ ion (see inset). Similar situations regarding the selectivity of the ions were found for all

In order to assess the quantitation capabilities of MEDUSA, two test solutions were prepared, which contained synthetic isopeptides spiked at a ratio of 1:2:10 and 1:5:10 (Δ0:Δ4:Δ8), respectively, into a background of E. coli digest peptides at 1:1:1 (Δ0:Δ4:Δ8). These samples were loaded at 600 ng of E. coli digest (200 ng of each label) containing 100:200:1000 fmol or 100:500:1000 fmol of the labeled synthetic isopeptides on-column. An example of the quantitation capabilities of the MEDUSA method is shown in Figure 6. MS-based quantitation was performed by extracting the precursor ion mono-isotopic peak and integrating using PeakView software. Figure 6a shows an overlay of the extracted ions for EGVKGGTENNDHINLK (+4) with m/z 537.0408 (Δ0), 540.0461 (Δ4), and 543.0514 (Δ8). The calculated peak areas for these were 115410, 615362, and 1231118, respectively, which is equivalent to a ratio of 1:5.3:10.7. The same isopeptide was also quantified in MS/MS using the

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

Figure 5. (a) TOF-MS extracted ion chromatogram overlay showing elution profiles for mTRAQ-labeled isopeptide, LIFAGKGGQLEDGR (+3); RT = 45.06 min, and the isomeric mTRAQ-labeled linear peptide, GGLIFAGKQLEDGR (+3); RT = 45.92 min. Blue trace corresponds to 1 pmol loading (mTRAQ Δ8), green trace corresponds to 200 fmol loading (mTRAQ Δ4), and red trace corresponds to 100 fmol loading (mTRAQ Δ4). (b) MS/MS spectrum for isopeptide LIFAGKGGQLEDGR showing important a/a′ and b/b′ ions. (c) MS/MS spectrum for linear peptide GGLIFAGKQLEDGR showing important a and b ions

dominant b2′ diagnostic ion [m/z 255.1452 (Δ0), 259.1523 (Δ4), and 263.1594 (Δ8)] (Figure 6b). The calculated peak areas for these were 3096, 20581, and 58319, respectively, which is equivalent to a ratio of 1:6.6:18.8. These data suggest that quantification in the MS mode is more accurate than in

MS/MS; however, the low intensity and poor peak shape of the Δ0, b2′ ion may be distorting this to some extent since the calculated ratio of Δ4:Δ8 is 1:2.8, which is close to the expected ratio of 1:2. Both approaches do show a clear difference between all three spiked levels of this synthetic

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

Figure 6. MS and MS/MS-based quantitation of isopeptides with the backbone sequence, EGVKTENNDHINLK, based upon extracted ion peak areas. Red traces correspond to Δ0-tagged (100 fmol), green traces relate to Δ4-tagged (500 fmol), and blue traces represent Δ8-tagged (1 pmol) isopeptides

ubiquitination-derived isopeptide (GG-tagged) and the data from MS and MS/MS analysis support one another. The same approach was also applied to TGG- and QTGG-tagged derivatives of the same backbone peptide (i.e., EGVKTGGTENNDHINLK (+4) and EGVKQTGGTENNDHINLK (+4). Figure 6c shows an overlay of the extracted ions for EGVKTGGTENNDHINLK (+4) with m/z 562.3027 (Δ0), 565.3080 (Δ4), and 568.3133 (Δ8). The calculated peak areas for these were 114100, 522100, and 1391700, respectively, which is equivalent to a ratio of 1:4.6:12.2. The same isopeptide was also quantified in MS/MS using the dominant b3′ diagnostic ion [m/z 356.1928 (Δ0), 360.1999 (Δ4), and 364.2070 (Δ8)] (Figure 6d). The calculated peak areas for these were 5074, 33979, and 82401, respectively, which is equivalent to a ratio of 1:6.7:16.2. Finally, for the same peptide backbone sequence, the QTGG-tagged isopeptide, EGVKQTGGTENNDHINLK (+4), was analyzed in the same way. Figure 6e shows an overlay of the extracted ions for EGVKTGGTENNDHINLK (+4)

with m/z 594.3173 (Δ0), 597.3226 (Δ4), and 600.3279 (Δ8). The calculated peak areas for these were 132730, 716690, and 1672200, respectively, which is equivalent to a ratio of 1:5.4:12.6. The same isopeptide was also quantified in MS/MS using the dominant b4′ diagnostic ion [m/z 484.2514 (Δ0), 488.2585 (Δ4), and 492.2656 (Δ8)] (Figure 6f). The calculated peak areas for these were 5583, 33139, and 70559, respectively, which is equivalent to a ratio of 1:5.9:12.6. All of the other synthetic isopeptides were subjected to the same quantitative assessment and the data are summarized in Table 2. A complete set of peak area data is included in Supplementary Tables S3 and S4. The data demonstrate that by using either MS precursor ion peak areas, or the most intense diagnostic product ion peak areas, quantitation is possible across three channels. Also, there is, in general, good agreement between the theoretical ratios and those calculated. Both Ub-derived isopeptides and SUMO-derived isopeptides are amenable to detection and quantitation by this approach. Taken together, the points above lead us to propose that a MEDUSA-driven

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

Table 2. Quantitation Ratios for MS and MS/MS Acquisitions of 1:2:10 and 1:5:10 Spikes Peptide GG-tag VFAGKQLEDFLK NSSYVLLKTGK YDFFILNKTGK FGAWYKTTVYR LIFAGKQLEDGR EGVKTENNDHINLK TGG-tag VFAGKQLEDFLK NSSYVLLKTGK YDFFILNKTGK FGAWYKTTVYR LIFAGKQLEDGR EGVKTENNDHINLK QTGG-tag VFAGKQLEDFLK NSSYVLLKTGK YDFFILNKTGK FGAWYKTTVYR LIFAGKQLEDGR EGVKTENNDHINLK LINEAR ISOMER GGLIFAGKQLEDGR

MS ratio (1:2:10)

MS/MS ratio (1:2:10)

MS ratio (1:5:10)

MS/MS ratio (1:5:10

1:2:14 1:2:8 1:4:13 1:3:11 1:2:13 1:2:10

1:1:7 * 1:3:11 1:2:7 1:1:10 1:2:12

* 1:6:11 1:7:11 1:8:14 1:6:15 1:5:11

1:5:10 1:2:6 1:11:15 1:7:12 1:5:11 1:7:19

1:2:10 1:2:8 1:3:10 1:2:9 1:2:14 1:1:10

1:2:8 1:2:8 1:1:8 1:2:6 1:2:9 1:2:13

1:6:13 1:5:8 1:5:8 1:6:10 1:7:19 1:5:12

1:5:8 1:3:8 1:4:9 1:6:8 1:4:10 1:7:16

1:2:11 1:2:10 1:3:12 1:2:10 1:2:12 1:3:19

1:2:8 1:2:6 1:3:12 1:2:9 1:2:11 1:3:6

1:8:15 1:6:12 1:8:14 1:5:10 1:7:15 1:5:13

1:5:8 1:5:6 1:10:15 1:6:10 1:5:10 1:6:13

1:2:11

1:2:10

1:7:15

1:6:11

* Unable to integrate peak due to poor peak shape

workflow for the simultaneous detection and quantification of both Ub and SUMO modifications would consist of: 1. SWATH acquisition of MS/MS data using 6 Da window size to capture all product ions generated upon CID in an unbiased fashion (DIA); 2. Post-acquisition diagnostic ion extraction to selectively detect potential modified isopeptides; 3. Integration of major diagnostic b′ ion to filter for differentially expressed isopeptides detected in 2. above; 4. Interrogate MS data at retention times of the differentially expressed putative isopeptides within the SWATH window m/z range of the diagnostics to obtain accurate precursor ion m/z and charge state. There should be perfect retention time overlay for all product ions associated with a specific precursor ion; 5. Interrogate MS/MS spectrum containing the diagnostic ions for backbone b and y ions and use in combination with precursor m/z determined from 4. above. Note: this stage is only required for analytes deemed to be putative isopeptides AND which are differentially expressed since non-changing isopeptides are of no interest (i.e., in a quantitative results-driven approach [33].

Conclusions A novel strategy for the simultaneous detection and quantification of Ub- and SUMO-derived isopeptides using mass spectrometry has been described. Using this approach, which we have termed Mass spectral Enhanced Detection of Ubls using SWATH Acquisition (MEDUSA), we have demonstrated an impressive data reduction from a total of 5200 MS precursors

(of charge states +2, +3, and +4) from an E. coli digest sample to only 23 detected GG-N-terminating peptides (0.44 %), which is entirely consistent with the in silico predicted percentage of 0.5 % for the proteome [14]. The method takes full advantage of the fact that upon linear collision cell CID fragmentation, m-TRAQ-labeled isopeptides release multiple modification-specific diagnostic ions from the isotag region: GG-tag a1′, b1′, and b2′; TGGtag a1′, b1′, and b3′, and QTGG-tag a1′, b1′, b3′, and b4′. These ions offer more diagnostic value than similar ions generated using reductive methylation (RUbI), and the technique was shown to be excellent at detecting a suite of synthetic isopeptides, designed to simulate the proteolytic products of ubiquitinated and SUMOylated proteins, when present in a complex background of E. coli digest peptides. The use of SWATH data-independent acquisition results in the data capture of virtually all product ions from all detectable precursors in a single scan cycle, thus enabling post-acquisition ion extraction of any product ions formed. Modification-specific diagnostic ions may be extracted and matched to the appropriate precursor and MS/MS sequence ions for identification of differentially expressed isopeptides. Crucially, MEDUSA allows relative quantitation to be performed in a triplex experiment, with quantitation readily available in both MS, using precursor ion abundance, and MS/MS, using diagnostic ion abundance, spectral domains. In effect, the method enables multiple intra-analysis modes of confirming relative quantitation. Since the charge states of mTRAQ-labeled isopeptide precursors are predominantly greater than +2, and the diagnostic ions are at reasonably large m/z values (i.e., greater than the low mass cut-off in many instances), we envisage that this approach is likely to have utility in instruments using ion trap CID [34].

J. R. Griffiths et al.: Ubl Modification Analysis Using DIA and mTRAQ

Acknowledgments N.C. thanks the EPSRC for financial support. Y.C, J.G, and D.S were funded by Cancer Research, UK. The authors thank Dr. Christie Hunter for valuable advice on using SWATH acquisition, and Craig Mageean for assistance with figure generation.

References 1. Finley, D., Chau, V.: Ubiquitination. Annu. Rev. Cell Biol. 7, 25–69 (1991) 2. Hay, R.T.: SUMO: a history of modification. Mol. Cell 18, 1–12 (2005) 3. Mukhopadhyay, D., Riezman, H.: Proteasome-independent functions of ubiquitin in endocytosis and signaling. Science 315, 201–205 (2007) 4. Bergink, S., Jentsch, S.: Principles of ubiquitin and SUMO modifications in DNA repair. Nature 458, 461–467 (2009) 5. Hochstrasser, M.: Origin and function of ubiquitin-like proteins. Nature 458, 422–429 (2009) 6. Baek, S.H.: A novel link between SUMO modification and cancer metastasis. Cell Cycle 5, 1492–1495 (2006) 7. Sun, Y.: E3 ubiquitin ligases as cancer targets and biomarkers. Neoplasia 8, 645–654 (2006) 8. Denuc, A., Marfany, G.: SUMO and ubiquitin paths converge. Biochem. Soc. Trans. 38, 34–39 (2010) 9. Hunter, T., Sun, H.: Crosstalk between the SUMO and ubiquitin pathways. Ernst Schering Found. Symp. Proc. 1, 1–16 (2008) 10. Yang, X.J.: Multisite protein modification and intracellular signaling. Oncogene 24, 1653–1662 (2005) 11. Peng, J., Schwartz, D., Elias, J.E., Thoreen, C.C., Cheng, D., Marsischky, G., Roelofs, J., Finley, D., Gygi, S.P.: A proteomics approach to understanding protein ubiquitination. Nat. Biotechnol. 8, 921–926 (2003) 12. Nielsen, M.L., Michiel Vermeulen, M., Tiziana Bonaldi, T., Cox, J., Moroder, L., Mann, M.: Iodoacetamide-induced artifact mimics ubiquitination in mass spectrometry. Nat. Methods 5, 459–460 (2008) 13. Warren, M.R.E., Parker, C.E., Mocanu, V., Klapper, D., Borchers, C.E.: Electrospray ionization tandem mass spectrometry of model peptides reveals diagnostic product ions for protein ubiquitination. Rapid Commun. Mass Spectrom. 19, 429–437 (2005) 14. Chicooree, N., Connolly, Y., Tan, C.-T., Malliri, A., Li, Y., Smith, D.L., Griffiths, J.R.: Enhanced detection of ubiquitin isopeptides using reductive methylation. J. Am. Soc. Mass Spectrom. 24, 421–430 (2013) 15. Hsu, J.L., Huang, S.Y., Chow, N.H., Chen, S.H.: Stable isotope dimethyl labeling for quantitative proteomics. Anal. Chem. 75, 6843– 6852 (2003) 16. Lamoliatte, F., Bonneil, E., Durette, C., Caron-Lizotte, O., Wildemann, D., Zerweck, J., Wenshuk, H., Thibault, P.: Targeted identification of SUMOylation sites in human proteins using affinity enrichment and Paralog-specific reporter ions. Mol. Cell Proteomics 12, 2536–2550 (2013) 17. Cooper, H.J., Tatham, M.H., Jaffray, E., Heath, J.K., Lam, T.T., Marshall, A.G., Hay, R.T.: Fourier transform ion cyclotron resonance mass spectrometry for the analysis of small ubiquitin-like modifier (SUMO) modification: identification of lysines in RanBP2 and SUMO targeted for modification during the E3 AutoSUMOylation reaction. Anal. Chem. 77, 6310–6319 (2005)

18. Chicooree, N., Griffiths, J.R., Connolly, Y., Tan, C.-T., Malliri, A., Eyers, C.A., Smith, D.L.: A novel approach to the analysis of SUMOylation with the independent use of trypsin and elastase digestion followed by database searching utilising consecutive residue addition to lysine. Rapid Commun. Mass Spectrom. 27, 127–134 (2013) 19. Chicooree, N., Griffiths, J.R., Connolly, Y., Smith, D.L.: Chemically facilitating the generation of diagnostic ions from SUMO(2/3) remnant isopeptides. Rapid Commun. Mass Spectrom. 27, 2108–2114 (2013) 20. Boersema, P.J., Raijmakers, R., Lemeer, S., Mohammed, S., Heck, A.J.: Multiplex peptide stable isotope dimethyl labeling for quantitative proteomics. Nat. Protoc. 4, 484–494 (2009) 21. DeSouza, L.V., Taylor, A.M., Li, W., Minkoff, M.S., Romaschin, A.D., Colgan, T.J., Siu, K.W.M.: Multiple reaction monitoring of mTRAQlabeled peptides enables absolute quantification of endogenous levels of a potential cancer marker in cancerous and normal endometrial tissues. J. Proteome Res. 7, 3525–3534 (2008) 22. Ross, P.L., Huang, Y.N., Marchese, J.N., Williamson, B., Parker, K., Hattan, S., Khainovski, N., Pillai, S., Dey, S., Daniels, S., Purkayastha, S., Juhasz, P., Martin, S., Bartlet-Jones, M., He, F., Jacobson, A., Pappin, D.J.: Multiplexed protein quantitation in Saccharomyces cerevisiae using amine-reactive isobaric tagging reagents. Mol. Cell Proteomics 3, 1154–1169 (2004) 23. Heumann, K.G.: Isotope dilution mass spectrometry. Int. J. Mass Spectrom. Ion Process. 118/119, 575–592 (1992) 24. Roepstorff, P., Fohlman, J.: Proposal for a common nomenclature for sequence ions in mass spectra of peptides. Biomed. Mass Spectrom. 11, 601 (1984) 25. Carr, S.A., Huddleston, M.J., Annan, R.S.: Selective detection and sequencing of phosphopeptides at the femtomole level by mass spectrometry. Anal. Biochem. 239, 180–192 (1996) 26. Mann, M., Hendrickson, R.C., Pandey, A.: Analysis of proteins and proteomes by mass spectrometry. Annu. Rev. Biochem. 70, 437–473 (2001) 27. Domon, B., Aebersold, R.: Mass spectrometry and protein analysis. Science 312, 212–217 (2006) 28. Plumb, R.S., Johnson, K.A., Rainville, P., Smith, B.W., Wilson, I.D., Castro-Perez, J.M., Nicholson, J.K.: UPLC/MSE; a new approach for generating molecular fragment information for biomarker structure elucidation. Rapid Commun. Mass Spectrom. 20, 1989–1994 (2006) 29. Geiger, T., Cox, J., Mann, M.: Proteomics on an Orbitrap benchtop mass spectrometer using all-ion fragmentation. Mol. Cell Proteomics 9, 2252–2261 (2010) 30. Gillet, L.C., Navarro, P., Tate, S., Rost, H., Selevsek, N., Reiter, L., Bonner, R., Aebersold, R.: Targeted data extraction of the MS/MS spectra generated by data-independent acquisition: a new concept for consistent and accurate proteome analysis. Mol. Cell Proteomics 11, (2012). doi:10.1074/mcp.0111.016717 31. Silva, J.C., Denny, R., Dorschel, C., Gorenstein, M.V., Li, G.Z., Richardson, K., Wall, D., Geromanos, S.J.: Simultaneous qualitative and quantitative analysis of the Escherichia coli proteome: a sweet tale. Mol. Cell Proteomics 5, 589–607 (2006) 32. Shi, Y., Bajrami, B., Morton, M., Yao, X.: Cyclophosphoramidate ion as mass defect marker for efficient detection of protein serine phosphorylation. Anal. Chem. 80, 7614–7623 (2008) 33. Zappacosta, F., Collingwood, T.S., Huddlestone, M.J., Annan, R.S.: A quantitative results-driven approach to analyzing multisite protein phosphorylation. Mol. Cell Proteomics 5, 2019–2030 (2006) 34. March, R.E.: An introduction to quadrupole ion trap mass spectrometry. J. Mass Spectrom. 32, 351–369 (1997)