Anal. Chem. 2003, 75, 3107-3114
Detection of Arginine Dimethylated Peptides by Parallel Precursor Ion Scanning Mass Spectrometry in Positive Ion Mode Juri Rappsilber,† Westley J. Friesen,‡ Sergey Paushkin,‡ Gideon Dreyfuss,‡ and Matthias Mann*,†
Center for Experimental Bioinformatics and Department of Biochemistry & Molecular Biology, University of Southern Denmark, DK-5230 Odense M, Denmark, and Howard Hughes Medical Institute and Department of Biochemistry & Biophysics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104-6148.
Dimethylation at arginine residues has been shown to be central in cellular processes such as signal transduction, transcription activation, and protein sorting. The two methyl groups are either placed symmetric or asymmetric on the ζ standing nitrogen atoms of the arginine side chain. Here, we introduce a novel method that enables the localization of dimethylarginine (DMA) residues in gelseparated proteins at a level of sensitivity of better than 1 pmol and that allows one to distinguish between the isomeric symmetric and asymmetric position of the methyl groups. The method utilizes two side-chain fragments of DMA, the dimethylammonium ion (m/z 46.06) and the dimethylcarbodiimidium ion (m/z 71.06), for positive ion mode precursor ion scanning. Dimethylcarbodiimidium ions (m/z 71.06) are produced by symmetric as well as asymmetric dimethylarginine but are observed more strongly for symmetric DMA. It is utilized here in the precursor of m/z 71 scan to indicate the presence of DMA in a peptide. The dimethylammonium ion (m/z 46.06) is specific for asymmetric DMA and is utilized here in the precursor of m/z 46 scan. The positive ion mode allows for the identification of the protein by peptide sequencing and simultaneous detection and localization of the modified residues. The analysis can be conducted on any mass spectrometer capable of precursor ion scanning. However, the high resolution of a quadrupole TOF instrument is beneficial to assign the accurate charge state of the often highly charged precursors. Using the precursor of m/z 71 scan, we found FUS/TLS and Sam68 to be DMAcontaining proteins. We discovered at least 20 DMA sites in FUS/TLS. In MS/MS, we observed neutral loss of dimethylamine (m/z 45.05) from which it follows that the dimethylation in FUS/TLS is asymmetric. Monitoring in parallel the fragments m/z 46.06 and 71.06 in precursor ion scans and peptide sequencing, we identified at least nine asymmetric DMA modifications in Sam68. The parallel monitoring of fragments in precursor ion scans * Corresponding author. Tel: +45 6550 2364. Fax: +45 6593 3929. E-mail:
[email protected]. † University of Southern Denmark. ‡ University of Pennsylvania School of Medicine. 10.1021/ac026283q CCC: $25.00 Published on Web 06/03/2003
© 2003 American Chemical Society
is a versatile tool to specify the nature of protein modifications in cases where a single fragment is not conclusive. Posttranslational modifications modulate or regulate the function of proteins. As the analysis of proteins becomes more routine, the importance of modifications becomes increasingly apparent. Methylation of arginine residues exemplifies this fact as was discovered more than 30 years ago,1 while only recent work has placed it in the functional context of signal transduction, transcription activation, and protein sorting.2 The addition of methyl groups onto the side chain of arginine increases its size and removes hydrogen atoms that might be involved in hydrogen bonds. As many RNA-binding proteins are shown to contain this modification3 it is worth noting that arginine hydrogens are central in the binding of arginine fork motifs to RNA.4 The methylation also affects packing of the side chain into binding pockets and is thought to thereby modulate protein folding and protein-protein interactions.3 The diverse functions of arginine-methylated substrates has led to the suggestion that this specifically eukaryotic modification may parallel phosphorylation in its level of complexity.2 Two forms of arginine dimethylation have been described: NGNG-dimethylarginine (asymmetric dimethylarginine, aDMA) and NGN′G-dimethylarginine (symmetric dimethylarginine, sDMA) (Figure 1A). Asymmetric DMA was found in histones, which play a role in DNA packing and transcription control, in heterogeneous nuclear ribonucleoproteins (hnRNPs), which play a role in mRNA processing and nucleocytoplasmic RNA transport, and a number of other RNA-binding proteins. Symmetric DMA was found in myelin basic protein and the spliceosomal snRNP proteins SmD1 and SmD3. Arginine dimethylation occurs in RG motifs, even though there are exceptions to this rule.5,6 Protein arginine N-methyltransferases (PRMTs) catalyze the posttranslational transfer of methyl groups from S-adenosyl-L-methionine (SAM) (1) Paik, W. K.; Kim, S. J. Biol. Chem. 1968, 243, 2108-14. (2) McBride, A. E.; Silver, P. A. Cell 2001, 106, 5-8. (3) Gary, J. D.; Clarke, S. Prog. Nucleic Acid Res. Mol. Biol. 1998, 61, 65-131. (4) Calnan, B. J.; Tidor, B.; Biancalana, S.; Hudson, D.; Frankel, A. D. Science 1991, 252, 1167-71. (5) Smith, J. J.; Rucknagel, K. P.; Schierhorn, A.; Tang, J.; Nemeth, A.; Linder, M.; Herschman, H. R.; Wahle, E. J. Biol. Chem. 1999, 274, 13229-34. (6) Mowen, K. A.; Tang, J.; Zhu, W.; Schurter, B. T.; Shuai, K.; Herschman, H. R.; David, M. Cell 2001, 104, 731-41.
Analytical Chemistry, Vol. 75, No. 13, July 1, 2003 3107
Figure 1. Mass spectrometric analysis of synthetic dimethylargininecontaining peptides. (A) Chemical structure of symmetric and asymmetric dimethylarginine. (B) Fragmentation spectrum of unmodified, symmetrically and asymmetrically arginine-dimethylated TWRGGEEK with the stable dimethylated arginine residue sequenced by the singly and doubly charged y5- and y6-ions (shaded triangle). Inset shows the low-m/z regions in which the unmodified, symmetrically, or asymmetrically dimethylated peptide differs in its fragmentation behavior. (C) Suggested chemical structure of the m/z 46 and 71 fragments.
to arginine residues.7 Type 1 PRMTs catalyze the formation of aDMA, while type 2 PRMTs catalyze the formation of sDMA. This suggests different functions of the two isomers as is also evidenced by the fact that so far they have not been observed to occur in the same protein. Both enzyme types can also form NG-monomethylarginine (MMA), presumably as an intermediate on the way to the dimethylated residue. On the basis of the regulative function of arginine dimethylation it has been proposed that this modification may be reversible and resemble protein phosphorylation as a molecular switch.2 Traditionally, aDMA and sDMA are identified after total hydrolysis of proteins.1 This allows us to distinguish between the (7) Lee, H. W.; Kim, S.; Paik, W. K. Biochemistry 1977, 16, 78-85.
3108 Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
two isomers but does not yield their position in the protein sequence. Modified arginine residues can be localized to individual peptides of a proteolytic digest using MALDI time-of-flight mass spectrometry.5,8 The mass shift introduced by the methylation is used as an indication to find peptides that potentially contain modified arginines. However, the identity of a peptide is not established to certainty based on the mass measurement alone. Furthermore, the position of the modified residue is not conclusively assigned in the case of several potential modification sites. Finally, the two arginine-dimethylated isoforms are isobaric and can therefore not be differentiated by peptide mass measurements. These problems can in principle be solved in MALDI by PSD8 or if the proteolytic digest is separated by HPLC and the dimethylarginine (DMA)-containing fractions are subjected to Edman degradation.9 The fractions containing putatively modified peptides are selected by comparison to the peptides of the in vitro expressed, unmethylated protein. Alternatively, the protein can be methylated in an in vitro assay using [3H]-SAM as methyl donor and radioactive fractions are then selected for sequencing. The location of the modified residue can be determined in this way, and aDMA and sDMA are differentiated. HPLC coupled to Edman requires large quantities of material and pure peptide fractions if compared to a purely mass spectrometric method, however. Therefore, it would be desirable to develop a method that would utilize mass spectrometry and allow differentiation of the two forms of arginine dimethylation as well as allow their localization in the protein sequence. Precursor ion scanning already allows the detection of a number of protein modifications: phosphorylation on serine, threonine, and tyrosine,11,12 sulfation on tyrosine,13 glycosylation on asparagine and serine,14,15 acetylation on lysine and arginine,16 the hydroxynonenal adduct of lysine,17 and palmitylation.13 It is advantageous to conduct precursor ion scanning experiments in the positive ion mode as this allows the direct selection of the candidate peptides for sequencing. Recently, this approach has been used to analyze tyrosine phosphorylation,18 tyrosine nitration19 and bromotryptophan and hydroxyproline modifications.20 It was furthermore shown that precursor ion scanning allows differentiating tyrosine phosphorylation from the isobaric tyrosine sulfation.21 (8) Kim, S.; Merrill, B. M.; Rajpurohit, R.; Kumar, A.; Stone, K. L.; Papov, V. V.; Schneiders, J. M.; Szer, W.; Wilson, S. H.; Paik, W. K.; Williams, K. R. Biochemistry 1997, 36, 5185-92. (9) Brostoff, S.; Eylar, E. H. Proc. Natl. Acad. Sci. U.S.A. 1971, 68, 765-9. (10) Bertram, J. S.; Peterson, A. R.; Heidelberger, C. In Vitro 1975, 11, 97106. (11) Huddleston, M. J.; Annan, R. S.; Bean, M. F.; Carr, S. A. J. Am. Soc. Mass Spectrom. 1993, 4, 710-7. (12) Neubauer, G.; Mann, M. Anal. Chem. 1999, 71, 235-42. (13) Bean, M. F.; Annan, R. S.; Hemling, M. E.; Mentzer, M.; Huddleston, M. J.; Carr, S. A. In Techniques in Protein Chemistry VI; Crabb, J., Ed.; Academic Press: San Diego, CA, 1995; pp 107-16. (14) Haynes, P. A.; Aebersold, R. Anal. Chem. 2000, 72, 5402-10. (15) Carr, S. A.; Huddleston, M. J.; Bean, M. F. Protein Sci. 1993, 2, 183-96. (16) Borchers, C.; Parker, C. E.; Deterding, L. J.; Tomer, K. B. J. Chromatogr., A 1999, 854, 119-30. (17) Bolgar, M. S.; Gaskell, S. J. Anal. Chem. 1996, 68, 2325-30. (18) Steen, H.; Kuster, B.; Fernandez, M.; Pandey, A.; Mann, M. Anal. Chem. 2001, 73, 1440-8. (19) Petersson, A. S.; Steen, H.; Kalume, D. E.; Caidahl, K.; Roepstorff, P. J. Mass Spectrom. 2001, 36, 616-25. (20) Steen, H.; Mann, M. Anal. Chem. 2002, 74, 6230-6. (21) Rappsilber, J.; Steen, H.; Mann, M. J. Mass Spectrom. 2001, 36, 832-3.
In this paper, a novel mass spectrometric approach is described for the detection of dimethylarginine-modified peptides. It allows us to differentiate between the symmetric and the asymmetric dimethylarginine isomers and to locate the modified residue in the protein sequence. The method is based on precursor ion scanning experiments in the positive ion mode using characteristic fragments of aDMA and sDMA. Analytical sensitivity in the subpicomole range for peptides from gel-separated, dimethylargininecontaining proteins has been achieved. MATERIALS AND METHODS Chemicals and Compounds. Chemicals were obtained from Sigma (St. Louis, MO). High-purity solvents used for nanoelectrospray experiments were purchased from Lab-Scan (Dublin, Ireland). The peptides SYRGLSAFTK, SY(sDMA)GLSAFTK, and SY(aDMA)GLSAFTK were custom-made by Sigma-Genosys (The Woodlands, TX). TWRGGEEK, TW(sDMA)GGEEK, and TW(aDMA)GGEEK were custom-made by the HHMI-Keck facility (Yale, New Haven, CT). The novel arginine-methylated protein FUS/TLS was the leftover of a Coomassie Blue-stained band that originally contained an estimated 2-5 pmol of protein and that had been subjected to the in-gel digestion procedure and stemmed from an investigation of proteins involved in RNP biogenesis that will be reported elsewhere. Approximately 2 pmol of Sam68 was isolated from Hela cell extracts using a monoclonal anti-Sam68 antibody (Paushkin et al., manuscript in preparation), further purified by SDS-PAGE, and stained by Coomassie Blue. The bands were excised and subjected to in-gel reduction, alkylation, and tryptic digestion as previously described.22 Mass Spectrometry. All experiments were performed on a QSTAR Pulsar quadrupole time-of-flight tandem mass spectrometer (AB/MDS-Sciex, Toronto, Canada) equipped with a nanoelectrospray ion source (Protana Engineering A/S, Odense, Denmark). Synthetic peptides were fragmented under identical conditions using a collision energy of 30 eV, determined by the Q0 voltage. Precursor ion scanning experiments were acquired using a dwell time of 50 ms at a step size of 0.5 Da and with the Q2-pulsing function turned on. Nitrogen was used as a collision gas. The collision energy was set to 60 eV. M/z 71.06 was used as reporter ion in the DMA, and m/z 46.06 was used in the aDMA precursor ion scans. Synthetic peptides were dissolved in 5% formic acid and desalted using novel C18-StageTips.23 Peptides were eluted using 50% methanol in 5% formic acid and appropriate concentrations obtained by dilution using the same solvent. The FUS/TLS digest was desalted and concentrated on a column of Poros R2 (Perseptive Biosystems, Framingham, MA) packed into GELoader tips (Eppendorf, Hamburg, Germany), as described24 and eluted directly into nanospray needles (Protana Engineering A/S). The Sam68 digest was treated analogously but using C18StageTips. NanoESI was performed as was described.25 Proteins were identified by searching peptide sequence tags, derived from fragment ion spectra of selected peptides, against the nonredun(22) Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Anal. Chem. 1996, 68, 850-8. (23) Rappsilber, J.; Ishihama, Y.; Mann, M. Anal. Chem. 2003, 75, 663-70. (24) Ishihama, Y.; Rappsilber, J.; Andersen, J. S.; Mann, M. J. Chromatogr., A 2002, 979, 233-9. (25) Wilm, M.; Shevchenko, A.; Houthaeve, T.; Breit, S.; Schweigerer, L.; Fotsis, T.; Mann, M. Nature 1996, 379, 466-9.
dant protein database maintained and updated regularly by the European Bioinformatics Institute (EBI, Hinxton, U.K.) using the program PepSea (MDS Proteomics, Odense, Denmark). The program GPMAW (Lighthouse data, Odense, Denmark) was used for the prediction of peptide masses and fragment masses to match the data with modified peptides of the identified proteins. RESULTS AND DISCUSSION Fragmentation Behavior of Dimethylarginine-Containing Peptides. To study the fragmentation behavior of peptides containing DMA residues and design a method to specifically detect DMA as well as distinguish the symmetric from the asymmetric form, we obtained synthetic peptides. TWRGGEEK and SYRGLSAFTK were synthesized in the unmodified form, as well as either symmetrically or asymmetrically dimethylated at the side chain of their arginine residue. Figure 1B shows the spectra of the unmodified, as well as the two forms of argininedimethylated peptide TWRGGEEK fragmented as doubly charged species. In general, the fragmentation is not effected by the modification. Dimethylarginine was sufficiently stable under CID conditions that result in cleavage of the peptide backbone that it could be located in the sequence of the peptide by the observed y5+- and y6+-ions. In addition, the y52+-ion and the much weaker y62+-ion were detected. Side-chain fragments analogous to the loss of 64 Da for oxidized methionine were observed, loss of 45 Da from aDMA and loss of 70 Da from sDMA. In the low-m/z range, TW(sDMA)GGEEK showed a fragment at m/z 71.06 that was absent in the unmodified peptide (inset in Figure 1B). The fragment was observed also with the asymmetric DMA peptide, albeit at lower intensity. In addition to the fragment at m/z 71.06 in the asymmetric DMA peptide, we observed a more intense signal at m/z 46.06 that was absent in the symmetric DMA peptide as well as in the unmodified peptide. The signals at m/z 46.06 and 71.06 were observed independent of the charge state of the precursor and increased in intensity with increasing collision energy within the accessible collision energy range (data not shown). The other synthetic peptide SYRGLSAFTK and its dimethylated isoforms fragmented similarly (data not shown). We therefore conclude that these fragments are related to side-chain fragmentation of dimethylarginine and propose that the dimethylammonium ion gives rise to the signal at m/z 46.06 and that either the N,N′-dimethylcarbodiimidium or the N,N-dimethylcarbodiimidium ion result in the m/z 71.06 signal, depending on whether the fragment arose from sDMA or aDMA (Figure 1C). With this knowledge, we focused once more on the upper m/z range for the possibility of neutral loss of dimethylamine (m/z 45.05) and dimethylcarbodiimide (m/z 70.05). This would be analogous to phosphorylation on serine and threonine that results in loss of H2PO3- (m/z 79) in the negative ion mode, however, and a neutral loss of phosphoric acid (m/z 80) in the positive ion mode. Also, for high-energy CID, arginine-derived immonium ions at m/z 112 and 87 are described26 that could be the result of such neutral losses of ammonia (m/z 17) and carbodiimide (m/z 43), respectively, in the case of unmethylated arginine. Reinvestigating the fragmentation spectra of the synthetic peptides, we found, indeed, the characteristic loss of m/z 45.05 in the aDMA peptides (26) Johnson, R. S.; Biemann, K. Biomed. Environ. Mass Spectrom. 1989, 18, 945-57.
Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
3109
that corresponds to ammonia in unmodified arginine. However, the signals were weak and partly obscured by background that is particularly high in the region just below the precursor. This was the reason we missed this characteristic ion in the initial comparison. We furthermore found m/z 70.05 as neutral loss in the sDMA peptides. Unlike the signal for the dimethylcarbodiimidium ion (m/z 71.06), the loss of dimethylcarbodiimide (m/z 70.05) was not observed in the aDMA peptides. Therefore, loss of dimethylamine (m/z 45.05) and dimethylcarbodiimide (m/z 70.05) could be used in neutral loss scans to detect specifically asymmetric and symmetric DMA peptides, respectively. We did not investigate this further as the QSTAR does not allow for direct neutral loss scans. Instead, we focused on the conclusion that a signal at m/z 71.06 in the product ion scan of a peptide is indicative of the presence of dimethylarginine in this peptide and a signal at m/z 46.06 is indicative of the presence of asymmetrical dimethylarginine. This knowledge can be utilized in precursor ion scans as shown below. Precursor of m/z 71 Scan Analysis of FUS/TLS. Using m/z 71.06 ((0.1) as a characteristic fragment in a precursor ion scan on a quadrupole TOF hybrid mass spectrometer, the tryptic peptide mixture of the human alternative splice factor FUS/TLS27 was analyzed for dimethylarginine-containing peptides (Figure 2A). The sensitivity for the low-mass fragment was increased by a factor of 20 by using the pulsing function of the QSTAR. While a large number of peptide signals were detected in the TOF mass spectrum (upper panel), only few selected signals were observed in the precursor of m/z 71 scan (lower panel), indicating the specificity of the scan. The signals of the precursor ion scan allowed a mass assignment of the precursor ion with an accuracy of m/z 1 because of the step size in the scan. In most cases, this was sufficient to match the signals of the precursor ion scan unambiguous to those of a high-resolution MS spectrum to obtain more exact m/z values. The MS spectrum furthermore allowed assigning the charge states that in this analysis ranged from 3 to 10. This is the result of dimethylarginine often being located in RG clusters related to their biological function. As trypsin only very inefficiently cleaves after dimethylarginine, peptides containing this modification tend to be large and often contain multiple dimethylarginines. Here, the high resolution of the quadrupole TOF hybrid mass spectrometer was very helpful for charge-state recognition and allowed the determination of the peptide mass with an error below 40 ppm. In some cases, however, several signals in the MS spectrum could have corresponded to the signal in the product ion scan preventing unambiguous assignment. At m/z 446, there are two overlapping signals, one with the charge state nine and the other doubly charged (Figure 2A, inset). Similarly, at m/z 654, a singly charged species overlaps with a septuply charged one. This ambiguous situation is resolved as the highly charged species in these two regions are different charge states of the same peptide for which furthermore the complete series of charge states from 6 to 10 is observed in the MS spectrum. All of these have a signal in the product ion scan and are attributed to a tryptic peptide of FUS/TLS with the mass 4006.122 Da that contains nine dimethylarginine residues (Table 1). As the singly and doubly charged peptides did match unmodified peptides, we attributed the signal in the product ion (27) Calvio, C.; Neubauer, G.; Mann, M.; Lamond, A. I. RNA 1995, 1, 724-33.
3110 Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
Figure 2. DMA analysis of FUS/TLS, part 1. (A) Mass spectrum (upper part) and precursor of m/z 71 scan (lower part) of the tryptic peptides of a Coomassie Blue-stained gel band containing FUS/TLS. Triangles mark those peaks in the MS spectrum that have been matched with a DMA-containing peptide of FUS/TLS as a result of their signal in the precursor of m/z 71 scan. Asterisks mark peaks in the MS spectrum that contribute to background in the precursor of m/z 71 scan. The inset shows an enlargement of those signals of the DMA peptide that are marked by shaded triangles in the mass spectrum of FUS/TLS peptides (A, upper part). N denotes singly charged chemical noise and the asterisks overlaying signals. (B) Fragmentation spectrum of m/z 501.77. The peptide sequence is shown as a inset with the dimethylated arginine residues marked in boldface type.
scan exclusively to the highly charged peptides in these two examples. The high resolution of the MS spectrum also allows us to resolve signals from the chemical noise and enables us to detect otherwise hidden signals such as m/z 401.63 (10+) or 409.89 (6+). It also allows mapping the observed DMA peptides to the sequence of FUS/TLS unambiguously. As most peptides that were observed in this analysis were dimethylated at all their arginines, the exact location of the modifications is clear in this case. One peptide with very low signal intensity was special in that the number of DMA residues it contained requires the C-terminal arginine to be modified, too. Although trypsin usually does not cleave the peptide bond at modified arginines, it has previously been reported to occur in individual cases for DMA.28 Furthermore, the exact composition of this peptide is not clear, as there are two isobaric, largely overlapping peptides (m/z 474495 and 477-498) that match the observed mass. (28) Belyanskaya, L. L.; Gehrig, P. M.; Gehring, H. J. Biol. Chem. 2001, 276, 18681-7.
Table 1. Modifications Found in FUS/TLS m/z
z
M
Mcalc
ppm
483.233
3
1446.676
1446.732
39
6.6
DMA 2 MMAa 1 deam
ADFNRGGGNGRGGR(G)
409.890 491.658
6 5
2453.292 2453.251
2453.290
1 16
0.8 2.4
6 DMA
GGRGGYDRGGYRGRGGDRGGFR(G)GGYDRGGYRGRGGDRGGFRGGR(G)b
435.079 507.422 608.711
7 6 5
3038.498 3038.485 3038.516
3038.497
0 4 6
3.0 26.5 36.1
5 DMA 1 Mox
SSGGYEPRGRGGGRGGRGGMGGSDRGGFNK
579.443
6
3470.611
3470.731
35
1.8
4 DMA 2 MMAa 1 Mox
GRGGDRGGFRGGRGGGDRGGFGPGKDSRGEHR
401.625 446.119 501.767 573.293 668.690
10 9 8 7 6
4006.172 4005.999 4006.048 4005.996 4006.093
4006.122
12 31 12 31 7
0.7 12.0 24.1 21.7 16.9
9 DMA
RGGRGGYDRGGYRGRGGDRGGFRGGRGGGDRGGFGPGK
654.749 763.703
7 6
4576.188 4576.171
4576.209
5 8
12.0 16.9
6 DMA 1 Mox 1 deam
ADFNRGGGNGRGGRGRGGPMGRGGYGGGSGGGGRGGFPSGGGGGGGQQR
592.551 677.041 789.710
8 7 6
4732.346 4732.232 4732.213
4732.303
9 16 20
4.2 7.2 7.2
6 DMA 1 Mox 1 deam
RADFNRGGGNGRGGRGRGGPMGRGGYGGGGSGGGGRGGFPSGGGGGGGQQR
421.760 518.264 567.997 712.840
2 3 4 2
841.504 1551.769 2267.957 1423.664
841.502 1551.805 2267.962 1423.677
a
rel I
mods
Background in Precursor of m/z 71 Scan 3 88.0 23 24.1 2 28.9 1 Mox 9 100.0 1 Mox
sequence
VATVSLPR (trypsin) EFSGNPIKVSFATR APKPDGPGGGPGGSHMGGNYGDDR TGQPMINLYTDR
Two MMA or one DMA. b The two peptides are isobaric.
In contrast to a simple TOF instrument in a quadruploe TOF hybrid instrument, it is possible to select the DMA-containing peptides for fragmentation to validate the peptide and to determine the accurate position of the modification. We have selected for fragmentation the most prominent, nine DMA residues containing peptide in its octuply charge state at m/z 501.77 (Figure 2B). It readily loses three times dimethylamine (m/z 45.05) from the precursor, revealing that the peptide contains at least three asymmetric dimethylarginines. DMA so far was found in proteins to be exclusively either in its symmetric or asymmetric form, as mentioned before. The remaining DMA residues in the selected peptide are, therefore, likely asymmetric as well and FUS/TLS is an aDMA-containing protein. As seen for the synthetic DMA peptides, dimethylcarbodiimide (m/z 70.05) is not lost from the aDMA peptide of FUS/TLS. The identity of the peptide was validated by a number of peptide fragments that contained either completely dimethylated arginines or partially residues that had lost dimethylamine. Due to limitations in sample amount, only this peptide was fragmented. The precursor of m/z 71.06 scan is selective as only the two most intense and two other intense signals of the MS spectrum gave a signal in the precursor of m/z 71 scan without actually carrying the modification. These signals corresponded to a known trypsin autolysis peptide, and three tryptic peptides of FUS/TLS by mass and were easily distinguished on the basis of their low charge states. The nature of this background signal is not known, as none of the typical peptide-derived fragments has a nominal mass of m/z 71.
In total, seven dimethylarginine-containing peptides with a mass of up to 4700 Da were identified in the FUS/TLS peptide mixture (Figure 2D and Table 1). This is the first report of arginine dimethylation in FUS/TLS, and in total, 20 arginine residues were found to be asymmetric dimethylated. Similarly, the homologous EWS protein was shown previously to contain asymmetric DMA by a combined approach of mass spectrometry and Edman sequencing.28 All arginines of FUS/TLS that are in RG repeats, the methylation consensus sites, were detected and shown to be modified with the exception of three that are located in a tryptic peptide of 23 kDa size and one that is in the C-terminal peptide that escaped detection and are possibly modified as well. Contrary to the case of the synthetic aDMA peptides, the loss of dimethylamine was very intense for the FUS/TLS peptide sequenced. We concluded from this difference in signal intensity that the loss of dimethylamine is peptide dependent with factors such as charge state and number of dimethylated arginine residues possibly playing a role. The very intense loss of dimethylamine (m/z 45.05) observed here in the FUS/TLS peptide accompanied by the absence of any loss of dimethylcarbodiimide (m/z 70.05) suggests that the loss of dimethylamine is the primary fragmentation path of the aDMA side chain. As stated earlier, these two fragmentations can therefore be used to specifically detect aDMA and sDMA peptides. However, the peptide dependence of the neutral loss observed here would require a detailed investigation before such scans could be generally employed. A further disadvantage of the neutral loss approach would be that only one neutral loss scan can be recorded at a time using Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
3111
instruments capable of such experiments. We therefore concentrated our efforts on the precursor ion scans in the positive ion mode. All DMA peptides we have analyzed up to date produced the expected DMA characteristic low-mass fragments. Precursor ion scans also can only be conducted serially on a triple quadrupole mass spectrometer; however, a quadrupole TOF instrument offers the possibility for simultaneous detection of multiple fragments. This allows conducting the scans fully in parallel using the dimethylammonium ion (m/z 46.06) and the dimethylcarbodiimidium ions (m/z 71.06) and monitoring for sDMA and aDMA simultaneously as demonstrated below. Parallel Precursor Scan Analysis of Sam68. For the analysis of the human protein Sam68, we extended the precursor of m/z 71.06 ((0.1) scan by a scan for precursors of m/z 46.06 ((0.1) to immediately determine whether detected dimethylarginine residues are symmetric or asymmetric methylated. Again, pulsed extraction of ions from the fragmentation cell allowed increased sensitivity. Pulsing is efficient, however, only in a certain m/z window.29 This window, ranging on our instrument from about 75% to 150% of the selected m/z, was optimized by adjusting the instrument using a synthetic peptide. We found that the fragments m/z 46.06 and 71.06 are sufficiently close that their detection can be maximized by a single extraction pulse. This allows for their simultaneous monitoring. Condensing two separate precursor ion scans into one parallel scan minimizes the consumption of sample and reduces the measurement time by a factor of 2. A number of signals were evident in the precursor of m/z 71 scan and allswith one exceptionswere unambiguously matched to a signal in the normal mass spectrum. This allowed determining the peptide masses very accurately and selecting the peptides for MS/MS. All signals of the precursor of m/z 71 scan had corresponding signals in the precursor of m/z 46 scan with the exception of three peaks that were three charge states of the same unmodified peptide and that are discussed below. We therefore concluded that the dimethylarginine residues we detected in Sam68 were all asymmetric methylated. This is in agreement with previous studies that used recombinant Sam68 expressed in Sf9 cells30 or modified in vitro.31 The precursors of the four most intense signals of the precursor of m/z 46 scan were selected for MS/MS. Together with their other charge states, they accounted for eight of the nine most intense signals in the precursor of m/z 46 scan (Figure 3A, bottom). Fragmentation of the peptide with the most intense aDMA signal (m/z 438.765) revealed the predicted sequence containing two aDMA residues, R(340) and R(346). The y-ion series could be followed up to the last arginine. It split twice due to loss of dimethylarginine at a dimethylated arginine. The peptide already lost at low collision energies the N-terminal dipeptide GV that is followed in the sequence by a stretch of four prolines. This gave rise to the peptide with the second most intense signal in the precursor of m/z 46 scan at m/z 399.741 (4+) and 532.656 (3+) as was verified by sequencing. An interfering signal at m/z 399.580 (3+) was well resolved due to the resolution of the instrument (Figure 3A, inset). Also, the third peptide sequenced (29) Chernushevich, I. V. Eur. J. Mass Spectrom. 2000, 6, 471. (30) Wong, G.; Muller, O.; Clark, R.; Conroy, L.; Moran, M. F.; Polakis, P.; McCormick, F. Cell 1992, 69, 551-8. (31) Bedford, M. T.; Frankel, A.; Yaffe, M. B.; Clarke, S.; Leder, P.; Richard, S. J. Biol. Chem. 2000, 275, 16030-6.
3112 Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
Figure 3. DMA analysis of Sam68. (A) Mass spectrum (upper part), precursor of m/z 71 scan (middle part), and precursor of m/z 46 scan (lower part) of the tryptic peptides of a Coomassie Blue-stained gel band containing Sam68. Triangles mark those peaks in the MS spectrum that have been matched with a DMA-containing peptide of Sam68 as a result of their signal in the precursor of m/z 71 and precursor of m/z 46 scans. Asterisks mark peaks in the MS spectrum that contribute to background in the precursor of m/z 71 scan. The shaded triangle marks the signal of the DMA peptide that is enlarged in the inset. (B) Fragmentation spectrum of m/z 417.997. The peptide sequence is shown as inset with the dimethylated arginine residues marked in boldface type.
(m/z 592.096, 4+) contained R(340) and R(346) and in addition R(331) in dimethylated form. The last peptide fragmented (m/z 417.997, 4+) contained dimethylated R(320) and R(325) and unmodified R(331) (Figure 3B). This peptide fragmented in an unusual manner as it did not allow a record of a balanced fragmentation spectrum. The precursor fragmented with a very high yield into low-m/z products. This resulted in its efficient detection by the precursor of m/z 46.06 ((0.1) scan. The peptide shows only minor loss of dimethylamine and resembles in this the synthetic DMA peptides described above. This further manifests that aDMA peptides can exhibit varying preferences for the competing fragmentation paths of neutral loss of dimethylamine or creation of the dimethylamonium ion. It also emphasizes that fragment signals do not necessarily give an accurate reflection of the intensity of their precursors. By matching the masses corresponding to minor signals of the precursor of m/z 46 scan with predicted DMA peptides of Sam68 via their corresponding signals in the MS spectrum, we furthermore observed that arginine residues in the region 284-315 are also dimethylated,
Table 2. Modifications Found in Sam68 m/z
z
M
Mcalc
ppm
rel I
mods
445.273
3
1332.795
1332.836
31
2.1
aDMA MMA
GRGVGPPRGALVR
449.941
3
1346.799
1346.863
47
2.3
2 aDMA
GRGVGPPRGALVR
527.985
3
1580.931
1580.915
10
1.3
aDMA MMA
PPPPTVRGAPAPRAR
399.741 532.656
4 3
1594.932 1594.944
1594.942
6 1
21.3 5.3
2 aDMA
PPPPTVRGAPAPRAR
PPP...AR
417.997 557.002
4 3
1667.956 1667.982
1667.980
15 1
1.4 1.9
2 aDMA
GTPVRGAITRGATVTR
...GATVTR
435.259
4
1737.004
1737.005
1
5.6
aDMA MMA
GVPPPPTVRGAPAPRAR
438.765 584.684
4 3
1751.028 1751.028
1751.032
2 2
26.8 5.0
2 aDMA
GVPPPPTVRGAPAPRAR
515.811
4
2059.212
2059.275
30
0.9
2 aDMA MMA
GVPVRGRGAAPPPPPVPRGR GRGVPVRGRGAAPPPPPVPR b
415.654 519.302
5 4
2073.231 2073.176
2073.302
34 60
0.5 0.6
3 aDMA
GVPVRGRGAAPPPPPVPRGR GRGVPVRGRGAAPPPPPVPR b
471.083 588.594
5 4
2350.376 2350.344
2350.357
8 5
3.3 1.0
2 aDMA MMA
GATVTRGVPPPPTVRGAPAPRAR
473.887 592.096
5 4
2364.396 2364.352
2364.387
4 15
3.3 1.4
3 aDMA
GATVTRGVPPPPTVRGAPAPRAR
489.024a 570.179
7 6
3415.113 3415.027
3415.005
31 6
1.3 0.4
4 aDMA MMA
GTPVRGAITRGATVTRGVPPPPTVRGAPAPRAR
491.022a 572.678
7 6
3429.099 3429.001
3429.032
20 9
1.0 0.6
5 aDMA
GTPVRGAITRGATVTRGVPPPPTVRGAPAPRAR
622.002 746.206 932.505
6 5 4
3725.965 3726.101 3725.988
3725.988
6 1 0
a
sequence
Background in Precursor of m/z 71 Scan 4.1 ASPATQPPPLLPPSATGPDAT7.8 VGGPAPTPLLPPSATASVK 10.3
MS/MS
GV...VRGAPAPRAR
GATV...PPP...R
ASPATQPPPLLPPSATGPDATVGGPAPTPLLPPSATASVK
Second isotope peak; first isotope peak is hidden in noise. b Both peptides are isobaric.
as summarized in Table 2. In addition, we find to a small extent monomethylation occurring in the entire range (Figure 4B). The intensities of different peptides in the MS spectrum are usually not indicative as a measure of relative abundances as the signal intensity is also influenced by peptide properties. However, summing the relative intensities of peaks from those DMA peptides containing a certain arginine residue, we find that signals of peptides with aDMA on R(340) and R(346) are 20-50 times more abundant than signals for peptides containing the other arginine residues. The large difference in signal intensity suggests that these are the major aDMA sites in Sam68 (Figure 4B). This conclusion is supported by the fact that all DMA peptides of Sam68 we could detect are composed almost exclusively of a set of only five amino acids, which may equalize their ionization efficiency. One set of signals was observed only in the precursor of m/z 71 scan and was therefore indicative of an sDMA-containing peptide. The signals coincided, however, with the signals of an unmodified peptide of Sam68 whose identity was verified by sequencing. This left two possible explanations. The signals of the unmodified peptide either obscured the signals of the sDMA peptide in the normal scan and in the MS/MS experiment, or the unmodified peptide itself gave rise to a fragment of m/z 71.06 and therefore is seen as background in the precursor of m/z 71 scan. Using recombinant Sam68, we did, indeed, observe the unmodified peptide both in the normal scan and in the DMA
analysis (data not shown). We therefore conclude that we have no evidence for symmetric dimethylarginine modification in Sam68 at this point. Neither during the analysis of the proteins reported here nor of a number of other samples did we observe a similar background in the precursor of m/z 46 scan and we conclude that this scan is very specific. Very narrow detection windows as used in phosphotyrosine analysis did not improve the S/N ratio in the DMA and precursor of m/z 46 scans during our work so far. Contrary to the phosphotyrosine immonium ion (PHI) scan,18 the DMA and precursor of m/z 46 scans can therefore be conducted on a triple quadrupole mass spectrometer extending the analytical capabilities of this machine type. This also means that work on high-resolution machines such as the QSTAR is simplified when compared to the PHI scan. However, as seen in the analysis of FUS/TLS and Sam68, the high resolution is clearly of advantage to determine the correct precursor in the MS spectrum of a signal in the product ion scan and to assign its charge state. The combination of precursors of m/z 71.06 ((0.1) and 46.06 ((0.1) scans is suitable to detect not only asymmetric but also symmetric DMA. From the absence of a signal in the m/z 46.06 ((0.1) scan, the conclusion can be drawn that symmetric DMA is present if the peptide is sequenced to exclude the possibility of misinterpreting a background signal. The fragmentation spectrum should also reveal the loss of dimethylcarbodiimide (m/z Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
3113
Figure 4. Summary of identified DMAs. (A) The DMA peptides found in FUS/TLS are underlined in the partial protein sequence. Filled circles denote dimethylarginine, half-filled circles monomethylarginine, squares oxidized methionine, and triangles deamination. The symbols are placed under the respective residues or, if the location was ambiguous, at the end of the line marking the peptide. A star denotes the fragmented peptide. Two monomethylarginines are equivalent to one dimethyl and one unmodified arginine. (B) The DMA peptides found in Sam68 are underlined in the partial protein sequence. Filled circles denote dimethylarginine, half-filled circles monomethylarginine, and empty circles unmodified arginine. The symbols are placed under the respective residues or, if the location was ambiguous, at the end of the line marking the peptide. A star denotes the fragmented peptides. Two monomethylarginines are equivalent to one dimethyl and one unmodified arginine. The numbers give the relative intensities (Rel. I.) of the respective peptide in the MS spectrum summed over all charge states.
70.05) but not dimethylamine (m/z 45.05). A synthetic aDMA peptide as a control should furthermore be added to the tryptic mixture to ensure that m/z 46.06 would have been detected in the scan.
3114
Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
CONCLUSION We have shown that the important protein modification dimethylarginine gives rise to five characteristic fragments in the positive ion mode: m/z 46.06 and 71.06 as well as a neutral loss of m/z 45.05 for asymmetric dimethylation and m/z 71.06 as well as a neutral loss of and m/z 70.05 for symmetric dimethylation. We monitored m/z 46.06 ((0.1) and 71.06 ((0.1) in precursor ion scans to detect peptides containing asymmetric dimethylation on arginine residues from tryptic digests of human proteins and detected 20 asymmetric dimethylarginines in FUS/TLS and 9 in Sam68. These findings were biologically significant as will be reported elsewhere. Precursor ion scans can be used to differentiate isobaric protein modifications. We have shown this previously utilizing stability differences in the case of the isobaric sulfo- and phosphotyrosine.21 Here, we monitored two fragments to detect dimethylarginine in peptides and to specify which isomer is present. Parallel monitoring of the characteristic fragments minimized time and sample consumption. Our new method could be a crucial technique in the rapidly expanding field of dimethylarginine modification. The diagnostic value of the here reported fragments can also be used in LC/MS-type experiments by monitoring the low-mass products of each fragmented peptide. ACKNOWLEDGMENT We thank members of our institutes and especially Dr. Hanno Steen for helpful discussions. J.R. is a Marie Curie Fellow. G.D. is an Investigator of the Howard Hughes Medical Institute. This work was supported by a grant from the National Institute of Health to G.D. and a grant from the Danish National Research Foundation to the Center of Experimental Bioinformatics. Received for review November 5, 2002. Accepted April 3, 2003. AC026283Q
