Delineating Anopheles gambiae coactivator associated arginine methyltransferase 1 automethylation using top–down high resolution tandem mass spectrometry

Abstract

Coactivator-associated arginine methyltransferase 1 (CARM1), originally defined as a coactivator for steroid receptors, is a member of the protein arginine methyltransferases. Here, we report the discovery and characterization of an automethylation event by AgCARM1, a CARM1 homologue in the mosquito Anopheles gambiae, using top–down high resolution tandem mass spectrometry, which allows fine mapping of modifications in the intact protein accurately and quantitatively without priori knowledge. Unexpectedly, we found that AgCARM1 has already been predominantly dimethylated during its expression in Escherichia coli. A single arginine methylation site, R485, was identified which is conserved among CARM1 in insects. No methylation was observed in the intact AgCARM1^R485K mutant where R485 is mutated to lysine, which confirms that R485 is the only detectable methylation site. Using AgCARM1 methyltransferase defective mutants, we confirmed that this is an automethylation event and show the automethylation of AgCARM1 occurs intermolecularly. This study represents the first comprehensive characterization of an automethylation event by top–down mass spectrometry. The unexpected high percentage of automethylated recombinant AgCARM1 expressed in E. coli may shed light on other bacterially expressed post-translational modifying enzymes, which could be modified but overlooked in biochemical and structural studies. Top–down high resolution tandem mass spectrometry thus provides unique opportunities for revealing unexpected protein modification, localizing specific modification to one amino acid, and delineating molecular mechanism of an enzyme.

Introduction

The methylation of arginine (R) residues is catalyzed by the protein arginine methyltransferase (PRMT) family of enzymes, which has been shown to play an important role in diverse cellular functions including transcription regulation, mRNA splicing, and DNA damage repair. The PRMTs are generally classified as either Type I or Type II enzymes: Type I PRMTs monomethylate and asymmetrically dimethylate the primary amine groups on the arginine side chain, whereas Type II PRMTs catalyze the formation of monomethyl arginine and symmetric dimethyl arginine.¹

Coactivator-associated arginine methyltransferase 1 (CARM1), also known as PRMT4, is a Type I PRMT which was originally identified in a yeast two-hybrid screen as an associated protein for glucocorticoid receptor interacting protein 1, the p160 family steroid receptor coactivator.² Mutation and structural studies have identified a number of functional domains in CARM1.^3–5 CARM1's central catalytic core is characteristic of PRMT family proteins,⁶ which contain an S-adenosylmethionine (AdoMet) binding domain and a substrate binding domain. Within this central core, the AdoMet binding domain contains a valine, leucine, aspartic acid (VLD, amino acids 189–191) sequence that is essential for enzymatic function.² This sequence is part of motif 1 (aliphatic-aliphatic-acidic-alphatic-G-X-G-X-G), the first of a set of shared motifs among methyltransferase enzymes.⁷ The central dimerization motif within the substrate binding domain makes up an “arm” that facilitates dimerization. The C-terminal catalytic core is the substrate binding domain, which is necessary for recognition of the arginine to be methylated. CARM1 has been shown to methylate both histone and non-histone substrates.^8–13 In contrast to most PRMTs that methylate glycine-rich and arginine-rich patches (GAR motifs) within their substrates, CARM1 does not methylate the GAR motif but rather has a unique set of methylation targets,¹ which have no apparent recognition sequence.

The function of CARM1 has recently been shown to be regulated by post-translational modifications (PTMs). Phosphorylation of serine 228 blocks CARM1 dimerization, likely interrupting proper binding of the dimerization “arm” in phosphorylated CARM1 and thus inhibits its methylation activity.¹⁴ Automethylation of CARM1 has been previously implicated in vitro¹⁰ in analog with other PRMT family members including PRMT1, PRMT6, and PRMT8,^15,16 however, the automethylation of CARM1 and PRMT1 appeared to be significantly less than PRMT6 and PRMT8. The site(s) of CARM1 automethylation are unknown and the role of PRMT automethylation remains to be elucidated.

Mass spectrometry (MS) is the only technique that can universally provide information about protein PTMs without a priori knowledge of the modification.¹⁷ In the conventional “bottom-up” MS approach proteins of interest are digested with an enzyme prior to MS analysis providing only partial coverage of the protein sequence.^18,19 In contrast, top–down MS analyzes intact protein directly, providing “a bird's eye view” of all possible protein modifications with full sequence coverage.^20–33 The top–down MS approach subsequently fragments the protein ions of interest in the mass spectrometer to locate the modification site(s), which allows fine mapping of modifications in the intact protein. Moreover, the top–down MS approach is especially attractive for estimating the relative abundance of protein species with specific modifications since the ionization efficiency of intact proteins is much less affected by the presence of modifying groups in comparison with peptides.^30,32,34,35 The newly developed MS/MS technique of electron capture dissociation (ECD)³⁶ greatly improves both efficiency and sequence coverage in top–down analyses. ECD cleaves NH-CHR bonds to produce mainly c and z^. ions, complementary to those from the well developed energetic dissociation methods such as collisionally activated dissociation (CAD).²¹ More importantly, ECD generates far more fragmentation ions as well as unique cleavages due to its nonergodic dissociation of covalent protein backbone bonds causing local fast (<10⁻¹² s) cleavages of covalent bonds before energy randomization.^21,23,37,38

In this study, we report the discovery and characterization of an unexpected automethylation event by AgCARM1, a CARM1 homologue in the mosquito model Anopheles gambiae, using top–down high resolution Fourier transform (FT) tandem mass spectrometry (MS/MS). AgCARM1 automethylation is shown to occur intermolecularly during expression in Escherichia coli as well as in vitro.

Results

AgCARM1 is the A. gambiae homologue of CARM1

CARM1 contains a conserved PRMT catalytic core,³⁹ which is highly conserved across species even between mammals and insects. The Drosophila CARM1 homologue CARMER (or DART4) has been reported to have similar substrate specificity to mouse CARM1 (mCARM1) but is lacking the C-terminal activation domain found in other CARM1 sequences (Supporting Fig. S1).^40,41 To develop an animal model for CARM1 function, we chose to study CARM1 in the mosquito model A. gambiae since this model organism has a sequenced genome and a simplified hormone signaling pathway. We have cloned a sequence that is highly similar to known CARM1 genes from the A. gambiae genome (AgCARM1, Genbank: FJ391182) and generated highly purified AgCARM1^1–509 protein in large quantities [Fig. 1(A)]. To verify AgCARM1^1–509 is the functional homologue of CARM1, we performed in vitro methylation assays using purified AgCARM1^1–509 and AgCARM1^VLD (catalytic inactive mutant where the amino acids in the AdoMet binding domain, VLD, were mutated to AAA) and compared their methylation specificity to mCARM1 and mPRMT1. These were incubated with known CARM1 substrates polyA binding protein 1 (PABP1), core histones from Hela cells, as well as GAR, a common substrate for PRMT family members including PRMT1 but not CARM1. AgCARM1^1–509 was found to methylate the CARM1 specific targets exclusively [Fig. 1(B,C)], while exhibiting no activity towards GAR, which is methylated by PRMT1 [Fig. 1(D)]. Analogous to the mCARM1^VLD mutant,² the AgCARM1^VLD mutant showed a complete loss of methylation activity [Fig. 1(B,C)]. AgCARM1^1–509 also exhibits automethylation activity in vitro [Fig. 1(E)] as reported previously for mCARM1.¹⁰ These results confirm that AgCARM1^1–509 is the CARM1 homologue in A. gambiae.

Figure 1.

In vitro methylation studies of AgCARM1^1–509 show similar substrate specificity to mCARM1. (A) Coomassie stain of Ni-NTA purified AgCARM1.^1–509 (B,C) In vitro methylation of core histones (B) and PABP1 (C) by CARM1 variants in the presence of ³H-AdoMet. Top panel is an autoradiograph showing radiolabel incorporation of the methyl group. Bottom panel is the commassie stained gel after exposure to film. (D) In vitro methylation of the fibronectin fragment GAR by various PRMTs in the presence of ³H-AdoMet. (E) Autoradiograph of AgCARM1^1–509 incubated without substrate in an otherwise identical in vitro methylation assay as seen in B–D. *Band that copurifies with PABP1, likely a degradation product. GST-GAR also exhibits significant degradation consistent with previous publications of this protein fragment. CS, Coomassie Stain; AR, Autoradiograph.

AgCARM1^1–509 overexpressed in E. coli is dimethylated at R485 in vivo

We used top–down MS to characterize AgCARM1^1–509 expressed in E. coli, later referred to as AgCARM1^WT. After purification, the AgCARM1^WT protein sample was directly analyzed by LTQ FT ultra high resolution mass spectrometer. Electrospray ionization (ESI)/FTMS spectrum revealed two molecular forms of AgCARM1^WT (Fig. 2, top). The most abundant molecular weight of the minor form is 58549.36–58549.38, which is consistent with the DNA predicted mass value for AgCARM1^WT with the removal of its N-terminal methionine (Calc'd M_r 58549.31–58549.38, 0.9 ppm). Surprisingly, the molecular weight of the major form, 58577.37–58577.38, is 28 Da more than the minor form suggesting a dimethylated form of AgCARM1^WT (Fig. 2, bottom). Eighty-three percentage of the total AgCARM1^WT was dimethylated and 17% was unmethylated, with no detectable amount of monomethylated AgCARM1^WT based on peak height. This relative quantification is based on the assumption that the ionization efficiency of intact proteins is much less affected by a modifying group in comparison to peptides.³⁰ Since the analyzed sample was freshly purified after over-expression, it is conceivable that this methylation event occurred in vivo after expression in E. coli.

Figure 2.

High resolution ESI/FTMS analysis of AgCARM1^WT expressed in E. coli. Top, ESI/FTMS spectrum of AgCARM1^WT intact protein ions (M⁴⁴⁺), suggesting AgCARM1^WT is predominantly dimethylated. Bottom, fragmentation map from 3 ECD and 6 CAD spectra of both un- and di-methylated AgCARM1^WT matched with assignments to the DNA-predicted sequence of AgCARM1^WT with the removal of N-terminal methionine, localizing the dimethylation site to R485 (highlighted in circle). The fragmentation ions containing the dimethylated form were indicated in dots.

To locate the specific site of dimethylation, multiple charge states of AgCARM1^WT were individually isolated and then fragmented by CAD and ECD. The six combined CAD spectra generated 39 b and 47 y ions whereas 3 ECD spectra generated 66 c and 12 z^. fragment ions representing 78 cleavages each. These covered 26% of the total 521 NH-CH available backbone bonds in AgCARM1^WT and confirmed removal of methionine at the N-terminus. Fragmentation ions were consistent with dimethylation of R485, showing a +28 Da shift in those fragments containing the R485 residue. MS/MS spectra of a mixture of unmethylated and dimethylated AgCARM1^WT show predominantly (∼83%) dimethylated form of fragment ions containing R485 (Supporting Fig. S2) suggesting R485 is the only site for dimethylation. Large b and y ions were observed in CAD spectra including several complementary pairs although fragmentation occurred most frequently in the terminal regions of the protein (Fig. 2, bottom). The overlapping b and y ions span the entire sequence giving confidence that the assigned primary structure is correct. The Asn39-Pro40 bond was very sensitive to CAD dissociation, resulting in not only the complementary ion pair (y₄₈₃ and b₃₉) but also a series of internal ions with Asn39 as the C-terminus (data not shown). This pattern was observed in all AgCARM1 mutants analyzed hereafter.

To validate that R485 is the sole methylation site on AgCARM1^WT, R485 was mutated to lysine to maintain the positive charge while blocking methylation. The AgCARM1^R485K mutant was expressed and purified as for wild-type protein. Only one molecular form was shown in ESI/FTMS spectrum, matching the mass of unmethylated AgCARM1^R485K (Fig. 3, top). Similarly, CAD and ECD data confirmed the lack of detectable methylation in AgCARM1^R485K, which is estimated to be <3% of the total protein ion population (Fig. 3, bottom). Two ECD spectra generated 63 c and 26 z^. fragment ions representing 89 cleavages and two CAD spectra generated 24 b and 26 y ions representing 39 cleavages. This result confirms that R485 is the only significant methylation site on CARM1 during its expression in E. coli. Similar to the AgCARM1^WT data, very large fragments were observed in broadband CAD spectra (see Fig. 4). The largest b/y fragmentation ions observed were b₄₉₅ (Calc'd 55576.66–55576.37; Expt'l 55576.57–55576.37) and y₄₈₃ (Calc'd 54169.99–54169.36; Expt'l 54169.94–54169.36), which is, to our knowledge, the largest fragmentation ions isotopically resolved in a MS/MS spectrum.

Figure 3.

High resolution ESI/FTMS analysis of AgCARM1^R485K mutant expressed in E. coli. Top, ESI/FTMS spectrum of AgCARM1^R485K intact protein ions (M⁴⁰⁺), suggesting AgCARM1^R485K is not methylated. Bottom, fragmentation map from 2 ECD and 2 CAD spectra of AgCARM1^R485K matched with assignments to the DNA-predicted sequence of AgCARM1^R485K without the N-terminal methionine.

Figure 4.

CAD spectrum of AgCARM1^R485K intact protein ions. (A) CAD spectrum of isolated single charge state of AgCARM1^R485K (M⁴⁰⁺). (B) Expanded spectrum of (A) showing large fragmentation ions observed at m/z 1500–1800. (C) Representative isotopically resolved large fragmentation ions.

AgCARM1^R485K exhibits normal methylation activity towards other CARM1 substrates

To examine whether methylation of R485 affects the methyltransferase activity of AgCARM1, either AgCARM1^WT or AgCARM1^R485K was incubated with substrates PABP1, p300 and core histones in the presence of ³H-AdoMet in an in vitro methylation assay. Methylation of substrates was determined to be at the same level between AgCARM1^WT and AgCARM1^R485K [Fig. 5(A)]. These results suggest that the methyltransferase activity of AgCARM1 is independent of its methylation status at R485.

Figure 5.

In vitro methylation assays characterizing AgCARM1 automethylation and its affect on enzyme activity. ³H-AdoMet was incubated with purified protein followed by SDS-PAGE, coomassie staining, and exposure to film. (A) In vitro methylation of known CARM1 substrates by AgCARM1^WT and AgCARM1^R485K. (B) Isolated and mixed automethylation reactions with 0.1 μg and/or 1 μg of AgCARM1. Top shows a cartoon representation of reaction contents and possible interactions. Looping arrows indicate possible intramolecular and/or intermolecular automethylation. Straight arrows indicate possible intermolecular automethylation. Bottom shows in vitro methylation results of these reaction conditions, with lane numbers corresponding to the same numbered cartoon above. *Band copurified with PABP1, likely due to degradation. CS, Coomassie Stain; AR, Autoradiograph.

AgCARM1^WT is automethylated during its expression in E. coli

There are no known arginine methyltransferases in E. coli. To ensure that methylation of AgCARM1 in bacteria is due to automethylation activity and not a previously unidentified endogenous bacterial enzyme, the methyltransferase defective AgCARM1^VLD mutant was bacterially expressed and analyzed by top–down MS. This mutation is not proximal to arginine 485, and so will likely maintain the sequence and structural integrity of the methylation site, thus serving as an effective control. A single molecular form was observed at 58433.83–58433.38 (Calc'd 58433.97–58433.38, 2 ppm), consistent with the unmethylated mass of AgCARM1^VLD with the removal of N-terminal methionine. The overlapping b and y ions fragmentation ions from CAD of AgCARM1^VLD span the entire sequence, which were consistent with an unmethylated parent ion (see Fig. 6). A single ECD spectrum generated 77 c and 38 z^. fragment ions representing 115 cleavages; in contrast, one CAD spectrum generated 24 b and 24 y ions representing 47 cleavages. ECD fragmentation data showed a very rich fragmentation pattern mainly located on the two termini (Supporting Fig. S3). Lack of fragmentation of the central region of the protein was consistent among the AgCARM1^VLD, AgCARM1^R485K, and AgCARM1^WT proteins. These results suggest that methylation is dependent on AgCARM1 activity and AgCARM1^1–509 is automethylated when expressed in bacteria.

Figure 6.

Top–down MS analysis of AgCARM1^VLD mutant. Top, ESI/FTMS spectrum of AgCARM1^VLD (M³⁹⁺), intact protein ions, suggesting AgCARM1^VLD is not methylated. Bottom, Fragmentation map from both 1 CAD and 1 ECD spectra with assignments to the DNA-predicted sequence of AgCARM1^VLD with the removal of N-terminal Met. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Automethylation of AgCARM1 occurs intermolecularly

In contrast to AgCARM1^WT, whose automethylation indicates that it acts as both enzyme and substrate, no automethylation was observed in vitro when AgCARM1^R485K or AgCARM1^VLD mutants were incubated alone with ³H-SAM [Fig. 5(B), compare lane 1 with lane 2 and 3]. AgCARM1^R485K shows a lack of auto-methylation activity due to its inability to act as substrate in this reaction, while AgCARM1^VLD has no enzymatic activity but still contains the methylation site arginine residue. To determine how the AgCARM1 mutants and wild-type proteins interact, a number of mixed in vitro methylation reactions were performed. Although the AgCARM1^R485K and AgCARM1^VLD mutants do not automethylate in isolated reactions, mixing AgCARM1^WT either with AgCARM1^R485K or AgCARM1^VLD at a 1:10 ratio in vitro results in increased methylation above that of AgCARM1^WT alone [Fig. 5(B), compare lanes 5 and 6 with lane 4]. Mixing of AgCARM1^R485K and AgCARM1^VLD shows a weak methylation in vitro even though neither form shows automethylation activity by itself [Fig. 5(B), lane 7]. These data support an intermolecular mechanism for the automethylation of AgCARM1 since these increases in methylation must be due to the interaction between different AgCARM1 proteins.

Discussion

We have biochemically characterized a novel CARM1 homologue from the mosquito A. gambiae. AgCARM1 shows conserved substrate specificity for CARM1 methylation targets [Fig. 1(B,C)] as well as exhibiting automethylation activity both in vivo during recombinant expression (see Fig. 2) and in vitro [Fig. 1(E)]. Unexpectedly, our top–down MS results indicate that AgCARM1^WT has already been predominantly automethylated during its expression in E. coli. Recombinant proteins expressed in E. coli are generally thought to be lacking methylation and other PTMs; however, our study clearly shows protein arginine methyltransferase can utilize bacterial AdoMet and be automethylated when expressed in bacteria, suggesting that the same principle may apply to other post-translational modifying enzymes. Since PTMs can affect the activity or binding specificity of a given protein,^34,42,43 the possibility of PTM during recombinant expression may create structural and functional diversity that would affect characterization of these proteins.⁴⁴ This phenomenon may apply to other methyltransferases and specifically PRMT enzymes. Previous observation of weakly methylated CARM1 and PRMT1 in vitro has been logically attributed to weak automethylation activity.¹⁶ Our results indicate that a weak in vitro automethylation signal can also be attributed to near saturated automethylation in vivo prior to purification of the enzyme. Thus, it may be inaccurate to compare the activity of these enzymes through in vitro methylation without prior characterization of their methylation states.

Mutation of R485 to lysine appears to have no effect on AgCARM1's enzymatic activity, as CARM1 substrates are methylated equally by AgCARM1^WT and AgCARM1^R485K. This suggests that dimerization, AdoMet binding, and substrate specificity are not affected by loss of automethylation. R485 is located in the C-terminal domain adjacent to the substrate-binding site and, consistent with the above results, the corresponding region in mCARM1 is unstructured and not essential for the enzymatic activity of CARM1.³ The AgCARM1^R485K mutant shows a loss of automethylation both in bacteria (see Fig. 3) and when isolated in vitro [Fig. 5(A)], suggesting that R485 is the only detectable automethylation site on AgCARM1. R485 is highly conserved in other sequenced insect species (Supporting Fig. S1).

The mechanism of automethylation has yet to be well characterized in PRMTs or other methyltransferases. An automethylation site on the lysine methyltransferase G9a has been determined,^45,46 and this work suggests that G9a automethylation may occur by an intramolecular mechanism. While we were not able to confirm or rule out intramolecular activity by AgCARM1, we do show direct evidence of intermolecular methylation using a series of mixed in vitro methylation reactions [Fig. 5(B)]. Mixing AgCARM1^WT with the catalytically inactive AgCARM1^VLD, which can act only as a substrate in an automethylation reaction, increases radiolabeling above that of AgCARM1^WT alone. Since AgCARM1^VLD cannot contribute to the enzymatic activity of the reaction, this result indicates that AgCARM1^VLD is being methylated by AgCARM1^WT. Similar experiments show that AgCARM1^R485K is able to methylate AgCARM1^VLD in a similar manner. Mixing AgCARM1^WT with AgCARM1^R485K also shows an increase in the intensity of radiolabeling above AgCARM1^WT alone, suggesting that AgCARM1^R485K is methylating AgCARM1^WT, since it does not appear to methylate itself. Altogether these data support an intermolecular mechanism for automethylation.

The ESI/FTMS analysis of AgCARM1 expressed in bacteria showed predominant dimethylated protein, while the remaining AgCARM1^WT was unmethylated. This observation was supported by fragmentation spectrum (Supporting Fig. S2), which showed a similar ratio of dimethylated and unmethylated forms with no detectable monomethylated form in fragmentation ions containing the dimethylation site, R485. We speculate that the unmethylated AgCARM1^WT represents the freshly expressed protein, while the dimethylated AgCARM1^WT may be expressed earlier and automethylated over time. This notion is supported by the top–down MS result that AgCARM1^WT can be further dimethylated when incubated with cold AdoMet in vitro (data not shown). The specific lack of monomethylated AgCARM1^WT suggests that AgCARM1 may bind its substrate continuously during methylation, which is known as a processive mechanism. If validated, this would be in contrast to data from other Type I PRMTs such as PRMT1, PRMT3, and PRMT6. PRMT6 has been shown to methylate synthetic peptide substrates using a distributive mechanism, where a majority of substrate is monomethylated before dimethylation occurs.⁴⁷ PRMT1's mechanism is still under debate, as recent works suggest both a partially processive mechanism,⁴⁸ and a distributive mechanism.⁴⁹ These studies focused on the methylation of peptides instead of intact proteins. Thus, the role of extended substrate structure and the unique character of automethylation in defining the mechanism of PRMT activity remain poorly defined. Although not conclusive, our results indicate that automethylation may have a different mechanism than other substrates' methylation by PRMT family members.

The top–down MS used in our study allows us to unambiguously determine any and all detectable modifications of AgCARM1. We were able to resolve the intact AgCARM1 protein (>58 kDa) in a LTQ/FT mass spectrometer and achieved extensive fragmentation by CAD and ECD. Extremely large fragmentation ions (>55 kDa) were isotopically resolved in CAD spectra. The overlapping b and y ions span the entire sequence giving confidence that the assigned primary structure is correct. ECD provided more backbone cleavages than CAD; here one ECD spectrum of one isolated single charge state of a 58 kDa protein generated as many as 115 cleavages out of 521 bonds. Therefore, top–down MS shows great promise in the elucidation and quantitation of large intact proteins.^20–33 Recently, Webb et al. characterized methylation of lysine residues in yeast ribosomal protein Rpl42ab (12 kDa) via such a top–down MS approach. Despite its enormous promise, top–down MS has yet to be popularized mainly due to the high-resolution instrument requirement and technical difficulties in handling large intact proteins. Desalting is required before a protein sample can be analyzed on a mass spectrometer, however, proteins are more likely to precipitate in a solution free of salt. To overcome this dilemma, we found that it is critically important to analyze proteins freshly purified and desalted. Another limitation to top–down MS is the sensitivity of currently commercially available mass spectrometers. We need to acquire 1000–3000 transients per spectrum to ensure high quality MS and MS/MS, which required a relatively large amount of high purity proteins (>20 μg) and a highly stable electrospray. Hence, improvements in the instrument (i.e., sensitivity and resolution of a mass spectrometer) and software for automatic data processing⁵⁰ will greatly facilitate the use of top–down MS for large proteins.

Materials and Methods

Cloning, expression, and purification

AgCARM1 was amplified from A. gambiae cDNA. The PCR product of the forward primer: (5′CGACTTGCTAGCATGAGTGAACAAATGGCTCG 3′) and reverse primer: (5′ATGGCAGCGGCCGCGTTGGTGTTGATGATTGTCGACG 3′) was restriction digested and ligated into pET21a, adding seven N-terminal amino acids (MARMSEQ) and an C-terminal His tag AgCARM1^1–509. AgCARM1^R485K, AgCARM1^VLD, GST-PABP1, and GST-GAR were expressed in BL21 cells and purified using tag affinity chromatography (Qiagen). mCARM1 and mPRMT1 were purified from insect cells as described previously.¹⁰

Site-directed mutagenesis

pET21a-AgCARM1^1–509 was mutated by amplification with complementary primers containing mutations to induce codon changes. AgCARM1^VLD forward primer (5′CGCAAGACTTCCAGAATAAGATCGCGGCAGCTGTGGGGGCCGGATCCG 3′) and AgCARM1^R485K forward primer: (5′CAGCTGGACGCACAGGGCGCCAAAAATGCGGTCAACCTCGTCAAC3′) (bold = mutated).

In vitro methylation assay

Various enzymes and substrates were incubated in 15 μL of 5 mM MgCl_2, 20 mM Hepes pH 7.9, 1 mM EDTA, 1 mM DTT, 10% glycerol containing 1 μL of ³H-S-adenosylmethionine (GE Healthcare) for 1 h. This was then separated by SDS-PAGE and fixed in methanol for 1 h. Fixed gels were incubated in “Amplify” (Amersham Biosciences) scintillation fluid for 30 min before exposure to film.

Top–down mass spectrometry

All the protein samples (20–100 μg) were desalted by ultrafiltration using microcon centrifugal filter devices (Millipore, Billerica, MA), dissolved in 10–50% ACN and 1% acetic acid in water at 0.2–1 μg/μL, and introduced to the mass spectrometer using an automated chip-based nanoESI source, the Triversa NanoMate (Advion BioSciences, Ithaca, NY) with a spray voltage of 1.25–1.6 kV versus the inlet of the mass spectrometer, resulting in a flow of 50–200 nL/min. Intact protein molecular ions were analyzed using a linear trap/FT-ICR (LTQ FT Ultra) hybrid mass spectrometer (Thermo Scientific, Bremen, Germany). Capillary temperature was 200°C, capillary voltage was 36 V, and the tube lens voltage was 190 V. Ion transmission into the linear trap and further to the FT-ICR cell was automatically optimized for maximum ion signal with the target values (the approximate number of accumulated ions) for a full MS scan linear trap (LT) scan, FT-ICR cell (FT) scan, MSⁿ FT-ICR, ECD scan were 3 × 10⁴, 10⁶, 10⁶, and 8 × 10⁶, respectively. The resolving power of the FT-ICR mass analyzer was set at 100,000 or 400,000 m/Δm_50% at m/z 400, resulting in one scan/s acquisition rate. Individual charge states of the protein molecular ions were first isolated and then dissociated by ECD using 2–3% “electron energy” and a 50–70 ms duration time with no delay. Isolated charge states were also dissociated by CAD using 15–20% collision energy. Up to 3000 transients were averaged per spectrum to ensure high quality ECD/CAD spectra. All FT-ICR spectra were processed with Xtract Software (FT programs 2.0.1.0.6.1.4, Xcallibur 2.0.5, Thermo Scientific, Bremen, Germany) using a signal to noise threshold of 1.5 and fit factor of 60% and validated manually. The resulting monoisotopic mass lists were further assigned using in-house “Ion Assignment” software (Version 1.0) based on the recombinant CARM1 protein sequence generated from DNA sequencing with consideration of the static (mono-, di-) methylation and 10 ppm monoisotopic precursor and fragment tolerance. The M_r value reported in the study are all most abundant masses and the mass difference (in units of 1.00235 Da) between the most abundant isotopic peak and the monoisotopic peak is denoted in italics after each M_r value. The top 10 most abundant isotopic peak heights were integrated to calculate the relative abundance of the intact protein.

Conclusions

We present, here, the discovery and comprehensive characterization of an automethylation event by AgCARM1 using top–down high resolution MS/MS. Unexpectedly, we found AgCARM1^WT has already been predominantly automethylated on a single site during its expression in E. coli via this top–down MS approach which could easily have been missed in traditional bottom–up MS. Mutation of this site reveals automethylation occurs intermolecularly and is consistent with a processive model for AgCARM1^WT automethylation. This study may shed light on other bacterially expressed post-translational modification enzymes, which could be modified but overlooked in biochemical and structural studies. Top–down high resolution MS/MS shows great promise for revealing unexpected protein modification and localizing such modification to one amino acid.

Glossary

Abbreviations:

AdoMet

S-adenosylmethionine

AgCARM1

A. gambiae CARM1

CAD

collisionally activated dissociation

ECD

electron capture dissociation

ESI

electrospray ionization

FT

Fourier transform

LTQ

linear ion trap

MS

mass spectrometry

MS/MS

tandem mass spectrometry

PABP1

polyA binding protein 1

PRMT

protein arginine methyltransferase

PTM

post-translational

modification R

arginine.