Specificity and mechanism of RNA cap guanine-N2 methyltransferase (Tgs1).

The 2,2,7-trimethylguanosine (TMG) cap structure is characteristic of certain eukaryotic small nuclear and small nucleolar RNAs. Prior studies have suggested that cap trimethylation might be contingent on cis-acting elements in the RNA substrate, protein components of a ribonucleoprotein complex, or intracellular localization of the RNA substrate. However, the enzymatic requirements for TMG cap formation remain obscure because TMG synthesis has not been reconstituted in vitro from defined components. Tgs1 is a conserved eukaryal protein that was initially identified as being required for RNA cap trimethylation in vivo in budding yeast. Here we show that purified recombinant fission yeast Tgs1 catalyzes methyl transfer from S-adenosylmethionine (AdoMet) to m7GTP and m7GDP. Tgs1 also methylates the cap analog m(7)GpppA but is unreactive with GTP, GDP, GpppA, m2,2,7GTP, m2,2,7GDP, ATP, CTP, UTP, and ITP. The products of methyl transfer to m7GTP and m7GDP formed under conditions of excess methyl acceptor are 2,7-dimethyl GTP and 2,7-dimethyl GDP, respectively. Under conditions of limiting methyl acceptor, the initial m2,7GDP product is converted to m2,2,7GDP in the presence of excess AdoMet. We conclude that Tgs1 is guanine-specific, that N7 methylation must precede N2 methylation, that Tgs1 acts via a distributive mechanism, and that the chemical steps of TMG synthesis do not require input from RNA or protein cofactors.

Many small noncoding eukaryal RNAs contain a distinctive hypermodified 2,2,7-trimethylguanosine (TMG) 1 cap structure (1,2). TMG caps are also found on nematode mRNAs that undergo trans-splicing of a 5Ј-capped leader sequence (3). It is generally assumed that TMG caps are formed by post-transcriptional methylation of m 7 G caps, but there is, as far as we know, no definitive evidence for an obligate relationship between prior N7 methylation and subsequent N2 methylation. Many studies, usually entailing RNA microinjections in amphibian oocytes, have focused on the question of how the TMG cap is targeted to specific cellular mRNAs in vivo. Spliceosomal snRNAs were shown to be trimethylated in the cytoplasm in a reaction that depended on an RNA element to which Sm protein components of the snRNP bind (4). Trimethylation of snoRNAs occurs in the nucleus and depends on RNA structural motifs specific to the snoRNAs (5,6). An analysis of snRNA trimethylation in an in vitro system derived from human cytosol indicated that: (i) AdoMet is the methyl donor in the reaction; (ii) snRNA pre-assembled with Sm proteins into a snRNP particle is an effective methyl acceptor; (iii) the Sm binding site in the RNA is required for trimethylation in vitro; and (iv) free snRNA does not serve as a methyl acceptor (7).
A major step forward in defining the genetic pathway of TMG cap formation was made when Bordonné and colleagues (8) identified the yeast Tgs1 protein in an interaction screen using a yeast Sm protein as bait. A tgs1 deletion mutant was viable even though the snRNAs and snoRNAs in the tgs1⌬ strain lack TMG caps, as gauged by precipitation with anti-TMG antibody (8). The presence of a putative AdoMet binding motif in the Tgs1 polypeptide, mutation of which affected TMG formation in vivo (8,9), suggested that Tgs1 might be involved directly in TMG formation. Although Tgs1 has consequently been dubbed a trimethylguanosine synthase, there is no evidence as yet that this protein is itself the catalyst of guanosine hypermethylation and, if so, what its specificity, mechanism, and requirements for RNA or proteins partners might be. To address these issues, we produced and characterized the Tgs1 ortholog for the fission yeast Schizosaccharomyces pombe.

EXPERIMENTAL PROCEDURES
Recombinant Tgs1-The open reading frame encoding Tgs1 was amplified from an S. pombe cDNA library by PCR and inserted into pET28-His 10 -Smt3. The pET-His 10 Smt3-Tgs1 plasmid was transformed into Escherichia coli BL21CodonPlus(DE3). A 500-ml culture amplified from a single transformant was grown at 37°C in Luria-Bertani medium containing 50 g/ml kanamycin and 50 g/ml chloramphenicol until the A 600 reached 0.6. The culture was adjusted to 2% ethanol and 0.2 mM isopropyl-1-thio-␤-D-galactopyranoside and then incubated at 17°C for 20 h with constant shaking. Cells were harvested by centrifugation, and the pellet was stored at Ϫ80°C. All subsequent procedures were performed at 4°C. Thawed bacteria were resuspended in 25 ml of buffer A (50 mM Tris-HCl, pH 8.0, 200 mM NaCl, 10% glycerol). Cell lysis was achieved by the addition of lysozyme to 100 g/ml. The lysate was sonicated to reduce viscosity, and insoluble material was removed by centrifugation. The soluble extract was applied to a 1-ml column of nickel-nitrilotriacetic acid-agarose resin (Qiagen) equilibrated with buffer A. The column was washed with 10 ml of the same buffer and then eluted stepwise with 2-ml aliquots of buffer A containing 50, 100, 250, and 500 mM imidazole. The polypeptide compositions of the column fractions were monitored by SDS-PAGE. The recombinant His 10 Smt3-Tgs1 polypeptide was recovered predominantly in the 250 mM imidazole fractions. The 250 mM imidazole eluate was dialyzed against buffer containing 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 2 mM DTT, 1 mM EDTA, 10% glycerol and then stored at Ϫ80°C. The protein concentration was determined by SDS-PAGE analysis of serial dilutions of the Tgs1 preparation and a BSA standard. The gel was stained with Coomassie Blue, and the staining intensities of the His 10 Smt3-Tgs1 and BSA polypeptides were quantified with a Fujifilm FLA-5000 digital imaging and analysis system. Tgs1 concentration was calculated by interpolation to the BSA standard curve.
Methyltransferase Assay-Reaction mixtures (20 l) containing 50 mM Tris-HCl (pH 8.0), 5 mM DTT, 50 M [ 3 H-CH 3 ]AdoMet, 5 mM m 7 GTP or m 7 GDP, and enzyme were incubated for 60 min at 37°C. The mixtures were spotted on DEAE-cellulose filters (25 mm, Whatman DE81), which were washed three times batchwise with 20 mM ammonium bicarbonate (modified from Ref. 10). The filters were dried, and the radioactivity adsorbed to the filter was quantified by liquid scintillation counting. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Recombinant Tgs1 Is an AdoMet-dependent Methyltransferase-
The 239-amino acid S. pombe Tgs1 protein was produced in E. coli as a His 10 -Smt3-tagged fusion and purified from a soluble bacterial extract by adsorption to nickel-agarose and elution with imidazole. SDS-PAGE revealed two predominant polypeptides, a larger species corresponding to His 10 Smt3-Tgs1 and a smaller polypeptide corresponding to the His 10 -Smt3 tag (Fig. 1A). The methyltransferase activity of Tgs1 was demonstrated by incubating the protein with 50 M [ 3 H-CH 3 ]AdoMet and 5 mM m 7 GTP at 37°C, which resulted in label transfer from AdoMet to the m 7 GTP to form an anionic methylated nucleotide product that was adsorbed to a DEAE filter and thereby separated from the cationic AdoMet substrate. The extent of methylation increased with time and was proportional to input enzyme (Fig. 1B). Methyl transfer displayed a bell-shaped pH dependence with optimal activity at pH 8.0 in Tris-HCl buffer (data not shown); activity was 60% of the optimum at pH 9.5 (Tris-HCl) and 25% of the optimum at pH 5.5 (Tris acetate).
The recombinant protein was subjected to zonal velocity sedimentation in a 15-30% glycerol gradient (Fig. 2). Marker proteins catalase (native size 248 kDa), BSA (66 kDa), and cytochrome c (12 kDa) were included as internal standards. His 10 Smt3-Tgs1 (calculated to be a 41-kDa polypeptide) sedimented as a discrete peak (fraction 17) between BSA and cytochrome c and was clearly resolved from the free His 10 Smt3 tag (fraction 19). The methyltransferase activity profile paralleled the abundance of the tagged Tgs1 polypeptide and peaked at fraction 17. We surmise from these results that the methyltransferase activity is intrinsic to Tgs1 and that the tagged enzyme is a monomer in solution.
The product of the Tgs1-catalyzed reaction of 5 mM m 7 GTP with 5.4 M [ 3 H-CH 3 ]AdoMet was analyzed by PEI-cellulose TLC and compared with a control reaction from which Tgs1 was omitted (Fig. 4A). The chromatogram was developed with 0.2 M ammonium sulfate, which allowed resolution of m 7 GTP from m 2,2,7 GTP and AdoMet. The positions of unlabeled nucleotide standards were identified by UV shadowing. The distribution of 3 H radioactivity was gauged by cutting the chromatogram into 1-cm strips and quantifying each fraction by liquid scintillation counting. Most of the radioactivity in the no enzyme control comigrated with AdoMet in fractions 13-14, and there was no radioactivity associated with the guanine nucleotides (Fig. 4A, lower panel). In the presence of Tgs1, nearly all  Cap Guanine-N2 Methyltransferase 4022 of the [ 3 H-CH 3 ]AdoMet was consumed, and the label transferred to two products: (i) a major species peaking at fraction 6 that migrated between the m 7 GTP and m 2,2,7 GTP standards; and (ii) a second species peaking at fraction 11, one fraction ahead of the m 7 GDP standard (Fig. 4A, upper panel). We surmise that the major reaction product is 2,7-dimethyl GTP and the minor product is likely to be 2,7-dimethyl GDP, formed by reaction of Tgs1 with the contaminating m 7 GDP nucleotide present in the commercial m 7 GTP preparation, which was visible on the UV-shadowed chromatogram.
We also analyzed the product of Tgs1 reaction with 5 mM m 7 GDP with 5.4 M [ 3 H-CH 3 ]AdoMet by PEI-cellulose TLC developed with 0.1 M ammonium sulfate (Fig. 4B). Most of the radioactivity in the no enzyme control comigrated with AdoMet in fractions 11-12 (Fig. 4B, lower panel). In the presence of Tgs1, the [ 3 H-CH 3 ]AdoMet was depleted and the label transferred to a single product peaking at fraction 6, between the m 7 GDP and m 2,2,7 GDP standards (Fig. 4B, upper panel). Formation of only one 3 H-labeled product migrating between m 7 GDP and m 2,2,7 GDP was confirmed by autoradiography of a duplicate TLC plate (not shown). We surmise that the reaction product is 2,7-dimethyl GDP.
We exploited the TLC assay to gauge the extent of methyl transfer from 50 M [ 3 H-CH 3 ]AdoMet to m 7 GDP as a function of m 7 GDP concentration (Fig. 5A). Cap Guanine-N2 Methyltransferase 4023 affinity for its substrate AdoMet and its product AdoHcy. Addition of up to 20 mM GDP had only a modest (40%) inhibitory effect on methylation of 1 mM m 7 GDP by Tgs1 (Fig. 5B). From the inhibition profile, we estimate that Tgs1 binds GDP with one-thirtieth the affinity that it binds m 7 GDP. TLC analysis confirmed that no 3 H-labeled m 2 GDP was formed by Tgs1 in the presence of 20 mM GDP (not shown).

Synthesis of Trimethylguanosine by a Distributive
Mechanism-The absence of 2,2,7-trimethylguanosine products in the experiments presented above implies that either: (i) Tgs1 is not responsible for the second methylation reaction; or (ii) Tgs1 does catalyze the second methylation reaction, but we are precluded from detecting it because the enzyme acts distributively, i.e. the labeled m 2,7 GTP or m 2,7 GDP product dissociates after a single round of catalysis and must compete with a Ͼ1,000-fold molar excess of unlabeled m 7 GTP or m 7 GDP for rebinding to Tgs1. To address these issues, we analyzed the products of methylation reactions containing low concentrations of the methyl acceptor. As outlined in Fig. 6, we incubated Tgs1 with 50 M m 7 GDP and 100 M [ 3 H-CH 3 ]AdoMet for 60 min and then supplemented the reaction mixture with 1 mM cold AdoMet and allowed the reaction to proceed. Aliquots were taken at 30-min intervals; the products were analyzed by TLC and the 3 H-methylated nucleotides visualized by autoradiography. The instructive finding was that the majority of the m 2,7 GDP formed during the pulse-labeling phase was subsequently converted to m 2,2,7 GDP during the chase phase in the presence of excess cold AdoMet (Fig. 6). These results show that Tgs1 is a bona fide trimethylguanosine synthase capable of catalyzing serial methylation reactions at the exocyclic amino nitrogen of m 7 G via a distributive mechanism.

DISCUSSION
The experiments presented here show that Tgs1, a protein implicated genetically in TMG cap formation (8), is indeed a catalyst of TMG synthesis. A key finding is that methylation of guanine N2 by Tgs1 in vitro is strictly dependent on prior methylation of guanine N7. This requirement is unique to Tgs1, i.e. no such requirement applies to two other RNA guanine-N2 methyltransferases, Trm1 and Trm-G10 (11,12), and it suffices to explain why Tgs1 is cap-specific. Previous in vivo studies of Tgs1 had not illuminated this specificity, because the genetic analysis is complicated by the fact that whereas Tgs1 is not required for growth of S. cerevisiae (8), the upstream enzyme (Abd1) that catalyzes cap guanine-N7 methylation is essential for growth (13). Moreover, the reliance on antibodies to characterize the state of the cap in mutant yeast strains (8) places limits on what can be measured, i.e. anti-TMG antibodies might not be of use in determining whether snRNAs synthesized at restrictive temperature in an abd1-ts strain are methylated at N2 in the absence of methylation at N7.
Our results argue against Tgs1 acting processively to add two methyl groups to the guanine N2 following a single event of m 7 GDP binding to the active site. In a processive mechanism, the AdoHcy product of the first methylation step would have to dissociate and the N-CH 3 rotate about the C2-N2 bond to the purine ring in order to clear the transferred methyl group out of the active site to make room for an incoming AdoMet donor for the second methylation step. Our findings favor a distributive mechanism in which the AdoHcy and m 2,7 GDP products dissociate from Tgs1 after the first methylation step, and the second methylation step entails rebinding of m 2,7 GDP from solution.
The finding here that guanine-N2 methylation of the cap by Tgs1 in vitro requires no RNA component and no protein cofactor forces reevaluation of the prevailing models that TMG synthesis depends strictly on cis-acting RNA signals or the assembly of specific RNP structures. Rather, our results instate a more conservative model in which RNP components might simply target Tgs1 to a particular subset of cellular RNAs that already have an m 7 G cap. This view of TMG formation, in which Tgs1 macromolecular interactions dictate access but not catalysis, mirrors the problem of enzymatic specificity versus cellular targeting encountered during the formation of the m 7 G cap itself. To wit, m 7 G capping in vivo is targeted specifically to cellular transcripts synthesized by RNA polymerase II, but the enzymes that perform the three capping reactions, starting from a triphosphate-terminated RNA, have no intrinsic ability to discriminate one RNA substrate from another as long as it has the requisite number of 5Ј-phosphates. Indeed, cap guanine-N7 methyltransferase is capable of methylating free guanine nucleotides (14) and thus does not even require an RNA moiety to achieve its specificity in catalysis (15). The capping apparatus is targeted to nascent RNAs by virtue of direct binding of the component enzymes to the RNA polymerase II transcription elongation complex (16).
In conclusion, the present biochemical characterization of Tgs1 as an autonomous cap-specific guanine-N2 methyltransferase should stimulate efforts to dissect substrate selectivity determinants in a defined reconstituted system and to probe the structural basis for catalysis of TMG formation.