Structure and Biophysical Characterization of the S-Adenosylmethionine-dependent O-Methyltransferase PaMTH1, a Putative Enzyme Accumulating during Senescence of Podospora anserina *

Background: PaMTH1, a putative O-methyltransferase protects Podospora anserina from oxidative stress during senescence and acts as a longevity assurance factor. Results: Crystal structures of PaMTH1/PaMTH1-SAM/SAH co-complexes and NMR-based characterization of enzymatic methylation of its substrate were obtained. Conclusion: PaMTH1 catalyzes methyl group transfer from the co-factor to the substrate in a cation-dependent manner. Significance: This work facilitates the identification of endogenous polyphenolic compounds directly involved in metal-induced oxidative stress. Low levels of reactive oxygen species (ROS) act as important signaling molecules, but in excess they can damage biomolecules. ROS regulation is therefore of key importance. Several polyphenols in general and flavonoids in particular have the potential to generate hydroxyl radicals, the most hazardous among all ROS. However, the generation of a hydroxyl radical and subsequent ROS formation can be prevented by methylation of the hydroxyl group of the flavonoids. O-Methylation is performed by O-methyltransferases, members of the S-adenosyl-l-methionine (SAM)-dependent O-methyltransferase superfamily involved in the secondary metabolism of many species across all kingdoms. In the filamentous fungus Podospora anserina, a well established aging model, the O-methyltransferase (PaMTH1) was reported to accumulate in total and mitochondrial protein extracts during aging. In vitro functional studies revealed flavonoids and in particular myricetin as its potential substrate. The molecular architecture of PaMTH1 and the mechanism of the methyl transfer reaction remain unknown. Here, we report the crystal structures of PaMTH1 apoenzyme, PaMTH1-SAM (co-factor), and PaMTH1-S-adenosyl homocysteine (by-product) co-complexes refined to 2.0, 1.9, and 1.9 Å, respectively. PaMTH1 forms a tight dimer through swapping of the N termini. Each monomer adopts the Rossmann fold typical for many SAM-binding methyltransferases. Structural comparisons between different O-methyltransferases reveal a strikingly similar co-factor binding pocket but differences in the substrate binding pocket, indicating specific molecular determinants required for substrate selection. Furthermore, using NMR, mass spectrometry, and site-directed active site mutagenesis, we show that PaMTH1 catalyzes the transfer of the methyl group from SAM to one hydroxyl group of the myricetin in a cation-dependent manner.

erated either endogenously during metabolic processes or exogenously due to environmental exposure (1). Accumulation of ROS can potentially damage proteins, lipids, carbohydrates, and DNA, resulting in several pathological occurrences like aging and age-associated diseases, including cancer (2,3). ROS interfere with cellular functions by either inactivating enzymes with sulfhydryl groups via oxidation, lipid peroxidation, and subsequent increase in membrane permeability; depolymerization of polysaccharides; or degradation of DNA (4,5). Although low concentrations of ROS (nitric oxide (NO), superoxide anion (O 2 . ), and hydrogen peroxide (H 2 O 2 )) facilitate important signaling and biological functions, high concentrations of H 2 O 2 , hydroxyl radicals (OH ⅐ ), and O 2 . can be damaging to biologically significant targets and constitute the basis for the free radical (oxidative stress) theory of aging (1,6). Among all three ROS, the hydroxyl radical is the most reactive and toxic, and no detoxification system exists for it (7). Some transition metals, in particular iron and copper, catalyze the formation of hydroxyl radicals from superoxide radicals or hydrogen peroxide (Haber-Weiss and Fenton's reaction) (4). Apart from generating ROS via the above reaction, metals like iron and copper are also known to interact with naturally occurring antioxidants like polyphenolic compounds (flavonoids) or vitamin C, thereby transforming an antioxidant into a prooxidant (8). Flavonoids, a frequent component of the human diet, in general exhibit their beneficial antioxidant properties by inhibiting several enzymes, including oxidases, lipases, and protein kinases (9 -13). However, apart from these beneficial effects, some flavonoids in vitro are known to be mutagenic, due to their prooxidant behavior (14). For example, quercetin, a plant flavonoid, in the presence of ferrous ions generates hydroxyl radicals and leads to enhanced sulfhydryl group oxidation of the enzyme glyceraldehyde 3-phosphate dehydrogenase and its subsequent inactivation. In the absence of metal ions, quercetin becomes protective in function and prevents sulfhydryl oxidation (15,16). Therefore, the presence/ absence of metal ions modulates the biological or pharmacological behavior of flavonoids to act as an antioxidant or prooxidant (17). Several studies in model organisms have established substantial correlation between age-related accumulation of metal ions and enhanced formation of ROS (18,19). Most of the polyphenols, including flavonoids, react with metals via their vicinal dihydroxyl system and lead to generation of ROS (20).
Several defense mechanisms (enzymes, vitamins, and metabolites) in the cells are involved in either preventing the formation of ROS or inactivating them (21). Particularly, in flavonoids, methylation of the vicinal hydroxyl groups and their subsequent conversion to methyl ethers prevent their interaction with metals, thereby abolishing their prooxidant activity (22). Methylation of biological compounds is also an integral part of various cellular processes, such as protein trafficking, signal transduction, biosynthesis, metabolism, and gene expression (23). Methyltransferases (MTs), via an S N 2-like nucleophilic substitution reaction mechanism, catalyze the transfer of a methyl group from a donor to an acceptor molecule, resulting in the formation of methylated products and a by-product. Most often, methylation is catalyzed by S-adenosyl-L-methio-nine (SAM)-dependent methyltransferases. SAM-dependent methyltransferases (SAM-MTs) are ubiquitously present throughout prokaryotic and eukaryotic organisms and are responsible for methylation of structurally distinct biological and natural product substrates (24,25). SAM, an essential metabolic intermediate in several organisms, functions classically as the electron-deficient methyl donor. S-Adenosyl homocysteine (SAH) is the by-product of methylation and is a potent feedback inhibitor of many SAM-MTs. Depending on the methyl-accepting atom on the substrates, MTs are categorized as O, N, C, or S. Hydroxyl group methylation (O-methylation) is performed by SAM-dependent O-methyltransferases.
Structurally, many SAM-MTs adopt a Rossmann-like fold with an ␣/␤ domain, a characteristic requirement for binding of nucleotide-containing co-factors. The Rossmann-like fold is characterized by an ␣␤␣ sandwich structure consisting of seven ␤-strands flanked by two layers of ␣-helices. The C-terminal regions of the ␤-strands and the adjoining loops form the core of the catalytic site and mediate important interactions with the co-factor SAM and the substrate. The GXGXG motif present at the end of the ␤1-strand in Rossmann fold-containing SAM-MTs is widely conserved and makes direct contact with the carboxypropyl group of SAM. In addition to the core fold, additional helices and extensions in the N terminus play a significant role in oligomerization and/or modulate the substrate specificity (25). For many hydroxyl-modifying SAM-MTs, the catalytic activity is dependent on the presence of divalent metal ions like Mg 2ϩ or Ca 2ϩ . These divalent cations play a key role in structural stabilization and typically participate directly in substrate binding to certain metal-dependent O-MTs (24).
For more than 5 decades, Podospora anserina, a filamentous ascomycete, has been a well studied and well established aging model (26). Interestingly, during senescence, an O-MT, PaMTH1, was reported to accumulate in total and mitochondrial extracts. This result suggested that PaMTH1 plays a key role during aging (27). Subsequent sequence analysis of PaMTH1 revealed significant homology to SAM-MTs. It was hypothesized that PaMTH1 has a protective function against oxidative stress. Activity and biochemical analysis revealed specificity of this protein for flavonoids containing vicinal hydroxyl groups in general and myricetin in particular (28), which in the presence of metals can generate ROS. Constitutive overexpression of PaMth1 in P. anserina resulted in increased resistance against exogenous oxidative stress and marked reduction in ROS-induced damage of proteins, thereby prolonging the life span (28). Deletion of PaMth1 resulted in increased sensitization for several stress factors, including metals (copper) and hydrogen peroxide (29).
PaMTH1, a 27-kDa protein, based on conserved amino acid sequence motifs, is classified as an enzyme belonging to the cation-dependent subclass (class I) of SAM-MTs. PaMTH1 shows a striking sequence similarity with catechol-O-methyltransferase (COMT), an enzyme belonging to the same superfamily, and is involved in the methylation of catechols in both plants and humans. However, the molecular architecture of PaMTH1 and the biophysical basis for the methyl transfer reaction remain unknown.
Here, we present crystal structures of the PaMTH1 apoenzyme, a PaMTH1-SAM (co-factor) co-complex, and a PaMTH1-SAH (by-product) co-complex refined to 2.0, 1.9, and 1.9 Å, respectively. The enzyme is a highly stable dimer both in solution and in the crystal due to swapping of the N termini. Structural analysis of the apo-and holoenzyme reveals an overall architecture similar to SAM-MTs but shows significant differences in the substrate binding region. The ␣2-␣3 loop, involved in substrate binding, adopts a "closed" conformation in the presence of SAM/SAH, as observed in other SAM/SAHbound structures, and an "open" conformation in the apoenzyme. In addition, using sequence and structural homology, residues important for substrate binding were predicted and probed using site-directed mutagenesis. Based on NMR chemical shift perturbations, we show that SAM and the substrate (myricetin) bind to PaMTH1 and catalyze the methyl group transfer from SAM to one hydroxyl group of myricetin in a cation-dependent manner. Furthermore, by using mass spectrometry, we confirmed the conversion of myricetin to monomethylated myricetin.

Experimental Procedures
Protein Expression and Purification-The PaMth1 gene was cloned into the expression vector pETTEV-16b and transformed in Escherichia coli BL21 (DE3) cells (Invitrogen) for overexpression of the fusion protein with an N-terminal His 7 tag. Transformed E. coli cells were grown at 37°C in autoinduction medium ZYM-5052 (30) containing 100 g/ml ampicillin to an A 600 of ϳ0.6, and then the temperature was reduced to 20°C to support soluble protein expression. After 48 -60 h of cell growth (due to the long induction time, hydrolysis of the antibiotic can occur, so ampicillin was replenished every 24 h), bacteria were harvested by centrifugation (20 min, 4000 rpm) and resuspended in lysis buffer (20 mM Tris-HCl, pH 8.0, 200 mM NaCl, 10 mM ␤-mercaptoethanol) with the addition of complete protease inhibitor (1 tablet/100 ml of EDTA-free; Roche Applied Science). Selenomethionine-derivatized protein was expressed according to standard EMBL protocols (see the EMBL Web site). The cells were lysed using microfluidizer (15,000 p.s.i., 2 cycles), and the cell debris was separated by centrifugation (30 min, 16,000 rpm). The supernatant was filtered and passed through a 5-ml nickel-nitrilotriacetic acid column (GE Healthcare), washed with 10 bed volumes of wash buffer (same as lysis buffer), and the His-tagged protein was eluted using an isocratic gradient (0 -100%) of wash buffer and elution buffer (20 mM Tris-HCl, pH 8.0, 200 mM NaCl, 500 mM imidazole, 15 mM ␤-mercaptoethanol) for 8 column volumes. The eluted fractions were analyzed by SDS-PAGE, and the ones containing PaMTH1 were pooled, incubated with tobacco etch virus protease to remove the N-terminal His tag (1 mg of tobacco etch virus protease (in-house produced) per 10 mg of PaMTH1) and dialyzed for 24 h at 4°C against 25 mM Tris, pH 8.0, 100 mM NaCl, 15 mM ␤-mercaptoethanol, and 5% glycerol. The cleaved protein was reloaded onto a nickel-nitrilotriacetic acid column for removal of the tag and the tobacco etch virus protease. Further purification of protein was achieved by gel filtration chromatography on a Superdex-200 column (GE Healthcare) equilibrated with 25 mM HEPES, pH 7.5, 100 mM NaCl, and 1 mM DTT. Fractions containing the target protein were pooled and concentrated up to ϳ5 mg/ml and stored at Ϫ80°C or immediately used for experiments. 15 N-Labeled protein was expressed in M9 minimal medium containing 15 NH 4 Cl (1 g/liter) as the sole nitrogen source.
Mutagenesis-Site-directed mutagenesis was used to introduce the catalytic mutations (K147A, D144A/K147A, D172A/ N173A) and N-terminal swapping region mutation (S4A/P7A) into the wild-type PaMTH1 plasmids following the protocol mentioned in QuikChange kit (Stratagene). Mutations were confirmed by sequencing (Eurofins Genomics). N-terminal swapping region deletion was achieved through PCR amplification of wild-type plasmid using 5Ј-gtgtgtcatatggagacggcggacc-3Ј (forward) and 5Ј-gtgtgtagatcttcaagcccgctgag-3Ј (reverse) primers and cloning of the restriction-digested PCR products in pETTEV-16b vector. All of the mutant proteins were purified following the same protocol as the wild-type PaMTH1. Circular dichroism and 1 H, 15 N HSQC (heteronuclear spin quantum correlation) spectra for the wild type and the mutants are mostly similar, indicating that there are no significant changes in the secondary structure due to mutation. Further, size exclusion chromatography (SEC) analysis of the mutants showed no change in their respective oligomeric state due to mutation (data not shown).
SEC Coupled with Multiangle Laser Light Scattering (SEC-MALS)-SEC-MALS was performed using a TSK-GEL G3000SWXL column (15 ml; Tosoh Bioscience, Stuttgart, Germany); a light scattering detector (TREOS) and refractometer (Optilab rEX) from Wyatt Technology (Dernbach, Germany); and a UV detector, HPLC pump, and degasser from Jasco (Gross-Umstadt, Germany). The system was equilibrated with 2 column volumes of SEC buffer (50 mM Tris, 150 mM NaCl, pH 6.9, filtered through 0.1-m pore size VVLP filters (Merck Millipore, Darmstadt, Germany)) following a recirculation through the system for at least 1 day at 0.5 ml/min to improve the baseline drastically by removing air bubbles and particles by the degasser and preinjection filter (0.1 m). Per measurement, 200 -300 g of protein in 200 l of SEC buffer were injected and analyzed at a flow rate of 0.5 ml/min at 4°C. The light scattering detector was calibrated every day by using monomeric BSA and carboanhydrase. The obtained signals were processed with the ASTRA software package version 5.3.4.13 (Wyatt Technology) to calculate the molecular mass using the UV signal for concentration determination. A refractive index increment dn/dc of 0.185 ml/g and an extinction coefficient ⑀ 280 of 51,910 liters/ (cm mol) was assumed.
Isothermal Titration Calorimetry-All titration experiments were performed at 25°C using a VP-ITC microcalorimeter (MicroCal Inc.). All titrations were performed using buffer containing 25 mM Tris, pH 7.5, 100 mM NaCl, 0. The protein concentrations were calculated from the UV absorption at 280 nm with a Nanodrop spectrophotometer (Thermo Fisher Scientific) using the extinction coefficient obtained from the amino acid sequence (51,910 liters/(cm mol)). Concentrations of compounds were obtained similarly using the known extinction coefficient (⑀ 260 of 15,400 liters/(cm mol) for both SAM and SAH). The raw ITC data were analyzed with the ITC-Origin version 7.0 software with a "one-site" binding model after correction for the dilution heat of the compounds.
NMR Spectroscopy-NMR experiments were performed at T ϭ 298 K on Bruker 600-, and 800-MHz spectrometers equipped with room temperature TXI-HCN probes. NMR samples were prepared with 10% D 2 O to lock the spectrometers, and 3-(trimethylsilyl)-2,2Ј,3,3Ј-tetradeuteropropionic acid (TSP-d 4 ; 1 mM) was used as an internal standard for spectral referencing. The processing and analysis of NMR spectra were done in Topspin version 2.1 (Bruker Biospin).
All ligands, including co-factor (SAM), by-product (SAH), and putative substrates (apigenin, kaempferol, pyrocatechol, and myricetin) were purchased from Sigma-Aldrich and used without further purifications. The solubility of ligands (in H 2 O and buffer 25 mM Tris, pH 7.5, 100 mM NaCl, 5 mM TCEP) were determined by a one-dimensional 1 H NMR experiment (Bruker, 600 MHz) using peak integrals of 1 H peaks of the ligands, which were normalized against the calibrated peak of 1 mM TSP at 0 ppm. The assignments of one-dimensional 1 H peaks corresponding to SAM (BMRB ID: bmse000059), SAH (BMRB ID: bmse000289), and pyrocatechol (BMRB ID: bmse000385) were obtained from the Biological Magnetic Resonance Bank (BMRB) metabolomics database, and the assignment for myricetin (HMDB ID: HMDB02755) was obtained from the Human Metabolomics Data Base (HMDB).
The affinity for each titration was determined by a simultaneous fit of the titration parameters to the observed CSPs. The affinity of the titrated ligand and the chemical shifts in the complex structure were used as fitting parameters. A single step binding mechanism in the fast or intermediate exchange regime was presumed. The differential equations were solved numerically to determine the equilibrium concentrations, presuming a given affinity. No statistical evidence for more complicated binding mechanisms could be found.
The enzymatic reaction of PaMTH1 (wild type and mutants) was monitored by recording a series of one-dimensional 1 H NMR experiments (Bruker 600 MHz) of the reaction mixture containing 100 M PaMTH1, 1 mM SAM (unlabeled), and 1 mM myricetin (1:10:10) in deuterated Tris buffer containing MgCl 2 (25 mM d-Tris, pH 7.5, 100 mM NaCl, 1 mM MgCl 2 , 5 mM TCEP). The methoxy peak of the product was identified by measuring one-dimensional 1 H NMR spectra of the reaction mixture after removal of the protein (denaturation by heat at 70°C for 5 min, followed by centrifugation at 13,000 rpm for 10 min). Two-dimensional 1 H, 13 C HSQC (44) and multiplicity-edited 1 H, 13 C HSQC (45) (Bruker 800 MHz) experiments were employed to confirm the methylation status of myricetin. The reaction mixture for the two-dimensional NMR experiments was prepared using 13 C, 15 N-labeled SAM (synthesized enzymatically from 13 C, 15 N-labeled ATP and 13 C, 15 N-labeled methionine as described previously (46)). To further corroborate the formation and methylation status of the product (methoxymyricetin), electron spray ionization-mass spectrometry of the reaction mixture was also performed.

Results
Overall Structure of PaMTH1-Recombinant PaMTH1 was overexpressed in E. coli and purified both in native and selenomethionine-substituted forms. The crystal structure refined to 2.0 Å was solved using experimental Se-MAD (Table 1). The protein crystallizes as a homodimer (Fig. 1A). On SEC, PaMTH1, a 27-kDa protein, eluted as a single peak and with an elution volume corresponding to a 44-kDa protein. To independently ascertain the exact molecular weight in solution, we performed in-line SEC-MALS. The measured molecular weight confirms that PaMTH1 exists as a dimer in solution (Fig. 1A,  inset).
Most of the structurally characterized plant O-MTs are dimers (47), including the caffoeyl-CoA O-methyltransferase (CCoAOMT), which displays significant sequence homology to PaMTH1 (Fig. 2). Previous structures of O-MTs have shown that the SAM-binding site is located in the vicinity of the dimer interface. In addition, many MTs contain an additional N-terminal extension or dimerization domain also important for substrate binding and specificity (24,25). PaMTH1 lacks such an additional dimerization domain, and the interaction between the monomers in the dimer appears to be unique. The dimerization interface buries 6270 Å 2 of surface area and accounts for 32% of the total available surface area of the dimer, consistent with previously reported small molecule O-MTs. Importantly, the first 10 N-terminal loop residues are swapped between the two monomers ( Fig. 1, A and B). Additionally, the interface includes interactions involving helices ␣1, ␣3, and ␣9 and ␤-strands ␤6 and ␤7. The N-terminal loop directly inserts into the catalytic pocket of the neighboring molecule and fixes the positions of helix ␣3 and the ␣8-␣9 loop in the substrate binding region via a hydrogen bond network, suggesting a critical role in substrate binding (Fig. 1B). The hydroxyl group of Ser 4 (chain A) in the loop hydrogen bonds with the hydroxyl group of Ser 50 (chain B) of helix ␣3, and the carbonyl oxygen of the Pro 7 (chain A) hydrogen bonds with the N⑀1 hydrogen of Trp 188 (chain B) in the ␣8-␣9 loop (Fig. 1B). An analysis of the PaMTH1 dimer using PISA (48)  CCoAOMT (Medicago sativa) (Protein Data Bank entry 1SUI) is one of the closest structural homologues of PaMTH1 and also crystallizes as a dimer. The dimerization interface buries 25% of the total available surface area of the dimer. In the context of catalysis, dimerization of CCoAOMT is not critical for its activity (49). However, similar to other plant O-MTs, dimerization of PaMTH1 might be required either for catalysis or to impart substrate specificity, because its N-terminal loop contributes to the active site of the dyad-related monomer. Interestingly, in the dimeric structures of two bacterial O-MTs (LiOMT from Leptospira interrogens (50) and Syn-OMT from cyanobacterium Synechocystis sp. (51)), the N-terminal loop (first 10 -13 residues) plays an important  Each monomer subunit of PaMTH1 consists of nine ␣-helices and seven ␤-strands, and folds into a globular tertiary structure consisting of a core ␣/␤ Rossmann fold (Fig. 1C), typical for many SAM-binding MTs. The seven-stranded ␤-sheet core is sandwiched between two helical regions. As in all Rossmann fold SAM-dependent MTs, the ␤-sheet adopts a strand topology ␤3, ␤2, ␤1, ␤4, ␤5, ␤7, ␤6, with all strands oriented parallel to each other except for ␤7, which is antiparallel (Fig. 1D).
PaMTH1 Binding to Co-factor (SAM), By-product (SAH), and Substrate-We used NMR spectroscopy to investigate the PaMTH1-SAM/SAH interaction in solution. 1 H, 15 N HSQC experiments can detect ligand binding to a target through CSPs and serve as a qualitative tool for identifying binders. In addition to the CSPs, the complete disappearance of signals can also occur upon complexation, which is also indicative of binding. An overlay of 1 H, 15 N HSQC spectra of PaMTH1 in the presence and absence of SAM/SAH shows that the addition of the ligand induces both CSPs and disappearance of a subset of NMR signals, indicating specific binding (Fig. 3, A and B). Previous studies have shown that both SAM and SAH occupy the same binding pocket in most methyltransferases. A comparison between SAM-and SAH-induced NMR-spectral changes reveals that there is a significant overlap, suggesting an identical or at least overlapping binding pocket for SAM and SAH in PaMTH1.
Subsequently, we co-crystallized SAM/SAH with PaMTH1 and solved the three-dimensional structures of the co-complexes refined to 1.9 Å ( Table 1). The structures of PaMTH1 in complex with SAM and SAH clearly reveal a conserved SAM/ SAH binding motif (Fig. 1E). Positional conservation of the amino acids involved in the co-factor/by-product binding is directly evident from the crystal structures of SAM/SAH bound to PaMTH1 as well as sequence alignments of the closely related MTs (Fig. 2). Analogous to other SAM-MTs, the binding pocket is located on the C-terminal end of the ␤-strands. SAM/SAH binding within the active site pocket of PaMTH1 is mediated through hydrogen bond networks and van der Waals andstacking interactions (Fig. 1E). The GXGXG motif is highly conserved in Rossmann fold SAM-dependent MTs and is considered to be a hallmark of the SAM-binding site (25). In PaMTH1, this region (GCYSG) is well conserved except that FIGURE 2. Multiple sequence alignment. Sequence alignment of the amino acid sequences of PaMTH1 from P. anserina, LiOMT from Leptospira interrogans, CCoAOMT from V. vinifera, and COMT from R. norvegicus and H. sapiens. The secondary structure of PaMTH1 is indicated above the sequences, and the percentage sequence identity of PaMTH1 to other O-MTs is shown in parenthesis. The ␣-helices and ␤-strands are indicated by cylinders (pink) and arrows (blue), respectively. The analysis was performed by multiple-alignment program ClustalW. Strictly conserved residues among the four proteins are marked by an asterisk, and highly conserved residues are indicated by a colon and dot. The highly conserved motifs (I-VIII) are highlighted in yellow boxes. Amino acids involved in the dimer interface are indicated by violet boxes, and the ones involved in SAM binding, substrate/metal binding, and catalysis are indicated by red, blue, and green, respectively. the central glycine of the GXGXG motif is substituted by a tyrosine (Tyr 75 ). In the apoenzyme, Tyr 75 points into the SAM/ SAH binding pocket, and upon SAM/SAH binding, it flips and reorients to establish direct hydrogen bonding with Asp 44 , thereby stabilizing the ␣2-␣3 loop (Fig. 1E). This reorientation of the Tyr 75 and the subsequent stabilization of the ␣2-␣3 loop in a closed conformation might possibly function to allow substrate specificity and binding.
Flavonoids are potential substrates of PaMTH1. Previous biochemical functional assays tested several phenolic putative PaMTH1 substrates and showed that the flavonoid myricetin was the best recognized substrate (28). However, the biophysical and structural evidence for methylation by PaMTH1 remains unknown. Three putative flavonoids (apigenin, kaempferol, and myricetin) and pyrocatechol were chosen for characterization of their interaction with PaMTH1 by NMR. Solubility of a compound is one of the essential factors for inter-action studies. One-dimensional NMR-based solubility analysis of the chosen compounds in the interaction buffer revealed that two compounds (apigenin and kaempferol) were insoluble under the conditions of the experiment, whereas myricetin and pyrocatechol were soluble (data not shown). 1 H, 15 N HSQC spectra of PaMTH1 in the presence and absence of myricetin or pyrocatechol show that the addition of the ligand induces both CSPs and disappearance of a subset of NMR signals, indicating specific binding (Fig. 3, C and D) even in the absence of co-factor.
Cation-dependent Methylation of Substrate by PaMTH1-Previously, based on the sequence and the size of PaMTH1 (27 kDa), it had been suggested that the protein belongs to the small cation-dependent class I and not to the cation-independent class II with a molecular mass of about 40 kDa (28). Sequence comparison shows that the residues Asp 144 , Lys 147 , Asp 172 , and Asn 173 , which are important for divalent cation binding, are FIGURE 3. PaMTH1 binding to co-factor and product. Overlay of two-dimensional 1 H, 15 N HSQC spectra of PaMTH1 without (black) and with ligands (protein/ligand ratio 1:2). A, SAM (orange); B, SAH (green); C, myricetin (red); and D, pyrocatechol (gray). The inset shows one-dimensional 1 H spectra of A. SAM with protons from the adenosine moiety (a and b) and the methyl group (d) B, SAH with protons from the adenosine moiety (a and b). C, myricetin. D, pyrocatechol is shown with protons from the catechol moiety highlighted. All spectra were acquired at 800 MHz (T ϭ 298 K) in the following buffer: 25 mM Tris, pH 6.5, 100 mM NaCl, 5 mM TCEP, 10% D 2 O, and 1 mM TSP-d 4 .
highly conserved in PaMTH1. Cation binding is important for substrate binding and orientation for the transmethylation reaction. We investigated the cation dependence of the myricetin O-methylation catalyzed by PaMTH1 using NMR spectroscopy. The protons of the methyl group bound to the oxygen of an aromatic ring (methoxy) resonate around at ϳ4 ppm, whereas protons of aliphatic methyl groups typically are found at ϳ1 ppm. SAM acts as a methyl group donor required for methylation of the substrate. In a proton one-dimensional NMR spectrum of SAM, the thiomethyl group appears as a singlet peak at 2.98 ppm (Fig. 3A, inset). We then took advantage of the typical chemical shifts of methyl/methoxy groups and based our analysis on the fact that the catalytic transfer of a methyl group from SAM to the aromatic hydroxyl oxygen of the substrate will result in a decrease in the intensity of the thiomethyl signal of SAM and the appearance of the methoxy methyl signal of the product at distinct chemical shifts.
To determine the metal requirements of the myricetin methylation reaction, we investigated the effect of the presence and absence of Mg 2ϩ and Ca 2ϩ ion on the catalytic activity of PaMTH1. The one-dimensional 1 H NMR spectrum of a reaction mixture of SAM, substrate (myricetin), and PaMTH1 shows that in the absence of cations, the intensity of the methyl group signal of SAM remains constant even after 8 h, indicating no catalysis (Fig. 4A). In the presence of Mg 2ϩ , we observe a fast decrease in the intensity of the methyl group signal, indicating catalysis of the methyl transfer reaction (Fig. 4B). In the presence of Ca 2ϩ , the enzymatic activity is significantly slowed down (Fig. 4C). This strongly suggests that PaMTH1 is a divalent cation (Mg 2ϩ )-dependent methyltransferase. The influence of the Mg 2ϩ ion on the catalytic activity of PaMTH1 may be directly associated with the enhanced ability of the enzyme to form a more stable complex with myricetin (see also below). Although pyrocatechol binds to PaMTH1 (Fig. 3D), even in the absence of Mg 2ϩ , the enzyme does not catalyze its methylation even in the presence of Mg 2ϩ (Fig. 4D), suggesting that flavonoids are the preferred substrates.
During enzymatic methyl group transfer to myricetin, SAM donates the methyl group and is converted to SAH. Comparison of the aromatic region of the one-dimensional 1 H NMR spectrum reveals a decrease in the signal intensity of the adenosine protons of SAM and a simultaneous increase in the signal intensity for SAH adenosine protons, confirming the formation of SAH (Fig. 5A). It should also be possible to directly detect the formation of methoxymyricetin by inspecting the spectral region at ϳ4 ppm typical for the methoxy signal. However, due to significant overlap with protein signals in this region, a methoxy signal could not be resolved. Upon removal of the protein by heat denaturation, however, a clear signal for the methoxy protons could be resolved at 3.93 ppm (Fig. 5B, inset), confirming the formation of methoxymyricetin. Further, using 13 C, 15 N-labeled SAM, we were able to detect and confirm the formation of methoxymyricetin by measuring multiplicity-edited two-dimensional 1 H, 13 C HSQC spectra in which the methyl group signals are inverted in comparison with CH and CH 2 signals. Only one clear cross-peak ( 1 H ϭ 3.93 and 13 C ϭ 58.9 ppm) for the methoxy group was detected, indicating the specific methylation of a single hydroxyl group of myricetin ( Fig. 5B). Further, mass spectrometry analysis of the reaction mixture confirmed that myricetin is monomethylated (Fig. 5C). However, several attempts to isolate methoxymyricetin by HPLC and characterize the position of methylation were unsuccessful, presumably due to its low stability.
Feedback Inhibition of PaMTH1 by SAH-SAH, the by-product of methionine transmethylation, acts as feedback inhibitor of many SAM-dependent MTs and plays a role in the control of the overall methyl transfer rates (52). The time-dependent onedimensional 1 H NMR spectra recorded for a reaction mixture of SAM, substrate (myricetin), and PaMTH1-SAH (PaMTH1 saturated with SAH, 1:20 ratio) shows a significant decrease in the reaction rate as monitored by the decrease in intensity of the thiomethyl group NMR signal of SAM. This result demonstrates that SAH saturation of PaMTH1 (Fig. 6A) significantly reduced the catalytic efficiency compared with free PaMTH1 (Fig. 5B). Further, the 1 H, 15 N HSQC spectrum of PaMTH1 bound to SAH shows no changes upon the addition of SAM, indicating that SAH has a higher binding affinity to PaMTH1 compared with SAM (Fig. 6B). We then used isothermal titration calorimetry to determine the binding affinities of SAM and SAH to PaMTH1, respectively. In line with the NMR results, ITC data show that SAH binds (K D ϭ 3.2 M) to PaMTH1  Table 2).
Structural Comparisons with Other O-MTs-The unavailability of structural information for the ternary complex of PaMTH1 bound to co-factor (SAM) and the substrate hinders a complete description of the binding pocket involved in cataly-sis. Instead, we used sequence homology and the Dali server to compare PaMTH1 with the structures of other O-MTs with known function.
PaMTH1 presents significant homology to various MTs. A DALI (53) search for three-dimensional structure similarity resulted in 40 structures with Z scores greater than 15. The first five structures with the highest Z scores are from bacterial O-MTs with Z scores ranging from 26.2 to 25.7, followed by CCoAOMT (25.6) from the plant M. sativa and human COMT with 24.3 (Table 3). It is interesting to note that most of the Schematic representation of the methyl group transfer reaction is shown above. B, overlay of two-dimensional 1 H, 13 C HSQC (blue) and multiplicity-edited (red) spectra of reaction mixture after protein precipitation. The negative peaks of multiplicity-edited spectra are shown in magenta. The one-dimensional projections (magenta) of two-dimensional cross-peaks of multiplicity-edited spectra are shown above corresponding peaks. Inset, overlay of one-dimensional 1 H spectra of reaction mixture before (blue) and after (unlabeled SAM (red) and 13  closest bacterial structural homologues are involved in antibiotic synthesis pathways, whereas the plant CCoAOMT and human-COMT are involved in lignin biosynthesis and the metabolism of catecholamine neurotransmitters, respectively. A superposition of each of the closest homologue structures with a monomer of PaMTH1 revealed an overlap with root mean square deviations (RMSDs) ranging from 1.3 to 1.8 Å for C␣ atoms (RMSDs generated using PyMOL (54)). ClustalWbased multiple-sequence alignment of the amino acid sequence shows that PaMTH1 contains eight motifs (motifs I-VIII) with a high degree of identity to the corresponding stretches of plant CCoAOMT, bacterial LiOMT, rat COMT, and human COMT (Fig. 2). The individual motifs in a tertiary fold together form the defining feature of the bipartite active site to accommodate SAM and in addition Mg 2ϩ that can bind catechol substrates. Motifs I-V are mainly involved in SAM binding. Asp 144 , Lys 147 , Asp 172 , and Asn 173 are important for divalent cation binding, which in turn is required for substrate binding and hydroxyl orientation before the transmethylation reaction of some O-MTs. All four residues are highly conserved in PaMTH1 and structurally adopt a conformation similar to that of CCoAOMT and human COMT, suggesting that PaMTH1 might share a similar metal-dependent catalytic mechanism (Fig. 1E). Mutation of these residues (K147A and double mutants D144A/ K147A and D172A/N173A) either abrogates or slows down the transmethylation reaction, suggesting a crucial role in catalysis (Fig. 8).
In the substrate binding region, conformational differences in the ␣2-␣3, ␤5-␣9, and ␤6-␤7 loops play an important role in mediating divalent cation and substrate binding at least in rat and mouse COMT (55). Further, conformational changes in the ␣2-␣3 loop upon ligand binding reposition the hydrophobic side chain of a conserved methionine (within the loop) onto the aromatic ring of the catechol and shield the Mg 2ϩ binding site, displaying a critical role during catalysis, at least in COMT. In PaMTH1, both apo-and holoenzyme superpose well with each other with an RMSD of 0.42 Å calculated over all C␣ atoms excluding the ␣2-␣3 loop region, which shows an RMSD of 1.7 (PaMTH1-SAM) and 2.3 Å (PaMTH1-SAH), respectively. The conformation of the ␣2-␣3 loop is not dominated by crystal contacts. Additionally, the side chain of the Met 48 in SAM/ SAH-bound PaMTH1 is swung 5.4 Å into the active site ("closed"), whereas it is solvent-exposed in the apoenzyme ("open") (Fig. 1F). The two conformational states observed for Met 48 indicate that the ␣2-␣3 loop in PaMTH1 could play a similar role as observed in rat and mouse COMT and could mediate substrate binding or specificity.

Discussion
Small molecule natural product MTs play an important role in modulating diverse biological processes, such as cell signaling and biosynthesis. Most of the natural product MTs belong to two (Class I and Class III) of 15 currently known protein fold superfamilies of SAM-binding proteins (25,56).   PaMTH1 shows significant sequence identity to other known O-MTs (bacterial LiOMT, alfalfa CCoAOMT, rat COMT, and human COMT). The highest degree of sequence identity is found for the stretches involved in the co-factor SAM-binding amino acids and shows subtle differences in the substrate binding regions. Further, a Dali search for structural homology identified 10 structures with a Z-score greater than 24. All 10 structures belonged to cation-dependent O-MTs and can be superimposed to PaMTH1 with an RMSD of less than 3 Å. Sequence and structural comparison demonstrates that a high degree of structural similarity is observed for the SAMbinding region in many species across all kingdoms of life, despite differences in the substrate specificity. Although the overall fold of the protein is conserved, the divergence between bacterial, plant, and animal MTs results in altered oligomerization states, different substrate recognition modes, or even altered catalytic mechanism of methyl transfer. PaMTH1 exists as a dimer both in solution and crystals. An SEC analysis shows that in the presence of putative substrate (myricetin) and cofactor (SAM), the oligomeric state of PaMTH1 is not altered (Fig. 5D), suggesting that dimerization might be a functional requirement. Within the crystal structure of the PaMTH1 homodimer, two features distinguish PaMTH1 from its homologues: the first 10 residues of the N-terminal region and an unusual insertion between ␤5 and ␣9 (Ala 186 -His 192 , "capping loop"; Fig. 9). Strikingly, the N-terminal loop swaps onto the dyad-related monomer and makes direct contacts with the capping loop before entering into the catalytic region, thereby FIGURE 8. Catalytic activity of wild-type and mutant PaMTH1. One-dimensional 1 H NMR spectra of SAM (methyl group) were used to monitor the enzyme activity of PaMTH1 (wild type and mutants). Wild-type PaMTH1 catalyzes the transmethylation reaction within 15 min, whereas the mutants showed either slow or no catalysis even after 120 min. All spectra were acquired at 800 MHz (T ϭ 298 K) in the following buffer: 25 mM d-Tris, pH 7.5, 100 mM NaCl, 5 mM TCEP, 10% D 2 O, and 1 mM TSP-d 4 .  forming a unique dimerization interface. The side-chain N⑀ of Trp 188 (chain B) in the capping loop hydrogen-bonds with the carbonyl oxygen of the Pro 7 (chain A) on the N terminus of the dyad-related molecule and stabilizes the interaction interface. Both mutation of key N-terminal residues (S4A/P7A double mutant) and the N-terminal deletion mutant (⌬ 1-10 PaMTH1) considerably slow down the catalysis, suggesting a definitive role in substrate binding (Fig. 8). These observations suggest that the capping loop and the residues therein might play a critical role in substrate selection and binding. Insertions to the common core Rossmann fold in the structures of MTs seem to be an evolutionary way of imparting unique features necessary for substrate selectivity (57,58). Furthermore, the conformational flexibility within the residues of the ␣2-␣3 loop and in particular a well conserved methionine whose side chain sequesters aromatic moieties of the substrates at least in other MTs (55) could also play a similar role in PaMTH1. The natural in vivo substrate of PaMTH1 is unknown. However, previous studies suggested that PaMTH1 specifically methylates flavonoids and not L-3,4-dihydroxyphenylalanine, a substrate of mammalian COMT (28). In this study, we have given direct biophysical evidence and show by NMR and mass spectrometry that PaMTH1 binds to the co-factor (SAM) and methylates the flavonoid myricetin in a cation-dependent manner. Taking into account the three-dimensional structure of PaMTH1, we suggest a catalytic mechanism; the side chains of conserved residues Asp 144 , Asp 172 , and Asn 173 and the hydroxyl groups of the flavonoid coordinate the Mg 2ϩ . Mg 2ϩ probably acts as weak Lewis acid, and during catalysis, the conserved Lys 147 can act as a Brønsted base. Further, the positive charge of Mg 2ϩ may also result in a decrease of the pK a for the Lys 147 side chain, followed by positioning of the side chain amino group of Lys 147 against the hydroxyl group to be methylated resulting in an abstraction of a proton and generation of a reactive phenolic oxyanion near the reactive methyl group of SAM. Such a metaldependent catalytic mechanism of methyl transfer persists in several related MTs (24,55).
Understanding the aging process and improving the life span of organisms has become a specific aim of many research activities, resulting in significant fundamental advances (59). Aging and several age-related diseases, including cancer, are fuelled by damage to the macromolecules in cells and tissues (60,61). Studies involving genetically tractable model organisms like the ascomycete Podospora anserina, yeast, Caenorhabditis elegans, Drosophila melanogaster, and mice have shown that specific genes are involved in the regulation of aging (62). PaMTH1 protects P. anserina from oxidative stress during senescence and acts as a longevity assurance factor. Identification of its natural substrates and thus elucidation of the underlying principle mechanisms contributing to life span increase still remains a challenge. To this end, the structures (apo-and holoenzyme) of PaMTH1 and biophysical characterization of its interaction with putative substrates will facilitate future studies involving the identification of endogenous polyphenolic compounds directly involved in metal-induced oxidative stress.