Phytochemistry 113 (2015) 130–139
Contents lists available at ScienceDirect
Phytochemistry journal homepage: www.elsevier.com/locate/phytochem
A catalytic triad – Lys-Asn-Asp – Is essential for the catalysis of the methyl transfer in plant cation-dependent O-methyltransferases Wolfgang Brandt a,⇑, Kerstin Manke b, Thomas Vogt b a b
Leibniz Institute of Plant Biochemistry, Dept. Bioorganic Chemistry, Weinberg 3, D-06120 Halle(Saale), Germany Leibniz Institute of Plant Biochemistry, Dept. Cell and Metabolic Biology, Weinberg 3, D-06120 Halle(Saale), Germany
a r t i c l e
i n f o
Article history: Received 8 May 2014 Received in revised form 10 December 2014 Available online 14 January 2015 This contribution is in honor of Professor Vincenzo de Luca’s 60th birthday Keywords: Ice plant Mesembryanthemum crystallinum Aizoaceae Methyltransferase Catalytic triad Molecular modelling Site-directed mutagenesis
a b s t r a c t Crystal structure data of cation-dependent catechol O-methyltransferases (COMTs) from mammals and related caffeoyl coenzyme A OMTs (CCoAOMTs) from plants have suggested operative molecular mechanisms. These include bivalent cations that facilitate deprotonation of vicinal aromatic dihydroxy systems and illustrate a conserved arrangement of hydroxyl and carboxyl ligands consistent with the requirements of a metal-activated catalytic mechanism. The general concept of metal-dependent deprotonation via a complexed aspartate is only one part of a more pronounced proton relay, as shown by semiempirical and DFT quantum mechanical calculations and experimental validations. A previously undetected catalytic triad, consisting of Lys157-Asn181-Asp228 residues is required for complete methyl transfer in case of a cation-dependent phenylpropanoid and flavonoid OMT, as described in this report. This triad appears essential for efficient methyl transfer to catechol-like hydroxyl group in phenolics. The observation is consistent with a catalytic lysine in the case of mammalian COMTs, but jettisons existing assumptions on the initial abstraction of the meta-hydroxyl proton to the metal stabilizing Asp154 (PFOMT) or comparable Asp-carboxyl groups in type of cation-dependent enzymes in plants. The triad is conserved among all characterized plant CCoAOMT-like enzymes, which are required not only for methylation of soluble phenylpropanoids like coumarins or monolignol monomers, but is also present in the similar microbial and mammalian cation-dependent enzymes which methylate a comparable set of substrates. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Methylation of aromatic compounds can be performed by two different classes of S-adenosyl-L-methionine (SAM)-dependent Omethyltransferases (OMTs) (EC 2.1.1): cation-dependent and cation-independent enzymes (Joshi and Chiang, 1998). Among the large group of OMTs, the cation-dependent enzymes occupy a small, yet relevant niche. In animals and microbes they are referred to as catechol-O-methyltransferases (COMTs). In both cases, the enzymes perform distinct regiospecific alkylation reactions of vicinal dihydroxy groups in the biosynthesis of natural products modulating their chemical reactivity and physiological properties (Vidgren et al., 1994; Hou et al., 2007). COMT in mammals essentially functions to inactivate potentially mutagenic catechols in Abbreviations: SAM, S-adenosyl-L-methionine; CCoAOMT, caffeoyl-coenzyme A dependent O-methyltransferase; OMT, O-methyltransferase; queg, quercetagetin. ⇑ Corresponding author. Tel.: +49 345 5582 1360; fax: +49 5582 1309. E-mail address:
[email protected] (W. Brandt). http://dx.doi.org/10.1016/j.phytochem.2014.12.018 0031-9422/Ó 2014 Elsevier Ltd. All rights reserved.
the liver, and also the neurotransmitter L-DOPA in the brain (Männistö and Kaakkola, 1999; Zhu, 2002; Bai et al., 2007). In plants, one class of enzymes are referred to as caffeoyl CoA Omethyltransferases (CCoAOMTs), due to their most relevant physiological substrate, caffeoyl coenzyme A (Ye, 1997). They are distinct from a class of cation-independent enzymes, which produce sinapoyl alcohol moieties in lignin formation in vascular tissues and isorhamnetin in Arabidopsis floral tissues (Gujon et al., 2003; Fellenberg et al., 2012). CCoAOMTs designate a group of dimeric bivalent cation-dependent, low molecular weight (23– 27 kDa) enzymes required for the methylation of monolignols in gymno- and angiosperms and, therefore, are of utmost importance to the structural integrity of this important polymer (Ye, 1997). This type of enzyme is either acceptor specific, as in the case of CCoAOMT required for guaiacyl lignin monomer formation, or is promiscuous like in case of a flavonoid and phenylpropanoid methylating enzyme (PFOMT) from the ice plant, Mesembryanthemum crystallinum, methylating a variety of aromatic compounds with high catalytic efficiency (Ibdah et al., 2003). With only a
131
W. Brandt et al. / Phytochemistry 113 (2015) 130–139
Table 1 Kinetic constants of recombinant PFOMT wild type and mutant enzymes with eriodictyol (1) and quercetagetin (3) (2–80 lM) at concentrations of 500 lM SAM, 200 lM MgCl2 in 50 mM Kpi pH 7.5. Structures of both substrates and corresponding products are also shown below. Enzyme
Compound
kcat [s1]
km [lM]
kcat/km [M1 s1]
PFOMT WT
Eriodictyol (1) Quercetagetin (3) Eriodictyol (1) Quercetagetin (3) Eriodictyol (1) Quercetagetin (3) Eriodictyol (1) Quercetagetin (3)
0.100 0.078 0.00159 0.106 n.d. 0.0021 0.00277 0.0116
3.01 0.33 184 249 n.d. 9.80 22.5 1.40
33,200 237,000 8.64 426 n.d. 214 123 8285
Eriodictyol (1) Homoeriodictyol (2)
R1, R2, R3 = H R1, R3 = H; R2 = Me R1, R2 = Me; R3 = H R1 = Me; R2, R3 = H R1, R2 = H; R3 = OMe R1, R3 = Me, R2 = H
Quercetagetin (3) Queg-5-OMe (4) Queg-5,30 -diOMe (5) Queg-30 -OMe (6) Queg-6-OMe (7) Queg-6,30 -diOMe (8)
K157A N181A D228A n.d. no activity detectable
R1 = H R1 = Me
Fig. 2. HPLC chromatogram illustrating structures of substrate eriodictyol (1) and meta-methylated product homoeriodictyol (2) after enzymatic action by 1 lg wild type PFOMT at 20 lM substrate concentration within 1 min, absorbance recorded at 287 nm.
Fig. 1. Docking arrangement of eriodictyol (1) (green carbon atoms) in the active site of PFOMT. The side-chain of Tyr51 is not displayed for better visibility of the important residues. Red dashed lines indicate the proton relay and the reaction paths studied. The final catalytic step consists of the methyl transfer from SAM (orange carbon atoms) to the meta-oxygen atom of the substrate. Red labelled amino acid residues form the catalytic triad. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
few exceptions, CCoAOMTs from plants show conserved position specificity towards the meta-hydroxyl group of catechols consistent with the vanilloid types of substitution patterns observed in plants (Anterola and Lewis, 2002; Wils et al., 2013; Widiez et al., 2012). The crystal structures of the CCoAOMT’s from Medicago sativa (alfalfa) and the PFOMT from M. crystallinum provide insights into the specificity and mode of action of this type of enzyme (Ferrer et al., 2005; Kopycki et al., 2008a). The overall structure and folds
of the dimeric plant enzymes, including the nucleotide-binding or Rossmann fold closely resemble the monomeric COMT from mammals, whereas positioning of the N-terminus and an insertion loop between strand b5 and helix a8, close to the active site resulted in an altered substrate specificity and substrate promiscuity in the case of PFOMT (Vogt, 2004; Kopycki et al., 2008a) further confirmed by crystal structure data and site-directed mutagenesis of a microbial CCoAOMT-like protein from Synechocystis sp. (Kopycki et al., 2008b). The reaction mechanism of CCoAOMTs is strictly dependent on a chelated bivalent metal-ion (Schmitt et al., 1991; Lee et al., 2008). Comparable activities are observed with Mg2+, the physiological ion in all CCoAOMT-like enzymes, and with the Ca2+ derived from the crystallization solution and identified as part of the CCoAOMT crystal structures in vitro (Lukacˇin et al., 2004; Ferrer et al., 2005; Kopycki et al., 2008a,b). A core a/b Rossmann fold for binding of SAM/SAH is observed in both structures. Tsao et al. (2011), based on molecular dynamics simulations, suggested that the metal cation is also essential for proper positioning of the methyl group of SAM which adopts a different docking pose in the absence of a metal cation versus the presence of Zn2+. In alfalfa, the side-chains of Asp163, Asp189, and Asn190 are considered to perform the
132
W. Brandt et al. / Phytochemistry 113 (2015) 130–139
O
O O-
O
-
NH
D180
O
O N H
HN O
HN
O
O O N H
D154
O
O-
N181
rc1
O
O
O
rc3
O
HN
O
O
S OH O
N
NH2
H N
H N
M52 N H
SAM O
H N
OH
N
HO
NH
O
O
D228
H
N H N
-
H
H
O
N O
H
rc2a
HO
HO
D102
H
O
N
O
Ca2+ O
-
N
rc5
NH3+
rc2 H
rc4
rc5a
O S+
HN
K157
rc3a
-
NH
O
HN
O O
O
Fig. 3. Molecular system of the active site of PFOMT used for semiempirical quantum mechanical PM7 calculations. All backbone atoms including the hydrogen atoms attached were fixed during calculations. Reactions studied are indicated by corresponding arrows and labelled accordingly. For representation of the 3D-structure compare with Fig. 1.
Fig. 4. The reaction coordinate (distance in Å) rc2 (continuous line) describes the lengthening of the distance between a proton of the amide nitrogen atom from Asn181 as a function of the energy. The dashed line (rc2a) represents the resulting change of the distance (in Å) between the carbonyl oxygen atom of Asp228 and the transferred proton as a function of rc2 but not to the energy. All data are listed in Table S2.
chelation of the calcium ion, with this coordination complex completed by the oxygen of the aromatic methoxy group of feruloyl CoA and a water molecule. This architecture is largely conserved
in PFOMT with equivalent residues Asp154, Asp180, Asn181 and guarantees a close proximity of the oxyanion to the reactive methyl group of the sulphur in SAM. Conserved residues Asp141, Asp169, and Asn170 which are essential for the catalysis are also present in the orthologous mammalian COMTs. It was hypothesized that in human COMT an active lysine side-chain (Lys144) close to the meta-hydroxyl of the vicinal dihydroxy group acts as a general base to accept a single proton from the hydroxyl group to be methylated in an SN2-type reaction (Vidgren et al., 1994; Männistö and Kaakkola, 1999). In addition, active site compression behind the sulphur bearing methyl group also directs hydrogen movement (Zhang and Klinman, 2011). The overall structure of cation-dependent enzymes is consistent with a universal mechanism of SAM dependent methyl transfer and suggests a common ancient ancestor of plant, microbial and mammalian enzymes (Schluckebier et al., 1995). Individual differences as extra loops or hydrophobic sandwiches provided by three aromatic residues Tyr51, Trp184, and Phe198 in case of PFOMT reduce acceptor specificity. Although this array of amino acids is already identified to influence the catalytic properties observed in plant CCoAOMTs, relevant details of the initial deprotonation performed by individual aspartates in plant CCoAOMT and the subsequent proton transfer appear energetically questionable and, therefore, a more plausible reaction mechanism within the active site remains to be established. In several theoretical studies (Zheng and Bruice, 1997; Lau and Bruice, 1998; Ovaska and Yliniemela, 1998; Kuhn and Kollman, 2000),
W. Brandt et al. / Phytochemistry 113 (2015) 130–139
Fig. 5a. Graphical representation of the reaction coordinate rc3 (lengthening of the N–H bond of the side-chain of Lys157 in Å) as a function of the energy (in kcal/mol) and rc3a the resulting distance of the moving proton to the carbonyl carbon atom of the side-chain of Asn181 (dashed line). All data are listed in Table S3.
Fig. 5b. Graphical representation of the reaction coordinate rc3 (lengthening of the N–H bond of the side-chain of Lys157) as a function of the energy (in kcal/mol) and rc4 the resulting distance of the simultaneous move of the proton from the metahydroxyl group of the substrate to the nitrogen atom of the side-chain of Lys157 (dashed line).
wherein the catalytic mechanism of OMTs is investigated in detail consider a lysine in the spatial neighbourhood to serve as a catalytic base for the proton abstraction from the hydroxyl group of the substrate. However, in each case, the side-chain nitrogen atom of this lysine is protonated (pKa = 10.53) which impairs its function as a catalytic base. Ovaska and Yliniemela (1998) performed semiempirical quantum mechanical studies for all possible protonation states of catechols used alternatively an aspartate anion to explain proton abstraction. In all studies, neither the entire catalytic mechanism nor the catalytic role of lysine was clearly established. In this report, in silico calculations and experimental verification, describe a universal and highly conserved proton relay, suggesting a catalytic triad to be essentially involved in optimized transfer of the methyl group to all types of catechol acceptors.
2. Results and discussion 2.1. The currently anticipated mechanism is theoretically unfavoured Vidgren et al. (1994) as well as Männistö and Kaakkola (1999), suggested that in mammalian COMTs a lysine in the active site
133
Fig. 6. Graphical representation of the reaction coordinate rc5 (lengthening of the S–C distance of the methyl group of SAM) as a function of the energy (in kcal/mol) and rc5a the resulting distance of the carbon atom of the moving methyl group from the meta-hydroxyl group of the substrate (dashed line). All data are listed in Suppl. Table S5.
serves as proton acceptor from the substrate. However, the aforementioned theoretical works and our own prediction of the protonation state based on the protonate-3D option of MOE clearly indicates that this lysine is protonated under physiological pH. Therefore, an immediate proton acceptance by the positively charged and protonated lysine side-chain can be ruled out. The possibility that one of the vicinal hydroxyl groups of the substrate is already deprotonated under physiological conditions seems unlikely due to its relatively high pKa of 9.8 (Ovaska and Yliniemela, 1998). Nevertheless, this was also tested by energy and structure optimization with DFT calculations of catechol in a neutral and a single anionic state. For the neutral form of catechol for the C–O bond lengths, a distance of 1.36 Å (the one that forms an intramolecular hydrogen bond to the neighbour oxygen atom) and 1.37 Å was calculated. In the case of the ionized form, the C–O distance of the negatively charged oxygen is only 1.285, i.e. considerably shorter than in the neutral form. The corresponding measurements in crystal structures with co-crystallized ligands (pdb-code: 3BWY, Rutherford et al. 2008; and 3CBG, Kopycki et al., 2008a,b) in case of the inhibitor, showed a C-O distance of 1.33 Å (3BWY) and in case of 3CBG for both, (2E)-3-(3-hydroxy-4-methoxyphenyl)prop-2enoic acid and 3-(4-hydroxy-3-methoxyphenyl)-2-propenoic acid 1.362 Å, respectively, in excellent agreement with the theoretical calculations for the neutral form. This clearly rules out that the substrate is bound and already deprotonated in the active site of the enzymes. Thus, a mechanism is required which either leads to the activation of the lysine side-chain to serve as a base or, alternatively, the proton transfer from the phenolic hydroxyl group takes another route. To elucidate this mechanism, the crystal structure of the active site of plant PFOMT was used as a template. For these calculations, all hydrogen atoms were added with the protonate-3D option of MOE. A suitable substrate with a single aromatic vicinal dihydroxy group, in this case the flavanone eriodictyol (1) (see Table 1 for structures), was docked into the active site (Fig. 1). Methylation of eriodictyol (1) by PFOMT resulted in a single meta-methylated product, homoeriodictyol (2) (Wils et al., 2013) (Fig. 2). In this pose, the meta-hydroxyl group is in close proximity to the methyl group to be transferred from SAM. Based on this docking arrangement, quantum mechanical calculations were performed including all important amino acid residues forming the catalytic active site like as shown in Fig. 3. The backbone atoms, including the Ca– hydrogens attached, were fixed during all calculations to conserve
134
W. Brandt et al. / Phytochemistry 113 (2015) 130–139
K157
K157 N181
N181
rc2a N
O N
S
SAM
OH
H
H
H
rc2
O
D228
H
H
O
rc4
S
O
SAM OH
H
N
N
rc3 H
H
H
O
OH
K157 N181
N O
H
N
H
b -23.5 # 6.3
K157
O
O
N
HO
rc5a
S
D228
rc5 SAM
O
O
N H
H
O
O
D228 O OH
-9.4 # 25.3
HO
H
H
H
OH
SAH
OH
N181
H
H
H
D228
HO
-44.6
S
O
O
O
HO
O
OH
-11.7 a
HO
OH
rc3a O
H
c
O
OH
HO O
OH
Fig. 7. Summary of the catalytic reactions for meta-methylation of eriodictyol (1) by PFOMT. All energies above the arrows are given in kcal/mol (# labels the calculated activation barriers) based on semiempirical PM7 calculations using the molecular system displayed in Fig. 2.
the backbone atoms exactly as shown in the crystal structure (Kopycki et al., 2008a,b). Subsequently, semiempirical PM7 calculations were performed to investigate and understand the catalytic mechanism of this class of enzymes. Initially, several alternative catalytic mechanisms (possibilities) were considered. Except for the recognition of the vicinal hydroxyl groups by a magnesium (or alternatively by a calcium divalent cation as apparent in the crystal structure), the essential prerequisite for the catalytic reaction is the abstraction of the meta-hydroxyl proton of the substrate by a catalytic base followed by the methyl transfer from SAM to the substrate. Based on the applied 3D-protonate option, the meta-hydroxyl hydrogen forms a hydrogen bond to the side-chain of Asp154. Therefore, initially it seemed convincing to transfer the metahydroxyl proton to this aspartate in PFOMT or an equivalent residue, Asp165, in the crystallized alfalfa CCoAOMT (Ferrer et al., 2005). The resulting energy profile is shown in Suppl. Fig. S1 and Table S1, respectively. However, theoretical calculations indicate that this proton transfer is highly unfavoured by 20 kcal/mol, regardless of the aromatic donor used. Both, side-chains of Asp154 and Asp180 recognise and fix the divalent cations (Mg2+ or Ca2+). Therefore, protonation of either aspartate accompanied with the compensation of the negative charge will lead to considerable reduced interaction with the cation and result in an unfavoured energy. This clearly rules out this option to be the first step in the catalytic mechanism exhibited by cation-dependent OMTs. 2.2. Alternative theoretical options Quite surprisingly, after energy optimization of this complex, a proton from the amide group of Asn181 was transferred to the
side-chain of Asp228. In this case, reaction coordinate calculations (rc2 in Fig. 4) for this transfer showed no activation barrier and an energy gain of 11.7 kcal/mol. This step leads to the formation of an imidic acid anion which is stabilized by forming an ionic interaction with the neighbouring side-chain of Lys157. Thermodynamic calculations for the formation of an imidic acid anion by a proton transfer from acetamide to acetate to form imidic acid and acetic acid as products were performed (Suppl. Fig. S2). Consistent with calculations based on the reaction coordinates of this dyad the reaction is unfavoured by 13.6 kcal/mol. In this case, the proton transfer to Asp228, in agreement with the simple thermodynamic calculation, is certainly unfavoured. However, in a scenario where this dyad consisting of Asn181 and Asp228 is extended to a triad by addition of a positively charged lysine side-chain (in this case Lys157, Fig. 1), identical results are obtained i.e. as shown for the reaction coordinate rc2 calculation of the entire active site model in Figs. 5a and b. Apparently, the positive charge of the ammonium group of Lys157 in close proximity to Asn181 is sufficient to stabilize the imidic acid anion to result in a thermodynamically favoured reaction. These rather surprising results prompted checking these with DFT-calculations. In good agreement with the semiempirical calculations the formation of the imidic acid anion is unfavoured by 24.2 kcal/mol in the AsnAsp dyad. Just by adding the lysine to the dyad exactly in a position like in the crystal structure, but fixing all atoms to avoid any proton transfer, the formation of the imidic acid becomes favoured by 0.3 kcal/mol. Subsequently, a full optimization without fixed side-chain atoms of lysine leads to the proton migration from the protonated nitrogen atom of lysine to the imidic acid anion with an energy gain of 32 kcal/mol. Thus, in the next step of the semiempirical theoretical calculations (rc3 in Figs. 5a and b), a proton was transferred from
W. Brandt et al. / Phytochemistry 113 (2015) 130–139
Lys157 to the imidic acid anion of Asn181. This reaction led to a considerable energy gain of 23.5 kcal/mol in general agreement with the DFT-calculations, although two low activation barriers occur when considering at the energy course for the lengthening of the N-H bond of Lys157 (Figs. 5a and b). Initially, a very low activation barrier (2 kcal/mol) is calculated by increasing the N–H bond length from 1.066 (energy minimum) to 1.166 Å. Subsequent lengthening of only 0.1 Å leads already to an energy gain of 16.9 kcal/mol. Further stepwise increase of the N–H distance causes a slight rise in energy until an activation barrier of 6.5 kcal/mol for the calculated transition state at an NH distance of 2.16 Å is reached. In the next step of the reaction, coordinate calculation the proton transfer is already complete with a high energy gain. Further inspection of this reaction coordinate calculation indicates that the proton from the meta-hydroxyl group of the substrate has already been transferred to the side-chain of Lys157 (Fig. 5b) when the distance of the reaction coordinate is extended from 2.06 Å to 2.16 Å in the last step. In summary, the proton release from the hydroxyl group to be methylated in PFOMT is considered to be facilitated by a catalytic triad consisting of Lys157-Asn181-Asp228 consistent with the requirement for the initial catalytic steps. A very low maximal activation barrier of 6.3 kcal/mol occurs only for the proton transfer from lysine to the imidic acid form of asparagine with a simultaneous proton migration from the hydroxyl group of the substrate to lysine. Compared with the previously discussed initiation of proton abstraction involving Asp154 of PFOMT, the newly calculated mechanism is energetically highly favoured. The initial abstraction
135
of a proton by a lysine was also proposed for the mammalian COMT (Männistö and Kaakkola, 1999), although no additional consideration or calculations of the anticipated proton relay to subsequent Asn and Asp were performed. The final step, rather independent from the first activation, is now the ‘‘real’’ catalytic step, i.e. the methyl transfer from SAM to the phenolic anion. The resulting energy course depends on the lengthening of the S-CH3 bond of SAM as shown in Fig. 6. The reaction energy for this step is calculated as 9.4 kcal/mol which favours the reaction thermodynamically. The calculated activation barrier in this case is 25.3 kcal/mol. In each case of an SN2 reaction a Walden-inversion formally occurs. This proceeds, in general, via the intermediate formation of a planar cation with high energy in comparison to both trigonal ground states. This transition state characterized by the planar methyl cation explains the apparently unavoidable rather high activation barrier (Suppl. Fig. S3). Altogether, the entire catalytic mechanism as summarized in Fig. 7 is thermodynamically favoured by 44.6 kcal/mol which should also contribute energetically to overcome the considerable activation barrier of 25.3 kcal/mol. This overall high energy gain does not include the reconstitution of the enzymes ground state i.e. the original protonated/deprotonated states of the residues in the catalytic triad. Thus, the calculated true energy gain of 9.4 kcal/mol is only the one for the last step in the reaction. In step a (rc2, lengthening of the N–H bond of the side-chain of Asn181) the formation of an imidic acid is favoured by 11.7 kcal/ mol. In the next step b (rc3, lengthening of the N–H bond of the
Fig. 8. Sequence alignment of 15 CCoAOMT-like enzymes from plants, microbes, and mammals. The conserved residues K157, N181, and D228 in case of PFOMT are illustrated by an arrow. GenBank accession numbers: PFOMT, AY145521; alfalfa CCoAOMT, Q40313; Arabidopsis AtTSM1_At1g67990, NM_105469; Arabidopsis_At4g26220, AF360317; Arabidopsis CCoAOMT1_At4g34050, AY062630; Vanilla planifolia CCoAOMT, JF344740; Petroselinum crispum CCoAOMT, Z54183; Vitis vinifera CCoAOMT, NP001268047; Pinus radiata CCoAOMT, HQ444753; Oryza sativa CCoAOMT, NP001062142; Zea mays CCoAOMT, AJ242980; Synechocystis SynOMT, 3CBG_A; Leptospira interrogans, 2HNK_A (pdb); human COMT, NP_009294, 3A7E (pdb); rat COMT, 2ZVJ_A (pdb).
136
W. Brandt et al. / Phytochemistry 113 (2015) 130–139
Fig. 9. HPLC chromatograms illustrating methylation of the substrate quercetagetin (3) to various methylated products after enzymatic reaction with 2 lg recombinant wild type PFOMT (A and B) compared to 10 lg K157A (C and D) at 20 lM substrate concentration within 5 min. UV-absorbance (A and C) was recorded at maxplot detection (300– 370 nm). Fluorescence (B and D): excitation 370 nm; emission 520 nm. Products are as follows: queg 5-OMe (4); queg 5,30 -diOMe (5); queg 30 -OMe (6); queg 6-OMe (7); queg 6,30 -diOMe (8). x, uncharacterized impurity, potential oxidation product of (3). Only (4) and (5) display characteristic fluorescence spectra (Vogt, 2004).
side-chain of Lys157) after the low activation barrier of 6.3 kcal/ mol is overcome the proton migrates simultaneously to the oxygen atom of the imidic acid (rc3a) and without a barrier from the metahydroxyl group of eriodictyol (1) to the side-chain of Lys157 (rc4). This reaction is favoured by an energy gain of 23.5 kcal/mol. Finally, the methyl transfer (rc5) is accompanied by an additional energy gain of 9.4 kcal/mol but has to pass a barrier of 25.3 kcal/mol. This activation energy is in the range of an estimated experimental barrier of 19.8 kcal/mol from the rate constant for an isoflavone O-methyltransferase (He and Dixon, 1996). In addition, the data are consistent with extensive quantum mechanical DFT-B3LYP calculations predicting activation barriers between 17 and 24 kcal/mol for this type of enzyme, as performed by Cui et al. (2011). In the case of the cation-independent isoflavone O-methyltransferase, instead of a Lys-Asn-Asp triad, they postulate a catalytic His-Glu dyad, but without a proton relay from His to Glu for proton acceptance from the phenolic hydroxyl group to be methylated. 2.3. Experimental evidence The results of the theoretical quantum mechanical calculations prompted us to carry out an experimental validation. First, more than 100 available sequences of cation-dependent O-methyltransferases from plants, microbes, and mammals were aligned and checked for conservation of the catalytic triad, of which for clarity only a representative set of sequences including gymno- and angiosperms, as well as mono and dicots, and two characterized microbial enzymes, is presented in Fig. 8. In all sequences, this triad, consisting of amino acids marked by an arrow is conserved. In mammalian catechol OMTs a gap in the sequence alignment
before Asp228 was manually introduced according to the 3Dsuperposition of the mammalian crystal structures (pdb: 3A7E, 2ZVJ) with PFOMT (pdb: 3C3Y). Finally, it was predicted that each single mutation of Lys157, Asn181, or Asp228 to Ala, should lead to considerable loss in enzyme activity of PFOMT. Site directed mutagenesis of these amino acids was performed by replacing each one with alanine. Recombinant proteins of wild type and mutant PFOMT were produced in Escherichia coli and purified (Suppl. Fig. S4). The purified enzymes were checked for enzyme activities. Kinetic constants were calculated for two substrates, eriodictyol (1) as used in the theoretical investigations (Table 1) and for quercetagetin (queg) (3), the highly hydroxylated in vivo substrate, where several methylation products were observed (Ibdah et al., 2003). As expected, all mutations lead to a considerably reduced catalytic activity of more than three orders of magnitudes. In contrast, to the wild type enzyme in case of eriodictyol (1), either no (Asn181Ala) or only residual enzyme activities were detectable after 10-fold prolonged incubation times and 10 fold increased enzyme amounts resulting in 10,000 fold reduced efficiency, comparing the specificity constants (kcat/km) to the wild type enzyme. This is a strong indication for the essential functional role of this asparagine in the catalytic mechanism proposed. In the case of quercetagetin (3), the Lys157Ala mutant showed surprising results. First, the km but not the kcat is affected, which is apparently unchanged if compared with the wild type enzyme. Second, the position specificity in this mutant is modified and nearly exclusively, a single, unusual product is formed, namely the yellow fluorescing queg 5-O-methyl ether (4), this on reversed phase HPLC eluting before the substrate (3) (Vogt, 2004). Methylation of the 3-OH position in the B-ring and the A-ring 6-OH position that is usually observed with this type of enzymes is
W. Brandt et al. / Phytochemistry 113 (2015) 130–139
Fig. 10. Most favoured docking arrangement of quercetagetin (3) in the K157A mutant of PFOMT. A hydrogen bond of the 5-hydroxy group is formed with the catalytically active Asn181 and a very short distance (2.34 Å) between the carbon atom of the methyl group from SAM and the oxygen atom at position 5 indicates a preferred catalytic methyl transfer to result in compound (4) in agreement with the experimental data.
virtually eliminated (Fig. 9). Formation of queg 6,30 -di-O-methylether (8), characteristic for the wild type enzyme is also not detectable in this mutant (Fig. 9c). Subsequent docking studies of the Lys157Ala mutant with (3) indicates for the most favoured docking arrangement methylation at these positions are in this 5-OH position. That is the proton of the 5-hydroxy group now forms a hydrogen bond with Asn181. The resulting distance between the carbon atom of the methyl group to be transferred and the 5-oxygen is only 2.34 Å, ideal for such a reaction to occur (Fig. 10). In the case of quercetagetin (3) the catalytic triad is simply reduced to an active dyad. This is consistent with the apparent reduction in 30 -O-methylation efficiency. 5-O-methylation activity is neither observed in the Asn181 nor in the Asp228 mutant. Replacing the bulky side-chain of Lys157 by Ala enlarges the existing space consistent with a new preferred docking arrangement with increased km in the case of (3) resulting in a remarkable switch of product specificity of the enzyme in comparison to the wild-type PFOMT. This observation is an additional strong indication for the existence of a catalytic triad in cation-dependent OMTs targeting the B-ring catechol moieties.
3. Concluding remarks The catalytic mechanism of plant cation-dependent OMTs has not been clarified in detail. Especially, activation of the substrate by deprotonation of the phenolic hydroxyl group to which a methyl group is transferred was questionable based on resolution of the corresponding crystal protein structures only. Initial theoretical assumptions appeared energetically unfavoured. In this report, based on quantum mechanical calculations, it is demonstrated that a proton transfer to any aspartate that complexes the bivalent cation can be ruled out for energetic reasons. Thus, it is highly plausible that the proposed conserved catalytic triad (Lys-Asn-Asp) present in all plant CCoAOMT-like OMT essentially supports efficient catalysis. Except for a rather small activation barrier of 6.3 kcal/mol for the proton transfer from lysine to
137
asparagine, the entire sequence of steps for the activation of the phenolic moiety and the formation of the product are thermodynamically favoured and therefore, realistic. The activation barrier of 25 kcal/mol for the final step in the catalysis, the methyl transfer, is chemically (and enzymatically) unavoidable due to the necessary Walden-inversion in SN2 type reactions that passes a planar CH3-species with rather high electronic energy. The results were initially only based on theoretical investigations and predictions, but subsequently indirectly supported by site-directed mutagenesis of all relevant amino acids of this catalytic triad, showing considerably reduced catalytic efficiencies in all mutants. However, an alternative scenario cannot be completely excluded, i.e. where the experimentally introduced mutations may influence the structure of the transition state by alterations or disruptions of relevant hydrogen bonding patterns. The drastic reduction of catalytic efficacy by more than three orders of magnitudes that was observed specifically for the D228A mutation, three amino acids away at a distance of more than 8 Å from the catalytic site, does not apparently form any hydrogen bonds with any of the interacting substrates (see Fig. 1). This would be more difficult to explain by only small conformational changes in the lysine side-chain conformation, which according to our model, participates immediately in the catalysis. The current experimental results thus indicate a previously unknown essential function of the three amino acids (LysAsn-Asp) in the catalytic mechanism of plant CCoAOMT-like enzymes. It is plausible that the same or a similar mechanism is also valid for animal and microbial cation-dependent OMTs. Although the sequence identity to these is quite limited, the current calculations do not contradict this assumption.
4. Experimental 4.1. Molecular modelling and quantum mechanical calculations The crystal structure (PDB-code 3C3Y) of PFOMT from the ice plant, M. crystallinum was used for quantum mechanical investigations of the catalytic mechanism (Kopycki et al., 2008a). Ca2+ ions were included in the calculations instead of Mg2+ ions consistent with crystal structure data from PFOMT and M. sativa CCoAOMT (PDB-code 1SUS) (Ferrer et al., 2005). Protons were added using the ‘‘3D-protonate’’ tool of MOE (Molecular Operating Environment). The docking arrangement of S-eriodictyol results from GOLD (Hartshorn et al., 2007; Verdonk et al., 2003) as described in Wils et al. (2013) for a related CCoAOMT-like enzyme from Arabidopsis thaliana. Semiempirical PM7 calculations were performed using MOPAC 2012 (MOPAC 2012; Stewart, 2013; Maia et al., 2012). These quantum mechanical calculations were performed including all important amino acid residues forming the catalytic active site exactly as depicted in Fig. 3. The 3D-coordinates of these amino acid residues were extracted from the crystal structure of PFOMT (Kopycki et al., 2008a) and used as starting geometry. All backbone atoms including the Ca–hydrogens attached were fixed during all calculations to conserve the enzymes 3D-structure. Remaining atoms of the molecular system were energy optimized in each calculation. Reaction coordinate calculations were performed by stepwise (0.1 Å) enlarging or diminishing distances between atoms (labelled as RCx in Fig. 3 and are listed in Supplemental material). Finally, the real transition state energies and geometries were determined using the TS keyword. DFT (TZVP/B3LYP) calculations of the dyad and triad model were carried out with the graphical interface TmoleX (Steffen et al., 2010) and Turbomole (Furche et al., 2014). As for the semiempirical calculations, the coordinates were extracted from the crystal structure. Each amino acid residue was acetylated at the N-terminus and C-terminally neutralized by the addition of a
138
W. Brandt et al. / Phytochemistry 113 (2015) 130–139
N-methyl group to mimic adjacent amino acid backbones. The backbone atoms were fixed during all energy optimizations. Docking studies for quercetagetin (3) were performed with GOLD (Hartshorn et al., 2007; Verdonk et al., 2003) using the ChemScore fitness function and for all other options with standard settings. 30 docking position have been calculated. For the definition of the binding site, the Ca2+ atom of the active site was chosen as origin with a radius of 15 Å. Four amino acid side-chains (Y51, M62, W184, and N202, Fig. 10) were set to be flexible applying the rotamer library included in GOLD. The docking results were analysed manually with MOE looking for conformations principally appropriate for a methyl transfer from SAM to the substrate. 4.2. Generation of mutants and purification of enzymes The PFOMT-sequence (GenBank accession number AY145521) was cloned into pQE30 (Qiagen), and mutagenized by a site-directed mutagenesis kit (Thermo Scientific, Dreieich, Germany) according to the manufactures instructions. Primers used were: L157A (two step procedure) K157T, for, 50 -GCTTTGTTGATGCGGA CACAC CTAACTACATCAAG-30 ; K157T, rev, 50 -CTTGATGTAGTTAGGTGTGT CCG CATCAACAAAGC-30 ; T157A, for 50 -GCTTTGTTGATGCGGACG CACCTAACTACAT CAAG-30 , T157A, rev 50 -CTTGATGTAGTTAGGTG CGTCCGCATCAACAAAGC-30 ; N181A, for 50 -CATAGTCGCTTATGACG CCACATTATGGGGTGGAAC-30 , rev 50 -GTTCCACCCCATAATGTGGCGT CATAAGCGACTATG-30 ; D228A, for 50 -GTACATC TTCCTTTGGGTG CTGGTATCACTTTCTG-30 , rev 50 -CAGAAAGTGATACC AGCACC CAAAGGAAGATGTAC-30 . Mutations of cDNAs were verified by sequencing (Eurofins-MWG, Ebersberg, Germany). Recombinant proteins were produced in each case from two litres each of 1 mM IPTG induced M15rep4 E. coli cells (OD600 nm 0.6) containing the different wild type and mutant constructs, grown further at 37 °C for four hours. Cells were centrifuged for 10 min at 10,000g, the pellet ultrasonicated in Buffer A (100 ml, 50 mM Kpi pH 7.5, 10% glycerol), centrifuged, and stirred with 0.05% protamine sulphate for 5 min to precipitate the nucleic acids and reduce viscosity of the extract. The suspension was centrifuged for 10 min at 20,000g and the supernatant directly applied to 25 ml of TALON affinity matrix (Clontech, Palo, Alto, CA, USA) equilibrated in Buffer A, at a flow rate of 2 ml/min. The column was washed with 10 mM and 30 mM imidazole in buffer A. Subsequently, active protein was eluted with 240 mM imidazole in buffer A in a total volume of 30 ml, concentrated to 2 mg/ml, and stored at 80 °C in a buffer containing 50 mM Kpi, 10 mM imidazole, and 10% glycerol for determination of substrate specificity and kinetic properties. Individual proteins were checked for purity by SDS–PAGE (Suppl. Fig. S4). 4.3. Enzyme assays of recombinant enzymes Kinetic and catalytic properties of the enzymes were determined with eriodictyol (1) and quercetagetin (3), consistent with the data of the modelling experiments. Assays were performed in a buffer containing 50 mM Kpi pH 7.5, 5% DMSO, 200 lM MgCl2, and 500 lM SAM, 1 lg to 20 lg of purified enzymes (depending on the enzymatic activity) in a time frame from 2 min (wild type) up to 60 min (Asn181Ala). Enzyme concentrations were calculated based on the molar extinction coefficient at 280 nm of wild type PFOMT 18730 ± 5%; 1 A280nm = 1.42 mg/ml; 1 lg PFOMT = 37.6 pmol with the Lasergene Software Package, DNASTAR, Madison, USA. Assays were started by the addition of either (1) or (3). Data are the average of two independent experiments each performed in triplicate at seven different concentrations of (1) and (3) from 2 lM to 80 lM. Assays were terminated by adding a mixture of MeCN: 10% trichloroacetic acid (TCA) in H2O (50:50, v/v), cooled for 5 min at 4 °C, centrifuged and immediately analysed by
reversed phase HPLC chromatography using either a Nucleosil C18 column (5 lM; 12.5 4 mm i.d) or a Nucleoshell (3 lM; 12.5 4 mm i.d. for optimized separation of individual methyl ethers of (3); both Macherey-Nagel, Düren, Germany). Separation of substrates and products was achieved using a 5.5 min linear gradient at a flow rate of 1 ml min1 from MeCN: 0.5% aq. H3PO4 (20:80 ? 70:30) (Nucleosil) or a 10 min linear gradient at a flow rate of 0.6 ml min1 from MeCN: 0.5% aq. H3PO4 (10:90 ? 80:20) (Nucleoshell). Compounds were detected photometrically between 280 and 320 (eriodictyol) and 300 and 370 nm (3–8) by a Waters 2996 photodiode array detector, and for fluorescence by a Waters 2475 detector in case of quercetagetin (3), excitation at 370 nm, emission at 520 nm. km- and kcat data were calculated based on Lineweaver–Burk plots and linear regression analysis. A racemic mixture of eriodictyol (1) was obtained from Roth (Karlsruhe, Germany), quercetagetin (3) was purchased from Extrasynthese (Genay, France), and SAM from Sigma (Munich, Germany). Standards of homoeriodictyol (2), queg 30 -O-methyl ether (6), and queg 6-O-methyl ether (7) were kindly provided by E. Wollenweber (Darmstadt, Germany). Acknowledgements The authors thank Eckhard Wollenweber (Darmstadt) for his kind gift of several eriodictyol and quercetagetin reference compounds. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.phytochem.2014. 12.018. References Anterola, A.M., Lewis, N.G., 2002. Trends in lignin modification: a comprehensive analysis of the effects of genetic manipulations/mutations on lignification and vascular integrity. Phytochemistry 61, 221–294. Bai, H.-W., Shim, J.Y., Yu, J., Zhu, B.T., 2007. Biochemical and molecular modeling studies of the O-methylation of various endogenous and exogenous catechol substrates catalyzed by recombinant human soluble and membrane-bound catechol-O-methyltransferases. Chem. Res. Toxicol. 20, 1409–1425. Cui, F.C., Pan, X.L., Liu, J.Y., 2011. Reaction mechanism of isoflavone Omethyltransferase: a theoretical investigation. Chem. Phys. Lett. 501, 502–507. Fellenberg, C., van Ohlen, M., Handrick, V., Vogt, T., 2012. The role of CCoAOMT1 and COMT1 in Arabidopsis anthers. Planta 236, 51–61. Ferrer, J.-L., Zubieta, C., Dixon, R.A., Noel, J.P., 2005. Crystal structure of caffeoyl CoA O-methyltransferase. Plant Physiol. 137, 1009–1017. Furche, F., Ahlrichs, R., Hättig, C., Klopper, W., Sierka, M., Weigend, F., 2014. Turbomole WIREs. Comput. Mol. Sci. 4, 91–100. Gujon, T., Sibout, R., Pollet, B., Maba, B., Nussaume, L., Bechthold, M., Lu, F., Ralph, J., Mila, I., Barriere, Y., Lapierre, C., Jouanin, L., 2003. Plant Mol. Biol. 51, 973–989. Hartshorn, M.J., Verdonk, M.L., Chessari, G., Brewerton, S.C., Mooij, W.T., Mortenson, P.N., Murray, C.W., 2007. Diverse, high-quality test set for the validation of protein-ligand docking performance. J. Med. Chem. 50, 726–741. He, X.Z., Dixon, R.A., 1996. Affinity chromatography, substrate/product specificity, and amino acid sequence analysis of an isoflavone O-methyltransferase from alfalfa (Medicago sativa L.). Arch. Biochem. Biophys. 336, 121–129. Hou, X., Wang, Y., Zhou, Z., Bao, S., Lin, Y., Gong, W., 2007. Crystal structure of SAMdependent O-methyltransferase from the pathogenic bacteria Leptospira interrogans. J. Struct. Biol. 159, 523–528. Ibdah, M., Zhang, X.H., Schmidt, J., Vogt, T., 2003. A novel Mg(2+)-dependent Omethyltransferase in the phenylpropanoid metabolism of Mesembryanthemum crystallinum. J. Biol. Chem. 278, 43961–43972. Joshi, C.P., Chiang, V.L., 1998. Conserved sequence motifs in S-adenosyl-Lmethionine dependent methyltransferases. Plant Mol. Biol. 37, 663–674. Kopycki, J.G., Rauh, D., Chumanevich, A.A., Neumann, P., Vogt, T., Stubbs, M.T., 2008a. Biochemical and structural analysis of substrate promiscuity in plant Mg2+-dependent O-methyltransferases. J. Mol. Biol. 378, 154–164. Kopycki, J.G., Stubbs, M.T., Brandt, W., Hagemann, M., Porzel, A., Schmidt, J., Schliemann, W., Zenk, M.H., Vogt, T., 2008b. Functional and structural characterization of a cation-dependent O-methyltransferase from the cyanobacterium Synechocystis sp. strain pcc 6803. J. Biol. Chem. 283, 20888– 20896.
W. Brandt et al. / Phytochemistry 113 (2015) 130–139 Kuhn, B., Kollman, P.A., 2000. QM-FE and molecular dynamics calculations on catechol O-methyltransferase: free energy of activation in the enzyme and in aqueous solution and regioselectivity of the enzyme-catalyzed reaction. J. Am. Chem. Soc. 122, 2586–2596. Lau, E.Y., Bruice, T.C., 1998. Importance of correlated motions in forming highly reactive near attack conformations in catechol O-methyltransferase. J. Am. Chem. Soc. 120, 12387–12394. Lee, Y.J., Kim, B.G., Chong, Y., Lim, Y., Ahn, J.H., 2008. Cation dependent Omethyltransferases from rice. Planta 227, 641–647. Lukacˇin, R., Matern, U., Hehmann, M., Specker, S., Vogt, T., 2004. Cations modulate the substrate specificity of a bifunctional class I O-methyltransferase from Ammi majus. FEBS Lett. 577, 367–370. Maia, J.D.C., Carvalho, G.A.U., Mangueira, C.P., Santana, S.R., Cabral, L.A.F., Rocha, G.B., 2012. GPU linear algebra libraries and GPGPU programming for accelerating MOPAC semiempirical quantum chemistry calculations. J. Chem. Theory Comput. 8, 3072–3081. Männistö, P.T., Kaakkola, S., 1999. Catechol-O-methyltransferase (COMT): biochemistry, molecular biology, pharmacology, and clinical efficacy of the new selective COMT inhibitors. Pharmacol. Rev. 51, 593–628. MOE, Molecular Operating Environment (MOE), 2012. Chemical Computing Group Inc., 1010 Sherbooke St. West, Suite #910, Montreal, QC, Canada, H3A 2R7, 2012. MOPAC2012, Stewart, J.J., 2013. Computational Chemistry. Colorado Springs, CO, USA, 2012, . Ovaska, M., Yliniemela, A., 1998. A semiempirical study on inhibition of catechol O-methyltransferase by substituted catechols. J. Comput. Aided Mol. Des. 12, 301–307. Rutherford, K., Le Trong, I., Stenkamp, R.E., Parson, W.W., 2008. Crystal structures of human 108V and 108M catechol O-methyltransferase. J. Mol. Biol. 380, 120–130. Schluckebier, G., O´Gara, M., Saenger, W., Cheng, X., 1995. Universal catalytic domain structure of Ado-Met dependent methyltransferases. J. Mol. Biol. 247, 16–20. Schmitt, D., Pakusch, A.E., Matern, U., 1991. Molecular cloning, induction and taxonomic distribution of caffeoyl-CoA 3-O-methyltransferase, an enzyme involved in disease resistance. J. Biol. Chem. 266, 17416–17423. Steffen, C., Thomas, K., Huniar, U., Hellweg, A., Rubner, O., Schroer, A., 2010. TmoleX - a graphical user interface for TURBOMOLE. J. Comput. Chem. 31, 2967–2970.
139
Stewart, J.J., 2013. Optimization of parameters for semiempirical methods VI: more modifications to the NDDO approximations and re-optimization of parameters. J. Mol. Model. 19, 1–32. Tsao, D., Diatschenko, L., Dokholyan, N.V., 2011. Structural mechanism of Sadenosyl methionine binding to catechol O-methyltransferase. PLoS One 6, e24287. Verdonk, M.L., Cole, J.C., Hartshorn, M.J., Murray, C.W., Taylor, R.D., 2003. Improved protein-ligand docking using GOLD. Proteins 52, 609–623. Vidgren, J., Svensson, L.A., Lijas, A., 1994. Crystal structure of catechol Omethyltransferase. Nature 368, 354–358. Vogt, T., 2004. Regiospecificity and kinetic properties of a plant natural product Omethyltransferase are determined by its N-terminal domain. FEBS Lett. 561, 159–162. Widiez, T., Hartmann, T.G., Dudai, N., Yan, Q., Lawton, M., Havkin-Frenkel, D., Belanger, F.C., 2012. Functional characterization of two new members of the caffeoyl CoA O-methyltransferase-like gene family from Vanilla planifolia reveals a new class of plastid-localized O-methyltransferases. Plant Mol. Biol. 76, 475–488. Wils, C.R., Brandt, W., Manke, K., Vogt, T., 2013. A single amino acid determines position specificity of an Arabidopsis thaliana CCoAOMT-like Omethyltransferase. FEBS Lett. 587, 683–689. Ye, Z.-H., 1997. Association of caffeoyl coenzyme A 3-O-methyltransferase expression with lignifying tissues in several dicot plants. Plant Physiol. 115, 1341–1350. Zhang, J., Klinman, J.P., 2011. Enzymatic methyl group transfer: role of an active site residue in generating active site compaction that correlates with catalytic efficiency. J. Am. Chem. Soc. 133, 17134–17137. Zheng, Y.-J., Bruice, T.C., 1997. A theoretical examination of the factors controlling the catalytic efficiency of a transmethylation enzyme: catechol Omethyltransferase. J. Am. Chem. Soc. 119, 8137–8145. Zhu, B.T., 2002. Catechol-O-methyltransferase (COMT)-mediated methylation metabolism of endogenous bioactive catechols and modulation by endobiotics and xenobiotics: importance in pathophysiology and pathogenesis. Curr. Drug Metab. 3, 321–349.