Y. Ebizuka, Department of Natural Product Chemistry, Graduate School of Pharmaceutical Sciences, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Fax: + 81 35 841 4744, E-mail: email@example.com
Chalcone synthase and stilbene synthase are plant-specific polyketide synthases. They catalyze three common consecutive decarboxylative condensations and specific cyclization reactions. They are highly homologous to each other, and are likely to fall into a family of polyketide synthases along with acridone synthase and bibenzyl synthase. Two cDNA clones (named HmC and HmS), both of which show high homology to the known chalcone synthases, were obtained from leaves of Hydrangea macrophylla var. thunbergii. They were expressed in Escherichia coli in order to determine their enzyme functions. Detection of chalcone formation clearly indicated that HmC encoded chalcone synthase, while HmS protein catalyzed the formation of neither chalcone nor stilbene. However, a novel pyrone, a lactonization product of a linear tetraketide was detected in reaction products of HmS protein. This proves that HmS encodes a novel polyketide synthase that catalyzes only chain elongation without cyclization.
gas chromatography electron ionization mass spectrometry
liquid chromatography-atmospheric pressure chemical ionization mass spectrometry
triacetic acid lactone
Chalcone synthase (CHS) and stilbene synthase (STS) are plant-specific polyketide synthases (PKSs) . CHS catalyzes a successive condensation of three two-carbon units from malonyl-CoA (2) with p-coumaroyl-CoA (1) as a starter to form a tetraketide intermediate (3) and a cyclization to give naringenin chalcone (4). p-Coumaroyl-CoA-specific STS (see below) catalyzes the same condensation reaction but cyclizes 3 in a different manner to form resveratrol (5). Terminal carboxyl carbon is lost during this cyclization (Fig. 1).
Genes of 40 CHSs and four STSs have been cloned. Some of them have been expressed functionally in Escherichia coli and their catalytic properties investigated. STSs from two Pinus plants prefer cinnamoyl-CoA as a starter [2,3]. Bibenzyl synthase (BBS) cloned from Phalaenopsis sp. utilizes m-hydroxyphenylpropionyl-CoA and synthesizes 3,3′,5-trihydroxybibenzyl by a similar reaction to STS . Acridone synthase (ACS) cloned from Ruta graveolens produces 1,3-dihydroxy-N-methylacridone from N-methylanthraniloyl-CoA and 2. It is believed that this reaction proceeds via a benzophenone-type intermediate which is obviously generated by CHS-type ring folding .
All of CHS, STS, BBS and ACS have ≈ 400 amino acid long polypeptide chains and share > 50% sequence identity. Because of the similarity not only in the reactions catalyzed but also in sequences, it is proposed that these enzymes should form a family called the CHS superfamily . It is of interest how these similar, but distinct, enzymes elaborate their specific products. It is reasonable to assume that conserved amino-acid residues of the active sites of these enzymes take part in common condensation reactions, and that characteristic residues of functionally different proteins determine the mode of cyclization and starter specificity. Identification of these important residues could be made by site-directed mutagenesis or domain-swapping experiments. Using site-directed mutagenesis experiments, Schröder’s group identified one cysteine residue (Cys169 in Sinapis alba CHS) which is conserved in all the members of the superfamily and proposed as an essential residue for the condensation reaction . However, nothing is known of the amino acid residues responsible for their substrate and/or product specificities.
In addition to a famous sweet dihydroisocoumarin, phyllodulcin, leaves of Hydrangea macrophylla var. thunbergii contain a stilbenecarboxylic acid, hydrangeic acid [3,4′-dihydroxystilbene-2-carboxylic acid, (6)], and secoiridoids such as hydramacroside B (8), which is likely to be biosynthesized from linear tetraketide 7 (Fig. 1)  with secologanin. Therefore, the presence of two novel PKSs belonging to the CHS superfamily, and responsible for the formation of stilbenecarboxylic acid and the synthesis of linear tetraketide are expected in this plant. This makes H. macrophylla an attractive source material for enzyme proteins and their cDNAs, and it was chosen in our studies. Sequence comparison of stilbenecarboxylic acid synthase and STS should provide information on the essential amino acid residues for decarboxylation during stilbene formation, whereas that of linear tetraketide synthase and CHS should lead to the identification of the indispensable residues for cyclization required for chalcone formation. To this end, cDNA cloning of PKSs from H. macrophylla var. thunbergii was attempted. In our previous paper , two cDNA clones, namely HmC and HmS, were obtained from leaves of this plant using PCR with degenerate primers designed from the highly conserved amino acid sequences of known CHSs and STSs. Sequence analysis showed that both of them belong to the CHS superfamily, although HmS lacks some of the consensus sequences (Fig. 2). In this paper, identification of HmS as a novel PKS, which catalyzes only chain elongation without cyclization, by heterologous expression in E. coli is presented.
Materials and methods
Malonyl-CoA was purchased from Sigma. p-Coumaroyl-CoA was synthesized according to a previously described method . Noryangonin was synthesized following the reported method .
Construction of expression plasmid
Oligonucleotides HmC-N2 (5′-CGGAGGATATCGTGACCGTCGAGGAAGTCCGT-3′) and HmC-C (5′-AATGGAAGCTTAAGTAGACACACCATGAAGAACCAC-3′) corresponding to the N-terminus and C-terminus of HmC cDNA, respectively, were synthesized. In HmC-N2, start codon ATG was substituted by Ile codon ATC (underlined) to generate an EcoRV site. The terminal codon TAG was substituted by another terminal codon TAA (underlined) to generate a HindIII site. These oligonucleotides were used to amplify the ORF of HmC cDNA by PCR using an H. macrophylla cDNA mixture  as a template. PCR was carried out with 0.1 nmol each of HmC-N2 and HmC-C, cDNA mixture and 2.5 U Ex-Taq DNA polymerase (TaKaRa, Shuzo) in a final volume of 0.1 mL according to the manufacturer’s protocol. PCR was performed in Program Temp Control System PC-700 (ASTEC) using the following profile, 30 cycles of 94 °C for 0.5 min, 60 °C for 0.5 min and 72 °C for 2 min with 10 min of final extension. The amplified product was purified with agarose gel electrophoresis, digested with EcoRV and HindIII and subcloned into pET-32a(+) (Novagen), giving the plasmid pET-HmC.
Plasmid pET-HmS was similarly constructed. Oligonucleotides HmS-N2 (5′-TAGTAGATATCGCAACAAAATCGGTAGCA-3′) having ATC instead of the start codon ATG and HmS-C (5′-TTGTTGAATTCAAATGGGGACACTGTGCAAGACAAT-3′), having an EcoRI site generated by alternate terminal codon TGA instead of TAA, were used.
These nucleotide and amino acid sequences are available under accession number AB011467 for HmC and AB011468 for HmS.
In a similar manner, Arachis hypogaea STS cDNA was cloned using reported sequences of N-terminus and C-terminus of A. hypogaea cDNA  from peanuts seeds, and ligated to pET-3d (Novagen; unpublished data).
Expression and purification of fusion proteins
E. coli BL21(DE3)pLysS harboring pET-HmC or pET-HmS was cultured to an A600 of 0.6 in Luria-Bertani medium containing 100 µg·mL−1 of ampicillin and 34 µg·mL−1 of chloramphenicol at 37 °C. After the culture had been cooled on ice, 1 mm isopropyl thio-β-d-galactoside (IPTG) was added to induce protein expression, and the culture was incubated further at 20 °C for 20 h. Cells were collected by centrifugation and resuspended in an adequate quantity (3 mL for each gram of cells) of NaCl/Pi (20 mm, pH 7.8) containing 0.5 m NaCl. Cell lysis was carried out using lysozyme . The lysate was diluted with the same buffer (12 mL·g−1 of cells) and centrifuged at 25 000 g for 20 min. The supernatant was used as crude extract.
Crude extract was passed through a column of Chelating Sepharose Fast Flow (Pharmacia) in which Ni2+ was retained as an affinity ligand. After applying, nonbinding proteins were washed out with NaCl/Pi (20 mm, pH 6.0) containing 0, 50 and 200 mm imidazole in a stepwise manner. Fusion proteins were eluted with the same buffer containing 500 mm imidazole and concentrated with Centriprep-30 (Amicon). The obtained proteins were designated as thioredoxin-HmC (TrxHmC) and thioredoxin-HmS (TrxHmS), for pET-HmC and pET-HmS, respectively.
Polyketide synthase assay
The reaction mixture consisting of purified protein, 0.185 mmp-coumaroyl-CoA and 0.30 mm malonyl-CoA in 50 µL of 0.1 m Tris/HCl pH 7.5 containing 1 mm EDTA, was incubated at 30 °C for 1 h. The mixture was then acidified by adding 5 mL of 50% AcOH and extracted by ethyl acetate. After evaporation of solvent, the residue was dissolved in 45% MeOH aq. and separated using reverse-phase HPLC on a ODS-80Ts column (4.6 mm ID × 150 mm, Tosoh) maintained at 40 °C. Gradient elution was performed with H2O and MeOH, both containing 1% AcOH. The gradient profile was as follows: 0–10 min, 45% MeOH; 10–22 min, linear gradient from 45 to 49% MeOH. The flow rate was 0.6 mL·min−1. Peaks were monitored at 254 nm.
Identification of structures of compounds A and C
A large-scale reaction was performed with 47 mg of TrxHmS, 37 µmol of p-coumaroyl-CoA (0.185 mm) and 60 µmol of malonyl-CoA (0.30 mm) in 200 mL of 0.1 m Tris/HCl pH 7.5 containing 1 mm EDTA at 30 °C for 1 h with gentle shaking. The mixture was acidified by addition of dilute HCl and extracted three times using ethyl acetate. Extracts were dried over sodium sulfate and then evaporated to dryness. The residue was dissolved in 45% MeOH aq. containing 1% AcOH and insoluble materials were removed using an ODS cartridge column (TOYOPAK ODS-S). This solution was divided into several portions and applied to preparative HPLC. The HPLC conditions were as follows; ODS-PREP column (20 mm ID × 250 mm, Tosoh) with 45% MeOH aq. containing 1% AcOH (flow rate, 4.0 mL·min−1; detection, UV 254 nm) at room temperature. Fractions containing compounds A and C were collected, evaporated and lyophilized.
The 1H-NMR spectra were measured in CD3OD with trimethylsilyane as the internal standard using GSX-400 (JEOL). Liquid chromatography-atmospheric pressure chemical ionization mass spectrometry (LC-APCIMS) was measured with LCQ (Thermo Quest). HPLC condition is the same as described in the PKS assay. Gas chromatography electron ionization mass spectrometry (GC-EIMS) of yangonin was carried out using 5890 (Hewlett Packard) as the GC apparatus and SX-102 A (JEOL) as the mass spectrometer. Conditions were as follows: column, DB5 (J&W, 30 m × 0.25 μm ID, film thickness, 0.25 μm); temperature, 200–250 °C (10 °C·min−1); ionization voltage, 70 eV.
Yangonin. The authentic sample of noryangonin and compound C was methylated with trimethylsilyldiazomethane (TOKYO KASEI) and the resulting products were applied to GC/MS. Both gave the same GC profile and MS pattern. Retention time on GC, 22 min; EIMS m/z (rel. int.), 258 (M+, 100), 230 (32), 187 (42).
In order to determine the enzyme functions of HmC and HmS, expression of cDNAs in E. coli was carried out. To obtain high solubility and easy purification of the expressed proteins, pET-32a(+), which introduces thioredoxin and His-tag upstream of the N-terminus of inserted cDNA, was employed as a vector. To obtain full-size inserts of cDNAs, PCR was carried out with two sets of specific oligonucleotide primers corresponding to the respective N-terminal and C-terminal sequences using H. macrophylla cDNA pool as a template. Amplified PCR products with an expected size (≈1.2 kbp) were integrated into pET-32a(+) to yield pET-HmC and pET-HmS, respectively. These plasmids were used to transform a host E. coli strain, BL21(DE3)pLysS.
Bacteria harboring either pET-HmC or pET-HmS were cultured and protein expression was induced by addition of IPTG to the medium. When cells were grown at 37 °C, all the induced proteins became insoluble. A similar phenomenon was observed in the expression of intact HmS (expressed with pET-3d). In our earlier experiments, CHS from Pueraria lobata was expressed in a soluble form even at 37 °C using pT7-7  or pET-3d (T. Akiyama, unpublished result). The thioredoxin part, which is expected to increase the solubility of the expressed proteins, did not work in the present experiments. Therefore, the culture temperature was lowered to 20 °C after induction to obtain soluble proteins. The reason why HmC and HmS were not expressed as soluble proteins at the higher temperature remains unclear. Thus, the obtained soluble fusion proteins, designated as TrxHmC and TrxHmS, were purified by affinity chromatography using a Ni2+ column (Fig. 3).
Enzyme activity of HmC and HmS
The functions of the purified fusion proteins were investigated using an in vitro enzyme reaction. The enzyme assay was carried out following the reported method  with some minor modifications. p-Coumaroyl-CoA and malonyl-CoA served as the substrates. After termination by acidification, the reaction products were extracted by ethyl acetate and the extracts then analyzed by HPLC. Because analyses using TLC and HPLC gave no qualitative or quantitative difference between the products with fusion protein and intact protein (obtained by enterokinase digestion), further analysis was carried out using fusion proteins.
Detection of naringenin, which was formed by nonenzymatic isomerization of naringenin chalcone, together with two unidentified compounds A and C as HmC products, proved that HmC is a CHS clone. However, HmS gave neither naringenin nor resveratrol and produced only compounds A and C.
Structures of reaction products
In order to elucidate the structures of compounds A and C, a large-scale reaction consisting of 47 mg of purified TrxHmS, 37 µmoles of p-coumaroyl-CoA and 60 µmoles of malonyl-CoA, was conducted. Compounds A and C were separated and purified by preparative HPLC and analyzed by 1H-NMR and LC-APCIMS.
Spectral data revealed that compound C was bis-noryangonin (10) (Fig. 4), which has been reported to be a CHS by-product in vitro. All the signals of the 1H-NMR spectrum could be assigned as shown in Table 1 except for one missing proton at position 3 of the α-pyrone ring, which was exchanged with solvent deuterium due to α-pyrone and γ-pyrone tautomerism. Multi-stage LC-APCIMS (negative-ion mode) gave a [M-H]– peak at m/z 229 in MS, [M-H-CO2]− at m/z 185 in MS/MS (precursor ion at m/z 229) and [M-H-CO2-CH2CO]− at m/z 143 in MS/MS/MS (transitions m/z 229 > 185), showing successive cleavage of the α-pyrone ring. Further identification of compound C as bis-noryangonin was obtained by methylation with trimethylsilyldiazomethane and GC-EIMS comparison of the resulting product with authentic yangonin.
Table 1. The 1H-NMR data for compounds A and C.
7.66 (1H, d, J = 16.0)
7.27 (1H, d, J = 16.1)
6.74 (1H, d, J = 16.0)
6.56 (1H, d, J = 16.1)
7.52 (2H, d, J = 8.8)
7.40 (2H, d, J = 8.4)
6.81 (2H, d, J = 8.8)
6.78 (2H, d, J = 8.4)
LC-APCIMS of compound A gave a parent peak at m/z 271 suggesting that its molecular mass was 272 Da, 42 mass units (CH2-CO) larger than that of bis-noryangonin. Multistage LC-APCIMS gave [M-H-CO2]− at m/z 227 in MS/MS and [M-H-CO2-CH2CO]− at m/z 185 in MS/MS/MS (transitions m/z 271 > 227), the same successive fragmentation pattern as bis-noryangonin, indicating that this compound also had an α-pyrone ring. Among the 1H-NMR signals, those for 1,4-substituted aromatic ring protons (6.81 and 7.52 p.p.m.), a pair of olefinic protons (6.74 and 7.66 p.p.m.) and a proton 5 of the α-pyrone ring (5.85 p.p.m.) are commonly found in those of compound C. A proton at position 3 of the pyrone ring might not be observed for the same reason in the case of compound C. From these spectral data, compound A was identified as a novel compound, 4-hydroxy-6-[4-(4-hydroxyphenyl)-2-oxo-3-butenyl]-2H-pyran-2-one, conventionally named as p-coumaroyltriacetic acid lactone (CTAL), as shown in Fig. 4. All the signals of the 1H-NMR spectrum were assigned as shown in Table 1. The absence of 1′ methylene signals could be explained by the extensive exchange with solvent D2O due to keto-enol tautomerism. CTAL is likely to be formed by nonenzymatic cyclization of a linear tetraketide, p-coumaroylacetoacetic acid, produced by HmS protein.
p-Coumaroyltriacetic acid synthase
It is interesting to note that CTAL, the major product of HmS, was detected not only in the reaction products of HmC, CHS of H. macrophylla, but also in those of P. lobata CHS and Arachis hypogaea STS when the reaction mixtures were acidified before extraction (data not shown). To date, formation of this compound as a by-product of CHS or STS has not been reported. The apparent reason for past failures to detect this compound could be the extraction of CHS and/or STS reaction mixtures without acidification. Indeed, when the reaction products of HmC, P. lobata CHS and A. hypogaea STS were extracted without acidification, the detected amount of bis-noryangonin was reduced drastically, CTAL could not be detected at all (data not shown), and the chromatograms looked as though naringenin or resveratrol was produced as the sole product. Only one research group, Zuurbier et al. , was aware of the presence of a hydrophilic product in the CHS reaction mixture, which could only be detected when the reaction mixture was acidified before extraction. Although its structure has not been reported, the above circumstantial evidence suggests that this hydrophilic product is CTAL. Judging from the fact that CTAL or its derivatives have not been reported from any plant sources other than H. macrophylla, it seems that CTAL is a common by-product of all CHSs and STSs but its production is limited to in vitro reactions. Recently, CHS and STS have been shown to produce 4-hydroxy-2-pyrone derivatives in vitro with the nonphysiological substrates, isovaleryl-CoA and isobutyryl-CoA .
Even though CTAL was found as a common by-product of CHS and STS in this study, there are three good reasons to believe that HmS protein is a distinct enzyme which catalyzes three consecutive decarboxylative condensations of p-coumaroyl-CoA and malonyl-CoAs without aromatic ring formation. The first is that leaves of H. macrophylla, from which this cDNA was cloned, contain CTAL derivatives such as hydramacroside B (8). The second is that this cDNA was cloned together with HmC, a normal CHS, and the third is that HmS protein produced CTAL as the major product in vitro. With all this evidence we now propose that HmS protein is a novel plant PKS responsible for the biosynthesis of hydramacroside B (8) in the mother plant and should be termed p-coumaroyltriacetic acid synthase (CTAS; Fig. 4).
As described in our previous paper , the percentage similarity of the amino acid sequences between HmS and HmC is 76%, and comparison of their sequences indicated that there are only 12 amino acid residues which show no apparent similarity in charge and size. Among them, some residues which might be responsible for the difference in enzyme function could be pointed out. Substitution of any of potentially nucleophilic residues in the CHS consensus into neutral amino acid Gly18 and amides Gln264, Asn321 and Gln324 in HmS (they are Glu or Asp, Glu, Asp and Glu in the CHS consensus, respectively) might have resulted in the loss of cyclization activity required for CHS reaction. In addition to these residues, His169 in HmS, which is Phe in the CHS consensus and adjacent to the putative active site Cys168, and a characteristic insertion of Met296–Ser301, which is only present in HmS, are hopeful targets for site-directed mutagenesis studies and may reveal the origin of product specificity.
As mentioned above, H. macrophylla produces deoxy-type stilbenecarboxylic acid, hydrangeic acid (6) as well as hydramacroside B (8). Despite our considerable efforts, the PKS responsible for stilbenecarboxylic acid biosynthesis in this plant has not been identified either at the enzyme or gene level. Some possible explanations for its biosynthesis could be suggested here. One is the presence of the third PKS clone specific for stilbenecarboxylic acid synthesis, which could not be obtained by PCRs under the conditions employed in this study. Even without assuming the presence of this third PKS clone, the presence of a specific cyclase which coacts with CTAS to produce stilbenecarboxylic acid or the presence of a reductase which coacts with CTAS to produce deoxy-type stilbenecarboxylic acid could account for the biosynthesis of hydrangeic acid. In the latter case, CTAS is assumed to produce deoxy-stilbenecarboxylic acid only by close association with the reductase protein and otherwise be unable to cyclize tetraketide intermediate into stilbenecarboxylic acid. Further investigations are required to uncover the actual mechanism of hydrangeic acid biosynthesis in this plant.
Among the members of CHS superfamily reported so far, only a few have been demonstrated for their enzyme functions. Quite recently, GCHS2 cloned from Gerbera hybrida (Asteraceae) has been proven to produce 6-methyl-4-hydroxy-2-pyrone and has been named 2-pyrone synthase . This enzyme, however, should be called as triacetic acid lactone (TAL) synthase to avoid confusion with 4-hydroxy-6-[4-(4-hydroxyphenyl)-2-oxo-3-butenyl]-2H-pyran-2-one (CTAL) synthase of this study. Now that these new members of the CHS superfamily have shown unique enzyme activities to catalyze only condensations without aromatization, it is tempting to speculate that some of the known members, whose enzyme functions have not been demonstrated, might show unexpected new enzyme activities.
Enzymes: chalcone synthase (EC 184.108.40.206); p-coumaroyltriacetic acid synthase (EC 2.3.1.-); stilbene synthase (EC 220.127.116.11, 18.104.22.168).Note: The novel nucleotide sequence data published here have been submitted to the EMBL sequence databank and are available under accession numbers AB011467 for HmC and AB011468 for HmS.