Novel Benzene Ring Biosynthesis from C3 and C4 Primary Metabolites by Two Enzymes*♦

The shikimate pathway, including seven enzymatic steps for production of chorismate via shikimate from phosphoenolpyruvate and erythrose-4-phosphate, is common in various organisms for the biosynthesis of not only aromatic amino acids but also most biogenic benzene derivatives. 3-Amino-4-hydroxybenzoic acid (3,4-AHBA) is a benzene derivative serving as a precursor for several secondary metabolites produced by Streptomyces, including grixazone produced by Streptomyces griseus. Our study on the biosynthesis pathway of grixazone led to identification of the biosynthesis pathway of 3,4-AHBA from two primary metabolites. Two genes, griI and griH, within the grixazone biosynthesis gene cluster were found to be responsible for the biosynthesis of 3,4-AHBA; the two genes conferred the in vivo production of 3,4-AHBA even on Escherichia coli. In vitro analysis showed that GriI catalyzed aldol condensation between two primary metabolites, l-aspartate-4-semialdehyde and dihydroxyacetone phosphate, to form a 7-carbon product, 2-amino-4,5-dihydroxy-6-one-heptanoic acid-7-phosphate, which was subsequently converted to 3,4-AHBA by GriH. The latter reaction required Mn2+ ion but not any cofactors involved in reduction or oxidation. This pathway is independent of the shikimate pathway, representing a novel, simple enzyme system responsible for the synthesis of a benzene ring from the C3 and C4 primary metabolites.

knowledge on the biosynthesis of aromatic compounds in nature.
Construction of Plasmids-Full details of the experimental methods are given in supplemental Methods; a summary is given below. A plasmid, pAYP20, conferring the production of a grixazone-like yellow pigment on strain M31, was isolated by shotgun cloning of a library of the chromosome of the wild-type strain. A 3.5-kb SphI fragment containing griI and griH was excised from pAYP20 and reintroduced into pIJ702 at the SphI FIGURE 1. Biosynthesis pathways of a benzene ring. A, a proposed pathway for biosynthesis of 3,4-AHBA and its derivatives in S. griseus. 3,4-AHBA is formed from ASA and DHAP through two reactions catalyzed by GriI and GriH. ASA is derived from the biosynthetic pathway for amino acids of the aspartate group and DHAP is from the glycolytic pathway. The brackets for compound 1 denote that this compound is the probable product but has not been independently characterized. B, the shikimate pathway leading to chorismate (upper) (2) and an alternative DHQ synthesis pathway, involving MJ0400 and MJ1249, established for M. jannaschii (lower) (4). DKFP, 6-deoxy-5-ketofructose-1-phosphate; E4P, erythrose-4-phosphate.
site in the same direction as that on pAYP20, resulting in pAYP25. The griI-griH, griI, and griH sequences were amplified by PCR and cloned in a pIJ702-derived plasmid, in which the melC1-melC2 sequence under the melC promoter was replaced by a short linker, resulting in pAYP26, pAYP27, and pAYP28, respectively. On these plasmids, griI and griH were under the control of the melC promoter. For protein preparation, the griI and griH sequences were also cloned in pIJ4123, resulting in pIJ4123-griI and pIJ4123-griH, in which griI and griH were both under the control of the thiostrepton-inducible tipA promoter. The co-translational griI-griH sequence was generated by PCR and cloned into pET-17b for expression in E. coli, resulting in pET-griIH, in which the co-translational griI-griH sequence was under the control of the T7 promoter.
HPLC and LC-ESIMS Analysis-HPLC analysis was carried out by using the Waters 600 HPLC system equipped with the Waters 996 photodiode array detector. Conditions for HPLC were as follows: column, Senshu Pak Docosil-B (4.6 ϫ 250 mm, Senshu Kagaku); column temperature, 30°C; flow rate, 1 ml/min. After 10 l of the reaction mixture had been injected into the column equilibrated with 0.1% trifluoroacetic acid in water, the column was initially developed isocratically for 3 min, followed by development by a linear gradient from 0 to 90% acetonitrile in water containing 0.1% trifluoroacetic acid for 15 min. LC-ESIMS analysis was carried out by using a HPLC system (model 1100 series, Agilent Technologies) equipped with a mass spectrometer (Bruker HCT plus, Bruker Daltonics) with the ESI-positive/negative mode. HPLC was conducted using a Senshu Pak Docosil-B (2.6 ϫ 250 mm, Senshu Kagaku) at a flow rate of 0.2 ml/min. After injection of the sample into the column equilibrated with 0.1% acetic acid in water, the column was initially developed isocratically for 10 min, followed by development by a linear gradient from 0 to 100% acetonitrile in water containing 0.1% acetic acid for 10 min.
Preparation of GriI and GriH Proteins-S. lividans TK21 [pIJ4123-griI] and S. lividans TK21 [pIJ4123-griH] were cultured at 30°C for 2 days in YEME medium supplemented with 5 g/ml kanamycin. The tipA promoter was induced by 5 g/ml thiostrepton and the cultivation was further continued for 2 days. All operations described below were carried out at 4°C. Cells (5 ml wet volume) were harvested by centrifugation and resuspended in 5 ml of buffer A (50 mM sodium phosphate, pH 8.0, 0.5 M NaCl, and 20% glycerol) containing 2 mg/ml lysozyme. After incubation of the mixture on ice for 30 min, the cell suspension was sonicated for 3 min and then centrifuged at 10,000 ϫ g for 10 min to remove cell debris. Polyethyleneimine was added to the supernatant to give a final concentration of 0.1% (w/v), followed by centrifugation of the mixture at 20,000 ϫ g for 20 min. The supernatant, to which imidazole was added to give a final concentration of 10 mM, was applied to a 1-ml HiTrap chelating HP column (Amersham Biosciences) equilibrated with buffer A containing 10 mM imidazole on fast protein liquid chromatgraphy. For GriI and GriH purification, the column was charged with Co 2ϩ ions and Ni 2ϩ ions, respectively. The column was washed successively with buffer A containing 10 mM imidazole and buffer A containing 50 mM imidazole, and proteins were then eluted with a 50 -250 mM linear gradient of imidazole in buffer A. The enzyme solution was concentrated by ultrafiltration and applied to a gel filtration column (HiLoad Superdex 200 16/60 prep grade, Amersham Biosciences) on fast protein liquid chromatgraphy with isocratic elution in buffer B (20 mM HEPES-NaOH, pH 7.2, 0.15 M NaCl, and 20% glycerol) at a flow rate of 1 ml/min. Proteins were quantified by measuring the absorbance of protein solution at 280 nm using the molar absorbance coefficients, 16,500 M Ϫ1 cm Ϫ1 for GriI and 23,400 M Ϫ1 cm Ϫ1 for GriH, calculated from their amino acid sequences.
In Vitro Assay of GriI and GriH-The standard reaction mixture (100 l) consisted of buffer B containing 0.1 mM MnCl 2 , 1 mM ASA, 1 mM dihydroxyacetone phosphate (DHAP, Fig. 1A), 0.1 mg/ml (3 M for monomer) GriI, and 0.4 mg/ml (13 M) GriH. After incubation at 30°C for 30 min, reaction products were analyzed by HPLC and LC-ESIMS. 3,4-AHBA produced was quantified by a colorimetric method using 4-dimethylami-nobenzaldehyde (20), as follows. The reaction was stopped by the addition of 100 l of 20% trichloroacetic acid (w/v) and then centrifuged. After 4-dimethylaminobenzaldehyde (5%, 400 l) in acetonitrile/water (9:1 (v/v)) had been added to the supernatant, the absorbance at 450 nm was measured by a spectrometer (Spectra Max plus; Molecular Devices). The amount of 3,4-AHBA was estimated on the basis of the results of the control reactions with authentic 3,4-AHBA.
The GriI reaction was examined using the standard reaction mixture in the absence of GriH and MnCl 2 . The GriI reaction product, from which GriI was removed by ultrafiltration, was used as a substrate for the GriH reaction. The concentration of the substrate for GriH in the GriI-removed reaction mixture was estimated on the basis of the amount of 3,4-AHBA produced by prolonged incubation of the mixture containing GriH and MnCl 2 . The kinetics of the GriI reaction was determined by the rate of increase in the 3,4-AHBA concentration in the presence of an excess of GriH; the reaction mixture contained 0.

Estimation of Primary Metabolites Serving as the Substrates for 3,4-AHBA Biosynthesis by Incorporation Experiments-
Gould and co-workers (11,22) and Floss and co-workers (23,24) proposed that 3,4-AHBA was formed from a C 4 unit (corresponding to C-7, C-1, C-6, and C-5 of 3,4-AHBA) derived from the tricarboxylic acid cycle and a C 3 unit (C-2, C-3, and C-4 of 3,4-AHBA) derived from the glycolytic pathway. Gould and co-workers (22) also proposed the condensation between oxalacetate and either PEP or pyruvate for 3,4-AHBA biosynthesis. To determine the primary metabolite(s) serving as the substrates for the GriI-GriH system, we examined incorporation of 13 C-labeled precursors into 3,4-AcAHBA by 13 C NMR (Table 1). When [4-13 C]aspartate and [2-13 C]aspartate were fed, C-5 and C-1 of 3,4-AcAHBA, respectively, were mainly enriched. We therefore assumed that ASA, in addition to oxalacetate, was the most probable candidate for the C 4 unit of 3,4-AHBA in agreement with the proposal of Gould and coworkers (22), because these compounds were direct derivatives from aspartate and could be served as a substrate for aldol condensation.
When sodium [3-13 C]pyruvate was fed in the absence of glucose or glycerol, C-2 of 3,4-AcAHBA, in addition to C-9 in its acetyl moiety, was mainly enriched. The acetyl moiety was derived probably from acetyl-CoA. This finding is also consistent with the observation by Gould et al. (11). When sodium [3-13 C]pyruvate was fed in the presence of glucose or glycerol, however, C-1 and C-6 of 3,4-AcAHBA were enriched, indicating that pyruvate was incorporated into 3,4-AcAHBA after being converted to a precursor of the C 4 unit through the tricarboxylic acid pathway. On the other hand, when [1-13 C]glucose or [6-13 C]glucose was fed, C-2 of 3,4-AcAHBA was highly enriched even in the presence of pyruvate. Similarly, a very high level of enrichment at C-3 of 3,4-AcAHBA was observed when [2-13 C]glycerol was fed even in the presence of pyruvate. These results suggested that the C 3 unit moiety of 3,4-AHBA was derived from a metabolite upstream from pyruvate in the glycolytic pathway but not pyruvate itself.
In the glycolytic pathway, glucose is converted to fructose 1,6-bisphosphate and cleaved into two C 3 units, DHAP and glyceraldehyde 3-phosphate. The carbon at the phosphorylated position of DHAP and that of glyceraldehyde 3-phosphate are derived originally from the C-1 and C-6 carbons of glucose, respectively, although DHAP is reversibly converted into glyceraldehyde 3-phosphate. In our incorporation experiments with [1-13 C]glucose and [6-13 C]glucose, the level (11.4%) of enrichment at C-2 of 3,4-AcAHBA from [1-13 C]glucose was higher than that (7.1%) from [6-13 C]glucose. Furthermore, the ratio (the calculated ratio, 11.4/5.5 ϭ 1.96) of enrichment at C-2 to C-9 of the labeled 3,4-AcAHBA derived from [1-13 C]glucose was much higher than that (7.0/6.9 ϭ 1.01) from [6-13 C]glucose. These results showed that the carbon of C-1 of glucose had been incorporated into C-2 of 3,4-AcAHBA before it was incorporated into acetyl-CoA more efficiently than that of C-6 of glucose. Therefore, in disagreement with the proposal of Gould and co-workers (22), we assumed that DHAP, but not PEP or pyruvate, was the most probable candidate for the C 3 unit of 3,4-AHBA. This idea was supported by the finding that C-2 of glycerol, which enters into the glycolytic pathway via DHAP, was incorporated into C-3 of 3,4-AcAHBA very efficiently.
Using the GriI and GriH proteins, we examined various conditions for the synthesis of 3,4-AHBA. When ASA and DHAP were incubated with GriI and GriH in the presence of 0.1 mM MnCl 2 at 30°C, 3,4-AHBA was produced at a constant rate of about 22 M/min (2.2 nmol/min) for 30 min (Fig. 2C, panel b). As shown in Fig. 3C, this reaction proceeded over a pH range of 6.0 -9.5 with a maximum rate at pH 8.0 at 40°C. GriI was stable between pH 6.5-10.0 (at 30°C for 1 h) and below 40°C (at pH 7.2 for 1 h). GriH was stable between pH 6.5-10.5 (at 30°C for 1 h) and below 30°C (at pH 7.2 for 1 h). We examined all combinations of possible C 4 units (ASA, oxalacetate, L-aspartate, and L-homoserine) and C 3 units (DHAP, PEP, pyruvate, and dihydroxyacetone) as substrates and confirmed that 3,4-AHBA was synthesized only from the combination of ASA and DHAP. Addition of Mn 2ϩ or some bivalent metal ions (0.1 mM) to the reaction mixture was essential for the synthesis of 3,4-AHBA (relative activity: MnCl 2 , 100%; CoCl 2 , 85%; FeSO 4 , 81%; MgSO 4 , 41%; NiSO 4 , 6%; ZnSO 4 , 4%; CaCl 2 and CuSO 4 , Ͻ1%). No Cofactor Requirement for the GriH Reaction-Mn 2ϩ was required for the GriH reaction but not for the GriI reaction, because addition of GriH and MnCl 2 to the GriI-removed mixture containing no Mn 2ϩ and incubation of the mixture at 30°C for 30 min yielded 3,4-AHBA. This experiment also showed that a compound produced as a result of aldol condensation of ASA and DHAP by GriI served as the substrate of GriH, resulting in 3,4-AHBA. The plots of the initial velocities of the production of 3,4-AHBA versus concentrations of Mn 2ϩ showed sigmoidal curves (Fig. 3B), indicating that more Mn 2ϩ was required for the reaction as the amount of GriH was decreased. This finding excluded the possibility that Mn 2ϩ bound tightly to GriH functioned as a normal cofactor. Mn 2ϩ may stabilize the substrate for GriH, i.e. the compound produced as a result of aldol condensation between ASA and DHAP by GriI, by chelating the compound, which was extremely unstable (see below).
Because MJ1249, a GriH homologue of M. jannaschii, requires NAD (4), we examined effects of cofactors on the formation of 3,4-AHBA by GriH. Addition of 0.1 mM each of NAD, NADH, NADP, NADPH, FAD, or FMN to the reaction mixture had negligible effects on the 3,4-AHBA formation (relative activity, 98 -107%), although the reaction was partially inhibited by 0.1 mM pyridoxal phosphate (relative activity, 9.4%). Furthermore, HPLC and absorption spectrum analyses of the purified GriH protein showed the absence of any cofactors (data not shown).
Kinetics Analyses of GriI and GriH Reactions-We determined the kinetics of the GriI reaction by measuring the rate of increase in the 3,4-AHBA concentration in the presence of an excess of GriH. Because the reciprocal plots of the GriI reaction showed apparently parallel lines (Fig. 3D, panel a), the kinetic parameters were determined by fitting on the equation for the double-displacement mechanism (k cat , 0.20 Ϯ 0.01 s Ϫ1 for monomer; K m for ASA, 5.6 Ϯ 1.5 M; K m for DHAP, 140 Ϯ 9 M). The kinetics of the GriH reaction was determined by the rate of increase in the 3,4-AHBA concentration using the GriIremoved reaction mixture (see "Experimental Procedures") as the substrate. The amount of the substrate for GriH (probably compound 1; see below) in the mixture was estimated on the basis of the amount of 3,4-AHBA produced by prolonged incubation of the mixture containing GriH and MnCl 2 . The reciprocal plots of the GriH reaction followed the Michaelis-Menten kinetics (Fig. 3D, panel b). The k cat and K m values were calculated as 0.025 Ϯ 0.002 s Ϫ1 and 12 Ϯ 2 M, respectively. This K m value for the GriH reaction might be underestimated because the concentration of the GriH substrate determined was perhaps lower than the actual concentration due to the instability of the substrate compound.
Proposed Reaction Pathway of 3,4-AHBA Formation-Although the product of the GriI reaction was extremely unstable, we detected a relatively stable product with absorption at 292 nm at pH 2.0 after prolonged GriI reaction (Fig. 2C, panel b,  inset). Because the compound did not serve as a substrate for GriH, it must be a shunt product. The compound was also detected in the culture broth of S. lividans [pIJ4123-griI] and S. griseus [pAYP27] (data not shown). This shunt product was identified as 5,2-APC. We assumed that the actual product of  4-AHBA (B). A, a proposed mechanism for the non-enzymatic reaction leading to 5,2-APC from the product of the GriI reaction, 1. B, a proposed mechanism for the GriH reaction yielding 3,4-AHBA from the product of the GriI reaction, 1. Mn 2ϩ is essentially required for the GriI-GriH reaction, although it is not illustrated.
the GriI reaction was 2-amino-4,5-dihydroxy-6-one-heptanoic acid-7-phosphate (1, Fig. 1A), since 5,2-APC was presumably derived from 1 through a rationalized pathway (Fig. 4A). The pathway we propose is as follows. 1 is converted to a Schiff base 1b via formation of 1a by an intramolecular reaction. In 1b, the double bond migrates to produce 3a, which results in dephosphorylation to yield a product 3b. 3b is in equilibrium with 3c and 3d. Ring opening of 3d and subsequent ring closure afford 3f with a pyrrolidine ring, which produces 3g by dehydration. 5,2-APC is formed from 3g by migration of the double bonds. GriI thus catalyzed an aldolase reaction between the aldehyde carbon of ASA and the hydroxylated carbon of DHAP (Fig. 1A).
On the other hand, we detected no intermediates during the GriH reaction, suggesting that several steps in the GriH reaction proceeded consecutively in a substrate-binding pocket of GriH. Considering no requirement of GriH for cofactors involved in oxidative and reductive reactions, we propose a chemically rationalized pathway from the intermediate 1 to 3,4-AHBA (Fig. 4B). In the pathway, 1 is converted to a Schiff base 1b via formation of 1a by an intramolecular reaction. In 1b, the double bond migrates to produce a tautomeric Schiff base 1c, which is in equilibrium with the hydrated form 1d. A ketone 1e formed by dehydration of 1d facilitates dephosphorylation to give an enone 1f. The formation of an enone along with dephosphorylation is known for some enzyme reactions (26,27). Ring opening of 1f and subsequent aldol condensation affords a carbocyclic compound 1h. A series of reactions from 1e to 1h seems to be reasonable, because similar reactions have been proposed for the DHQ synthase (27) (6 to DHQ in Fig. 1B). 3,4-AHBA is readily formed from the imino-ketone 1h by dehydration.
Comparison of GriI-GriH System with an Alternative DHQ Synthesis Pathway of Archaea-Recently, MJ0400 (a GriI homologue) and MJ1249 (a GriH homologue) of M. jannaschii were reported to be involved in an alternative pathway for DHQ biosynthesis (4). In this pathway, MJ0400 forms 2-amino-3,7dideoxy-D-threo-hept-6-ulosonic acid (7, Fig. 1B) via a transaldol reaction between the dihydroxyacetone fragment of 6-deoxy-5-ketofructose-1-phosphate and ASA. DHQ is then formed by MJ1249 via NAD-dependent oxidative deamination of 7, producing 8, and subsequent cyclization. Because most Archaea lack the first two key enzymes involved in the synthesis of DHQ (3), MJ0400 and MJ1249 are thought to supply DHQ to the shikimate pathway (4). Interestingly, the reaction catalyzed by GriH is totally different from that catalyzed by MJ1249, despite sequence similarity between GriH and MJ1249. Much less conserved amino acid sequences in the N-terminal portions of GriH and MJ1249 may explain the difference of the reactions catalyzed by these enzymes. On the other hand, the reactions catalyzed by GriI and MJ0400 are analogous. Thus, the biosynthetic pathway of 3,4-AHBA may have evolved from the alternative DHQ synthesis pathway.
The genome data bases predict that griI and griH homologues are present in several bacteria (supplemental Figs. 1s and   2s). These griI and griH homologues are probably involved in the biosynthesis of 3,4-AHBA that serves as a building block of the respective secondary metabolites. Furthermore, griH homologues are found in the genomes of higher plants, such as Arabidopsis thaliana and Oryza sativa, giving the possibility that the benzene ring biosynthetic pathway involving GriH homologues is distributed widely in nature.