Alkylresorcinol synthases expressed in Sorghum bicolor root hairs play an essential role in the biosynthesis of the allelopathic benzoquinone sorgoleone

Abstract

Sorghum bicolor is considered to be an allelopathic crop species, producing phytotoxins such as the lipid benzoquinone sorgoleone, which likely accounts for many of the allelopathic properties of Sorghum spp. Current evidence suggests that sorgoleone biosynthesis occurs exclusively in root hair cells and involves the production of an alkylresorcinolic intermediate (5-[(Z,Z)-8',11',14'-pentadecatrienyl]resorcinol) derived from an unusual 16:3Delta(9,12,15) fatty acyl-CoA starter unit. This led to the suggestion of the involvement of one or more alkylresorcinol synthases (ARSs), type III polyketide synthases (PKSs) that produce 5-alkylresorcinols using medium to long-chain fatty acyl-CoA starter units via iterative condensations with malonyl-CoA. In an effort to characterize the enzymes responsible for the biosynthesis of the pentadecyl resorcinol intermediate, a previously described expressed sequence tag database prepared from isolated S. bicolor (genotype BTx623) root hairs was first mined for all PKS-like sequences. Quantitative real-time RT-PCR analyses revealed that three of these sequences were preferentially expressed in root hairs, two of which (designated ARS1 and ARS2) were found to encode ARS enzymes capable of accepting a variety of fatty acyl-CoA starter units in recombinant enzyme studies. Furthermore, RNA interference experiments directed against ARS1 and ARS2 resulted in the generation of multiple independent transformant events exhibiting dramatically reduced sorgoleone levels. Thus, both ARS1 and ARS2 are likely to participate in the biosynthesis of sorgoleone in planta. The sequences of ARS1 and ARS2 were also used to identify several rice (Oryza sativa) genes encoding ARSs, which are likely involved in the production of defense-related alkylresorcinols.

Figure 1.

Proposed Biosynthetic Pathway of the Allelochemical Sorgoleone. Dihydrosorgoleone, the hydroquinone produced in vivo, is thought to undergo autooxidation once secreted into the rhizosphere to yield sorgoleone, a more stable benzoquinone. The involvement of one or more cytochrome P450s in the formation of dihydrosorgoleone from 5-pentadecatrienyl resorcinol-3-methyl ether is at present speculative, as indicated by a question mark. DES, fatty acid desaturase; OMT, O-methyltransferase; P450, cytochrome P450.

Figure 2.

Comparison of 5-Pentadecatrienyl Resorcinol and PKS-Like Transcript Accumulation in Various S. bicolor Tissues. (A) The 5-Pentacedatrienyl resorcinol levels were determined by GC-MS analysis of methanol extracts prepared from isolated root hairs (top panel) and total roots (bottom panel) of 8-d-old etiolated seedlings of S. bicolor genotype BTx623. Extracted ion chromatograms are shown defined at m/z 314, and 5-pentacedatrienyl resorcinol peaks (retention time 14.8 min) are indicated by arrows. The corresponding mass spectrum for 5-pentacedatrienyl resorcinol is shown as an inset in the bottom panel. (B) The relative expression levels of five PKS-like contig sequences identified in root hair ESTs were determined by quantitative real-time RT-PCR using gene-specific primers. Data were normalized to an internal control (18S rRNA), and the ΔΔC_T method was used to obtain the relative expression levels for each sequence, expressed as mean ± sd from assays performed in triplicate.

Figure 3.

Enzymatic Activities of Recombinant ARS1 and ARS2. Relative 5-alkylresorcinol-forming activities were determined for recombinant ARS1 and ARS2 in assays using acyl-CoA starter units varying in chain length and degree of saturation. Triketide pyrone derailment products were also produced by ARS1 and ARS2 in assays containing C6 to C14 acyl-CoA starter units; however, the corresponding alkylresorcinolic products represented between ∼70 and 95% of the total moles product formed in these cases. Data are expressed as relative mean ± sd from assays performed in triplicate. The 100% relative activity corresponds to 81.0 pkat mg⁻¹ for ARS1 and 60.1 pkat mg⁻¹ for ARS2, defined as pmol of alkylresorcinol formed/s per mg of protein.

Figure 4.

Evaluation of S. bicolor RNAi Transformant Events. (A) Relative ARS1 and ARS2 endogenous transcript levels in 10-d-old S. bicolor hpRNA “+” and hpRNA “−” seedlings (representing six independent transformant events) were determined by quantitative real-time RT-PCR using gene-specific primers. Data were normalized to an internal control (18S rRNA), and the ΔΔC_T method was used to obtain the relative expression levels for each sequence, expressed as mean ± sd from assays performed in triplicate. Events numbered 1, 3, and 4 were generated using the vector pARS1-RNAi, and events numbered 2, 5, and 6 were generated using pARS2-RNAi (see Supplemental Figure 1 online). (B) Genomic DNAs isolated from the six S. bicolor RNAi transformant events, and control seedlings (genotype Tx430) were digested with either BamHI or SphI and then probed using radiolabeled Arabidopsis thaliana FAD2 gene intronic sequences present in the RNAi cassette. C, control; B, BamHI; S, SphI. (C) Sorgoleone levels were determined by GC-MS analysis of root exudates prepared from 10-d-old hpRNA “+” and hpRNA “−” seedlings representing the six RNAi transformant events. Data are expressed as mean ± sd from four measurements. The limit of quantitation (LOQ), determined to be ∼0.003 μg/mg fresh weight, is also indicated by a dashed line.

Figure 5.

Phylogenetic Relationships of ARS1, ARS2 Relatives. Strongly supported nodes (posterior probability >0.95) are indicated by shaded circles. Shaded boxes are included to highlight the placement of S. bicolor CHS and non-CHS-type sequences, and the bar at bottom represents the distance corresponding to 0.2 substitutions per amino acid. The M. polymorpha stilbene carboxylate synthase 2 (STCS2) sequence served as the outgroup for this analysis. ACS, acridone synthase; ALS, aloesone synthase; BAS, benzalacetone synthase; BBS, bibenzyl synthase; BPS, benzophenone synthase; CHS-LK, chalcone synthase-like (unknown function); CURS, curcumin synthase; DCS, diketide CoA synthase; OKS, octaketide synthase; OLS, olivetol synthase; PCS, pentaketide chromone synthase; PSS, pinosylvin synthase; STCS, stilbene carboxylate synthase; VPS, valerophenone synthase.

Figure 6.

Enzymatic Activities of ARSs Encoded by O. sativa LOC_Os05g12180, LOC_Os10g08620, and LOC_Os10g07040. Relative 5-alkylresorcinol–forming activities were determined for all three recombinant enzymes in assays using acyl-CoA starter units varying in chain length and degree of saturation. Triketide pyrone derailment products were also produced by Os05g12180 in assays containing C8 to C12 acyl-CoA starter units, by Os10g08620 in assays containing C8 and C10 starter units, and by Os10g07040 in assays containing C8 to C14 acyl-CoA starter units. The corresponding alkylresorcinolic products represented between ∼80 and 95% of the total moles product formed in these cases. Data are expressed as relative mean ± sd from assays performed in triplicate. The 100% relative activity corresponds to 66.6 pkat mg⁻¹ for Os05g12180, 48.6 pkat mg⁻¹ for Os10g08620, and 13.2 pkat mg⁻¹ for Os10g07040, defined as pmol of alkylresorcinol formed/s per mg of protein.

Figure 7.

Molecular Modeling of ARS1 and ARS2 Active Sites. The three-dimensional active site structures of M. sativa CHS2 and G. hybrida 2-PS, as well as homology models of the ARS1 and ARS2 active site structures, are shown as ribbon diagrams. Selected residues contributing to the shape/size of the active site architecture are shown in space-filling representation. (A) M. sativa (Ms CHS2). (B) G. hybrida 2-PS (Gh 2-PS). (C) and (D) Models for ARS1 and ARS2, respectively.

Figure 8.

Alignment of S. bicolor and O. sativa Sequences Exhibiting ARS Activity. The sequences for S. bicolor ARS1 and ARS2 and the ARSs encoded by O. sativa LOC_Os05g12180, LOC_Os10g08620, and LOC_Os10g07040 were aligned with M. sativa CHS2, G. hybrida 2-PS, S. bicolor CHS2, S. bicolor 0_1848, and O. sativa CHS1 using ClustalW. Residues associated with PKS functional diversity, catalysis (catalytic triad), and CoA binding are indicated based on previous crystallography studies (Ferrer et al., 1999; Jez et al., 2000) and by computational homology modeling of ARS1 and ARS2. Numbering shown above the catalytic triad positions, as well as several key residues potentially contributing to active site architecture, is based on the M. sativa CHS2 sequence. Also indicated by boxes are atypical residues identified within the 0_1848-encoded polypeptide that could account for the lack of enzymatic activity observed in recombinant enzyme studies.

Figure 9.

Alignment of Key Sequence Motifs Underlying the Aldol Switch Cyclization Mechanism. A partial alignment was generated using functionally diverse type III PKS enzymes from various sources, highlighting regions critical to the aldol switch (STS-type) cyclization mechanism first described by Austin et al. (2004a). For simplicity, numbering is shown according to the M. sativa CHS2 sequence. Positions 131 to 137 correspond to the displaced area 2 loop (outlined by rectangle) critical for the formation of the hydrogen bond network formed by Thr-132, Glu-192, and Ser-338 (highlighted in blue) in P. sylvestris STS1. Only three residues in area 2 (positions 132, 133, and 137; indicated by closed circles) are in direct contact with the active site cavity. The five area 2 positions mutated in M. sativa CHS2 to generate a functional STS-type enzyme (Austin et al., 2004a) are highlighted in yellow. Deviations from the consensus in positions Thr-132, Glu-192, and Ser-338 are boxed.