CYP701A8: A rice ent -kaurene oxidase paralog diverted to more specialized diterpenoid metabolism 1

All higher plants contain an ent -kaurene oxidase (KO), as such a cytochrome P450 CYP701 family member is required for gibberellin (GA) phytohormone biosynthesis. While gene expansion and functional diversification of GA biosynthesis derived diterpene synthases into more specialized metabolism has been demonstrated, no functionally divergent KO/CYP701 homologs have been previously identified. Rice ( Oryza sativa ) contains five CYP701A subfamily members in its genome, despite the fact that only one (OsKO2/CYP701A6) is required for GA biosynthesis. Here we demonstrate that one of the other rice CYP701A sub-family members, OsKOL4/CYP701A8, does not catalyze the prototypical conversion of the ent -kaurene C4 α -methyl to a carboxylic acid, but instead carries out hydroxylation at the nearby C3 α position in a number of related diterpenes. In particular, under conditions where OsKO2 catalyzes the expected conversion of ent -kaurene to ent -kaurenoic acid required for GA biosynthesis, OsKOL4 instead efficiently reacts with ent -sandaracopimaradiene and ent -cassadiene to produce the corresponding C3 α -hydroxylated diterpenoids. These compounds are expected intermediates in biosynthesis of the oryzalexin and phytocassane families of rice antifungal phytoalexins, respectively, and can be detected in rice plants under the appropriate conditions. Thus, it appears that OsKOL4 plays a role in the more specialized diterpenoid metabolism of rice, and our results provide novel evidence for divergence of a KO/CYP701 family member from GA biosynthesis. This further expands the range of enzymes recruited from the ancestral GA primary pathway to the more complex and specialized labdane-related diterpenoid metabolic network found in rice.


INTRODUCTION
The GAs are phytohormones required for normal plant growth and development in all higher plants (Yamaguchi, 2008). Accordingly, the genes encoding the relevant biosynthetic enzymes must be present in all plant genomes. These have provided a genetic reservoir from which other, more specialized diterpenoid metabolism has evolved (Peters, 2010). Indeed, such derivation of more specialized metabolism from hormone biosynthesis appears to be an emerging theme in plant metabolism (Chu et al., 2011). GA biosynthesis is initiated by cyclization of the general diterpenoid precursor (E,E,E)geranylgeranyl diphosphate (GGPP) 1 to ent-labdadienyl/copalyl diphosphate (ent-CPP), leading to designation of the derived natural products as labdane-related diterpenoids (Peters, 2010).
Intriguingly, the rice genome contains multiple paralogs of KO/CYP701 as a five-gene tandem array on chromosome 6 ( Figure 2A), despite the fact that only one (CYP701A6, termed OsKO2, as enumerated by chromosomal order) is apparently involved in GA biosynthesis (Sakamoto et al., 2004). Notably, almost all of the rice labdane-related diterpenoid phytoalexins contain oxygen at C3, and in many cases this can be traced back to insertion/hydroxylation at the C3α (pro-R) position (Peters, 2006), which is very close to the C4α-methyl (i.e., C19) that KO acts upon ( Figure 2B). In addition, the two more phylogenetically distant rice KO paralogs (thus termed KO-like -KOL, hence OsKOL4/CYP701A8 and OsKOL5/CYP701A9), exhibit inducible transcription in response to elicitation with the fungal cell wall component chitin oligosaccharide, and OsKOL4 is not able to complement the GA biosynthetic deficiency of an OsKO2 mutant, although the more closely related OsKO1/CYP701A7 can .
Together, this suggests that the OsKOL might be involved in more specialized metabolism (Peters, 2006).
Here we demonstrate that under conditions where OsKO2 exhibits KO activity, OsKOL4 does not, instead catalyzing C3α-hydroxylation of ent-sandaracopimaradiene and ent-cassadiene, as well as ent-kaurene. We further report that in planta OsKOL4 transcripts accumulate in response to methyl jasmonate, followed by accumulation of 3α-hydroxy-entsandaracopimaradiene and 3α-hydroxy-ent-cassadiene, which are expected intermediates in the similarly inducible production of oryzalexins A-F and phytocassanes A-E, consistent with a role for this CYP in early oxidative steps of the biosynthetic pathways for these phytoalexins.

Recombinant Expression
Based on previously reported evidence suggesting that OsKOL4 and OsKOL5 might be involved in rice labdane-related diterpenoid phytoalexin biosynthesis , we were interested in characterizing their biochemical activity. This was originally attempted using the native genes, obtained from the KOME rice cDNA database (Kikuchi et al., 2003). Due to our previous successful expression of plant CYP in E. coli (Swaminathan et al., 2009;Morrone et al., 2010;Wang et al., 2011;Wu et al., 2011), this was carried out using our bacterial modular metabolic engineering system (Cyr et al., 2007). In particular, by co-expression with the requisite CYP reductase (CPR) and functional pairings of upstream CPS and KS(L) to produce all of the labdane-related diterpenes found in rice (see Figure S1). Unfortunately, this did not lead to any further transformation, even when these native genes were modified at their Ntermini to optimize functional bacterial expression.
Codon optimization has been reported to increase expression efficiency of plant CYP (Chang et al., 2007), and we have found that complete gene recoding to optimize codon usage for expression in E. coli can lead to activity when none was observed with the native gene sequence Wu et al., 2011). Thus, we had such gene constructs synthesized for OsKOL4 and OsKOL5. For comparative purposes, we also had such a gene construct synthesized for OsKO2 as well. Each of these was further N-terminally modified for bacterial expression, and both full-length and modified constructs were again incorporated into our metabolic engineering system as described above. The N-terminally modified synthetic OsKO2 construct exhibited the KO activity expected from the previously reported genetic studies Sakamoto et al., 2004), as well as biochemical characterization of the native gene recombinantly expressed in Pichia pastoris (Ko et al., 2008), selectively reacting with entkaurene to produce ent-kauren-19-oic acid via sequential oxidation to ent-kauren-19-ol and entkauren-19-al intermediates ( Figure S2). By contrast, while no activity was observed for either OsKOL5 construct, the N-terminally modified synthetic OsKOL4 construct reacted with three diterpenes (ent-kaurene, ent-sandaracopimaradiene, and ent-cassadiene), with reasonable conversion of these diterpene olefins to apparently hydroxylated diterpenoids (MW=288 Da) in each case ( Figure 3). Furthermore, the ent-kaurene product was not ent-kauren-19-ol, demonstrating a change in regiochemistry relative to KO.

Product Identification
In order to produce enough of the OsKOL4 products for structural characterization by NMR, we increased the yield of our metabolic engineering system by incorporation of the "bottom half" of the yeast mevalonate dependent (MEV) pathway, enabling production of isoprenoid precursors from fed mevalonate, which significantly increases accumulation of the diterpenoid end product (Morrone et al., 2010). It was then possible to produce and purify several milligrams of each product by extraction from reasonable quantities of these recombinant cultures (3-L each). From the subsequent NMR analysis (Figures S3 and Tables S1-3), it was found that each product contained a C3α-hydroxyl [i.e., (R)-3-OH] group ( Figure 4).

Enzymatic Characterization
The hydroxylase activity of OsKOL4 was further characterized by in vitro enzymatic analysis. This was accomplished via co-expression with a rice CPR (OsCPR1), with determination of the level of functional CYP present by measurement of the CO difference binding spectra from the resulting clarified lysates ( Figure S4). However, it should be noted that such spectra are rather inconsistent, with some preparations exhibiting activity in the absence of the characteristic peak at 450 nm. Thus, the resulting catalytic rates must be viewed with some caution. Nevertheless, steady-state kinetic analysis of enzymatic activity was carried out with preparations for which CO difference binding was apparent, which indicated that OsKOL4 exhibits similar catalytic efficiency with ent-sandaracopimaradiene and ent-cassadiene, but a ~3fold reduction in affinity for ent-kaurene.

Physiological Relevance
C3α-hydroxy-ent-sandaracopimaradiene and C3α-hydroxy-ent-cassadiene already have been suggested as potential intermediates in the production of rice oryzalexin and phytocassane phytoalexins (Peters, 2006). Transcription of the genes encoding the relevant diterpene synthases, as well as production of the diterpene precursors themselves, is inducible by methyl jasmonate Morrone et al., 2011). Thus, we investigated the effect of such induction on transcription of OsKOL4 and accumulation of its enzymatic products. Similar to the upstream diterpene synthases, OsKOL4 mRNA levels were dramatically increased in rice leaves by induction with methyl jasmonate, accumulating to more than 10-fold higher levels within 12 hrs ( Figures 5 and S5). By contrast, OsKO2 mRNA levels were relatively unchanged.
In addition, it was possible to detect increased accumulation of both C3α-hydroxy-entsandaracopimaradiene and C3α-hydroxy-ent-cassadiene following induction, although C3αhydroxy-ent-kaurene was not detected at any point (

DISCUSSION
The mutually exclusive activity reported here for OsKOL4 relative to OsKO2 demonstrates biochemical divergence between these rice KO paralogs. Specifically, while the OsKO2 recombinant construct used here exhibited the expected KO activity Sakamoto et al., 2004;Ko et al., 2008), the analogous OsKOL4 construct was found to only catalyze C3αhydroxylation, even with their common substrate ent-kaurene ( Figure 4). Accordingly, although much more distantly related CYP701 family members retain KO activity -e.g., PpKO/CYP701B1, which falls into a separate sub-family and shares less than 42% amino acid (aa) sequence identity with any member of the CYP701A sub-family (Miyazaki et al., 2011) -OsKOL4, which shares 71% aa sequence identity with OsKO2, has evolved novel enzymatic function.
While we were unable to identify any activity for OsKOL5, it should be noted that, although its transcription is similarly inducible, OsKOL5 mRNA accumulates at a much lower level than does that of OsKOL4 . Thus, it is possible that OsKOL5 may be an inactive pseudogene. However, OsKOL5 is rather divergent, sharing only 79% aa sequence identity with OsKOL4, its closest relative. Accordingly, OsKOL5 may function at some later step in rice diterpenoid, or in some other type of natural products biosynthesis.
In any case, our results further indicate that OsKOL4 acts in rice diterpenoid phytoalexin biosynthesis. In particular, the C3α-hydroxylation of ent-sandaracopimaradiene and entcassadiene catalyzed by OsKOL4 is correlated with the presence of C3(α)-oxy moieties in the respectively derived oryzalexins and phytocassanes (Figure 6), and induction with the plant defense signaling molecule methyl jasmonate elicits both increased levels OsKOL4 mRNA and subsequent accumulation of its enzymatic products ( Figure 5). Moreover, uv-irradiation similarly increases OsKOL4 mRNA levels , which further has been shown to lead to accumulation of at least C3α-hydroxy-ent-sandaracopimaradiene (Kato et al., 1995).
Perhaps more critically, consistent with suggestions that the oryzalexins and phytoalexins serve as phytoalexins against the blast pathogen Magnaporthe oryzae (Peters, 2006), increased mRNA levels of not only OsKOL4, but also the relevant upstream diterpene synthases OsCPS2, OsKSL7, and OsKSL10 are observed upon infection with this fungus (Marcel et al., 2010).
While the role of OsKOL4 in oryzalexin biosynthesis seems straightforward, its exact role in phytocassane production is not as clear ( Figure 6). These phytoalexins all contain C3 ( C11α hydroxylation of ent-cassadiene, with C11α-hydroxy-ent-cassadiene also being detected in induced rice leaf extracts (Swaminathan et al., 2009). Hence, both C3α-and C11α-hydroxylated ent-cassadienes are present in rice, which would seem to suggest the possibility of a bifurcated biosynthetic network ( Figure 6B). However, at least in vitro, OsKOL4 does not react with C11α-hydroxy-ent-cassadiene, and nor does CYP76M7 react with C3α-hydroxy-ent-cassadiene, leaving the exact roles of these CYP in phytocassane biosynthesis unclear at this time.
The ability of OsKOL4 to react with ent-kaurene offers potential competition to the use of this intermediate in GA biosynthesis by OsKO2. However, C3α-hydroxy-ent-kaurene is not observed in planta, and OsKOL4 exhibits a lower affinity for ent-kaurene than its other substrates (Table 1). Moreover, while ent-kaurene is constitutively produced in rice plants, it is a minor component of the diterpenes produced following elicitation (Wickham and West, 1992;Mohan et al., 1996;Morrone et al., 2011). Given that OsKO2 is constitutively expressed throughout rice plants, while OsKOL4 exhibits a very limited expression pattern, with significant expression levels only observed upon elicitation , the ability of OsKOL4 to react with ent-kaurene does not seem to be physiologically relevant.
Regardless of exact role, the biochemical divergence of OsKOL4 provides a clear example of a KO/CYP701 family member associated with more specialized, rather than GA metabolism. While stevioside biosynthesis seems likely to proceed via ent-kaurenoic acid presumably produced by one of the two distinct KO family members found in Stevia rebaudiana, it remains unclear which, or even if one or the other is dedicated to stevioside versus GA biosynthesis, and these both catalyze the prototypical KO reaction in any case (Humphrey et al., 2006). By contrast, the novel enzymatic activity exhibited by OsKOL4 precludes any function in GA biosynthesis (c.f., Figures 1 and 4), and the transcriptional regulation of this gene also is more clearly associated with more specialized (i.e., phytoalexin) metabolism ( Figure 5) .
The organization of the rice KO/CYP701 family members as a five-gene tandem array with distinct division between phylogenetically related and co-regulated KO and KOL has suggested a plausible evolutionary history . In particular, this array presumably originated with tandem gene duplication of the ancestral KO required for GA metabolism. This enabled neo-functionalization, including at least divergence in transcriptional regulation of one of these KO genes (i.e., to form an inducible KOL), with subsequent gene duplication/expansion of this ancestral KO and derived KOL pair to yield the currently observed five-gene tandem array. The data presented here indicates divergence of the enzymatic activity of at least one of the KOL (i.e., the targeting of C3α instead of C19 by OsKOL4). Notably, although the roles of the rice KO(L) family members other than OsKO2 and OsKOL4 remain unknown, at least one of these presumably then provided a selective advantage -i.e., in order for this later gene expansion to sweep through the rice population. For example, while no activity was found here for OsKOL5, it still remains possible that this acts later in rice diterpenoid, or in other natural products biosynthesis.
Intriguingly, examination of the recently sequenced Brachypodium distachyon genome (TIBI, 2010), demonstrates the existence of a similar tandem array of KO/CYP701A sub-family members (BdKO1-3). However, molecular phylogenetic analysis further indicates that this arose separately from the rice KO cluster (Figure 7). Thus, it seems likely that similar functional diversification as that we report here for OsKOL4 may have occurred multiple times, at least in the grass plant family where diversion of GA biosynthesis to the production of more specialized metabolism seems to be widespread (Peters, 2006). This latter point perhaps also indicates why it has been possible to divert OsKOL4 to more specialized metabolism with relatively few changes (i.e., this is still a member of the CYP701A sub-family, sharing 65-79% aa sequence identity with the other OsKO). By contrast, diversion of other CYP from primary to secondary metabolism has been associated with the establishment of novel sub-families, which by definition share less than 55% identity with other family members -e.g., more specialized triterpenoid biosynthesis has been shown to variously utilize a novel KAO/CYP88D sub-family member diverted from GA metabolism (Seki et al., 2008), or a novel CYP51H sub-family member derived from the C14-demethylase operating in sterol metabolism (Qi et al., 2006). Accordingly, the close relationship between GA and other labdane-related diterpenoid biosynthesis, in particular the use of structurally similar multi-cyclic diterpene intermediates, presumably enabled the relatively facile diversion of OsKOL4 to more specialized metabolism observed here.

CONCLUSION
In summary, our results demonstrate that rice contains at least one functionally divergent KO/CYP701 family member. The distinct enzymatic activity of OsKOL4 identified here precludes its action in GA metabolism, and it instead appears to play a role in rice diterpenoid phytoalexin biosynthesis. This then further expands the range of enzymes recruited from GA/primary metabolism to play functionally distinct roles in more specialized labdane-related diterpenoid metabolism beyond the diterpene synthases, and supports the emerging paradigm that the requirement for phytohormone production provides a biosynthetic reservoir that often is tapped in the evolution of secondary metabolism.

General procedure
Unless otherwise noted, chemicals were purchased from Fisher Scientific (Loughborough, Leicestershire, UK), and molecular biology reagents from Invitrogen (Carlsbad, CA, USA).
Gene mapping was based on the annotated rice genome sequence at GenBank, along with the previously assigned OsKO(L) nomenclature Sakamoto et al., 2004). CYP nomenclature was determined via BLAST searches at the Cytochrome P450 Homepage maintained by Dr. David Nelson (http://drnelson.uthsc.edu/CytochromeP450.html). Gas chromatography (GC) was performed with a Varian (Palo Alto, CA) 3900 GC with Saturn 2100 ion trap mass spectrometer (MS) in electron ionization (70 eV) mode. Samples (1 µL) were injected in splitless mode at 50 °C and, after holding for 3 min. at 50 °C, the oven temperature was raised at a rate of 14 °C/min. to 300°C, where it was held for an additional 3 min. MS data from 90 to 600 m/z were collected starting 12 min. after injection until the end of the run.

Recombinant constructs
The OsKOL4 and OsKOL5 native cDNA were obtained from the KOME rice cDNA databank (GenBank accessions AY579214 and AY660664). Synthetic constructs for OsKO2, OsKOL4, and OsKOL5, codon-optimized for expression in E. coli, were obtained from Genscript (see supplemental information for the corresponding nucleotide sequences). All of these were sub-cloned into the Gateway vector pENTR/SD/D-TOPO by directional topoisomerization. N-terminal modification of these genes for improved functional bacterial expression was performed in a two-stage PCR process, first removing the N-terminal transmembrane helix (OsKO2, 39 aa; OsKOL4, 42 aa; and OsKOL5, 36 aa) from the 5' end of the open reading frame and then adding ten new codons (encoding the amino acid sequence "MAKKTSSKGK"), based on the modifications used for bacterial expression of the mammalian CYP2B sub-family (Scott et al., 2001). All constructs were verified by complete gene sequencing and then transferred via directional recombination into a modified pCDF-Duet vector (Novagen, Madison, WI), which contains a DEST cassette in the first multiple cloning site and a rice CPR (OsCPR1) in the second multiple cloning site, as previously described (Swaminathan et al., 2009).

Recombinant expression
All CYP were expressed in the C41 Overexpress strain of E. coli (Lucigen, Middleton, WI) using our previously described modular diterpene metabolic engineering system (Cyr et al., 2007). Briefly, these CYP were co-expressed with not only OsCPR1 (i.e., using the constructs described above), but also with a GGPP synthase and CPS carried on co-compatible pGGxC vectors, as well as OsKSL expressed from the additionally co-compatible pET-based pDEST14 or pDEST15 (i.e., for expression as a fusion to glutathione-S-transferase). The resulting diterpenoids were extracted from liquid cultures (media and cells), typically 50-mL volumes grown for 72 hrs at 16 °C after induction, with an equal volume of hexane and analyzed by GC-MS, including samples that were methylated to examine potential acid formation. In every case the expected diterpene olefin product (i.e., given the co-expressed diterpene synthases) was observed, with hydroxylated diterpenoids detected as described above (Figure 3), such that the overall yields were consistent.

Diterpenoid production
The novel enzymatic products were obtained in sufficient amounts for NMR analysis by both increasing flux into isoprenoid metabolism and scaling up the culture volumes. Flux towards isoprenoid biosynthesis was increased by incorporation of the "bottom half" of the mevalonate dependent isoprenoid precursor pathway from yeast, using the previously described pMBI (Martin et al., 2003). This enables production of the isoprenoid precursors isopentenyl diphosphate and dimethylallyl diphosphate from mevalonate, such that feeding of 20 mM mevalonolactone significantly increases diterpenoid production, as previously described (Morrone et al., 2010). The resulting diterpenoids were extracted from 3 L of culture (media and cells) with an equal volume of a 1:1 mixture of ethyl acetate and hexanes, and the organic extract then dried by rotary evaporation. The resulting residue was dissolved in 5 mL 45% methanol/45% acetonitrile/10% dH 2 O, and the hydroxylated diterpenoids purified by HPLC. This was carried out using an Agilent 1100 series instrument equipped with autosampler, fraction collector, and diode array UV detection, over a ZORBAX Eclipse XDB-C8 column (4.6 x 150 mm, 5 μ m) at a 0.5 mL/min flow rate. The column was pre-equilibrated with 20% acetonitrile/dH 2 O, sample loaded, then the column washed with 20% acetonitrile/dH 2 O (0-2 min), and eluted with 20%-100% acetonitrile (2-7 min), followed by a 100% acetonitrile wash (7-27 min). Following purification, each compound was dried under a gentle stream of N 2 , and then dissolved in 0.5 mL deuterated methanol (CD 3 OD; Sigma-Aldrich), with this evaporation-resuspension process repeated two more times to completely remove the protonated acetonitrile solvent, resulting in a final estimated ~5-10 mg of each novel diterpenoid.
www.plantphysiol.org on August 23, 2017 -Published by Downloaded from Copyright © 2012 American Society of Plant Biologists. All rights reserved.

Chemical structure identification
NMR spectra for the diterpenoids were recorded at 25 °C on a Bruker Avance 500 spectrometer equipped with a cryogenic probe for 1 H and 13 C. Structural analysis was performed using 1D 1 H, 1D, DQF-COSY, HSQC, HMQC, HMBC and NOESY experiment spectra acquired at 500 MHz and 13 C (125.5 MHz) and DEPT135 spectra using standard experiments from the Bruker TopSpin v1.3 software. All samples were placed in NMR tubes purged with nitrogen gas for analyses, and chemical shifts were referenced using known methanol [ 13 C 49.15 (7), 1 H 3.31(5) ppm (m)] signals offset from tetramethylsilane (Tables S1-3). Correlations from the HMBC spectra were used to propose a partial structure, while COSY correlations between protonated carbons and HSQC spectra were used to complete the partial structure and assign proton chemical shifts. The configuration of the A and B rings (C1-C10) is predetermined by the configuration of the CPP enzyme substrate, since chemical bonds in that portion of the molecule are not altered. Thus, nuclear Overhauser effect (NOE) dipole-dipole signals observed between the C5 proton and the C3 alcohol methine proton ( Figure S4) could be used to assign the alpha (R) configuration of the C3 hydroxyl group (Table S1-S3).

Kinetic analysis
Kinetic analysis was carried out using the synthetic and N-terminally modified OsKOL4 construct expressed in E. coli using the OsCPR1 co-expression construct described above.
Expression cultures in TB medium were supplemented with 1 mM thiamine, 5 mg/L riboflavin, and 75 mg/L 5-aminolevulinic acid, along with induction with 1 mM IPTG at A 600 of 0.8-1.0.
After 72 hours at 16 °C, the cells were harvested and clarified lysates prepared for in vitro kinetic assays with quantification of CYP by reduced CO-binding difference spectra using an extinction co-efficient of 91 mM -1 cm -1 (Omura and Sato, 1964). Kinetic assays were performed using 220 pM active OsKOL4 in each 1 mL reaction. The three diterpenes were used as substrates with varying concentrations (0.8-196 µM ent-sandaracopimaradiene, 1-200 µM ent-cassadiene, and 1-240 µM ent-kaurene), but otherwise these assays were carried out as previously described (Swaminathan et al., 2009). After 30 minutes, 50 µL 1M HCL was added to stop the reaction, enzymatic products extracted by ethyl acetate, confirmed by GC-MS, and quantified by GC with flame ionization detection, using an external standard curve of ent-kauren-19-ol.

Plant analyses
The rice plants (Orzya sativa L. ssp. Nipponbare) and subsequent mRNA and metabolite analyses were largely carried out as previously described . Briefly, the plants were cultivated in growth chambers under 12 hr light (28 °C) and 12 hr dark (24 °C  (D): Mass spectra for ent-kaurene derived product (3α-hydroxy-ent-kaurene).