Rice contains a biosynthetic gene cluster associated with production of the casbane-type diterpenoid phytoalexin ent-10-oxodepressin

Summary

• Diterpenoids play important roles in rice microbial disease resistance as phytoalexins, as well as acting in allelopathy and abiotic stress responses. Recently, the casbane-type phytoalexin ent-10-oxodepressin was identified in rice, but its biosynthesis has not yet been elucidated.

• Here ent-10-oxodepressin biosynthesis was investigated via co-expression analysis and biochemical characterization, with use of the CRISPR/Cas9 technology for genetic analysis.

• The results identified a biosynthetic gene cluster (BGC) on rice chromosome 7 (c7BGC), containing the relevant ent-casbene synthase (OsECBS), and four cytochrome P450 (CYP) genes from the CYP71Z subfamily. Three of these CYPs were shown to act on ent-casbene, with CYP71Z2 able to produce a keto group at carbon-5 (C5), while the closely related paralogs CYP71Z21 and CYP71Z22 both readily produce a keto group at C10. Together these C5 and C10 oxidases can elaborate ent-casbene to ent-10-oxodepressin (5,10-diketo-ent-casbene). OsECBS knock-out lines no longer produce casbane-type diterpenoids and exhibit impaired resistance to the rice fungal blast pathogen Magnaporthe oryzae.

• Elucidation of ent-10-oxodepressin biosynthesis and the associated c7BGC provides not only a potential target for molecular breeding, but also, given the intriguing parallels to the independently assembled BGCs for casbene-derived diterpenoids in the Euphorbiaceae, further insight into plant BGC evolution, as discussed here.

Introduction

Crop yield loss caused by microbial pathogens is a major challenge in global agricultural production (Peterson & Higley, 2000; Dresselhaus & Hückelhoven, 2018). Part of the plant response to such infection is production of an arsenal of natural products (Bednarek & Osbourn, 2009), often termed phytoalexins (Kuc, 1995). Terpenoids are of particular importance in cereal crop plants (Peters, 2006; Schmelz et al., 2014; Murphy & Zerbe, 2020), where the vast majority fall into the labdane-related diterpenoid superfamily (Zi et al., 2014), which are defined by the initiation of their biosynthesis from the general diterpenoid precursor (E,E,E)-geranylgeranyl diphosphate (GGPP) by class II diterpene cyclases (Peters, 2010). These natural products have been shown to play important roles in resistance to microbial pathogens (Toyomasu et al., 2014; Lu et al., 2018), as well as in allelopathy (Xu et al., 2012), and even in abiotic responses (Vaughan et al., 2015; Zhang et al., 2021), with some evidence that these also act against insect herbivory (Schmelz et al., 2011).

Nevertheless, rice also produces a diterpenoid phytoalexin that does not fall into this superfamily, specifically ent-10-oxodepressin (Inoue et al., 2013). This is a casbane-type diterpenoid derived from direct cyclization of GGPP to the olefin ent-casbene by a class I diterpene synthase, followed by further elaboration that generates the carbon-5 (C5) and C10 keto groups that form the bioactive ent-10-oxodepressin (ent-5,10-diketo-casbene). Such introduction of oxygen is predominantly catalyzed by cytochrome P450 monooxygenases (CYPs), which also can further oxidize initially produced hydroxyl groups to form oxo moieties in certain cases (Bathe & Tissier, 2019).

Notably, among some of the earliest biosynthetic gene clusters (BGCs) identified in plant genomes were two from rice that are involved in the production of many of its labdane-related diterpenoid phytoalexins (Prisic et al., 2004; Wilderman et al., 2004). In addition to the initiating class II diterpene cyclases and subsequently acting class I diterpene synthases whose co-clustering led to their discovery, these BGCs also contain CYPs involved in further elaboration (Zi et al., 2014).

Intriguingly, the rice BGC associated with momilactone production was the first for which genetic evidence was provided indicating that BGC assembly might involve negative selection pressure against partial pathway inheritance (Xu et al., 2012) – e.g., due to phytotoxicity of certain intermediates, as previously postulated (Swaminathan et al., 2009). Indeed, consistent with such negative selection pressure, an independently evolved BGC recently was associated with momilactone production in the bryophyte Calohypnum plumiforme (Mao et al., 2020; Zhang & Peters, 2020). In addition, barnyard grass (Echinochloa crus-galli) contains a BGC with significant orthology to the rice momilactone BGC (Guo et al., 2017; Kitaoka et al., 2020), which may have arrived in this species via hybridization and introgression (Peters, 2020). Such association of momilactones with BGCs seems to support the hypothesis that negative (phytotoxicity), as well as positive (bioactivity), selection pressure are required to drive BGC assembly in plants (Swaminathan et al., 2009; Takos & Rook, 2012).

Casbane-type diterpenoids have long been known from plants in the Euphorbiaceae family (Shi et al., 2008), and their biosynthesis therein has been well-studied. Indeed, the casbene synthase (CBS) from castor bean (Ricinus communis), RcCBS, was one of the first terpene synthases (TPSs) to be identified (Mau & West, 1994). Additional such CBSs and paralogs that produce the structurally-related macrocyclic olefin neocembrene also have been identified (Kirby et al., 2010). Notably, the genes for these diterpene producing TPSs were found to anchor BGCs that contain those encoding subsequently acting CYPs as well (King et al., 2014; Luo et al., 2016).

Here investigation of ent-10-oxodepressin biosynthesis led to discovery of a third diterpenoid BGC in rice that is responsible for production of this casbane-type phytoalexin, whose physiological relevance was demonstrated by genetic analysis. Discovery of a BGC in rice associated with casbane-type diterpenoid metabolism parallels those found for such purpose in the Euphorbiaceae, the implications of which are discussed below.

Materials and Methods

Rice plants were grown as previously described (Lu et al., 2018). Cloning and co-expression analysis by quantitative real-time PCR (qRT-PCR) were carried out with the rice cultivar (cv.) Nipponbare, using the primers listed in Table S1. Castor bean diterpene synthases were cloned from a previously described cDNA library (Jackson et al., 2014), and pseudomature constructs made as previously described (Kirby et al., 2010). All constructs were verified by complete gene sequencing. Subcellular localization was determined as previously described (Fu et al., 2016). Biochemical characterization was carried out using a modular metabolic engineering system in Escherichia coli developed for such studies (Cyr et al., 2007). Product extraction, analysis by gas chromatography with mass spectral detection (GC-MS), and purification for NMR analysis were carried much as previously described (Lu et al., 2018). Transgenic rice lines were generated with the rapidly propagated cv. Kitaake, again much as previously described (Zhang et al., 2021). Targeted metabolite analysis was carried out via GC-MS with plant extracts following elicitation with CuCl₂, also much as previously described (Lu et al., 2018). Infection assays with M. oryzae were accomplished via punch-incubation with detached leaves and, after incubation for 5 d, lesion areas were measured, and fungal growth quantified via qRT-PCR. Further details can be found in the Supporting Information (Methods S1).

Results

While the rice casbane-type diterpenoid phytoalexin ent-10-oxodepressin is expected to be derived from casbene that is enantiomeric to that found in Euphorbiaceae plants (i.e., is ent-casbene, 1), it was hypothesized that the rice ent-casbene synthase might be homologous to the known Euphorbiaceae CBSs. These all fall within the TPS-a subfamily, although most other members are involved in sesquiterpenoid biosynthesis instead (Chen et al., 2011). Consistent with localization of such initial steps of diterpenoid biosynthesis to plastids, the Euphorbiaceae CBSs contain an N-terminal plastid targeting sequence (Mau & West, 1994; Kirby et al., 2010). Accordingly, the rice genomic data available at Phytozome was searched with RcCBS (XP_002513343) as the query sequence and the resulting homologs were further screened for the presence of the N-terminal plastid targeting sequence necessary for a role in diterpenoid biosynthesis. Two genes were identified from this effort, OsTPS10 (Os03g22634) and OsTPS28 (Os07g11790). Although these have limited homology to the known CBSs, just 29-32% amino acid identity, and are only ~55% identical to each other, both seemed to have putative plastid targeting sequences. In addition, while both contained the aspartate-rich DDxxD motif involved in binding the divalent magnesium ion co-factors required for terpene synthase activity, the second motif required for this purpose, usually conserved as (N,D)Dxx(S,T)xxxE (Christianson, 2017), is essentially no longer present. While some variability has been shown to be tolerated in this motif, specifically at the (S,T) position (Zhou & Peters, 2009; Jackson et al., 2014), the corresponding sequence in both OsTPS10 and OsTPS28 is DDxxAxxxG, with the glycine found at the last position particularly leaving their activity in question (Fig. S1).

Regardless, cDNA was cloned for both from rice leaves following infection with the rice fungal blast pathogen Magnaporthe oryzae, although that obtained for OsTPS10 was a version lacking the N-terminal plastid targeting sequence (XP_015628760). A pseudomature version of OsTPS28 for recombinant biochemical characterization was constructed by removal of the first 46 residues. Both were then characterized using a previously described metabolic engineering system in which each was (separately) co-expressed with a GGPP synthase (GGPS) in E. coli (Cyr et al., 2007). Extracts of cultures in which OsTPS10 was co-expressed with GGPS were found to produce neocembrene (2), as identified by GC-MS based comparison to that of a previously identified neocembrene synthase from castor bean (Kirby et al., 2010), and this was not otherwise investigated here. On the other hand, extracts from cultures in which OsTPS28 was co-expressed with GGPS were found to primarily produce casbene, as identified by comparison to the previously identified RcCBS (Mau & West, 1994), along with minor amounts of neocembrene (Fig. 1a). For both OsTPS10 and OsTPS28 the major product was purified and NMR spectral data was collected to confirm their identity (Figs. S2 and S3, and Table S2). Although the minor product observed with both (3) was not further analyzed its mass spectra also is reported here (Fig. S4, which contains mass spectra for all products not shown in Fig. 1). As NMR analysis does not differentiate between enantiomers, to verify that the OsTPS28 product was ent-casbene (1), the optical rotation of this was compared to that the casbene (4) produced by RcCBS, revealing that these exerted opposing effects, with 1 exhibiting specific rotation of $[\alpha {]}_D^{26}$ −162.68 (c 0.16, CHCl₃), while 4 exhibited specific rotation of $[\alpha {]}_D^{26}$ −260.61 (c 0.29, CHCl₃). Further consistent with a role for OsTPS28 in diterpenoid biosynthesis, subcellular localization analysis indicated that the full-length protein is directed to the chloroplast (Fig. S5). Hence, OsTPS28 was determined to be an ent-casbene synthase and renamed OsECBS.

Fig. 1.

Recombinant biochemical characterization.

(a) Identification of OsECBS. (i) Chromatograms of extracts from E. coli cultures co-expressing GGPS and OsTPS28/OsECBS, RcCBS or OsTPS10, as indicated (peaks are numbered as described in the text). (ii) Mass spectra of peaks 1, 2, 2’ and 4 (those for peaks 3 and 3’ are shown in Fig. S4). (iii) Chemical structure of ent-casbene (1), with carbon numbering.

(b) Oxidation of ent-casbene by CYP71Z subfamily members. (i) Chromatograms of extracts from E. coli cultures co-expressing GGPP synthase, OsECBS and indicated CYP (peaks are numbered as described in the text). (ii) Mass spectra of peak 5, 6, 9 and 10 (those for other peaks are shown in Fig. S4). (iii) Chemical structures of 5, 6, 9 and 10.

(c) Substrate feeding assays. (i-ii) Chromatograms of extracts from E. coli cultures expressing the indicated CYP and fed with (i) 5-keto-ent-casbene (5) or (ii) 10-keto-ent-casbene (9), as also indicated (peaks are numbered as described in the text). (iii) The mass spectra of the three major products 20, 21 and 26 (those for other peaks are shown in Fig. S4). (iv) The chemical structure of 20 (ent-10-oxodepressin).

(d) Pathway reconstruction. Chromatograms of extracts from the cultures expressing OsECBS and indicated pair of CYPs (peaks are numbered as described in the text, with mass spectra shown in Fig. S4).

Strikingly, it has previously been noted that the gene for OsECBS was located on chromosome 7 (c7) in the rice genome near one encoding a CYP, specifically CYP71Z2, indicating a potential biosynthetic pairing (Boutanaev et al., 2015). Indeed, it has previously been shown that CYP71Z subfamily members serve roles in rice, as well as maize, terpenoid phytoalexin biosynthesis (Wu et al., 2011; Kitaoka et al., 2015b; Mao et al., 2016; Mafu et al., 2018; Ding et al., 2019). Moreover, further inspection of the genomic region revealed the presence of three additional CYP paralogs from the CYP71Z subfamily, namely CYP71Z21, CYP71Z22 and CYP71Z30, suggesting that this region is a potential BGC, which is termed here the c7BGC. In particular, the ent-casbene product of OsECBS requires hydroxylation and subsequent oxidation at both carbon-5 (C5) and C10 to form ent-10-oxodepressin, suggesting a need for multiple CYPs. Additionally, phylogenetic comparison to other monocot CYP71Z subfamily members indicates that the four from the c7BGC are closely related to CYP71Z6 and CYP71Z7, which are already known to operate in rice diterpenoid phytoalexin biosynthesis (Wu et al., 2011; Kitaoka et al., 2015b). These four also form an isolated clade (Fig. S6, Table S3), with CYP71Z21 and CYP71Z22 sharing > 90% amino acid sequence identity, suggesting that they arose from localized gene duplication.

True functionality for BGCs has been reported to depend on co-expression of the relevant genes (Wisecaver et al., 2017). Notably, queries of the RiceXPro database supported co-expression of the genes from the c7BGC across different tissues and developmental stages (Fig. S7). To provide further support for the hypothesis that these co-clustered genes forming a functional BGC, particularly for phytoalexin biosynthesis, their expression also was analyzed by qRT-PCR in both roots and shoots, at various time points following a variety of treatments that elicit production of such natural products (Fig. S8). While expression of CYP71Z30 could not be detected, OsECBS appeared to be co-expressed with the other three (CYP71Z2, CYP71Z21 and CYP71Z22). In particular, all four of these genes were induced by not only UV-irradiation and infection with M. oryzae, which both elicit production of ent-10-oxodepressin (Inoue et al., 2013), but also treatment with methyl jasmonate, salicylic acid or copper chloride, all similarly known to lead to other diterpenoid phytoalexin production (Lu et al., 2018).

To determine if the CYP71Z subfamily members from this putative BGC can react with ent-casbene (1), CYP71Z2 and CYP71Z21 were cloned from rice leaves 36 h after infection with M. oryzae, while synthetic genes were obtained for CYP71Z22 and CYP71Z30 (sequences listed in Table S4). In each case the open reading frame was modified to replace the N-terminal transmembrane helix with a sequence that enables functional CYP expression in E. coli, much as previously described for other plant CYPs (Kitaoka et al., 2015a). These were then incorporated into the metabolic engineering system to enable each of these four CYP71Z subfamily members to be (separately) co-expressed with a CYP reductase (CPR) in E. coli also engineered to produce 1 – i.e., from additional co-expression of OsECBS and GGPS. GC-MS analysis of extracts from such cultures with CYP71Z30 indicated that this did not react with any OsECBS product (Fig. 1b). On the other hand, CYP71Z2 produced three products (5-7), among which the major product 6 and minor product 7 both had apparent molecular ions of m/z = 288, while that for minor product 5 was m/z = 286, suggesting hydroxy- or oxo- modifications, respectively (Fig. 1b-i, ii). Product purification and NMR spectral analysis, with comparison to previously identified casbane-type diterpenoids (Inoue et al., 2013; Yin et al., 2013; King et al., 2014; Horie et al., 2016), identified 5 as ent-5-keto-casbene and 6 as 5S-hydroxy-ent-casbene (Figs. 1b-iii, S9 and S10, and Table S2). However, significant amounts of the OsECBS products 1-3 also were still evident. By contrast, CYP71Z21 reacted more efficiently with ent-casbene to generate three oxidized products (8-10), among which the major product 9 was identified as ent-10-keto-casbene, while minor product 10 was identified as 10S-hydroxy-ent-casbene (Fig 1b-iii), again by comparative NMR analysis (Figs. S11 and S12, and Table S2). CYP71Z22 generated a similar but even more complex product profile as CYP71Z21, including the same 8-10 (again with 9 as the major product), as well as six other minor oxidized products (11-16).

The ability of CYP71Z2 to introduce oxygen at C5 of 1, and that of the closely related CYP71Z21 and CYP71Z22 to introduce oxygen at C10, suggests that these C5 and C10 oxidases (respectively), together might suffice for production of 10-oxodepressin. To explore this biosynthetic hypothesis, as well as the order of their reactivity, ent-5-keto-casbene (5) or ent-10-keto-casbene (9) were fed to cultures (separately) co-expressing CPR with the four CYP71Z subfamily members from the c7BGC (Fig. 1c).

No product was detected with CYP71Z30 with either substrate (Fig. 1c-i, ii). While CYP71Z2 does not further react with 5, CYP71Z21 and CYP71Z22 produced the same seven products (17-23) from this, among which product 20 exhibited a mass spectrum similar to that previously reported for ent-10-oxodepressin (Inoue et al., 2013). The major product 21 exhibits an apparent molecular ion of m/z = 302, and presumably is the 10S-hydroxylated derivative – i.e., 5-keto-10S-hydroxy-ent-casbene (Fig. 1c-i, iii). When fed 9, only CYP71Z2 further elaborates this, yielding a few products, including 20, albeit during GC-MS analysis this closely elutes with new product 26, whose apparent molecular ion of m/z = 302 suggests it might be 10-keto-5S-hydroxy-ent-casbene (Fig. 1c-ii, iii). To obtain sufficient amount of product 20 for chemical structure identification, ent-5-keto-casbene (5) was fed in higher amounts to E. coli cultures co-expressing CYP7Z22 and CPR, and the partially purified products were fed again into fresh culture for further oxidization, by which process 21 was largely transformed to 20. A sufficient amount of 20 could then be obtained for comparative NMR analysis, verifying assignment as ent-10-oxodepressin (Fig. 1c-iv, S13, Table S2). Accordingly, it appears that the C5 oxidase CYP71Z2 can act together with a C10 oxidase, either CYP71Z21 or CYP71Z22, to transform ent-casbene (1) to ent-10-oxodepressin (20).

To further investigate the sufficiency of OsECBS along with the C5 and C10 oxidases for production of ent-10-oxodepressin, pairwise combinations of the four CYP71Z subfamily members were co-expressed in the metabolic engineering system along with CPR, as well as OsECBS and GGPS. A series of oxidation products, including new compounds (27-30), were detected in these cultures (Fig. 1d). While only a trace amount of ent-10-oxodepressin (20) was detected, two of the more abundant products are of particular interest, with the major product 30 exhibiting a molecular ion of m/z = 304 (Fig. S4), suggesting that this is the 5,10-dihydroxy derivative, as well as appearance of the putative 10-keto-5S-hydroxy derivative 26 (Fig. 1c-iii). This suggests that further oxidation of the 5S-hydroxy group by CYP71Z2 is particularly inefficient.

To investigate the importance of ent-10-oxodepressin for rice fungal blast disease resistance, OsECBS was manipulated in cv. Kitaake rice. Specifically, OsECBS was disrupted by CRISPR/Cas9 genome editing to generate two knock-out lines in which OsECBS was premature terminated at 235 and 234 aa, respectively (Fig. S14), and three over-expression lines also constructed (Fig. 2). As previously reported, 10-keto-ent-casbene and ent-10-oxodepressin (Horie et al., 2016), as well as ent-casbene (Lu et al., 2018), can be detected in rice. These were analyzed here, with none found in either knock-out line, while increased amounts were found in all three over-expression lines relative to wild-type cv. Kitaake rice (Fig. S15). Perhaps more importantly, the knock-out (ko) lines exhibited decreased resistance to the fungal blast pathogen M. oryzae, with larger infection lesions and more fungal pathogen growth, while the over-expression (OE) lines exhibited increased resistance, with shortened lesion length and decreased pathogen growth, although no significant reduction in lesion size (Figs. 2c-e and S16). These changes in resistance are consistent with metabolite accumulation and a role for ent-10-oxodepressin in rice blast disease resistance.

Fig. 2.

Effects of OsECBS manipulation in cv. Kitaake rice. (a) OsECBS knock-out (ko) by CRISP-Cas9. (b) Expression of OsECBS in over-expression (OE) lines. Asterisks indicate significant difference compared with WT (Student’s t-test, *P<0.05, **P<0.01, n=3). (c-e) Resistance of OsECBS transgenic plants to M. oryzae infection. (c) Leaf infection of wild-type (WT) and OsECBS-transgenic plants by punch-inoculation with M. oryzae for 5 d. (d) Lesion area measurement (n=45). (e) Fungal growth analyzed by qRT-PCR (n=3). Different letters indicate significant difference (one-way ANOVA, P<0.05.). Error bars indicate SE.

Discussion

Here investigation of the biosynthesis of the rice diterpenoid phytoalexin ent-10-oxodepressin led to identification of the relevant ent-casbene synthase OsECBS. The observed activity was somewhat surprising given the apparent loss of the usual (N,D)Dxx(S,T)xxxE magnesium co-factor binding motif (Figure. S1). Nevertheless, genetic analysis demonstrated the physiological relevance of the characterized activity, as well as the effectiveness of the derived phytoalexins against the fungal blast pathogen M. oryzae (Fig. 2). Arguably more interesting is the resulting realization that OsECBS anchors a BGC on rice chromosome 7 (c7BGC), which also contains four members of the CYP71Z subfamily, at least three of which were shown here to exhibit activity indicative of roles in ent-10-oxodepressin biosynthesis (Fig. 3).

Fig. 3.

Biosynthetic gene cluster (BGC) for ent-10-oxodepressin biosynthesis. (a) Schematic for BGC on rice chromosome 7. (b) Proposed biosynthetic pathway for production of ent-10-oxodepressin.

While this manuscript was in preparation another report appeared describing similar results (Zhan et al., 2020). In particular, guided by genome-wide association studies, the c7BGC was identified, with co-expression of the clustered genes shown, along with characterization of OsTPS28 as a casbene synthase, along with CYP71Z21 as an initially acting C10 oxidase and CYP71Z2 as a subsequently acting C5 oxidase. However, CYP71Z22 and CYP71Z30 were not investigated, nor was the stereochemistry of the enzymatic products. Accordingly, the results reported here provide further insight into rice casbane-type diterpenoid biosynthesis.

For example, although the lack of expression and activity with CYP71Z30 reported here is somewhat inconclusive, this appears to be severely truncated in cv. Kitaake rice (Fig. S17), suggesting loss of function, and CYP71Z30 may be a pseudogene. On the other hand, CYP71Z22 exhibits C10 oxidase activity that seems to be redundant to that shown for CYP71Z21. Together this clarifies the components of the c7BGC necessary for production of ent-10-oxodepressin, the importance of which is highlighted by the recently reported highly heterogeneous nature of this BGC in different varieties of rice (Zhan et al., 2020).

In addition, it was found here that, while CYP71Z2 efficiently carries out 5S-hydroxylation, it only inefficiently mediates the subsequent oxidation of this to a 5-keto group. This contrasts with the recent report (Zhan et al., 2020), and may be a function of the reduced metabolic complexity of the E. coli host used here relative to the yeast (Saccharomyces cerevisiae) host used in that study, specifically endogenous alcohol dehydrogenases as has been previously noted for other casbene-type diterpenoid biosynthesis (Wong et al., 2018). Thus, the results reported here suggest the possibility that at least one additional oxidase may be involved in ent-10-oxodepressin biosynthesis. Indeed, short chain alcohol dehydrogenases from rice have been shown to exhibit activity indicative of roles in other diterpenoid phytoalexin biosynthesis (Kitaoka et al., 2016; Kitaoka et al., 2020). Nevertheless, the rice c7BGC is largely responsible for production of ent-10-oxodepressin.

It has been hypothesized that ent-casbene is preferentially first elaborated to the 10-keto derivative (9) before hydroxylation and subsequent further oxidation to ent-10-oxodepressin (Fig. 3b), which is supported by the observation of 9 in planta (Horie et al., 2016), and the previously reported biochemical characterization (Zhan et al., 2020). However, the more extensive biochemical studies reported here demonstrated the ability of both CYP71Z2 and the functionally analogous CYP71Z21 and CYP71Z22 to act upon ent-casbene, suggesting that production of ent-10-oxodepressin may actually occur via a more complex biosynthetic network (Fig. S18).

Observation of a BGC associated with production of ent-10-oxodepressin in rice evokes intriguing parallels to the BGCs similarly associated with such casbane-type diterpenoid phytoalexin biosynthesis in Euphorbiaceae plant species. These were clearly independently assembled, with the incorporated CYPs derived from separate families – i.e., CYP726 in Euphorbiaceae (King et al., 2014) versus the CYP71 found in the rice c7BGC. In addition, other than falling within the same TPS-a subfamily, OsECBS also is not otherwise closely related to the Euphorbiaceae CBSs. This raises the question of why production of casbane-type diterpenoids seems to be associated with BGCs. Given the hypothesis that BGC assembly requires negative selection pressure against partial pathway inheritance (Swaminathan et al., 2009; Takos & Rook, 2012), it can be speculated that this could be due to the previously demonstrated phytotoxicity of casbene (Sitton & West, 1975), particularly since this has been suggested to arise from membrane disruption (Islam et al., 2016), which might still be exerted by ent-casbene. Regardless, identification of the c7BGC and its responsibility for production of ent-10-oxodepressin, as well as demonstration that this is an effective phytoalexin against M. oryzae, particularly combined with the previous report that the c7BGC is non-functional in many rice cultivars (Zhan et al., 2020), indicates that this loci may be a productive target for molecular breeding efforts against this devastating fungal pathogen and, thus, these findings should be of significant agricultural relevance.

Supplementary Material

SI

Fig. S1 Sequence alignment of OsTPS28/OsECBS, RcCBS and OsTPS10.

Fig. S2 ¹³C NMR spectrum for neocembrene (2’).

Fig. S3 ¹³C NMR spectrum for ent-casbene (1).

Fig. S4 Mass spectra of unidentified products.

Fig. S5 Subcellular localization of OsECBS.

Fig. S6 Phylogenetic tree for monocot CYP71Z subfamily.

Fig. S7 Heat map of the gene expression in tissues.

Fig. S8 qRT-PCR analysis of gene expression of c7BGC.

Fig. S9 ¹³C NMR spectrum for 5-keto-ent-casbene (5).

Fig. S10 ¹³C NMR spectrum for 5S-hydroxy-ent-casbene (6).

Fig. S11 ¹³C NMR spectrum for 10-keto-ent-casbene (9).

Fig. S12 ¹³C NMR spectrum for 10S-hydroxy-ent-casbene (10).

Fig. S13 ¹³C NMR spectrum for ent-10-oxodepressin (20).

Fig. S14 Sequence alignment of OsECBS protein in cv. Kitaake and CRISPR-lines.

Fig. S15 Casbane-type diterpenoid metabolite analysis of OsECBS transgenic lines.

Fig. S16 Lesion length of OsECBS transgenic rice and cv. Kitaake.

Fig. S17 Sequence alignment of CYP71Z30 from cv. Nipponbare and Kitaake.

Fig. S18 Potential biosynthetic network for ent-10-oxodepressin.

Table S1 Primer sequences.

Table S2 ¹³C chemical shift data of products identified in this study.

Table S3 Monocot CYP71Z subfamily members in the phylogenetic tree analysis.

Table S4 Synthetic gene sequences.