Gene discovery of modular diterpene metabolism in nonmodel systems

Abstract

Plants produce over 10,000 different diterpenes of specialized (secondary) metabolism, and fewer diterpenes of general (primary) metabolism. Specialized diterpenes may have functions in ecological interactions of plants with other organisms and also benefit humanity as pharmaceuticals, fragrances, resins, and other industrial bioproducts. Examples of high-value diterpenes are taxol and forskolin pharmaceuticals or ambroxide fragrances. Yields and purity of diterpenes obtained from natural sources or by chemical synthesis are often insufficient for large-volume or high-end applications. Improvement of agricultural or biotechnological diterpene production requires knowledge of biosynthetic genes and enzymes. However, specialized diterpene pathways are extremely diverse across the plant kingdom, and most specialized diterpenes are taxonomically restricted to a few plant species, genera, or families. Consequently, there is no single reference system to guide gene discovery and rapid annotation of specialized diterpene pathways. Functional diversification of genes and plasticity of enzyme functions of these pathways further complicate correct annotation. To address this challenge, we used a set of 10 different plant species to develop a general strategy for diterpene gene discovery in nonmodel systems. The approach combines metabolite-guided transcriptome resources, custom diterpene synthase (diTPS) and cytochrome P450 reference gene databases, phylogenies, and, as shown for select diTPSs, single and coupled enzyme assays using microbial and plant expression systems. In the 10 species, we identified 46 new diTPS candidates and over 400 putatively terpenoid-related P450s in a resource of nearly 1 million predicted transcripts of diterpene-accumulating tissues. Phylogenetic patterns of lineage-specific blooms of genes guided functional characterization.

Figure 1.

Diterpene biosynthesis. Schematic of proposed pathways leading from GGPP 1 to pseudolaric acid B 2, cis-abienol 3, triptolide 4, oridonin 5, carnosol 6, marrubiin 7, forskolin 8, grindelic acid 9, ingenol-3-angelate 10, and jatrophone 11. These diterpenes are formed by bifunctional class I or class I/II enzymes or pairs of monofunctional class II and class I diTPSs, catalyzing the cycloisomerization of GGPP into distinct diterpene scaffolds. Products of diTPSs can undergo various functional modifications primarily through the activity of P450 enzymes. Considering the modular organization of diterpene specialized metabolism (e.g. Hall et al., 2013), we showed the variable diTPS and P450 enzyme modules as “LEGO-like” blocks. This is conceptually modified from Baldwin (2010), who referred to the five-carbon units of terpenoids as “LEGO” blocks. Here, the different colors of the diTPS modules indicate variations of the γβα three-domain structure of plant diTPSs; a red “X” indicates loss of function of class I or class II activity. Recombining diTPS and P450 enzyme modules of diterpene pathway may allow for production of known and new diterpenes through metabolic engineering or synthetic biology.

Figure 2.

Diterpene profiling. Tissue-specific abundance of diterpene metabolites in the 10 target species of this study was evaluated by GC-MS (A) and LC-MS (B) of tissue extracts with compound verification by comparison to reference mass spectra of the Wiley Registry MS Libraries and, where available, authentic standards as detailed in Supplemental Materials and Methods S1. LC-MS chromatograms of tissue extracts (black lines) are compared with authentic standards (red lines), with asterisks indicating the respective diterpenes of interest.

Figure 3.

Transcriptome coverage of core terpenoid pathway genes. Coverage of MVA and MEP pathway genes as well as short chain isoprenyl synthases in the transcriptome assemblies was assessed by searches against databases of the KEGG. Blue lines highlight the predominant diterpenoid pathway. AACT, acetyl-CoA-C-acetyl transferase; HMGS, 3-hydroxy-3-methylglutaryl-CoA synthase; HMGR, 3-hydroxy-3-methylglutaryl-CoA reductase; MK, mevalonate kinase; MPK, mevalonate-5-P kinase; MDD, diphospho-mevalonate kinase; DXS, 1-deoxy-d-xylulose-5-P synthase; DXR, 1-deoxy-d-xylulose-5-P reductoisomerase; CMS, 2-C-methyl-erythritol-4-P cytidyl transferase; CMK, 4-(cytidine-5′-diphospho)-2-C-methyl-d-erythritol kinase; MCS, 2-C-methyl-D-erythritol-2,4-cyclo diphosphate synthase; HDS, 1-hydroxy-2-methyl-2-(E)-butenyl-4-diphosphate synthase; IDS, isopentenyl diphosphate synthase; IDI, isopentenyl diphosphate isomerase; GPPS, geranyl diphosphate synthase; FPPS, farnesyl diphosphate synthase; GGPPS, geranylgeranyl diphosphate synthase.

Figure 4.

Phylogeny of candidate diTPSs. Maximum likelihood tree, illustrating phylogenetic relationships of diTPS candidates relatively to previously known diTPSs. P. patens copalyl diphosphate synthase/kaurene synthase (PpCPS/KS) was used as the tree root. For abbreviations and accession numbers see Supplemental Table S3.

Figure 5.

Phylogenetic analysis of P450 candidates of four select species. Maximum likelihood trees of CYP88 members in T. wilfordii (A), CYP716 candidates of C. forskohlii (B), CYP76 candidates of P. amabilis (C), and members of the CYP71D and CYP726A families in E. peplus (D). Sequences of previously characterized P450s are underlined. Asterisks indicate bootstrap support of greater than or equal to 80%. Phylogenetic trees were rooted with ancestral representatives. Subfamilies represent select manually curated P450s, including all available subfamilies. For abbreviations and accession numbers see Supplemental Table S4.

Figure 6.

In vitro characterization of recombinant diTPSs from G. robusta and P. amabilis. GC-MS traces of reaction products from single or coupled in vitro assays of recombinant proteins with 15 μm GGPP 1 as substrate. Major products are compared with authentic standards or reference mass spectra of the Wiley Registry MS Libraries. A, Diterpene alcohol and olefin formation by PxaTPS4. B, Geranyllinalool production by GrTPS5. C, Formation of LPP with manoyl oxide and epi-manoyl oxide byproducts by GrTPS1 verified through identification of the dephosphorylated reaction product (labd-13-en-8,15-diol). D, Activity of GrTPS6, forming manoyl oxide and traces of epi-manoyl oxide in combination with GrTPS1, and abietadiene when coupled with ECPS (ZmECPS; Harris et al., 2005) or (+)-CPS (PaLAS:D611A; Zerbe et al., 2012a). IS, Internal standard 1.6 µm eicosene; peak a, palustradiene; peak b, levopimaradiene; peak c, abietadiene; peak d, neoabietadiene; peak e/f, epimers of 13-hydroxy-8(14)-abietene; peak g, geranyllinalool; peak h, manoyl oxide; peak i, epi-manoyl oxide; peak j, labd-13-en-8,15-diol; peak k, abietadiene; peak l, pimaradiene-type diterpene. AbCAS:D621A, A. balsamea CAS variant, producing LPP (Zerbe et al., 2012a).

Figure 7.

In vivo characterization of diTPSs expressed in N. benthamiana. GC-MS traces of A. tumefaciens-infiltrated N. benthamiana leaf extracts after transient expression of GrTPS1, GrTPS6, and combination of both diTPSs, in the presence of the P19 silencing suppressor strain (A); EpTPS3 (B); and EpTPS1, EpTPS7, CfTPS14, and the combinations EpTPS7/EpTPS1 and EpTPS7/CfTPS14 (C). Transformation with P19 alone served as a control. Compound accumulation was analyzed 4 d post infiltration. Peak m, Manoyl oxide; peak n, epi-manoyl oxide; peak o, casbene; peak p(a, b), ent-kaurene; IS, internal standard 0.2 mg^–l octadecane.