Discovery of non-squalene triterpenes

Abstract

All known triterpenes are generated by triterpene synthases (TrTSs) from squalene or oxidosqualene¹. This approach is fundamentally different from the biosynthesis of short-chain (C₁₀-C₂₅) terpenes that are formed from polyisoprenyl diphosphates^2-4. In this study, two fungal chimeric class I TrTSs, Talaromyces verruculosus talaropentaene synthase (TvTS) and Macrophomina phaseolina macrophomene synthase (MpMS), were characterized. Both enzymes use dimethylallyl diphosphate and isopentenyl diphosphate or hexaprenyl diphosphate as substrates, representing the first examples, to our knowledge, of non-squalene-dependent triterpene biosynthesis. The cyclization mechanisms of TvTS and MpMS and the absolute configurations of their products were investigated in isotopic labelling experiments. Structural analyses of the terpene cyclase domain of TvTS and full-length MpMS provide detailed insights into their catalytic mechanisms. An AlphaFold2-based screening platform was developed to mine a third TrTS, Colletotrichum gloeosporioides colleterpenol synthase (CgCS). Our findings identify a new enzymatic mechanism for the biosynthesis of triterpenes and enhance understanding of terpene biosynthesis in nature.

Fig. 1. Triterpene biosynthesis.

The classical pathway for triterpenes proceeds through squalene. This study describes three fungal bifunctional TSs that convert DMAPP and IPP through HexPP into the triterpenes talaropentaene (1), macrophomene (2) and colleterpenol (3). FPPS, FPP synthase; HexPPS, hexaprenyl diphosphate synthase; SQE, squalene epoxidase; SQS, squalene synthase.

Fig. 2. Characterization of TvTS and MpMS.

a, Engineering of S. cerevisiae for production of 1 (TvTS) and 2 (MpMS). b, Construction of T. verruculosus RC177. c, Ion chromatograms from high-resolution electrospray ionization mass spectrometry (EI-MS; m/z 505.3452) of extracts from incubation of DMAPP and IPP with (i) TvTS-PT and (ii) MpMS-PT D114A/N115A. d, EI-MS ion chromatograms of extracts from (i) incubation of DMAPP and IPP with TvTS-PT and TvTS-TC, (ii) S. cerevisiae XM139 expressing the gene for TvTS, (iii) incubation of DMAPP and IPP with MpMS, (iv) S. cerevisiae XM018 expressing the gene for MpMS, (v) engineered T. verruculosus RC177 with the TvTS gene under the control of the amyB promoter and (vi) wild-type T. verruculosus TS63-9 (negative control; Supplementary Fig. 13). e, f, HexPP cyclization to 1 (e) and 2 (f) with carbon numbering as in HexPP. Box, structurally related diterpenes.

Fig. 3. Structures of TvTS and MpMS.

a–c, Active sites in the docking model of TvTS-TC with 2,3-dihydro-HexPP (two possible conformers were docked on the basis of the observed electron density and are shown as yellow and purple sticks) (a), PaFS-TC (b) and MpMS-TC (c) (modelled by AlphaFold2). d, Cryo-EM map and reconstructed structure of non-cross-linked hexameric MpMS-PT (monomers in blue and cyan; map resolution of 3.17 Å; density map contoured at 0.065 using Chimera). e, Cryo-EM map and reconstructed structure of cross-linked MpMS-PT hexamer (map resolution of 4.00 Å; density map contoured at 0.030 using Chimera). The cross-linked TC domain helix is shown in green. f, Reconstructed structure of the MpMS-PT hexamer with a TC domain homology model (based on FgGS; Protein Data Bank (PDB), 6W26) docked into the cryo-EM map.

Fig. 4. AlphaFold2-based genome mining of CgCS.

a, AlphaFold2-based screening of chimeric class I TSs. b, Predicted structure of CgCS-TC docked with HexPP (purple spheres, Mg²⁺). In CgCS, small residues at the bottom of the active site (V222, N226, A313, A316 and S320) form a similar tunnel for two non-reacting isoprene units as observed for TvTS. c, EI-MS ion chromatograms of extracts from (i) S. cerevisiae RC181 expressing the gene for CgCS and (ii) S. cerevisiae YZL141 (negative control). d, Proposed mechanism for the cyclization of HexPP to 3 through syn addition of C1 and water to the C14=C15 double bond.

Extended Data Fig. 1. The structure of cyclase domain of TvTS.

a, the apo-form of TvTS-TC (green). The DDXXD, NSE, and RY motifs (red box) are conserved, and the active site forming regions, including aspartic-rich metal binding DDXXD motif, the region (V234-V243) after the NSE motif, and the A156-C183 region are disordered; b, PaFS-TC in complex with neridronate (cyan cartoon with magenta sticks, PDB: 5ER8, in comparison to TvTS-TS: rmsd values of 1.6 Å for Cα-atoms, 47% amino acid sequence identity); c, partially closed conformation of TvTS-TC. The disordered regions in the apo structure, especially the DDXXD motif and the active site loop D173-D182 (shown in salmon), appear clearly structured after soaking with 2,3-dihydro-HexPP. The docking model of 2,3-dihydro-HexPP was constructed based on the observed density and two possible conformers are shown by yellow and purple sticks.

Extended Data Fig. 2. Structure based mutagenesis studies with TvTS.

a, Active site cavity of TvTS-TC. The docking model of 2,3-dihydro-HexPP was constructed based on the observed density and two possible conformers were shown as yellow and purple sticks. b, In vitro activities of wild-type TvTS and its variants for production of 1. Full-length TvTS and its variants were expressed with a maltose binding protein (MBP) fused at the N-terminus. Peak integrals for the ion chromatogram of m/z 408 were used for quantification. Wild-type production was set to 100%, bars and error bars show mean and s.d. from three biological replicates, respectively.

Extended Data Fig. 3. Active site of prenyltransferase domain of MpMS.

a, noncrosslinked MpMS-PT; b, crosslinked MpMS-PT; c, crystal structure of PaFS-PT complexed with cobalt ions and pamidronate (PDB ID: 5ERO); d, superimposed crosslinked MpMS-PT (cyan) and PaFS-PT (wheat) crystal structure; e, active site cleft of MpMS-PT; f, active site cavity of PaFS-PT.

Extended Data Fig. 4. Close-up view of the interface between PT and TC domains.

a, The residues on α5 of TC domain (green) interact with the residues on α2 (419-429) and C-terminal α14 (686-695) of the PT domain (cyan); b, close-up view of the interface between the PT domain and fitted model of TC domain. Distances indicated by grey dashed lines are given in Å.

Extended Data Fig. 5. The active site of MpMS-PT (cyan) and TC (green) domains face to each other.

a, top view; b, side view.

Extended Data Fig. 6. Mutagenesis study of MpMS.

Ion chromatograms (m/z = 408) of wildtype MpMS (black) and its enzyme variants V206F (blue) and A207W (magenta).

Extended Data Fig. 7. Product detection of triterpenes produced by chimeric class I TrTS in S. cerevisiae.

Engineered S. cerevisiae containing i) PTTC027, ii) PTTC035, iii) PTTC044, iv) PTTC58, v) PTTC59, vi) PTTC60, vii) PTTC62, viii) PTTC74, ix) PTTC114, x) Cgl13855 overexpressed under the control of GAL10 promoter; xi) blank, engineered efficient terpene precursor providing chassis S. cerevisiae YZL141. Red asterisks indicated the presence of compound 3, white asterisks indicate the similar peak produced by PTTC074.