A novel terpene synthase controls differences in anti-aphrodisiac pheromone production between closely related Heliconius butterflies

Abstract

Plants and insects often use the same compounds for chemical communication, but not much is known about the genetics of convergent evolution of chemical signals. The terpene (E)-β-ocimene is a common component of floral scent and is also used by the butterfly Heliconius melpomene as an anti-aphrodisiac pheromone. While the biosynthesis of terpenes has been described in plants and microorganisms, few terpene synthases (TPSs) have been identified in insects. Here, we study the recent divergence of 2 species, H. melpomene and Heliconius cydno, which differ in the presence of (E)-β-ocimene; combining linkage mapping, gene expression, and functional analyses, we identify 2 novel TPSs. Furthermore, we demonstrate that one, HmelOS, is able to synthesise (E)-β-ocimene in vitro. We find no evidence for TPS activity in HcydOS (HmelOS ortholog of H. cydno), suggesting that the loss of (E)-β-ocimene in this species is the result of coding, not regulatory, differences. The TPS enzymes we discovered are unrelated to previously described plant and insect TPSs, demonstrating that chemical convergence has independent evolutionary origins.

Fig 1. Pathway of terpene biosynthesis.

(A) IPP and DMAPP are first formed from the mevalonate pathway. IPP and DMAPP are the substrates for isoprenyl disphosphate synthases (GPPS, FPPS, and GGPPS). IDSs produce isoprenyl diphosphates of varying lengths, depending on the number of IPP units added. Isoprenyl diphosphates (GPP, FPP, and GGPP) are themselves the substrates used by TPSs to make terpenes of various sizes. For example, monoterpene synthases produce monoterpenes, such as ocimene, from GPP. For illustration, (E,E)-α-farnesene is used as a representative sesquiterpene, and phytol as a diterpene. (B) Proposed biosynthetic pathway in H. melpomene. Reciprocal best BLAST hits are highlighted in bold. IDSs are in red and their products, isoprenyl diphosphates, in blue. BLAST, basic local alignment search tool; DMAPP, dimethylallyl diphosphate; FPP, farnesyl diphosphate; FPPS, farnesyl diphosphate synthase; GGPP, geranylgeranyl diphosphate; GGPPS, geranylgeranyl diphosphate synthase; GPP, geranyl diphosphate; GPPS, geranyl diphosphate synthase; IDS, isoprenyl diphosphate synthase; IPP, isopentenyl diphosphate; TPS, terpene synthase.

Fig 2. QTL and gene expression analyses to identify candidate genes for (E)-β-ocimene production.

(A) The 2 species used in the crosses, H. melpomene which produces (E)-β-ocimene and H. cydno which does not. (B) Genome-wide scan for QTL underlying (E)-β-ocimene production. (C) QTL on chromosome 6 for (E)-β-ocimene production. CIs as well as the positions of candidate genes (subunit 1 of DPPS (PDSS1) and the GGPPS cluster) in the region are marked. Black lines above x-axis represent genetic markers, and horizontal line shows genome-wide significance threshold (alpha = 0.05, LOD = 2.97). (D) HMELOS in H. melpomene shows male abdomen-biased expression (for expression of other genes, see S3 Fig). (E) HMELOS and HMEL037108g1 both show greater male-biased expression in H. melpomene than H. cydno (for expression of other genes, see S4 Fig). Full model statistics in S2 and S3 Tables. N = 5 for each boxplot. Gene expression is given in log2 of normalised counts per million (using the TMM transformation). Sequencing data used to make linkage maps are available from the ENA study PRJEB34160. RNA-seq data of H. cydno and H. melpomene heads and abdomens was obtained from GenBank BioProject PRJNA283415. Processed data and scripts are available from OSF (https://osf.io/3z9tg/). CI, confidence interval; ENA, European Nucleotide Archive; LOD, log odds ratio; QTL, quantitative trait locus; RNA-seq, RNA sequencing; TMM, trimmed mean of M values.

Fig 3. Functional characterisation of TPS activity of HmelOS and HMEL037108g1 from H. melpomene and HcydOS from H. cydno.

(A) Total ion chromatograms of enzyme products in the presence of different precursor compounds. HmelOS produces high amounts of (E)-β-ocimene in the presence of GPP, with trace amounts found in the treatment with DMAPP + IPP and none with FPP. HMEL037108g1 produces large amounts of linalool with GPP and nerolidol with FPP. HcydOS does not exhibit TPS activity, showing no difference from control treatments (see S6 Fig). 1, (E)-β-Ocimene; 2, Linalool; 3, Geraniol; 4, Nerolidol; 5, Farnesol; IS, internal standard. Abundance is scaled to the highest peak of all treatments per enzyme. For quantification of peaks, see S4–S6 Tables. (B) Confirmation of identity of (E)-β-ocimene by comparison of mass spectra of (E)-β-ocimene produced in experiments and a standard. Chromatograms and mass spectra of all standards can be found in S12 Fig. (C) Pathway of how (E)-β-ocimene and linalool are formed from GPP. Raw data are available from OSF (https://osf.io/3z9tg/). DMAPP, dimethylallyl diphosphate; FPP, farnesyl diphosphate; GPP, geranyl diphosphate; IPP, isopentenyl diphosphate; TPS, terpene synthase.

Fig 4. Phylogram of GGPPS, FPPS, and TPS proteins of animals, fungi, and plants.

The phylogeny was constructed by PhyML using LG model of amino acid evolution. Bootstrap (n = 1000) values are illustrated. The tree was rooted with the ocimene synthase of Citrus unshiu. Full species names in S14 Table. Script is available from OSF (https://osf.io/3z9tg/). FPPS, farnesyl diphosphate synthase; GGPPS, geranylgeranyl diphosphate synthase; TPS, terpene synthase.