Sensory-Directed Genetic and Biochemical Characterization of Volatile Terpene Production in Kiwifruit

Abstract

Terpene volatiles are found in many important fruit crops, but their relationship to flavor is poorly understood. Here, we demonstrate using sensory descriptive and discriminant analysis that 1,8-cineole contributes a key floral/eucalyptus note to the aroma of ripe 'Hort16A' kiwifruit (Actinidia chinensis). Two quantitative trait loci (QTLs) for 1,8-cineole production were identified on linkage groups 27 and 29a in a segregating A. chinensis population, with the QTL on LG29a colocating with a complex cluster of putative terpene synthase (TPS)-encoding genes. Transient expression in Nicotiana benthamiana and analysis of recombinant proteins expressed in Escherichia coli showed four genes in the cluster (AcTPS1a-AcTPS1d) encoded functional TPS enzymes, which produced predominantly sabinene, 1,8-cineole, geraniol, and springene, respectively. The terpene profile produced by AcTPS1b closely resembled the terpenes detected in red-fleshed A chinensis AcTPS1b expression correlated with 1,8-cineole content in developing/ripening fruit and also showed a positive correlation with 1,8-cineole content in the mapping population, indicating the basis for segregation is an expression QTL. Transient overexpression of AcTPS1b in Actinidia eriantha fruit confirmed this gene produced 1,8-cineole in Actinidia Structure-function analysis showed AcTPS1a and AcTPS1b are natural variants at key TPS catalytic site residues previously shown to change enzyme specificity in vitro. Together, our results indicate that AcTPS1b is a key gene for production of the signature flavor terpene 1,8-cineole in ripe kiwifruit. Using a sensory-directed strategy for compound identification provides a rational approach for applying marker-aided selection to improving flavor in kiwifruit as well as other fruits.

Figure 1.

Genetic mapping of 1,8-cineole and analysis of genes at the complex TPS locus on LG29a. A, cv Red5 LG27 and LG29a QTL regions for 1,8-cineole indicating the 95% (light gray) and 99% (dark gray) confidence intervals. Markers with the highest LOD scores on LG29a (LOD score 6.1) and LG27 (LOD score 3.3) are highlighted. B, Genomic organization at the complex TPS locus on Chr29. Gene models in gray correspond to putative TPS genes. Full descriptions and positions are given in Supplemental Table S4. C, Eighteen terpene synthase gene models identified in the cvRed5 genome (Acc numbers) and previously characterized Actinidia TPS genes were aligned using ClustalW in Geneious (v. R10) with a gap opening penalty of 15 and gap extension penalty of 0.3. Trees were generated from manually realigned sequences using the neighbor-joining method (Saitou and Nei, 1987) and visualized using MEGA6 software (Tamura et al., 2013). Percentage bootstrap values are shown based on 1000 replicates. GenBank accessions for published sequences are as follows: AcTPS1/MyS (myrcene synthase, KF319035), AaTPS1/TlS (terpinolene synthase, KF319036), AdGDS1 (germacrene D synthase AY789791), AcNES1 (nerolidol synthase, JN242243), AcLS1 (linalool synthase, GQ338153), and AdAFS1 (α-farnesene synthase, FJ265785). The four characterized TPS on Chr29 are marked with black symbols. D, Amino acid identities of ACTPS1a–AcTPS1d and published A. chinensis myrcene synthase AcTPS1 (MyS) and A. arguta terpinolene synthase AaTPS1 (TlS; Nieuwenhuizen et al., 2015).

Figure 2.

Volatile terpene production and expression of AcTPS1a–d during development and ripening in A. chinensis. Terpene volatiles were collected by SPME and analyzed by GC-MS from developing cv Red5 (A) and cv Hort22D (B) fruit flesh and cv Red5 fruit peel (C) 30 to 50 daa and at four ripening stages (H1–H4). Quantitative data for all terpenes found at levels > 1 ng·g⁻¹ is given in Supplemental Table S6. AcTPS1a–AcTPS1d expression was measured by RT-qPCR in a range of different cv Red5 tissues (D): eating ripe fruit stage H3 (Fr), main root (R), stem (S), mature leaf (L), flower (Fl), and expanding bud (B). AcTPS1a (E) and AcTPS1b (F) expression was measured in the same fruit flesh material used for GC-MS analysis. The gene-specific primers used are given in Supplemental Table S5. Data are means ± se (n = 3). Expression is relative to the fruit (Fr) sample of AcTPS1c (set as 1 in D) and to the H3 sample of AcTPS1b (set as 1 in E and F).

Figure 3.

Subcellular localization of AcTPS1a–AcTPS1d in N. benthamiana protoplasts. Translational fusion constructs of AcTPS1a–AcTPS1d to GFP were transiently expressed in N. benthamiana leaves and analyzed with a confocal laser-scanning microscope. GFP fluorescence (GFP) and chlorophyll auto-fluorescence (chlorophyll) are shown, as well as combined visible and fluorescence signals (merged) and light-microscopy images (brightfield) of the intact mesophyll protoplasts. From the top AcTPS1a-GFP targeted to the chloroplast. AcTPS1b-GFP targeted to the chloroplast. AcTPS1c-GFP targeted to the chloroplast. AcTPS1d-GFP targeted primarily to the cytoplasm. RFP shows fluorescence of red fluorescence protein from the vector (pMDC43) localized to the cytoplasm. Scale bars = 5 μm.

Figure 4.

Volatile terpenes produced by expression of AcTPS1a and AcTPS1b in vitro. Volatile terpenes produced by recombinant AcTPS1a and AcTPS1b enzymes in vitro compared to those produced in eating ripe (stage H3) A. chinensis 'Red5’ flesh (F) and peel (P), as well as in ripe A. chinensis 'Hort22D’ flesh. The recombinant enzymes were purified by Ni²⁺ affinity and gel filtration chromatography, and were incubated with GDP as substrate. Terpene volatiles were collected by SPME and analyzed by GC-MS in triplicate. Each terpene is expressed as a percentage of total volatile terpenes identified in that sample. Red, 1,8-cineole; green, sabinene; gray, other abundant terpenes. For details of all terpenes detected in each sample, see Supplemental Table S7.

Figure 5.

Volatile terpenes produced by transient expression in N. benthamiana. N. benthamiana leaves were infiltrated with Agrobacterium suspensions containing pHEX2_AcTPS1a–d or the negative control pHEX2_GUS (in combination with pHEX2-AcDXS or pEAQ-tHMGR-2A-BCCP1). Volatiles were collected by SPME and analyzed by GC-MS 7 d postinfiltration. Experiments were performed in triplicate, and a single representative trace is shown (based on the single ion m/z 93 + other typical ions for each chemical). The concentrations of all terpene volatiles measured after inoculation with pHEX2_AcTPS1a–AcTPS1d are presented in Supplemental Table S8. A, Top trace, AcTPS1a + DXS; bottom trace, GUS + DXS control. B, Top trace, AcTPS1b + DXS; bottom trace, GUS + DXS control. C, Top trace, AcTPS1c + DXS; bottom trace, GUS + DXS control. D, Top trace, AcTPS1d + HMGR; bottom trace, GUS + HMGR control.

Figure 6.

Transient expression of AcTPS1b in kiwifruit. Mature A. eriantha fruit were infiltrated with Agrobacterium suspensions containing pHEX2_AcTPS1b or the negative control pHEX2_GUS (both in combination with pHEX2-AcDXS + pBIN61-p19). Volatiles were collected by SPME and analyzed by GC-MS 7 d postinfiltration. The experiments were performed in triplicate and a single representative trace is shown. Authentic limonene and 1,8-cineole were run as standards, top trace; AcTPS1b, middle trace; GUS, lower trace.

Figure 7.

Correlation of 1,8-cineole production with AcTPS1b expression in mapping population fruit. 1,8-cineole production (squares) was measured in eating ripe fruit from 12 selected mapping population lines by SPME GC-MS. AcTPS1b expression (black bars) was measured by RT-qPCR in the same fruit material using the primers shown in Supplemental Table S5. Data are presented as mean ± se (n = 3 biological replicates).

Figure 8.

Protein structure of AcTPS1b. The homology of AcTPS1b model (A) was constructed with modeler using the 3D structure of Salvia fruticosa 1,8-cineole synthase (Kampranis et al., 2007) as template. The inset B shows the three residues (Ile-340, Asn-344, and Ser-345) that vary between AcTPS1a and AcTPS1b in the GDP binding region. The bound GDP, Mg ions (green spheres), and water molecule (red sphere) from the 1,8-cineole synthase structure were superimposed onto the model.