Coriandrum sativum, L.

Galata, Mariana, Sarker, Lukman S. & Mahmoud, Soheil S., 2014, Transcriptome profiling, and cloning and characterization of the main monoterpene synthases of Coriandrum sativum L., Phytochemistry 102, pp. 64-73 : 66-70

2.2. Essential oil analysis of C. sativum mericarps

Analysis of the EO steam distilled from C. sativum mericarps established that, as previously reported ( Bhuiyan et al., 2009; Misharina, 2001; Msaada et al., 2009a; Potter, 1996; Sriti et al., 2009), the major component of this EO is (S)-linalool (1). After (S)-linalool (1), mericarps from the C. sativum plants used in this study were most abundant in cymene (2), ocimene (3), camphor (4) and Ƴ- terpinene (5), listed in order of abundance. Those volatile terpene products present in amounts less than 2% of total EO terpenes include limonene (6), linalool oxide (7), geraniol (8), β- phellandrene (9), sabinene (10), camphene (11), terpinene-4-ol (12), borneol (13), Oi- terpineol (14), terpinolene (15), 1,8-cineole (16)and citronellene (17) ( Table 1 , Fig. 1 ).

2.3. Transcriptome sequencing and annotations

Transcriptome sequencing yielded a total of 33,330,312 raw reads of 65 bp length each, 10,638,013 from Sample 1 ( S 1; small mericarps), 12,513,426 from Sample 2 ( S 2; medium mericarps) and 10,178,873 from Sample 3 ( S 3; large mericarps). De novo assembly yielded 65,306 transcripts of a median length of 519 bp. All unique transcript sequences were aligned against sequences in The Arabidopsis Information Resource TAIR (v.2.2.8) and UniProtKB databases, resulting in 55,689 sequences with blast hits.

Of the 65,306 transcripts, 35,928 were assigned at least one gene ontology (GO) term. Among these, 26,882 (41.16%), 19,025 (29.13%) and 32,405 (49.62%) sequences were assigned at least one GO term in the biological processes, cellular component and molecular function categories, respectively. Distribution of the most abundant GO terms for biological processes, molecular functions, and cellular components is summarized in Supplementary Figure S2 . Transcript sequences with no BLASTx hits are likely novel and involved in functions specific to C. sativum . The study of these genes may lead to uncovering evolutionary or speciesspecific processes including adaptation and speciation.

KEGG annotations categorized 1508, 760 and 442 transcripts under metabolism, genetic information processing and cellular processes, respectively. Within those transcripts grouped under metabolism, 287 corresponded to carbohydrate metabolism, 235 to amino acid metabolism, 197 to lipid metabolism (includes all genes involved in fatty acid biosynthesis/metabolism), 189 to energy metabolism, 179 to secondary products metabolism (includes all genes involved in terpenoid, phenylpropanoid, flavonoid, alkaloid and polyketide metabolism), 151 to nucleotide metabolism, 101 to cofactors/vitamins metabolism and 169 to other processes (e.g., other amino acids and glycan metabolism) (Supplementary Figure S3 and Table S2 ).

Novel full-length protein and nucleotide sequences of CsƳTRPS and CsLINS generated for this study were deposited in Genbank with the following accession numbersKF700699 and KF700700, respectively.

2.4. Terpene biosynthetic gene expression analysis

Differential transcript abundance data for each of the DXP and MVA pathway genes, as well as related prenyl transferases and other terpene biosynthetic genes, are represented in Table 2 and Supplementary Table S3 . The data was analyzed using Reads Per Kilobase per Million mapped reads (RPKM), and thus data in Table 2 and Supplementary Table S3 are represented as RPKM-normalized counts. The fold change (differential expression) is represented by log 2 ratio transformations. A negative log 2 ratio represents down-regulation, a positive value, up-regulation and a ratio of 0 is indicative of no differential expression between two samples.

Transcripts for DXS and HMGR, which are considered key regulatory enzymes in EO biosynthesis ( Munoz-Bertomeu et al., 2006; Rodriguez-Concepcion, 2010) demonstrated relatively constant levels of transcript abundance throughout mericarp development, with the exception of DXS2 which exhibited an approximately 2-fold decrease in transcript abundance from S 2 to S 3. Further, transcripts for all isoprenoid synthesis-related prenyltransferase genes were present.

In coriander, because the EO is vastly dominated by monoterpenes, rather than sesquiterpenes ( Msaada et al., 2009a), it is likely that farnesyl diphosphate synthase (FPPS) feeds the biosynthesis of large amounts of non-EO related metabolites such as triterpenes, which are also derived from farnesyl diphosphate (FPP). All plants generate triterpenes, many of which are precursors to important plant sterols and other growth regulators ( Benveniste, 2004; Clouse and Sasse, 1998). The high GGPPS transcript expression suggests that coriander mericarps also produce tetraterpenes, which are commonly found in plant mericarps as precusors to important growth regulators, photoprotective quenching compounds and as accessory pigments in the photosynthetic system ( Maluf et al., 1997).

In a study by Lane et al., 2010, it was found that Lavandula angustifolia flowers exhibit clear differential expression of the DXS and HMGR genes with DXS expressing 7-fold more than HMGR, leading to the conclusion that the flower terpene content was primarily produced via the DXP pathway. The mericarps of C. sativum exhibit a relatively constant and constitutive pattern of both DXS and HMGR gene expression, with DXS more strongly expressed (2-fold) than HMGR. This suggests that, like in L. angustifolia flowers, in C. sativum mericarps , terpene content of the EOs are primarily produced through the DXP pathway. However, differential expression between DXS and HMGR genes was not as pronounced in coriander mericarps as was the case in the lavender flowers, suggesting that the MVA pathway may also contribute to EO production in coriander. This would occur via metabolic exchange of the phorphorylated intermediates (IPP and DMAPP) between the plastid and cytosol ( Hemmerlin et al., 2003).

Upon analyzing the transcript abundance of terpene synthase genes ( Table 2 ), it was found that there were many tetraterpene biosynthetic genes actively expressed in the coriander mericarp, which correlated with the high expression of GGPPS. The majority of these genes (57%) exhibited peak expression at S 3. The two most abundantly expressed tetraterpene genes, 15-cis-phytoene desaturase and zeaxanthin epoxidase are involved in β- carotene and abscisic acid biosynthesis, respectively. Genes involved with diterpene production had substantially lower transcript abundance than tetraterpene biosynthetic genes. The two diterpene biosynthetic genes with greatest expression are directly involved in gibberellin biosynthesis, diterpene-derived plant hormones which play roles in fruit/seed senescence.

Two putative sesquiterpene synthase genes (sTPS1 and sTPS2) were identified that, according to GO annotations, encoded for enzymes responsible for the production of β- caryophyllene, Oi- humulene, and germacrene D. Both genes were KEGG annotated as part of the β- caryophyllene and Oi- humulene biosynthetic pathways. Given that coriander EO contains all three sesquiterpenes, it is likely that one or both of those genes predominantly converts FPP to more than one of those three sesquiterpenes; more work is required to conclusively establish this.

Transcript levels for the ( S)-linalool synthase gene were the most abundant of the four mTPS candidates, and peaked at S 2 (2-fold upregulation from S 1 to S 2 and 5-fold downregulation from S 2 to S 3).In addition to the high levels of the ( S)-linalool synthase transcript in coriander, another likely reason for the large amount of linalool (1) content in coriander EO by late maturity is due to accumulation of the monoterpene over time, rather than increased ( S)-linalool synthase expression at S 3.In the case of Ƴ- terpinene synthase, transcript abundace also peaked at S 2 but did not parrallel Ƴ- terpinene oil content, which dips lowest at middevelopment. This may be due to the various factors involved with post-transcriptional and post-translational regulation of EO biosynthetic genes. It is known that transcriptome sequencing data only suggest changes in the transcript abundance, and does not necessarily represent ‘‘protein levels’’ since translation to active protein may be post-transcriptionally and post-translationally regulated ( Barrett et al., 2005; Valasek and Repa, 2005).

Given that 17 monoterpenes were identified in C. sativum mericarp EO, and only 4mTPS candidate genes were identified in this study, it is likely that as with many plant TPSs, coriander terpene synthases are multiproduct enzymes that can produce several monoterpene products from a single GPP substrate.

2.5. Cs γ TRPS and CsLINS candidate gene selection

Transcripts with the GO annotation ‘‘monoterpene biosynthetic process’’, as well as transcripts with BLASTx hits to known monoTPS genes having a cutoff e-value of <10 — 60 were selected as putative coriander mTPS gene candidates. Two of these candidates, Cs γ TRPS and CsLINS, were chosen based on sequence homology to known mTPS genes, especially the presence of conserved motifs shared by all known TPS genes, DDXXD, (N,D)D(L,I,V)X( S,T)XXXE and RRX 8 W, as well as one partially conserved motif, LQLYEASFLL. Protein sequence alignments between Cs γ TRPS, CsLINS and Citrus limon Ƴ- terpinene synthase (E2E2P0.1), Lavandula angustifolia linalool synthase (ABB73045.1), Salvia fruticosa 1,8-cineole synthase(ABH07677.1), Cannabis sativa limonene synthase (ABI21837.1), L. angustifolia β- phellandrene synthase (ADQ73631.1) and Salvia officinalis sabinene synthase (AAC26018.1) were performed with ClustalW2 (Supplementary Figure S4 ). This analysis indicated that Cs γ TRPS and CsLINS contained all highly conserved motifs found in monoterpene synthases including the RRx8W motif involved in cyclization of the GPP substrate, the DDxxD and (N,D)D(L,I,V)x( S,T)xxxE motifs involved in divalent metal ion coordination, and the LQLYEASFLL motif involved in substrate binding (Supplementary Figure S4 ). Sequence alignment via the BLASTx algorithm against NCBI non-redundant protein sequences demonstrated Cs γ TRPS to share 49% conserved identity with Ƴ- terpinene synthase from Citrus unshiu (BAD27259.1), and CsLINS to share 51% conserved identity with (d)-limonene synthase from Citrus unshiu (BAD27257.1).

2.6. Bacterial expression and functional characterization of Cs γ TRPS

and CsLINS

The N-terminal signal peptide of TPSs, which is necessary for the pseudo-mature TPS to be transported to the plastid, where it becomes fully mature mono-TPS, has been found to render TPSs expressed in bacteria insoluble, thus inactive. Therefore, the signal peptides are generally eliminated from mono-TPS gene sequences during cloning work ( Vonheijne et al., 1989). The complete ORF of Cs γ TRPS and CsLINS were 1833 bp and 1773 bp, of which 186 bp and 114 bp corresponding to the putative signal peptides, were removed to improve protein solubility during expression. The truncated genes tagged with eight C-terminal histidine residues encoded a 558 and 562 amino acid proteins for CsƳTRPS and CsLINS, respectively, with predicted masses of 65.16and 65.91 kDa. The ORFsof CsyTRPS and CsLINS, excluding the transit peptides, were expressed in bacterial cells and the recombinant proteins purified and assayed for activity with GPP. Incubation of the bacterially produced CsƳTRPS with GPP (18) yielded Ƴ- terpinene (5) as major product (91.1%) in addition to a number of minor products, including sabinene (10) (6.97%), Oi- terpinene (19) (1.18%), terpinene-4-ol (12) (0.533%), and Oi- terpineol (14) (0.246%) ( Fig. 3 and Supplementary Figure S5 ). Incubation of bacterially produced CsLINS with GPP (18) yielded a single product, ( S)- linalool (1) ( Fig. 3 ). The stereoisomerism of ( S)-linalool (1) was confirmed by gas chromatography using a chiral column (Agilent, Missisauga, ON, CAN) ( Fig. 4 ). In Fig. 5C , there is a slight (R)-linalool (20) peak in addition to the ( S)-linalool (1) peak. This (R)-linalool (20) presence is likely due to solvolysis of the GPP substrate, the stock of which was made up in water. Coriander EO contains only the ( S) isomer of linalool (1), as opposed to the lavender flowers previously studied by Lane et al., 2010, which only produce the (R) isomer (20). In future, work could be done to investigate the structural reason behind the stereospecificity of these linalool synthases. Together CsƳTRPS and CsLINS describe the majority of coriander’s EO terpenoid content.

The linear kinetics ranged from 2.5 to 15 (CsƳTRPS) and 10 to 90 (CsLINS) min. The optimum pH for both CsƳTRPS and CsLINS ranged from 6.0 to 6.5, while the optimum temperature ranged from 30 to 35 ° C with a slight activity peak at 32 ° C (CsƳTRPS), 33 ° C (CsLINS). The Michaelis–Menten enzyme saturation curve was prepared for each enzyme using the hyperbolic enzyme analysis module in the SigmaPlot software (v.10.0) (Systat Software, Erkrath, Germany) ( Fig. 5A and B ). The Km, Vmax and catalytic efficiency for CsƳTRPS were calculated to be 66.25 ± 13.32 µM, 2.24 ± 0.16 pkat /mg and 2.228 × 10 — 6 s — 1 µM — 1, respectively; and these parameters for CsLINS were 2.5 ± 0.63 µM, 19.63 ± 1.05 pkat /mg and 5.40 × 10 — 4 s — 1 µM — 1, respectively. From these values, it can be seen that CsƳTRPS had a low affinity for its substrate, GPP (18), and was saturated at a low substrate concentration, while CsLINS was saturated at a substrate concentration 10-fold higher than CsƳTRPS and its affinity for GPP was much greater than was the case with CsƳTRPS. Turnover and efficiency data shown in Table 3 for CsƳTRPS and CsLINS indicate that the catalytic efficiency of CsƳTRPS is much less than observed in CsLINS. It is expected that CsƳTRPS be a slower enzyme than CsLINS because, Ƴ- terpinene (5) is only a minor component of coriander EO while ( S)-linalool (1) makes up approximately 79% of the total EO terpene content ( Table 1 ). These data do not exclude other possible regulatory elements that determine the composition of coriander EO. No enzymatic activity was detected upon incubation with FPP. The major product of CsƳTRPS, Ƴ- terpinene (5), is a precursor of the second most abundant monoterpene in the C. sativum mericarps used for this study, cymene (2) (6.38% of total EO terpene content) ( Poulose and Croteau, 1978). Oxygenation of Ƴ- terpinene (5) carried out by cytochrome p450 mono-oxygenases is likely responsible for the conversion of Ƴ- terpinene (5) to cymene (2) in these mericarps, as has been shown to be the case in other plants ( Lupien et al., 1999). Camphor (4), the fourth most abundant monoterpene in the mericarps used here (3.62% total EO terpene content), is the product of a borneol dehydrogenase, converting borneol (13) to camphor (4), rather than the action of a monoterpene synthase ( Okamoto et al., 2011; Sarker et al., 2012). Thus, the two monoterpene synthases CsLINS and CsƳTRPS are together responsible for the biosynthesis of the most abundant EO constituents (( S)-linalool (1), Ƴ- terpinene (5), and cymene (2 )) and produce the bulk of the EO in the C. sativum mericap.

3. Conclusions

C. sativum is both an important culinary herb and EO crop. Coriander EO has been shown to exhibit medicinal activity, for example as an anti-hyperlipedemic and an anxiolytic ( Dhanapakiam et al., 2008; Mahendra and Bisht, 2011). Further, coriander mericarps contain large quantities of the monounsaturated fatty acid, petroselinic acid which is useful in the production of detergents and nylon polymers ( Msaada et al., 2009b). Although coriander is clearly an important crop, genomic resources for this plant have not been developed. This study is the first to develop a transcript library for coriander and to clone two mTPS genes which make up the majority of the total EO monoterpene content in this plant.

In addition to facilitating gene discovery, the de novo transcriptome assembly of this non-model plant contributes to the advancement of genetics and plant breeding research for this underutilized crop. For example, the genetic composition and gene functionality information provided by this research can lead to identification of molecular markers such as STRs and SNPs ( Li et al., 2012, 2013). Also, future development of molecular markers will allow this specialty oil crop plant to be industrially improved via markerassisted selective breeding. Finally, these results can be used to improve oil yield and quality of coriander through metabolic engineering.

Figures

Fig. 1. Structures for volatile terpenes found in C. sativum mericarps use in this study.

Fig. 2. Scanning electron microscope images of C. sativum mericarp tissue. (A) Whole mericarp, (B) cross-section with four vittae visible, (C) stomata on mericarp surface, (D) close-up of a vitta.

Fig. 3. Conversion of geranyl diphosphate to (S)-linalool by CsLINS, and to other monoterpene products by CsƳTRPS.

Fig. 4. Gas chromatogram using chiral column to separate (S)-linalool (1) and (R)-linalool (20). (A) (S/R)-Linalool stereoisomer blend standard (peak 1 is (R)- and peak 2 is (S)-linalool, (B) (S)-linalool standard and (C) CsLINS with geranyl diphosphate assay product.

Fig. 5. Kinetics data for CsƳTRPS (A) and CsLINS (B) with GPP substrate. Each enzymatic assay consisted of two biological and two technical replicates. CsƳTRPS and CsLINS proteins catalyzed the conversion of geranyl diphosphate to Ƴ-terpinene and (S)-linalool, respectively, with apparent Vmax and Km values of 2.24 ± 0.16 (CsƳTRPS) and 19.63 ± 1.05 (CsLINS) pkat/mg, and 66.25 ± 13 (CsƳTRPS) and 2.5 ± 0.6 (CsLINS) µM, respectively.