O-methyltransferases (OMT) are important enzymes that are responsible for the synthesis of many small molecules, which include lignin monomers, flavonoids, alkaloids, and aroma compounds. One such compound is guaiacol, a small volatile molecule with a smoky aroma that contributes to tomato flavor. Little information is known about the pathway and regulation of synthesis of guaiacol. One possible route for synthesis is via catechol methylation. We identified a tomato O-methyltransferase (CTOMT1) with homology to a Nicotiana tabacum catechol OMT. CTOMT1 was cloned from Solanum lycopersicum cv. M82 and expressed in Escherichia coli. Recombinant CTOMT1 enzyme preferentially methylated catechol, producing guaiacol. To validate the in vivo function of CTOMT1, gene expression was either decreased or increased in transgenic S. lycopersicum plants. Knockdown of CTOMT1 resulted in significantly reduced fruit guaiacol emissions. CTOMT1 overexpression resulted in slightly increased fruit guaiacol emission, which suggested that catechol availability might limit guaiacol production. To test this hypothesis, wild type (WT) and CTOMT1 that overexpress tomato pericarp discs were supplied with exogenously applied catechol. Guaiacol production increased in both WT and transgenic fruit discs, although to a much greater extent in CTOMT1 overexpressing discs. Finally, we identified S. pennellii introgression lines with increased guaiacol content and higher expression of CTOMT1. These lines also showed a trend toward lower catechol levels. Taken together, we concluded that CTOMT1 is a catechol-O-methyltransferase that produces guaiacol in tomato fruit.
Tomato (Solanum lycopersicum) is a major crop that is produced world-wide. Tomato breeding programs have focused mainly on resistance to biotic and abiotic stress, improved yield, storage quality, and color (Bai and Lindhout, 2007). The improvement of tomato flavor through breeding has proven to be a much more daunting task, as flavor is a complex trait that is comprised of a mixture of sugars, acids, and about 30 different volatiles (Tieman et al., 2006; Baldwin et al., 2008; Klee, 2010). The volatiles are synthesized from multiple precursors, which include amino acids, fatty acids, and carotenoids (Goff and Klee, 2006) and a large number of quantitative trait loci (QTLs) that affect their synthesis have been identified (Tieman et al., 2006; Mathieu et al., 2009).
Little information is known about how guaiacol is synthesized in plants. Based upon its structure, we hypothesized that it could be synthesized by methylation of catechol. This type of reaction is catalyzed by orthodiphenol-O-methyltransferases (Pellegrini et al., 1993). Orthodiphenol-O-methyltransferases are part of a class of O-methyltransferases (OMTs) that transfer a methyl group from S-adenosyl-l-methionine (AdoMet) to a hydroxyl or carboxyl group on an acceptor molecule (Noel et al., 2003). Many plant OMTs with important functions in phenylpropanoid biosynthesis have been identified. These enzymes synthesize secondary metabolites such as lignin, flavonoids and aromatic molecules (Gang, 2005). OMTs with activity on various orthodiphenolic substrates, including catechol, have been characterized previously in Fragaria × ananassa, Nicotiana tabacum and Ocimum basilicum (Collendavelloo et al., 1981; Maury et al., 1999; Gang et al., 2002; Wein et al., 2002), although a direct in vivo role for any enzyme involved in guaiacol synthesis has not been demonstrated.
Here, we identify and characterize a tomato catechol OMT (CTOMT1) responsible for synthesis of guaiacol. We demonstrate that this enzyme is able to produce guaiacol from catechol in vitro. We also show that down-regulation and overexpression of CTOMT1 in planta decrease and increase guaiacol emissions in fruit.
Identification of a catechol OMT from S. lycopersicum
Potential S. lycopersicum catechol OMT candidates were selected by identification of coding sequences with a similarity to small molecule OMTs characterized previously (Figure 1). Five candidate genes were selected, SGN-U582403, SGN-U565623, SGN-U319245, SGN-U575022, and SGN-U321686. Full-length cDNAs were synthesized from S. lycopersicum cv. M82 ripe fruit RNA. Candidate genes were cloned into pET160 plasmids for expression in E. coli. Initial screens were performed by addition of catechol directly to bacterial cultures that express recombinant protein and by measurement of guaiacol production. Only SGN-U582403 converted catechol to guaiacol (data not shown). The activity of SGN-U582403 on substrates with similar structure to catechol was measured to further test the specificity of SGN-U582403 for catechol. Recombinant enzyme purified from E. coli was incubated with [14C]-AdoMet and the following potential substrates: catechol, guaiacol, salicylic acid, benzoic acid, orcinol, caffeic acid, protocatechuic aldehyde, 2,5-dimethyl-4-methyoxy-3(2H)-furanone, and pyrogallol. SGN-U582403 had a relatively high activity with catechol as a substrate, much lower activity on protocatechuic aldehyde and a slight activity on orcinol, caffeic acid, and pyrogallol (Table 1).
Table 1. Relative activity of CTOMT1 on substrates with similar structure to catechol.
The specific activity of SGN-U582403 with catechol as substrate was also measured. Purified enzyme was incubated with excess [14C]-AdoMet and various concentrations of catechol. The enzyme was determined to have a Km of 8.36 ± 1.78 μm and Kcat value of 9.67 ± 2.42 sec−1 (Figure 2). Guaiacol was confirmed as the product by GC-MS. Based on these results, the gene that encoded the SGN-U582403 protein was renamed CTOMT1.
Characterization of CTOMT1 in planta
To further test the function of CTOMT1 in planta, a full-length CTOMT1 cDNA was cloned into pHK1001 for constitutive overexpression. The construct was transformed into S. lycopersicum cv. Flora-Dade. Seventeen independent lines were screened initially for transgene expression. Based on this screen, the four best lines were analyzed further for CTOMT1 mRNA levels (Figure 3a) and guaiacol synthesis in ripe fruits (Figure 3b). Guaiacol production was increased significantly in two of the four lines overexpressing CTOMT1. However, the increased guaiacol was not proportional to the increased RNA levels; up to a 26-fold increase in transcript resulted in only a two-fold increase in guaiacol. A similar lack of correlation between CTOMT1 transcript abundance and guaiacol production was observed throughout fruit development. Although CTOMT1 expression was the highest in immature green fruit, guaiacol production was much higher in ripe fruit (Figure 4). Although CTOMT1 is expressed in leaves or flowers, we could not detect guaiacol in these tissues.
CTOMT1 was also cloned into pK2WG7 (Karimi et al., 2002) for antisense knockdown and transformed into cv. Flora-Dade. Twenty-five lines were screened initially for knockdown of CTOMT1 RNA using leaf tissue. The four lines with greatest RNA reduction were further screened for CTOMT1 mRNA levels in ripe fruit (Figure 3c). Volatile emissions were also determined (Figure 3d). The guaiacol levels were reduced significantly in all four antisense lines, and confirmed the role of CTOMT1 in guaiacol synthesis.
Catechol feeding of fruit pericarp discs
While antisense lines produced significantly less guaiacol than controls, overexpression of CTOMT1 had much less effect on guaiacol levels. These results suggested that while CTOMT1 is essential for guaiacol synthesis, it may not be rate-limiting under normal circumstances. Rather, synthesis of catechol might limit the production of guaiacol in CTOMT1-overexpressing plants. We tested this hypothesis by feeding catechol to fruit pericarp discs of Flora-Dade (WT) and CTOMT1-overexpressing lines. Volatiles were collected after incubation for 4 h. Both WT and CTOMT1 discs produced more guaiacol when supplied with exogenous catechol. However, while WT catechol-fed discs exhibited a 36-fold increase in guaiacol synthesis, CTOMT1 discs exhibited a 52-fold increase (Table 2). These results indicated that while CTOMT1 catalyzes the conversion of catechol to guaiacol, the availability of catechol probably limits guaiacol synthesis in fruit tissue.
Table 2. Catechol feeding of tomato discs. Pericarp discs of Flora-Dade and CTOMT1 overexpressing (OE1683) ripe fruit treated with water or 1 m catechol
Guaiacol emitted (nm disc−1 h−1)
Volatiles were collected and analyzed using GC-MS.
0.04 ± 0.01
0.36 ± 0.04
0.54 ± 0.30
4.53 ± 0.83
We also observed that transgenic fruit that overexpress the salicylic hydrolase gene, NahG, produce 38-fold more guaiacol than the non-transgenic cv. Ailsa Craig control (data not shown). Historically, NahG plants have been used in pathogen-response studies because much of the salicylic acid pool is converted to catechol, which accumulates (Gaffney et al., 1993; Van Wees and Glazebrook, 2003). NahG plants provide an in vivo confirmation that when catechol levels are increased guaiacol production is also increased.
A QTL associated with guaiacol production
An introgression population developed by crossing the green-fruited S. pennellii with S. lycopersicum cv. M82 (Eshed and Zamir, 1995) was used to identify a guaiacol QTL. Previous analysis of introgression lines (ILs) grown in multiple locations over five different seasons identified a guaiacol QTL in IL 10-1 (Tieman et al., 2006). Further analysis of the complete data set indicated the presence of a guaiacol QTL on the overlapping IL 10-1-1 (fdr = 0.000220175) near the top of chromosome 10. Using the S. lycopersicum genome sequence database (http://solgenomics.net/), a cleaved amplified polymorphic sequence (CAPS) marker was developed (details in Experimental procedures) to map the position of CTOMT1 (Figure S1). The gene was located within the S. pennellii segment of IL 10-1 but not 10-1-1. IL 10-1 contains approximately 60 milion base pairs of S. pennellii DNA. IL 10-1-1 is a much smaller segment contained within IL10-1 and adjacent to the position of CTOMT1. To further elaborate the nature of the QTL, guaiacol was collected from ripe IL10-1, IL10-1-1, and M82 fruits (Figure 5a). Both IL 10-1 and IL 10-1-1 had elevated guaiacol levels relative to M82. However, IL 10-1-1 produced significantly more guaiacol than IL 10-1. Both ILs also produced significantly more methylsalicylate in addition (Figure S2).
In order to determine the underlying cause for the increased guaiacol production in these ILs, the orthologous S. pennellii CTOMT1 was cloned and sequenced. Differences in the promoter and the coding sequences were found (Figures S3 and S4). However, when the specific activity of recombinant S. pennellii CTOMT1 was tested there was no significant difference the specific activity between the S. pennellii and the M82 enzymes (Figure 2). CTOMT1 mRNA levels were also measured from ripe IL10-1, IL10-1-1, and M82 fruits (Figure 5b). RNA expression analysis indicated that CTOMT1 RNA levels were increased significantly in both ILs.
Quantification of catechol and salicylic acid
In order to understand better how the catechol synthesis pathway is affected in ILs 10-1 and 10-1-1, catechol and salicylic acid were quantified in ripe fruit. Catechol and salicylic acid were extracted from ground tissue and silylated for GC-MS analysis (Figure 6). Lower catechol and salicylic acid levels were observed in both ILs, although the levels were not significantly different from the M82 parent.
Although little information is known about how guaiacol is synthesized in tomato fruits, we hypothesized that guaiacol could be made by the methylation of catechol by an OMT. We identified potential candidates by screening for catechol methylation with tomato homologs of previously characterized orthodiphenol OMTs. Of the five candidate S. lycopersicum proteins that were screened for catechol methylation activity, only CTOMT1 was capable of converting catechol to guaiacol, which probably indicates that this enzyme is solely responsible for guaiacol synthesis in vivo. The closest homolog of this protein in sequence databases (81% identity) is an enzyme with in vitro catechol-OMT activity from N. tabacum (Collendavelloo et al., 1981; Pellegrini et al., 1993; Maury et al., 1999). The N. tabacum CTOMT gene is highly inducible by pathogen infection (Pellegrini et al., 1993). In vivo effects on catechol and guaiacol pools have not been reported.
The activity of CTOMT1 on catechol was confirmed by recombinant enzyme assays. The Km and Kcat values were similar to those of other characterized diphenol-O-methyltransferases (http://www.brenda-enzymes.org). While CTOMT1 preferentially methylates catechol, it does have some activity on protocatechuic aldehyde, orcinol, caffeic acid, and pyrogallol. All these molecules have a similar basic structure of a benzene ring with at least two hydroxyl groups. CTOMT1 was unable to methylate molecules that lack at least two hydroxyl groups, which indicated that the diphenol structure is important for substrate recognition. While plant OMTs usually have a high degree of selectivity, a few are promiscuous and catalyze methylation of structurally related compounds (Wein et al., 2002; Lam et al., 2007). However, CTOMT1 exhibited a strong preference for catechol over other tested diphenol compounds.
In order to confirm that CTOMT1 is a catechol OMT in vivo, its expression was increased or reduced in transgenic tomato plants. Knockdown of the endogenous gene caused guaiacol emission to be reduced significantly, which indicated that CTOMT1 is the major, if not only, enzyme responsible for guaiacol synthesis. Although there were high levels of CTOMT1 expression in overexpressing plants, there were not correspondingly large increases in guaiacol production. It is probable that catechol levels limit guaiacol production, as normal endogenous levels of catechol must be low and high levels have been shown to be toxic to plants (Van Wees and Glazebrook, 2003; Morse et al., 2007). This hypothesis was further supported by the fact that in Flora-Dade, CTOMT1 expression decreased with ripening, while guaiacol production increased.
When we tested the hypothesis that catechol might be limiting guaiacol production by supplying fruit pericarp discs with exogenous catechol, we were able to increase guaiacol emission significantly. Both WT and CTOMT1-overexpressing discs produced more guaiacol when fed with catechol. However, the increase in guaiacol synthesis was much greater in CTOMT1-overexpressing discs than in WT. These results indicated that, under certain circumstances, CTOMT1 expression influenced the rate of guaiacol synthesis but that there must be a level of control that precedes catechol.
Through screening of an introgression population we were able to identify a QTL associated with higher guaiacol production near the top of chromosome 10. CTOMT1 was mapped to the region covered by IL 10-1, with high guaiacol and high CTOMT1 expression. However, another IL, 10-1-1, adjacent to but not including the CTOMT1 gene also synthesized more guaiacol and had significantly higher CTOMT1 expression. While CTOMT1 expression is associated with both QTLs, it is not likely that CTOMT1 is a guaiacol QTL itself. This conclusion is supported by the result that overexpression of the CTOMT1 alone is not sufficient to increase guaiacol production. There must be a genetic element within the S. pennellii-derived 10-1-1 segment that directs higher guaiacol production and higher expression of CTOMT1. Recently, it has been shown that many QTLs are in trans to structural genes that encode enzymes that contribute to the phenotype (Steinhauser et al., 2011). This finding supports the hypothesis that a transcription factor, metabolite, or chromosomal rearrangement associated with IL10-1-1 is regulating expression of CTOMT1. The most probable explanation for the guaiacol increase in IL10-1 and IL10-1-1 is the existence of a trans-acting regulatory element contained in IL10-1-1. Additionally, the increased emissions of methylsalicylate, a salicylic acid derived volatile, supports the hypothesis that the entire catechol synthesis pathway is up-regulated in these ILs.
In conclusion, we have demonstrated that guaiacol is synthesized from catechol in tomato fruits by the action of CTOMT1. Expression of the CTOMT1 gene can significantly affect the levels of guaiacol synthesis. However, under some circumstances, steps leading up to catechol synthesis can limit the ability of the fruit to synthesize guaiacol. As reduced expression of CTOMT1 results in reduced guaiacol synthesis, it should be possible to obtain fruits with significantly reduced guaiacol synthesis by a variety of transgenic and non-transgenic techniques.
Phylogenetic tree of small molecule methyltransferases
Solanum lycopersicum OMT candidates were identified by a TBLASTN search of the sol genomics network Lycopersicon combined (tomato) unigene database using O. basilicum chavicol OMT and eugenol OMT amino acid sequences (Q93WU3; Q93WU2). Other similar proteins were identified by conducting a BLASTP search of the NCBI non-redundant protein sequences using candidate SlCTOMTs. Twenty-four OMT amino acid sequences were used to generate a protein alignment. The homology of OMTs was inferred using the Neighbor-Joining method. The evolutionary distances were computed using the Poisson correction method. All positions that contained gaps and missing data were eliminated. Both the protein alignment and the evolutionary analyses were conducted in MEGA5 (Tamura et al., 2011).
CTOMT1 in vitro expression and purification
SGN-U582403 (CTOMT1) was amplified by polymerase chain reaction (PCR) from S. lycopersicum and S. pennellii fruit cDNA. The products were cloned into pENTR/D/TOPO vectors and sequenced (CHUL Research Center, http://www.sequences.crchul.ulaval.ca). The coding regions were then cloned into vector pET160 by recombination and transformed into E. coli BL21-DE3 (Invitrogen, http://www.invitrogen.com) for inducible protein expression. Bacteria were precultured for 16 h at 37°C in Luria-Bertani broth that contained 50 μg ml−1 carbenicillin and the culture was used to inoculate 100 ml of the same medium. Cells were grown at 24°C to an OD600 of 0.5. Protein expression was induced by adding isopropyl-β-d-1-thiogalactopyranoside to the medium at a final concentration of 0.1 mm. Induced cultures continued growing at 25°C for 16 h.
Cells were harvested by centrifugation (10 min, 4420 g) and resuspended in 6 ml of lysis buffer [1× phosphate-buffered saline (PBS)], lysozyme, 10% v/v glycerol, and Bacterial Protease Inhibitor Cocktail [Sigma, http://www.sigmaaldrich.com/] and lysed with sonication. Protein was purified using Ni-Talon® (Clontech, http://www.clontech.com/) affinity chromatography. The column was washed with 1× PBS containing 5 mm imidazole. Imidazole concentration was increased to 150 mm in the elution buffer. Protein levels were quantified using Bradford Reagent (BioRad, http://www.bio-rad.com/). Protein was stored in 16% glycerol at –80°C.
For relative activity assays, 2.675 μg purified enzyme was assayed at 25°C in a 100 μl reaction containing 50 mm Tris–HCl, pH 7.5, 100 mm KCl, 2.8 mmβ-mercaptoethanol, 15 μm substrate, 10 mm AdoMet, 0.4 mm [methyl-14C]-AdoMet (specific activity 50.4 mCi mmol−1; Amersham). Substrates were diluted in 50% ethanol. Assays were done in triplicate, including boiled enzyme controls. After 30 min at 25°C the reactions were stopped by adding equal volumes of hexanes. The methylated substrate was extracted on a vortex mixer for 15 sec and centrifuging (5 min, 13 200 g). Next, 50 μl of the organic layer was counted for 10 min in 3 ml Ready Gel Scintillation Fluid (Beckman Coulter, http://www.beckmancoulter.com). Counts for the boiled enzyme controls were subtracted from the sample counts, and activity for catechol was normalized to 100%. For specific activity on catechol, procedures were the same as above, except for catechol concentrations. Concentrations used were: 0, 1, 5, 15, 25, and 50 μm. The conversion of catechol to guaiacol was validated by repeating the experiment using only cold AdoMet and injecting the organic layer on GC-MS.
Production of transgenic plants
The full-length open reading frame of CTOMT1 was cloned into a vector, pHK1001, that contained the constitutive FMV 35S promoter (Richins et al., 1987) followed by the nos 3′ terminator, for overexpression. S. lycopersicum cv. Flora-Dade cotyledons were transformed by Agrobacterium-mediated transformation (McCormick et al., 1986) with the kanamycin selectable marker, NPTII. Antisense constructions were made by cloning a full-length CTOMT1 into pK2WG7 (Karimi et al., 2002). Antisense constructs were made by the Plant Transformation Core Research Facility at the University of Nebraska (http://unlcms.unl.edu/biotech/plant-transformation).
Volatiles were collected from tomato fruits according to Tieman et al. (2006). Briefly, air was passed over the samples and volatiles were collected on a SuperQ Resin for 1 h. Five μl of nonyl acetate were added to each column as an internal control of column recovery. Volatiles were eluted off the column with methylene chloride and run on a GC/MS and GC for analysis as described in Tieman et al. (2006).
Tomato fruit was chopped and quickly frozen in liquid nitrogen. Samples were stored at –80°C until further use. RNA was extracted using Plant RNeasy kit (Qiagen, http://www.qiagen.com). Possible genomic DNA contamination was removed by on column DNase I treatment for 15 min at room temperature. Quantitative PCR was performed with StepOnePlus™ Real-Time PCR System using total amount of 325 ng total RNA, Taqman® 1-step kit (Applied Biosystems, http://www.appliedbiosystems.com), 500 nm forward and reverse primer. A total reaction volume of 25 μl was used. A standard curve was generated using pENTR-OMT1 ranging from 105 to 109 copies per 5 μl.
CTOMT1 expression and guaiacol quantification through Flora-Dade ripening
RNA was extracted and volatiles were collected from Flora-Dade fruit at the following stages: immature green, mature, turning, and red ripe. RNA extraction and volatile collection were performed as above described.
Tomato discs were cut from pericarp tissue of ripe Flora-Dade and CTOMT1-overexpressing fruit using a size 10 borer. One hundred discs were used for each sample treatment. Discs were placed in petri dishes and an ‘X’ was cut in the top of each with a razor blade. Next, 10 μl of either water or 1 m catechol dissolved in water were pipetted into each disc. Covers were placed on petri dishes and discs were left to incubate for 4 h. Discs were then placed in glass tubes and volatiles were extracted as described previously. Guaiacol was quantified on GC/MS using a guaiacol standard curve.
A guaiacol QTL on the overlapping ILs 10-1 and 10-1-1 was identified as previously described (Tieman et al., 2006). Guaiacol contents for each of five seasons as well as summary values for the combined seasons are available at http://ted.bti.cornell.edu/cgi-bin/TFGD/metabolite/metabolite_info.cgi?ID=M0000025. To determine the map position of CTOMT1, a marker that distinguishes between the S. lycopersicum and S. pennellii alleles was developed. The following primers were used: forward (F) ATTAATGCTTTCCTGTCGAACC and reverse (R) ACCTCCAACATCAACCAAAGTT. The product size was 3.7 kb. Amplification products were digested with DdeI (New England Biolabs, http://www.neb.com). Genomic sequence alignments of S. lycopersicum and S. pennellii were performed with ClustalW using genomic sequence data provided by Alisdair Fernie and Bjoern Usadel (personal communication). Volatiles were collected and RNA was quantified from ripe M82, IL10-1, and IL10-1-1 fruit as described above.
Catechol and salicylic acid quantification
Catechol was extracted from ripe M82, IL10-1-, and IL10-1-1 fruits (n ≥3). After grinding tissue in liquid nitrogen, 3 g were measured and catechol was extracted with 3 ml of acetonitrile. As an internal control, 500 ng of 4-nitrophenol were added to each sample. Samples were vortexed and centrifuge for 10 min at 25 000 g. The supernatant was placed in a glass vial and dried under nitrogen gas. A standard curve was also made using catechol, salicylic acid, and 4-nitrophenol using each standard at the following quantities: 0, 10 ng, 50 ng, 100 ng, 500 ng, 1000 ng, 5000 ng, 10 000 ng. Samples were resuspended in 200 μl of anhydrous acetonitrile. For derivatization, 100 μl of the resuspended sample was place in a new vial with 100 μl of N-methyl-N-trimethylsilyltrifluoroacetamide (MSTFA) with 1% trimethylchlorosilane (TMCS) (Thermo Scientific, http://www.thermoscientific.com). Reaction vials were place at 70°C for 1 h. Samples were then analyzed using GC-MS on a Agilent 5975 GC/MSD (http://www.chem.agilent.com) (He carrier gas; 0.7 ml min−1; splitless injector 250°C, injection volume 2 μl) with a Agilent DB-5 ms column [(5% phenyl)-methylpolysiloxane; 30 m long, 250 μm i.d., 1 μm film thickness]. The temperature was programmed from 100°C (4 min hold) at 8°C min−1 to 300°C. Source and quadrupole temperatures were 230°C and 150°C respectively. Ions selected for detection were as follows: catechol 136, 166, 239 (Figure S5); salicylic acid 135, 193, 267; 4-nitrophenol 150, 196, 211. Compounds were identified by retention times with standards and specific ions.
We would like to thank Dr Charles Goulet for his helpful discussions and Dawn Bies and Peter Bliss for technical assistance. Also, we thank Dr Alisdair Fernie for providing sequence information. This work was funded by National Science Foundation grant IOS-0923312 to HK and the Florida Agricultural Experiment Station.