C. chinense cultivars Habanero, No.3341, No.3582, No.3614, No.3616, No.3628, No.3639, No.3641, No.3643, No.3645, and No.3646 were used. An inbred line of Habanero was prepared by self-fertilizing six times. F₁ and F₂ progeny were obtained by crossing Habanero with No.3341. All plants were grown on the Kyoto University Experimental Farm, from March to October 2012 and 2013.

After the fruits were freeze-dried, the capsaicinoids were extracted and quantified according to the method described by Koeda et al. [16] . Capsaicinoid content was calculated as the sum of capsaicin and dihydrocapsaicin.

Pepper fruits were harvested at three stages, and the placenta was separated for RNA extraction. Total RNA was extracted and reverse transcribed according to the method described by Koeda et al. [22] . CaActin (AY572427) was used as a positive internal control for reverse transcription-polymerase chain reaction (RT-PCR). The partial Comt1 cDNA sequence was amplified using Comt1-F and Comt-R1 primer sets, and Comt2 was amplified using Comt2-F and Comt-R1 primer sets (Table 1). The full-length Comt1 cDNA sequence was amplified using Comt1-F and Comt1-R primer sets, and Comt2 was amplified using Comt2-F and Comt2-R primer sets (Table 1). A Gene Racer kit (Invitrogen, Carlsbad, CA, USA) was used to clone truncated Comt1 of No.3341. PCR was performed using KOD Plus Neo (Toyobo, Osaka, Japan). The reaction mixtures for all PCR reactions were initially denatured at 94˚C for 2 min, followed by 35 cycles at 94˚C for 30 s, 60˚C for 30 s, and 72˚C for 2 min, terminating with a 3 min extension at 72˚C. Electrophoresis using 1.0% (w/v) agarose gels was performed on the amplified PCR products. Three biological replicates of RT-PCR were performed for all treatments using independently prepared total RNA, and similar results were obtained. The full-length Comt1 and Comt2 sequences amplified by RT-PCR were cloned into the pTaq1 cloning vector (BioDynamics Laboratory, Tokyo, Japan). Nucleotide sequencing was performed in an ABI PRISM 3100 genetic analyzer with an ABI PRISM BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA).

Genomic DNA was extracted from young leaves of pepper plants using Nucleon PhytoPure (GE Healthcare,

Buckinghamshire, UK). The Comt1 genomic region was amplified using Comt1-F and Comt1-R primer sets, and Comt2 was amplified using Comt2-F and Comt2-R primer sets (Table 1). A Straight Walk kit (BEX Co., Ltd., Tokyo, Japan) was used for Comt1 genome walking. PCR, electrophoresis, and sequencing were performed as described above.

The Comt1 and Comt2 nucleic acid sequences were translated to their corresponding peptide sequences using EMBOSS Transeq (The European Bioinformatics Institute) [23] . BLAST (National Center for Biotechnology Information) [24] was used to search for similar sequences in the GenBank database. The sequences were aligned using Clustal W and a phylogenetic tree was generated using a Molecular Evolutionary Genetic Analysis software ver. 6.0 with 1000 bootstrap replicates [25] .

Comt1-F and Comt1-R primer sets and Comt2-F and Comt2-R primer sets were used for cloning the two OMTs from Habanero (C. chinense) placenta tissue. Two OMT sequences were cloned from Habanero and named Comt1 and Comt2 in this study. Both Comt1 and Comt2 had 1080 bp coding regions (CDS), and had 92% similarity in nucleic acid sequence. Comt1 had 99% similarity with Comt (AF081214.1), and Comt2 had 99% similarity with CaOMT1 (U83789.1). Sequencing the genomic regions of Comt1 and Comt2 clarified that both genes had four exons, and the CDS comprised 407, 311, 65, and 300 bp. In contrast to the exons, the sequences and the length of the introns differed between the two genes. Comt1 consisted of three introns, each comprising 241, 105, and 423 bp, whereas Comt2 comprised 1164, 251, and 440 bp. The predicted protein sequences were 359 amino acids for COMT1 and COMT2. The two protein sequences showed 88% residue identity (315/359), and showed high similarity with other OMTs, particularly with COMTs reported from other plants species (Figure 2).

A phylogenetic tree was constructed with COMT1, COMT2, and SAM methyltransferases (SAM-Mtases) of other plant species (Figure 3). Twenty-nine SAM-Mtases were clustered into three groups: group 1 comprised COMT, chalcone OMT (CHOMT), and catechol OMT (CTOMT); group 2 comprised various SAM-Mtases including myricetin OMT (MOMT), flavonoid OMT (FOMT), and isoflavone OMT (IOMT); group 3 comprised jasmonic acid carboxyl methyltransferase (JAMT), salicylic acid carboxyl methyltransferase (SAMT), and indole-3-acetate OMT (IAOMT). COMT1 and COMT2 were located close to group 1.

In the present study, 18 critical amino-acid residues of the SAM-binding domain, the catalytic domain, and the substrate binding domain were compared between 16 OMTs of groups 1 and 2 (Figure 4). Five of eight amino- acid residues in the SAM-binding domain were conserved in all OMTs, and all amino-acid residues were con-

served in group 1. The catalytic domain was also highly conserved in all OMTs, and all amino-acid residues were conserved in group 1. In contrast, only two (Phe-172 and Met-176) of seven amino-acid residues were conserved in the substrate binding domain of all OMTs, and the other five residues varied. In particular, group 2 OMT amino-acid residues were highly variable, suggesting that the difference in these amino-acid residues is critical for substrate specificity. The amino-acid residues were highly conserved in group 1 compared to group 2, and all residues in the COMTs were conserved except for Val-315 of COMT2.

The Comt1 and Comt2 expression analysis was conducted with RNA samples derived from the placenta tissue (stage 1: immature green, stage 2: mature green and stage 3: mature red) (Figure 5(a)), leaf, stem, and root by RT-PCR. The accumulation of capsaicinoids in Habanero increased drastically beginning at stage 2 (Figure 5(b)). Comt1 and Comt2 showed the same expression pattern in placenta tissue. Both genes were highly expressed during stages 1 and 2 and then decreased during stage 3 (Figure 5(c)). Although Comt1 expression was not detected in leaf, stem, or root, Comt2 was highly expressed in these tissues (Figure 5(c)).

Abraham-Juarez et al. [6] silenced Comt (AF081214.1) by VIGS, which has 99% similarity with Comt1, and showed that the accumulation of capsaicinoids decreased drastically. This result indicates that Comt1 loss-of- function mutation could induce loss of pungency in Capsicum. No.3341 (C. chinense) is a non-pungent cultivar found in our previous study that Pun1 and p-AMT could not account for its non-pungency [26] . No expression was detected when Comt1-F was designed to anneal the 5’ untranslated region (UTR) and Comt1-R was designed to anneal the 3’UTR primer set for detecting Comt1 in RNA samples derived from the placenta of No.3341 (Figure 6). As it was possible that primers could not anneal to the cDNA of No.3341 because of single nucleotide polymorphisms (SNPs) located at the annealing position, an additional four reverse primers were designed and used for PCR. Comt1 was detected by all the primer sets in Habanero (Figure 6). In contrast, only Comt-R1, which was designed at almost the middle of the coding region, could detect the Comt1 transcript in No.3341 (Figure 6). The same result was obtained when genomic PCR was conducted using the same primers (Figure 6).

The race-method was used to obtain the full sequence, because expression of a truncated Comt1 transcript was suggested in No.3341. Sequencing the full-length Comt1 cDNA of No.3341 clarified that the truncated transcript was expressed. The predicted protein sequences were 283 amino acids for COMT1 of No.3341 compared to 359 amino acids for Habanero (Figure 7). COMT1 of No.3341 differed from 262 amino acids residue compared to

Habanero, lacking almost half of the O-methyltransferase domain. The genomic region was sequenced by genome walk and clarified that Comt1 of No.3341 possessed a 2.3 kbp large deletion from the 5’ region of the fourth exon to the intergenic region (Figure 8).

The co-dominant DNA markers del-F3 and del-R4 were designed to confirm the relationship between the Comt1 deletion mutation and non-pungency in No.3341 (Table 1). The pungent and non-pungent phenotype segregated 3:1 in the F₂ population obtained by crossing Habanero with No.3341 (n = 98, chi-square = 0.340, P- value = 0.560). Contrary to our expectations, the Comt1 mutation did not co-segregate with non-pungency of No.3341 (Figure 9(a)). Thus, the Comt1 mutation was not related to non-pungency of No.3341. This result was confirmed by screening several pungent accessions harboring the same Comt1 deletion mutation (Figure 9(b)).

The CaOMT1 (U83789.1) from C. annuum [21] and Comt (AF081214.1) from C. chinense [7] are two OMTs ex- pressed in placenta tissue that have been cloned previously. Because CaOMT1 (U83789.1) and Comt (AF081214.1) were cloned by different research teams and from different Capsicum species, it is unclear whether the two transcripts are allelic or genes coded by different loci. In the present study, two transcripts were cloned from the Habanero inbred line. Comt1, cloned from Habanero, had 99% similarity with Comt (AF081214.1) and Comt2 had 99% similarity with CaOMT1 (U83789.1). In addition, the length and sequences of the Comt1 and Comt2 introns differed. Moreover, whole genome sequencing of Capsicum clarified that Comt has undergone extensive gene duplication [27] . Searching the whole genome sequence of Comt1 and Comt2 revealed that the two genes are coded in chromosome 3 with a 13 kbp distance. Thus, it can be concluded that Comt1 and Comt2 are not allelic but different genes coded by different loci.

Plant SAM-Mtases are key enzymes in phenylpropanoid, flavonoid, and many other metabolic pathways [28] . OMT is one of the SAM-Mtases and can be classified into two groups, a group including COMT and another group that includes other OMTs with various substrate specificities [19] [28] -[30] . In the present study, a phylogenetic analysis of 29 SAM-Mtases was conducted and showed that two Habanero OMTs were located closely with COMTs in group 1 of other plant species (Figure 3). These results suggest that COMT1 and COMT2 have similar substrate preferences to COMTs of other plant species.

Zubietaet al. [20] studied the precise structural basis of enzyme activity using CHOMT and IOMT of Medicago sativa (alfalfa). Plant OMTs use SAM as a methyl source, yielding S-adenosyl-L-homocysteine and the methyl ether derivatives as products. Critical residues for the SAM-binding domain and the catalytic domain were highly conserved in the 16 OMTs clustered in groups 1 and 2 of our phylogenetic tree (Figure 3 and Figure 4). In contrast, substrate binding domain diversity was observed for critical residues (Figure 4). As sequential diversity was observed particularly in the OMTs of group 2, these differences seemed to reflect substrate specificity. Six out of seven residues among the five COMTs of group 1 were conserved, suggesting that these OMTs use the same substrate. However, as amino acid 315 was a valine only in COMT2, but an isoleucine in other COMTs, further enzymological study is needed to clarify how this difference affects substrate preference.

Capsaicinoids start to accumulate in Capsicum beginning 20 days post-anthesis [11] , which coincides with the accumulation pattern observed in Habanero (Figure 5(b)). As Comt1 and Comt2 are expressed in placenta tissue where the capsaicinoids are synthesized and expressed prior to capsaicinoid accumulation, the two genes could be candidates for capsaicinoid biosynthesis (Figure 5(c)). The expression of the two genes differed in other organs. Only Comt2 was expressed in the leaf, stem, and root (Figure 5(c)), suggesting that Comt2 has broad functions in addition to capsaicinoid synthesis. Considering gene duplication of Comt [27] and our phylogenetic analysis (Figure 3), Comt1 may have been duplicated from Comt2.

Based on our phylogenetic and expression analyses, Comt1 and Comt2 could be candidates for capsaicinoid biosynthesis. Because Abraham-Juarez et al. [6] silenced Comt (AF081214.1) by VIGS, which had 99% similarity with Comt1, and showed that the accumulation of capsaicinoids decreases drastically, it was assumed that the Comt loss-of-function mutation would cause a loss of pungency in Capsicum. As transformation of Capsicum is technically difficult [31] , a natural mutant of a specific gene is valuable for predicting gene function. In the present study, non-pungent No.3341 was screened for the expression of the truncated Comt1 transcript which was caused by the large deletion mutation in the genomic region (Figure 8), lacking almost half of the OMT domain. Zubieta et al., [20] have clearly shown that histidine in CHOMT (His-278) catalytic domain and IOMT (His-257) are absolutely necessary catalytic residues. The predicted COMT1 of No.3341 lacks His-265 due to a mutation, which would diminish enzyme activity. F₂ populations obtained from crossing No.3341 and Habanero were prepared to elucidate the relationship between mutated Comt1 and capsaicinoid synthesis. Contrary to our expectations, the Comt1 mutation did not co-segregate with the non-pungency trait (Figure 9(a)). With our previous study [26] , it was shown that loss of pungency in No.3341 is controlled by a single recessive gene that is neither Pun1, p-AMT nor Comt1. For further confirmation, Comt1 mutants were screened from the Capsicumgermplasm and nine pungent accessions harboring the same Comt1 deletion mutation were screened (Figure 9(b)). Thus it was concluded that Comt1 loss-of-function mutation is insufficient for loss of pungency in Capsicum. Because Comt1 loss-of-function mutants were frequently found, those might have been selected artificially for other valuable traits caused by mutation of Comt1. Further study is needed to clarify this point.

In the present study, Comt1 and Comt2 were cloned from the Habanero pungent pepper (C. chinense). Moreover, an analysis of Comt1-mutated peppers clarified that Comt1 loss-of-function mutation was insufficient for loss of pungency. In contrast, Abraham-Juarez et al. [6] silenced Comt (AF081214.1) by VIGS, which had 99% similarity with Comt1, and showed that the accumulation of capsaicinoids decreased drastically. Expression of Comt2 in placenta tissue must be considered to explain these contradictory findings. In the study of Abraham Juarez et al. [6] , Comt(Comt1)-silenced plants had undetectable levels of Comt mRNA or a reduction of an 8 - 24 folds by northern blotting and RT-PCR. Our results showed that the Comt1 and Comt2 sequence similarity was 92% in the coding regions. Thus, even Comt1 was specifically silenced by VIGS, Comt2 would still be detected by the DNA probes or primers designed to anneal coding regions of Comt1. Two possibilities should be discussed to explain these results. When two or more genes have high sequence similarity, a specific sequence of siRNA induces simultaneous silencing such as in petunia, Trenia plants, yellow beans (Glycine max L.), and dahlia [32] -[37] . Silencing of Comt1 by VIGS may have simultaneously silenced Comt2; thus, leading to decrease capsaicinoid accumulation. The second possibility is that serrano chili pepper (C. annuum cv. Tampiqueño 74) used in the study of Abraham Juarez et al. [6] had a Comt2 expression defect, such as the deletion of the promoter region. If Comt1 was silenced in such a pepper plant, both Comt1 and Comt2 would not function and capsaicinoid accumulation would decrease drastically. In the present study, 280 pepper germplasms were screened for a Comt2 loss-of-function mutant, which will be valuable for asking the relation of Comts to capsaicinoid synthesis, using high-throughput sequencing technology, but no mutant was found (data not shown). It can be concluded that the Comt1 mutation is insufficient for the loss of pungency in Capsicum. Participation of COMT in phenylpropanoid-mediated capsaicinoid biosynthesis has long been proposed [5] [7] [18] [19] . Mozourek et al. [9] proposed the participation of caffeoyl-CoA O-methyltransferase (CCoAOMT) instead of COMT in the phenylpropanoid pathway of capsaicinoid biosynthesis. Thus, further study is needed to clarify which genes are critically important for capsaicinoid biosynthesis and could be a target for controlling fruit pungency in Capsicum.