The Antioxidant Cyclic 3-Hydroxymelatonin Promotes the Growth and Flowering of Arabidopsis thaliana

In plants, melatonin is metabolized into several compounds, including the potent antioxidant cyclic 3-hydroxymelatonin (3-OHM). Melatonin 3-hydroxylase (M3H), a member of the 2-oxo-glutarate-dependent enzyme family, is responsible for 3-OHM biosynthesis. Although rice M3H has been cloned, its roles are unclear, and no homologs in other plant species have been characterized. Here, we cloned and characterized Arabidopsis thaliana M3H (AtM3H). The purified recombinant AtM3H exhibited Km and Vmax values of 100 μM and 20.7 nmol/min/mg protein, respectively. M3H was localized to the cytoplasm, and its expression peaked at night. Based on a 1,1-diphenyl-2-picrylhydrazyl (DPPH) assay, 3-OHM exhibited 15-fold higher antioxidant activity than melatonin. An Arabidopsis M3H knockout mutant (m3h) produced less 3-OHM than the wildtype (WT), thus reducing antioxidant activity and biomass and delaying flowering. These defects were caused by reduced expression of FLOWERING LOCUS T (FT) and gibberellin-related genes, which are responsible for flowering and growth. Exogenous 3-OHM, but not exogenous melatonin, induced FT expression. The peak of M3H expression at night matched the FT expression pattern. The WT and m3h exhibited similar responses to salt stress and pathogens. Collectively, our findings indicate that 3-OHM promotes growth and flowering in Arabidopsis.

Melatonin is ubiquitous in plants with chloroplasts, likely because the ability to synthesize melatonin may derive from cyanobacteria, the precursors of chloroplasts [23,24]. Melatonin levels in healthy plant tissues range from picograms to a few nanograms per gram of fresh weight (FW) [25] but are increased several hundred-fold by stress [26]. The low melatonin level in plants may be attributed to the absence of the first two genes in melatonin biosynthesis-tryptophan decarboxylase (TDC) and tryptamine 5-hydroxylase (T5H), especially in Arabidopsis [25]. Other factors include the low activity of the last two enzymes in melatonin biosynthesis-serotonin N-acetyltransferase (SNAT) and Nacetylserotonin O-methyltransferase (ASMT)-and rapid melatonin metabolism [25].

Measurement of In Vitro M3H Activity and Enzyme Kinetics
Enzyme reaction medium contains 10 mM Tris/HCl (pH 8.0), 1 mM ascorbate, 100 µM FeSO 4 , 160 µM α-ketoglutarate, 100 µM melatonin, and 400-µg catalase. The reaction (100-µL) was incubated in the presence of the purified recombinant AtM3H protein at 30 • C for 1 h and stopped by adding 50-µL methanol. The 10 µL reaction products were analyzed using an HPLC system equipped with a fluorescence detector (Waters, Milford, MA). We employed an Atlantis C18 column (Waters; 3.9 × 150 mm) with a methanol gradient from 30 to 45% within 12 min and isocratic elution of 45% for 30 min at a flow rate of 0.25 mL/min. 3-OHM was detected with excitation at 280 nm and emission at 348 nm. Under these conditions, 3-OHM was eluted at 19 min. For detection of 2-OHM, aliquots were subjected to HPLC with a UV detector system (Waters) with a Sunfire C18 column (4.6 × 150 mm; Waters) by isocratic elution with 35% methanol in 0.3% trifluoroacetic acid for 25 min at a flow rate of 0.25 mL/min. All measurements were conducted in triplicate. The substrate affinity (K m ) and maximum reaction rate (V max ) values for AtM3H were measured using Lineweaver-Burk plots. inhibitor of 2-ODD enzymes, was obtained from Fluka (Buchs, Switzerland). Melatonin, 3-OHM, and 2-OHM were initially dissolved in 1% methanol.

Quantification of Cyclic 3-Hydroxymelatonin (3-OHM) from Arabidopsis
To quantify the 3-OHM from Arabidopsis tissues, frozen Arabidopsis samples (100 mg) were ground to a powder in liquid nitrogen using a Tissuelyser II (Qiagen, Tokyo, Japan) and extracted with 1 mL of chloroform. The chloroform extracts were evaporated until dry and dissolved in 100 µL of 40% MeOH. These samples were subjected to HPLC as described above. Quantification of 3-OHM was performed as described previously [30].

Measurement of Radical Scavenging Activity
The radical scavenging activity of 3-OHM was measured using the 1,1-diphenyl-2-picrylhy drazyl (DPPH) method as described previously [33]. In brief, Arabidopsis leaves (0.1 g) were extracted with 1 mL methanol and spun down for 5 min at 12,000× g. The resulting supernatants of methanolic extracts (0.1 mL) or standard compounds of 3-OHM, 2-OHM, and melatonin dissolved in methanol were mixed with 0.9 mL of 0.15 mM DPPH solution at 27 • C for 20 min. The control was prepared as above without any extract. Radical scavenging activity was calculated using the following numerical formula: % radical scavenging activity = (control optical density [OD] − sample OD)/control OD) × 100.

Arabidopsis Plants and Growth Conditions
A. thaliana (L.) Heynh. Columbia-0 wild type (WT) and an M3H knockout mutant (m3h) thereof were grown in plastic pots containing a commercial horticulture substrate (coco peat, 47%; peat moss, 35%; vermiculite, 10%; zeolite, 7% (FarmHannong, Seoul, Korea): perlite (SJ Company, Ulsan, Korea); 3:1 ratio). Plants were incubated in a growth room under a 14 h light and 10 h darkness photoperiod at 23 • C with 60% humidity and a photon flux density of 50 µmol m −2 s −1 . Each pot was covered with transparent film for 4 days to prevent water loss. Plants were illuminated with fluorescent light from OSRAM (Seoul, Korea) at 6500 K (865 FPL36EX-D) and 4000 K (840 FPL 36EX-W) at a 50:50 ratio. The m3h mutant (SALK-144842) was obtained from the Arabidopsis Biological Resource Center (ABRC, Columbus, OH, USA) and had a T-DNA insertion in the second exon from the 5 -end of M3H (AT1G17020). To confirm the mutant line as homozygous, PCR screening was conducted using the genomic DNA of m3h and gene-specific primers (forward, 5 -ATG GAA GCA AAA GGG GCA GC-3 ; reverse, 5 -GAT TCT CAA AGC ATC TAG AT-3 ).

Determination of Flowering Time
Flowering times were defined based on the days to bolting and the total number of rosette leaves after the main stem had bolted by 1 cm. For each replicate, the flowering time and number of rosette leaves were recorded per genotype. Leaf area was measured using Fuji ImageJ software [34]. Rosette leaves were photographed adjacent to a ruler; pixels were converted into metric units. Data were processed using Microsoft Excel 2010 (Microsoft Corp., Redmond, WA, USA) and are the means of three replicates from three independent plants.

Subcellular Localization Analysis of Arabidopsis thaliana Melatonin 3-Hydroxylase (AtM3H)
The pER8-mCherry for transgene localization analysis and pBIN61-GFP-HA (P35s:GFP-HA) plasmid for the cytoplasmic fluorescence marker were kindly provided by Dr. HG Kang (Texas state university, San Marcos, TX, USA). As for the subcellular localization study of AtM3H, a full-length AtM3H cDNA was amplified with PCR using a primer set containing an AscI site (forward, 5 -GGC GCG CCA TGG AAG CAA AAG GGG CAGC-3 , reverse, 5 -GGC GCG CCG GAT TCT CAA AGC ATC TAG-3 ). The amplified PCR product was first introduced into the T&A vector (RBC Bioscience, New Taipei City, Taiwan), from which the AscI insert of AtM3H was prepared and ligated into the AscI site of the binary vector pER8-mCherry to generate pER8-AtM3H-mCherry. The resulting plasmids were transformed into the Agrobacterium Tumefaciens strain GV2260 using the freeze-thaw method. Leaves from four-week-old tobacco (Nicotiana benthamiana) plants, native Australian species, were infiltrated with Agrobacterium strain (about 0.2 OD), followed by infiltration with 10 µM β-estradiol, and then incubated for 24 h in a growth room. Images were obtained using a Leica TCS-SP5 confocal microscope (Leica, Wetzlar, Germany) using a ×63 oil-immersion objective. Images were processed using the Leica LAS-AF software version 1.8.2 (Leica, Wetzlar, Germany).

RNA Extraction and Gene Expression Analysis
For various chemical-induced gene expression analyses, leaves of five-week-old Arabidopsis plants were syringe infiltrated with 4 µM of various chemicals in buffer containing 2 mM MgCl 2 in 5 mM MES of pH 5.7 under regular dark-light conditions, and leaf samples were collected 2 and 4 h after infiltration. A control solution was 0.001% methanol in 2 mM MgCl 2 in 5 mM MES (pH 5.7). Total RNA was extracted using a Nucleospin RNA Plant Kit (Macherey-Nagel, Duren, Germany). The RNA (800 ng) was used to prepare cDNA using the RNA to cDNA EcoDry TM premix system (TaKaRa Biotechniques, Shiga, Japan). The cDNA was diluted 2-fold, and 1-uL was used as a template in each reverse transcriptionpolymerase chain reaction (RT-PCR) analysis. The PCR reaction was conducted under the following conditions, with initial denaturation at 95

Bacterial Growth Assay
The bacterial strain used in this study was virulent Pseudomonas syringae pv. Tomato DC3000 (Pst DC3000) strain and the avirulent avrRpm1-containing Pst DC3000 strain (Pst-avrRpm1) were obtained from Dr. CS Oh (Kyunghee University, Yongin, Korea). Bacterial culture was grown in King's medium B in the presence of 50 µg rifampicin and 50 µg kanamycin per milliliter. Five-to six-week-old mature plants were spray inoculated with bacterial solutions containing 5 × 10 6 CFU/mL or 1 × 10 8 CFU/mL (in 0.01% [v/v] Silwet L-77) for Pst DC3000 or Pst-avrRpm1, respectively. Bacterial titers were assayed at 5 d after spray inoculation. For each measurement, extracts from 7 leaf discs (from three plants) were serially diluted 10 times, then 10 µL of each dilution was plated on King's medium B agar plates containing both 50 µg rifampicin and 50 µg kanamycin per milliliter. Each data point represents three replicates.

Salt Tolerance Assay
To determine tolerance to salt stress, at least 30 seeds of the WT and m3h per experiment were sown on MS, without or with 100 mM NaCl. To evaluate primary root growth under salt stress, seeds on MS medium were incubated vertically in a growth room for 3 weeks. Primary roots were photographed, and their lengths were measured using Fuji ImageJ software [34].

Statistical Analysis
The data were analyzed by analysis of variance using IBM SPSS Statistics 25 software (IBM Corp. Armonk, NY, USA). Means with different letters or asterisks indicate significantly different values at p < 0.05 according to a post-hoc Tukey's honest significant difference (HSD) test. All data are presented as mean ± standard deviation.

Sequence Analysis of Arabidopsis and Rice M3H
A BLAST search [35] using rice (Oryza sativa) M3H (AK067086) [30] as a query led to the identification of one M3H-like gene in Arabidopsis thaliana (AtM3H; AT1g17020). AtM3H encoded a protein of 358 amino acid residues, and the AtM3H polypeptide shared a 40% amino acid identity with rice M3H (Figure 1). Similar to rice M3H, neither chloroplast nor mitochondria transit peptide was present in AtM3H according to TargetP analysis [36], suggesting a cytoplasmic protein. Notably, the 2-ODD conserved domains of the OsM3H and AtM3H polypeptides showed 60% homology. Arabidopsis AtM3H is the sole homolog of rice M3H.
weeks. Primary roots were photographed, and their lengths were measured using F ImageJ software [34].

Statistical Analysis
The data were analyzed by analysis of variance using IBM SPSS Statistics 25 softwa (IBM Corp. Armonk, NY, USA). Means with different letters or asterisks indicate signi cantly different values at p < 0.05 according to a post-hoc Tukey's honest significant d ference (HSD) test. All data are presented as mean ± standard deviation.

Sequence Analysis of Arabidopsis and Rice M3H
A BLAST search [35] using rice (Oryza sativa) M3H (AK067086) [30] as a query led the identification of one M3H-like gene in Arabidopsis thaliana (AtM3H; AT1g1702 AtM3H encoded a protein of 358 amino acid residues, and the AtM3H polypeptide shar a 40% amino acid identity with rice M3H (Figure 1). Similar to rice M3H, neither chlor plast nor mitochondria transit peptide was present in AtM3H according to TargetP an ysis [36], suggesting a cytoplasmic protein. Notably, the 2-ODD conserved domains of t OsM3H and AtM3H polypeptides showed 60% homology. Arabidopsis AtM3H is the so homolog of rice M3H.

M3H Activity of Recombinant AtM3H
Due to the low homology (40%) between OsM3H and AtM3H, we examined wheth AtM3H has M3H activity. Full-length AtM3H was expressed in-frame with the codi

M3H Activity of Recombinant AtM3H
Due to the low homology (40%) between OsM3H and AtM3H, we examined whether AtM3H has M3H activity. Full-length AtM3H was expressed in-frame with the coding sequence of the amino-terminal hexahistidine in Escherichia coli and induced by IPTG (Figure 2A). Soluble recombinant AtM3H was purified using a nickel affinity column. The recombinant proteins had 98% homogeneity, as determined by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie staining. AtM3H exhibited M3H activity of 125 pkat/mg protein but no M2H activity ( Figure 2B). By contrast, rice OsM3H showed M3H and M2H activities, although its M2H activity was low [30]. Chromatograms of 2-OHM and 3-OHM authentic compounds, and AtM3H reaction products, are shown in Figure 3. recombinant proteins had 98% homogeneity, as determined by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie staining. AtM3H exhibited M3H activity of 125 pkat/mg protein but no M2H activity ( Figure 2B). By contrast, rice OsM3H showed M3H and M2H activities, although its M2H activity was low [30]. Chromatograms of 2-OHM and 3-OHM authentic compounds, and AtM3H reaction products, are shown in Figure 3.   recombinant proteins had 98% homogeneity, as determined by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie staining. AtM3H exhibited M3H activity of 125 pkat/mg protein but no M2H activity ( Figure 2B). By contrast, rice OsM3H showed M3H and M2H activities, although its M2H activity was low [30]. Chromatograms of 2-OHM and 3-OHM authentic compounds, and AtM3H reaction products, are shown in Figure 3.

Characteristics of Recombinant AtM3H
The optimum pH of M3H was pH 7.3, but M3H activity was high at pH 6.8-8.8 (Figure 4). The optimum temperature was 30 • C, followed by 37 • C and 25 • C; activity was absent at 70 • C. The optimum pH and temperature of AtM3H are similar to those of rice OsM3H, which also exhibited the optimum pH (7.4-8.0) and temperature (30 • C) [27,30]. AtM3H required α-ketoglutarate as a cosubstrate and several cofactors (ascorbate, Fe 2+ , and catalase); the absence of these factors markedly reduced or abolished AtM3H activity (Figure 4). At 100 and 500 µM, the 2-ODD inhibitor prohexadione-Ca decreased AtM3H activity by 10% and 50%, respectively. By contrast, 10 µM prohexadione-Ca reduced OsM2H Antioxidants 2022, 11, 1157 7 of 13 activity by 90% [27]. Therefore, AtM3H was more tolerant of prohexadione-Ca than M2H ( Figure 4D). Using the above optimum conditions, the K m and V max values of AtM3H were 100 µM and 20.7 nmol/min mg protein, respectively ( Figure 5). The K m of AtM3H was sevenfold lower than that of OsM3H, whereas the V max was 10-fold lower. The catalytic efficiency (V max /K m ) of AtM3H was 1.3-fold lower than that of OsM3H. Therefore, AtM3H is a member of the 2-ODD superfamily and has a catalytic efficiency similar to rice OsM3H. absent at 70 °C. The optimum pH and temperature of AtM3H are similar to those of rice OsM3H, which also exhibited the optimum pH (7.4-8.0) and temperature (30 °C) [27,30]. AtM3H required α-ketoglutarate as a cosubstrate and several cofactors (ascorbate, Fe 2+ , and catalase); the absence of these factors markedly reduced or abolished AtM3H activity (Figure 4). At 100 and 500 μM, the 2-ODD inhibitor prohexadione-Ca decreased AtM3H activity by 10% and 50%, respectively. By contrast, 10 μM prohexadione-Ca reduced OsM2H activity by 90% [27]. Therefore, AtM3H was more tolerant of prohexadione-Ca than M2H ( Figure 4D). Using the above optimum conditions, the Km and Vmax values of AtM3H were 100 μM and 20.7 nmol/min mg protein, respectively ( Figure 5). The Km of AtM3H was sevenfold lower than that of OsM3H, whereas the Vmax was 10-fold lower. The catalytic efficiency (Vmax/Km) of AtM3H was 1.3-fold lower than that of OsM3H. Therefore, AtM3H is a member of the 2-ODD superfamily and has a catalytic efficiency similar to rice OsM3H.  absent at 70 °C. The optimum pH and temperature of AtM3H are similar to those of rice OsM3H, which also exhibited the optimum pH (7.4-8.0) and temperature (30 °C) [27,30]. AtM3H required α-ketoglutarate as a cosubstrate and several cofactors (ascorbate, Fe 2+ , and catalase); the absence of these factors markedly reduced or abolished AtM3H activity (Figure 4). At 100 and 500 μM, the 2-ODD inhibitor prohexadione-Ca decreased AtM3H activity by 10% and 50%, respectively. By contrast, 10 μM prohexadione-Ca reduced OsM2H activity by 90% [27]. Therefore, AtM3H was more tolerant of prohexadione-Ca than M2H ( Figure 4D). Using the above optimum conditions, the Km and Vmax values of AtM3H were 100 μM and 20.7 nmol/min mg protein, respectively ( Figure 5). The Km of AtM3H was sevenfold lower than that of OsM3H, whereas the Vmax was 10-fold lower. The catalytic efficiency (Vmax/Km) of AtM3H was 1.3-fold lower than that of OsM3H. Therefore, AtM3H is a member of the 2-ODD superfamily and has a catalytic efficiency similar to rice OsM3H.

Subcellular Localization of AtM3H
TargetP analysis predicted that AtM3H is localized to the cytoplasm due to the absence of transit sequences. AtM3H-mCherry fusion protein was transiently induced in tobacco leaves and observed by confocal microscopy. The red fluorescence of AtM3H-mCherry colocalized with the green fluorescence of green fluorescent protein (GFP) in the cytoplasm, suggesting that AtM3H is localized to the cytoplasm as is rice M3H (Figure 6).

Subcellular Localization of AtM3H
TargetP analysis predicted that AtM3H is localized to the cytoplasm due to the absence of transit sequences. AtM3H-mCherry fusion protein was transiently induced in tobacco leaves and observed by confocal microscopy. The red fluorescence of AtM3H-mCherry colocalized with the green fluorescence of green fluorescent protein (GFP) in the cytoplasm, suggesting that AtM3H is localized to the cytoplasm as is rice M3H (Figure 6).

The Knockout Mutant of M3H (m3h) had Low Antioxidant Activity
3-OHM exerts an antioxidant effect by scavenging hydroxyl radical (·OH) and peroxyl radical (·OOH) [37,38]. A 1,1-diphenyl-2-picrylhydrazyl (DPPH) assay showed (Figure 7A) that 3-OHM exhibited 15-fold higher activity than melatonin; 2-OHM showed the lowest antioxidant activity, as reported previously [39]. m3h did not express M3H over 24 h, whereas the WT exhibited peak M3H expression at 10 pm (4 h after the light-to-dark transition) and declined thereafter to the level observed under light conditions ( Figure  7B,C). Therefore, the Arabidopsis M3H transcript level shows a diurnal rhythm that peaks at night, consistent with rice M3H [31]. Due to the lack of M3H expression, m3h produced less 3-OHM than the WT ( Figure 7D), resulting in lower leaf antioxidant activity ( Figure  7E). The moderate 3-OHM level in m3h suggests the presence of several M3H-like genes in Arabidopsis, as in rice [30].

The Knockout Mutant of M3H (m3h) Had Low Antioxidant Activity
3-OHM exerts an antioxidant effect by scavenging hydroxyl radical (·OH) and peroxyl radical (·OOH) [37,38]. A 1,1-diphenyl-2-picrylhydrazyl (DPPH) assay showed ( Figure 7A) that 3-OHM exhibited 15-fold higher activity than melatonin; 2-OHM showed the lowest antioxidant activity, as reported previously [39]. m3h did not express M3H over 24 h, whereas the WT exhibited peak M3H expression at 10 pm (4 h after the light-to-dark transition) and declined thereafter to the level observed under light conditions ( Figure 7B,C). Therefore, the Arabidopsis M3H transcript level shows a diurnal rhythm that peaks at night, consistent with rice M3H [31]. Due to the lack of M3H expression, m3h produced less 3-OHM than the WT ( Figure 7D), resulting in lower leaf antioxidant activity ( Figure 7E). The moderate 3-OHM level in m3h suggests the presence of several M3H-like genes in Arabidopsis, as in rice [30].

Subcellular Localization of AtM3H
TargetP analysis predicted that AtM3H is localized to the cytoplasm due to the a sence of transit sequences. AtM3H-mCherry fusion protein was transiently induced tobacco leaves and observed by confocal microscopy. The red fluorescence of AtM3H mCherry colocalized with the green fluorescence of green fluorescent protein (GFP) in th cytoplasm, suggesting that AtM3H is localized to the cytoplasm as is rice M3H (Figure 6

The m3h Had Delayed Flowering and Retarded Growth
We next investigated the growth and development of WT and m3h plants. m3h plants were smaller than the WT and possessed more rosette leaves, indicating delayed flowering (Figure 8). In addition, the total biomass per plant of m3h was lower than the WT. The stunted growth was caused by decreased expression of the gibberellin (GA) biosynthesisrelated genes ent-kaurene synthase (KS), GA3 oxidase1 (GA3ox1), and GA3 oxidase2 (GA3ox2) (Figure 9). The expression of the GA-responsive positive transcription factor Myb33 was also decreased. Regarding delayed flowering, the expression of the key flowering inducer, FLOWERING LOCUS T (FT), was decreased in m3h compared to the WT. Unlike KS, FT expression was rapidly and significantly induced by exogenous 3-OHM, and to a lesser degree by melatonin. Therefore, 3-OHM-mediated suppression of FT expression contributes to the delayed flowering in m3h, as does decreased GA synthesis. tonin (100 μM). (E) Antioxidant activities of lower leaves of 30-day-old WT and m3h, as revealed by DPPH assay. Error bars are standard deviations of three biological replicates. Different letters and asterisks (*) indicate significant differences (Tukey's post hoc HSD test; p < 0.05). RSA, radical scavenging activity.

The m3h had Delayed Flowering and Retarded Growth
We next investigated the growth and development of WT and m3h plants. m3h plants were smaller than the WT and possessed more rosette leaves, indicating delayed flowering ( Figure 8). In addition, the total biomass per plant of m3h was lower than the WT. The stunted growth was caused by decreased expression of the gibberellin (GA) biosynthesisrelated genes ent-kaurene synthase (KS), GA3 oxidase1 (GA3ox1), and GA3 oxidase2 (GA3ox2) (Figure 9). The expression of the GA-responsive positive transcription factor Myb33 was also decreased. Regarding delayed flowering, the expression of the key flowering inducer, FLOWERING LOCUS T (FT), was decreased in m3h compared to the WT. Unlike KS, FT expression was rapidly and significantly induced by exogenous 3-OHM, and to a lesser degree by melatonin. Therefore, 3-OHM-mediated suppression of FT expression contributes to the delayed flowering in m3h, as does decreased GA synthesis.

Stress Tolerance of m3h
The m3h and WT Arabidopsis seeds were sown on an MS medium containing 100 mM NaCl for 3 weeks. There was no difference in root length between the WT and m3h, indi-

Stress Tolerance of m3h
The m3h and WT Arabidopsis seeds were sown on an MS medium containing 100 mM NaCl for 3 weeks. There was no difference in root length between the WT and m3h, indicating that a decreased 3-OHM level is not significantly associated with the response to salt stress ( Figure 10A). A virulent pathogen (Pseudomonas syringae pv. tomato DC3000) and its avirulent variant containing avrRpm1 (Pst-avrRpm1) were used to evaluate pathogen resistance. Leaves of m3h and the WT exhibited similar responses to the two pathogens ( Figure 10B,C), indicating that 3-OHM is not implicated in the plant response to pathogen infection. These data were in contrast with those of the snat1 knockout mutant, which showed the increased susceptibility to the avirulent pathogen of Pst-avrRpm1 [40], indicative of the different role of 3-OHM compared to melatonin concerning pathogen response. levels were analyzed by RT-PCR using cDNA from leaves of 30-day-old mature plants befo bolting stage. Plants were sampled at the indicated times. EF-1α was used as the loading c Numbers in parentheses are numbers of PCR cycles. Effector solutions were infiltrated into axial leaf side at 8 pm and incubated for 2 and 4 h in darkness. Accession numbers: FT, AT1G KS, AT1g15550; GA3ox1, AT1g80340; GA3ox2, AT5g06100; MYB33, AT5g60390; and AT5g60390.

Stress Tolerance of m3h
The m3h and WT Arabidopsis seeds were sown on an MS medium containing 10 NaCl for 3 weeks. There was no difference in root length between the WT and m3h cating that a decreased 3-OHM level is not significantly associated with the respo salt stress ( Figure 10A). A virulent pathogen (Pseudomonas syringae pv. tomato DC and its avirulent variant containing avrRpm1 (Pst-avrRpm1) were used to evaluate ogen resistance. Leaves of m3h and the WT exhibited similar responses to the two p gens ( Figure 10B,C), indicating that 3-OHM is not implicated in the plant response to ogen infection. These data were in contrast with those of the snat1 knockout mutant, showed the increased susceptibility to the avirulent pathogen of Pst-avrRpm1 [40], ative of the different role of 3-OHM compared to melatonin concerning pathog sponse.

Discussion
In animals, melatonin is rapidly metabolized into various metabolites, including 6-hydroxymelatonin, by cytochrome P450 enzymes, followed by 6-sulfatoxymelatonin for urinary excretion [41]. Other animal melatonin metabolites include 3-OHM, 2-OHM, 4-hydroxymelatonin, N 1 -acetyl-N 2 -formyl-5-methoxykynuramine (AFMK), and N 1 -acetyl-5-methoxykynuramine (AMK). The first reported melatonin metabolite in plants was AFMK, the level of which in water hyacinth peaked during the late light phase; AFMK may be useful for phytoremediation of toxic chemicals or pollutants [42]. The first M2H gene in plant melatonin metabolism from the 2-ODD family was cloned from the rice genome [27]. Compared to SNAT, M2H has a lower K m and higher V max (145 µM and 18.5 pkat/mg protein, respectively). Consequently, the 2-OHM levels are higher than melatonin in numerous plant species [43,44]. M3H, also a member of the 2-ODD family, was cloned from the rice genome [30]. The catalytic efficiency (V max /K m ) of M3H was higher than that of M2H, and the 3-OHM level was fourfold higher than the 2-OHM level in rice leaves challenged with 1 mM melatonin for 24 h. These data suggest that, in plants, melatonin is rapidly metabolized to 3-OHM followed by 2-OHM. Also, melatonin triggered AFMK and AMK production in rice seedlings. Surprisingly, neither 6-OHM nor 4-OHM was detected in rice seedlings, suggesting the difference of melatonin catabolic pathway between plants and animals [30].
2-OHM conferred resistance to cold-plus-drought stress on rice, tobacco, tomato, and cucumber [28,45]. Also, 2-OHM mediates the response to cadmium of cucumber by increasing the levels of antioxidants such as glutathione, and antioxidant enzymes such as superoxide dismutase [29,46]. 2-OHM induced reactive oxygen species (ROS) production, thereby accelerating leaf senescence [47] and promoting seed germination [48]. Conversely, M2H suppression in rice by RNAi enhanced melatonin production and salt tolerance, indicating that melatonin catabolism regulates the endogenous melatonin level in rice [49]. By contrast, the functions of 3-OHM in plants are unclear, although it reportedly scavenges radicals [37] and peroxyl radicals [38].
We report for the first time that 3-OHM has different functions than melatonin and 2-OHM in Arabidopsis. The lower K m value of AtM3H than rice M3H reflected the lower melatonin level in Arabidopsis than in rice. The catalytic efficiency (V max /K m ) of AtM3H was comparable to that of rice M3H. Suppression of rice M3H expression did not significantly affect growth, but decreased biomass and grain yield, thus implicating 3-OHM in growth and reproduction in rice [31]. The Arabidopsis m3h knockout mutant showed similar responses to salt and pathogen stresses compared to the WT, but exhibited lower antioxidant activity, delayed flowering, and stunted growth. A vitamin C-deficient Arabidopsis mutant showed early flowering [50], and 3-OHM deficiency delayed flowering in m3h. Delayed flowering was caused by the suppression of FT expression because 3-OHM is positively associated with FT expression. Although ascorbic acid and 3-OHM are antioxidants, they have different biological functions in flowering. Further studies of melatonin and its metabolites (e.g., 2-OHM and 3-OHM) will shed light on their biological functions and interactions in plants [7,25,46,[51][52][53][54].

Conclusions
3-OHM is a potent antioxidant in vitro, but its physiological function in plants was previously unclear. In this study, Arabidopsis M3H had M3H activity in vitro. Furthermore, an Arabidopsis M3H knockout mutant (m3h) exhibited lower 3-OHM production than the WT, reducing leaf antioxidant activity. The m3h mutant also showed delayed flowering and less biomass than the WT. We report for the first time that 3-OHM is not only a potent antioxidant but also a signaling molecule that induces FT expression in vitro and in vivo, regulating flowering in Arabidopsis.