Comprehensive transcriptional and functional analyses of melatonin synthesis genes in cassava reveal their novel role in hypersensitive-like cell death

Melatonin is a widely known hormone in animals. Since melatonin was discovered in plants, more and more studies highlight its involvement in a wide range of physiological processes including plant development and stress responses. Many advances have been made in the terms of melatonin-mediated abiotic stress resistance and innate immunity in plants, focusing on model plants such as rice and Arabidopsis. In this study, 7 melatonin synthesis genes were systematically analyzed in cassava. Quantitative real-time PCR showed that all these genes were commonly regulated by melatonin, flg22, Xanthomonas axonopodis pv manihotis (Xam) and hydrogen peroxide (H2O2). Transient expression in Nicotiana benthamiana revealed the subcellular locations and possible roles of these melatonin synthesis genes. Notably, we highlight novel roles of these genes in hypersensitive-like cell death, as confirmed by the results of several physiological parameters. Moreover, transient expression of these genes had significant effects on the transcripts of reactive oxygen species (ROS) accumulation and defense-related genes, and triggered the burst of callose depositions and papillae-associated plant defense, indicating the possible role of them in plant innate immunity. Taken together, this study reveals the comprehensive transcripts and putative roles of melatonin synthesis genes as well as melatonin in immune responses in cassava.

In 1958, N-acetyl-5-methoxytryptamine (melatonin) was first discovered in the pineal gland of cow, thereafter melatonin was widely identified in multiple animals 1 . In 1995, melatonin was identified by two research groups in plants 2,3 . So far, using gas chromatography/mass spectrometry (GC/MS) and radioimmunoassay, melatonin has been identified in more and more plant species with different levels, including multiple edible plants (banana, cucumber, apple, coffee, corn) 4 , lupin 5 , tomato [6][7][8][9] , rice 9-15 , sweet cherry 16 , Arabidopsis 17,18 , bermudagrass 19 , etc. Moreover, it was found that different plant species, plant organs, plant stages, plant location and treatments had significant effects on endogenous melatonin levels [19][20][21][22] . The wide distribution of melatonin in plants especially in popular beverages and crops makes people can daily take in melatonin from the related products, and this may be benefit to human for the well-known beneficial effects of melatonin on human health [23][24][25] .
Cassava (Manihot esculenta) can be cultivated under adverse environmental (drought and heat) and nutrient-limited conditions (low phosphorus and low nitrogen), together with its high photosynthetic efficiencies and starch enrichment, making it be considered as an energy crop and important tropical crop [72][73][74][75] . So far, only two manuscripts reported the in vivo role of melatonin in cassava 49,50 . Melatonin delays the postharvest physiological deterioration of cassava storage root, through modulation of ROS metabolism, starch metabolism, calcium signaling and mitogen-activated protein kinase (MAPK) cascades 49,50 . To extend our understanding of melatonin in cassava stress responses, it is essential to reveal the function of melatonin synthesis genes in cassava. In this study, 7 melatonin synthesis genes were cloned and functionally analyzed, especially their possible involvement in immune response. The results may help us in understanding the putative roles of these genes as well as melatonin in immune responses in cassava.
Through quantitative real-time PCR, we found that the transcript levels of 7 melatonin synthesis genes (MeTDC1, MeTDC2, MeT5H, MeSNAT, MeASMT1, MeASMT2 and MeASMT3) were significantly affected after treatment with flg22, Xanthomonas axonopodis pv manihotis (Xam), melatonin or hydrogen peroxide (H 2 O 2 ) for 1, 3 and 6 h ( Fig. 2A-D). Notably, all these gene transcripts were commonly down-regulated by exogenous melatonin treatment, but were largely up-regulated by Xam and H 2 O 2 at least in one time-point ( Fig. 2B-D). However, the transcript levels of these genes were differentially regulated by flg22 treatment (Fig. 2A). Generally,   Fig. S1). The common expression profile of 7 melatonin synthesis genes in response to flg22, Xam, melatonin or H 2 O 2 treatments, indicates the possible role of them as well as melatonin in immune response and reactive oxygen species (ROS) signaling in cassava. Thus, melatonin synthesis pathways may play some roles in cassava immune response.

Subcellular localization of melatonin synthesis genes in cassava.
To investigate the subcellular location of 7 melatonin synthesis genes in cassava, the CDS of these genes were fused in the in-frame with green fluorescent protein (GFP) reporter gene and express in tobacco (N. benthamiana) leaves. After 2 days post infiltration (dpi), we found that the fusion proteins of GFP and 7 melatonin synthesis proteins (MeTDC1, MeTDC2, MeT5H, MeSNAT, MeASMT1, MeASMT2 and MeASMT3) exhibited obvious green signals in both nucleus and cell membrane in the leaf cells (Fig. 3).

Transient expression of melatonin synthesis genes triggers hypersensitive response-like cell death.
Interestingly, we observed that the transient expression of all these genes resulted in obvious cell death and hypersentive response (HR) symptoms after infiltration in tobacco (N. benthamiana) leaves in comparison to the overexpression of GFP (control) (Fig. 4A). HR is a common type of programmed cell death (PCD) as well as a marker feature of plant immune response, which are largely related to ROS accumulation 77,78 . In this study, tobacco leaves expressing melatonin synthesis genes exhibited significantly higher levels of H 2 O 2 and superoxide radical (O 2 • − ) than those from leaves expressing GFP alone, as evidenced by DAB, NBT staining ( Fig. 4B,C) and the quantifications of endogenous H 2 O 2 and O 2 • − (Fig. 4D,E). Moreover, the electrolyte leakage (EL) of leaf discs from tobacco leaves expressing these genes was significantly higher than those from leaves expressing GFP (Fig. 4F), indicating the significant increase of ion leakage triggered by these genes transient expression.
Consistently, malondialdehyde (MDA), which is a lipid peroxidation caused by intracellular ROS accumulation, displayed high level in tobacco leaves expressing these genes than GFP (Fig. 4G). These results indicate the in vivo roles of melatonin synthesis genes in HR, immune response and underlying ROS modulation.

Modulation of melatonin synthesis genes expression regulates melatonin content.
To reveal the possible relationship between melatonin synthesis genes-regulated HR and endogenous melatonin level, we investigate the effect of these genes transient expression on melatonin level. Through quantification by ELISA, we found that tobacco leaves expressing melatonin synthesis genes exhibited significantly higher levels of endogenous melatonin than those from leaves expressing GFP alone (Fig. 5), suggesting the possible involvement of melatonin in these genes -mediated HR and immune response. Although MeT5H gene is missing at least 70 aa at the N-terminus when compared to other MeT5H homologs such as OsT5H 55 , it had significant effect on melatonin synthesis, indicating the difference of T5H in different plant species.   PR2 and PR5) (Fig. 7). As common feature of defense response, ROS accumulation and PCD are involved in immune response. These results indicate the involvement of melatonin synthesis genes in both ROS and defense signalings. Additionally, the effects of melatonin synthesis genes transient expression on callose depositions were also analyzed. As shown in Fig. 8, the tobacco leaves expressing melatonin synthesis genes exhibited significantly more callose depositions than those from leaves expressing GFP alone. This result indicates that melatonin synthesis genes might be involved in the modulation of callose-associated cell wall and papillae-associated plant defense.

Discussion
In the long period of evolution, plants have developed complicated mechanisms to survive and thrive in response to various environmental stresses and microbial pathogens 26 . Briefly, different stress signals are first perceived by membrane receptors, thereafter are activated by several secondary messengers such as calcium, abscisic acid (ABA), nitric oxide (NO) and H 2 O 2 . Then the signal transduction by secondary messengers leads to the activation of protein kinases, transcription factors, stress-responsive genes and physiological responses, eventually resulting in protective responses 18,19,26,[66][67][68][69][70] . The significant elevations of endogenous melatonin in plant early stress signaling indicated that melatonin may serve as an important early messenger in plant stress response [20][21][22][23][24][25][26][27] .
Recently, our studies together with other studies provided some clues for the molecular mechanisms of melatonin-mediated stress responses in plants 18,19,26,[62][63][64][65][66][67] . ROS burst and associated changes such as the transcripts of defense genes play important roles in plant immune response, especially in plant-pathogen interaction 77,78 . However, the involvement of melatonin in hypersensitive-like cell death and underlying ROS accumulation remain unknown. Herein, the identification and functional analysis of melatonin synthesis genes in cassava provided direct link between melatonin and immune response, as well as the underlying mechanism of these genes in programmed-like cell death and ROS accumulation. ROS is important signal molecules in signal transduction, serving as second messengers, whereas ROS overproduction under stress conditions results in serious cell damage. Interestingly, melatonin also has dual roles in regulating ROS. On one hand, stress induced melatonin relieves oxidative stress damage by decreasing excess ROS 18,19,26,[62][63][64][65][66][67] . On the other hand, melatonin induces the ROS level to activate the downstream responses in the early stress response. Thus, the dual roles of melatonin in regulating ROS further indicate the protective effect of melatonin in various stress response.
In this study, we successfully identified 7 melatonin synthesis genes (2 MeTDCs, 1 MeT5H, 1 MeSNAT and 3 MeASMTs) in cassava and cloned the coding sequences of them ( Fig. 1 and Table 1). The responses of these genes under different treatments (flg22, Xam, melatonin and H 2 O 2 ) were analyzed through quantitative real-time PCR, the common expression profile of these genes indicates the possible role of them as well as melatonin in immune response and ROS signaling in cassava (Fig. 2). As a rate limited enzyme of melatonin synthesis, gene expression of MeSANT was not up-regulated by the flg22 and H 2 O 2 . However, the endogenous melatonin levels were elevated under the treatments of flg22 and H 2 O 2 , as well as the response in hypersensitive-like cell death. On one hand, the transcript level is not always consistent with enzyme activity of MeSNAT, and the post-transcriptional regulation and post-translational regulation may also result in the issue. On the other hand, MeSNAT may only responsible for part melatonin synthesis in cassava under these stress conditions, and more other rate limiting enzymes of melatonin synthesis need to be further isolated. More importantly, we identified the novel role of transient expressing melatonin synthesis genes in hypersensitive-like cell death in leaves of N. benthamiana, depending on ROS accumulation and endogenous melatonin (Figs 4 and 5). SOD and CAT are major antioxidant metabolic enzymes by catalysing O 2 −. into H 2 O 2 and O 2 78,79 . RbohA and RbohB are also important regulators of not only H 2 O 2 accumulation, but also plant immunity. PR1, PR2 and PR5 are widely known marker genes of innate immune response 66,79,80 . Further investigation of gene expression indicated that transient expressing melatonin synthesis genes induced the transcription of both ROS-and defense-related marker genes (Figs 6  and 7), suggesting that these genes might exert its function through ROS accumulation and PCD as well as immune response. Additionally, the tobacco leaves expressing these genes triggered the burst of callose depositions (Fig. 8), suggesting that these genes might be involved in the modulation of callose-associated cell wall and papillae-associated plant defense.
Pathogen associated molecular patterns (PAMPs)-triggered immunity (PTI) can resist the incidence of most pathogenic microbes, and plays important roles in plant immune response [77][78][79] . In this study, melatonin synthesis genes expression plays common roles in some PTI responses, including the transcript profile in response to flg22 and Xam, ROS accumulation, hypersensitive-like cell death, defense-related gene expression and callose deposition. HR symptoms with accumulation of ROS usually occur in biotic or abiotic stresses in plants, but it also happen as a constitutive activated protective mechanism in recent studies 79 . The induction of PTI response as well as HR by melatonin synthesis genes indicates the basal immunity triggered by them.
Taken together, this study shows the comprehensive transcripts and the putative roles of melatonin related genes as well as melatonin in immune responses in cassava. The results may also provide important candidate genes for genetic breeding to improved disease resistance of cassava.

Methods
Plant materials and growth conditions. In this study, cassava plants of South China 124 (SC124) variety were grown in pots with soil and vermiculite (1:1) (pH 5.7) in the green house, which was controlled under 12 h light/28 °C and 12 h dark/26 °C cycles, at the irradiance of about 120-150 μ mol quanta m −2 s −1 . Hoagland's solution was watered twice every week to keep the well growth of cassava plants.
RNA isolation and quantitative real-time PCR. Plant leaves were harvest for total RNA isolation using AxyPrep TM Multisource Total RNA Miniprep Kit (AYXGEN-09113KD1, Santa Clara, California, USA) according to the manufacturer's instruction 81 . First-strand cDNA synthesis was performed using reverse transcriptase (Thermo-K1622, Waltham, Massachusetts, USA) as Shi et al. described previously 19 . Thereafter, the diluted cDNA and SYBR ® Premix DimerEraser TM (TaKaRa Biotechnology-RR091A, Dalian city, Liaoning, China) were used for quantitative real-time PCR in LightCycler ® 96 Real-Time PCR System (Roche, Basel, Switzerland). All gene transcripts were normalized to elongation factor 1α (EF1α) using comparative 2 −ΔΔCT method. The primers used in this study were listed in Supplemental Table S1. plasmids and P19 were co-infected into 30-day-old tobacco leaves as previously described 83 . For each vector construct, 20 independent leaves of five independent tobacco (N. benthamiana) plants (four leaves per plant) were used for GFP and physiological assays. After two days, GFP signals in the transformed tobacco leaves were observed using a confocal laser-scanning microscope (TCS SP8, Leica, Heidelberg, Germany). Physiological assays were performed at different time points after infection.
Determination of physiological parameters. For the ROS staining, H 2 O 2 and O 2 • − were stained in diaminobenzidine (DAB) solution (pH 3.8) and nitro blue tetrazolium (NBT) solution (pH 7.8) and further decolorized in 70% (v/v) ethanol as Shi et al. described 19 . Moreover, H 2 O 2 and O 2 • − contents were quantified by examining the peroxide-titanium from the titanium sulphate regent and antibody-antigen-enzyme-antibody complex from the plant O 2 • − ELISA Kit as Shi et al. described 19 .
The EL was calculated as the relative value of initial conductivity (C i ) to the maximum conductivity (C max ), which were assayed in the mixture of plant leaves-deionized water after 6 hr of shaken and after boiled and cooling, respectively.
The MDA was determined by examining the absorbance of the supernatant of leaves samples in the thiobarbituric acid (TBA) regent as Shi et al. described 19 . Callose staining. Tobacco leaves were harvest after infiltration GV3101 cell culture harbouring the construction, cleared in alcoholic lactophenol solution and staining with 0.01% (w/v) aniline blue as Hauck et al. described 84 . Images were examined using a fluorescence microscope (DM6000B, Leica, Heidelberg, Germany), and the callose depositions were quantified using the ImageJ software.

Significant difference analysis.
In this study, all experiments were repeated with three biological experiments. All data were shown as average value ± SD of three biological experiments. ANOVA and student's t-test were performed to analyze the significant difference in comparison to mock treatment, and asterisk symbols (*) were shown as significant difference at p < 0.05.