Novel flavin-containing monooxygenase protein FMO1 interacts with CAT2 to negatively regulate drought tolerance through ROS homeostasis and ABA signaling pathway in tomato

Abstract Drought stress is the major abiotic factor that can seriously affect plant growth and crop production. The functions of flavin-containing monooxygenases (FMOs) are known in animals. They add molecular oxygen to lipophilic compounds or produce reactive oxygen species (ROS). However, little information on FMOs in plants is available. Here, we characterized a tomato drought-responsive gene that showed homology to FMO, and it was designated as FMO1. FMO1 was downregulated promptly by drought and ABA treatments. Transgenic functional analysis indicated that RNAi suppression of the expression of FMO1 (FMO1-Ri) improved drought tolerance relative to wild-type (WT) plants, whereas overexpression of FMO1 (FMO1-OE) reduced drought tolerance. The FMO1-Ri plants exhibited lower ABA accumulation, higher levels of antioxidant enzyme activities, and less ROS generation compared with the WT and FMO1-OE plants under drought stress. RNA-seq transcriptional analysis revealed the differential expression levels of many drought-responsive genes that were co-expressed with FMO1, including PP2Cs, PYLs, WRKY, and LEA. Using Y2H screening, we found that FMO1 physically interacted with catalase 2 (CAT2), which is an antioxidant enzyme and confers drought resistance. Our findings suggest that tomato FMO1 negatively regulates tomato drought tolerance in the ABA-dependent pathway and modulates ROS homeostasis by directly binding to SlCAT2.


Introduction
The adverse effects of various abiotic stresses on crop yield have become a major challenge in changing climates. Drought is a major environmental stress that dramatically affects agriculture worldwide [1]. Therefore, the molecular mechanism of droughtresponsive genes needs to be characterized and understood to increase plant drought resistance. Plants develop drought stress tolerance through the integration of hormones and environmental signals that regulate different drought-responsive genes.
Hormones that regulate gene expression are essential to plants under drought stress [2]. As one of the most effective endogenous hormones in plants, abscisic acid (ABA) is a key regulator of the physiological processes of plants' response to abiotic stresses [3]. For instance, ABA induces the accumulation of late embryogenesis abundant (LEA) protein and participates in the response of plants to drought stress [4]. LEA proteins widely exist in the cytoplasm and nucleus of plants and respond to various abiotic stress signals. ABA responses to drought stress occur through ABAdependent and ABA-independent pathways [5]. ABA activates the SnRK2 S III response to drought in the ABA-dependent pathway [6].
In the ABA-dependent pathway, SnRK2s III, ABA receptor (ABA-PYR/PYL/RCAR) complexes, and protein phosphatases 2C (PP2Cs) are core components in ABA sensing [6]. SnRK2 III is the key positive regulator of the downstream signal of the ABA receptor. When drought stress occurs, ABA is largely stored in plant cells. Then, the ABA receptor competes with PP2Cs to combine with SnRK2s III; as a result, the inhibitory effect of PP2Cs on SnRK2s III is released [7][8][9]. Activated SnRK2s III can phosphorylate transcription factors, such as AREB or ABF, to increase the expression of abiotic stress response genes [6].
NCED (9-cis-epoxycarotenoid dioxygenase) is a crucial gene that regulates ABA synthesis [10]. Overexpression of SlNCED1 can increase the concentration of ABA in leaves [11]. Aluminum can upregulate the expression of ClNCED3 to accelerate the accumulation of ABA in roots and stomatal closure under aluminum stress [12]. The silencing of NCED4 changes the ABA biosynthesis and signaling pathways related to gene expression [13].
Flavin-containing monooxygenases (FMOs) are a family of f lavoenzymes that often contain an FAD and NADPH interaction motif [14]. They were discovered in mammals, and they function in catalyzation for oxygenation [15]. More FMO genes have been found in plants than in animals. For instance, 29 putative FMO genes have been discovered in Arabidopsis, whereas only five have been found in humans; meanwhile, only a single FMO gene has been identified in yeast [15]. Many recent reports have verified that FMO is involved in auxin biosynthesis, glucosinolate metabolism, and abiotic and biotic stresses. For instance, YUCCA proteins belong to a family of FMOs that catalyze critical steps in auxin biosynthesis, and they contribute to plant development and abiotic stress resistance [16][17][18]. The induction of FMO GS-OX2 changes the glucosinolate profile of Arabidopsis leaves [19]. Overexpression of FMO GS-OX1 can cause a 5-fold increase in the levels of glucosinolates, which are broken down into isothiocyanates to enhance the cancer-preventive activity of cruciferous vegetables [20]. Furthermore, the induced expression of FMO GS-OX by glucose can increase endogenous ABA accumulation in Arabidopsis thaliana, and the trend is significantly inhibited in ABA-deficient (aba2-1, aba3-1) and ABA-insensitive (abi1-1, abi2-1) mutants [19]. FMO can use cysteine as a nitrogen source to produce nitrogenous aroma volatiles in tomato [21].
Tomato is one of the most important vegetable crops and plant models in the world. It has the characteristics of simple diploid inheritance, easy transformation, high reproductive efficiency, and short life cycle [22]. Tomato is often exposed to drought stress, which poses a critical threat to tomato growth and sustainable agriculture worldwide [23]. Transgenic methods have been successfully used to improve the drought resistance of tomato. Overexpression of the tomato WHIRLY transcription factor SlWHY2 increases the drought and Pseudomonas solanacearum resistance of tobacco [24], and the transcription factor SlbHLH22 enhances tomato salt and drought resistance [25]. Knockout of a protein kinase gene SlOST1 reduces tomato drought resistance and exhibits a late f lowering phenotype [26]. Moreover, tomato genes SpPKE1, SlDREB, SlWRKY81, SlERF1, and SlWRKY70 play a positive role in drought resistance [27][28][29][30], and SlbZIP38 and HyPRP1 negatively regulate drought tolerance [31,32]. In addition, knockout of SlNPR1 reduces tomato drought resistance [33]. Therefore, discovering new drought resistance genes is important.
In this study, we studied tomato FMOs, such as FMO1, which is a drought-responsive gene screened via drought transcriptome analysis [34]. We demonstrated that inhibiting the expression of FMO1 could significantly improve the drought tolerance of tomato, whereas overexpression of FMO1 could make it increasingly sensitive. RNA-seq transcriptome analysis revealed that FMO1 regulated the expression levels of ABA signaling and drought-responsive genes. FMO1 physically interacted with catalase 2 (CAT2) via yeast-two-hybrid (Y2H) analysis, which can promote reactive oxygen species (ROS) scavenging, thereby conferring drought tolerance by maintaining ROS homeostasis. This work presents evidence on the critical role of FMO1, which negatively regulates drought tolerance by interacting with SlCAT2 and modulates the expression of drought-responsive genes in the ABA-dependent pathway.

Gene features of FMO1
The full-length cDNA of the tomato FMO1 sequence was 1293 bp and encoded with 430 amino acids (Fig. 1a). The most homologous proteins were retrieved through NCBI pBLAST FMO1. All of the highly homologous FMOs were from plants. All FMO1 sequences showed high conservation (Fig. 1a), suggesting that FMO1 may have similar functionality in different species. Compared with previously reported structures of FMOs [14,15], the 3D protein structure of FMO1 was generated by SWISS-MODEL homology. Tomato FMO1 was composed of two conservated binding domains. Located close to the N-terminus of the protein was the FAD-binding 'GxGxxG' motif, and the FMO-identifying motif 'FxGxxxHxxxY/F' was located near the center of the protein ( Fig. 1a and b). However, the less well-conserved NADP-binding motif was not found in all the plant FMOs. A phylogenetic tree was constructed based on the protein sequences of FMO1 and the retrieved highly homologous plant FMOs. It showed that all Solanaceae FMOs clustered together (Fig. 1c).

Expression patterns of FMO1 under plant hormone and abiotic stress treatment
The tissue-specific expression of FMO1 was analysed by quantitative reverse transcription polymerase chain reaction (qRT-PCR). FMO1 was expressed in all the detected tissues, among which the expression in the f lowers was the highest, followed by that in the fruits and leaves (Fig. 2).
To determine whether these hormones regulate the expression of FMO1, we investigated the effect of exogenous plant hormone application on the expression of FMO1 via qRT-PCR. We found that the expression of FMO1 was rapidly downregulated at 1 h and at 6 h or 12 h. Then, it was upregulated at 24 h under ABA and GA treatments. However, under the Eth and JA treatments, the FMO1 expression was induced 1 h after treatment and was the highest at 24 h. After SA treatment, the FMO1 expression was also upregulated at 6 h (Fig. 2). These results indicate that FMO1 was downregulated by ABA and GA but upregulated by Eth, JA, and SA.
The expression of FMO1 in the various stress treatments was determined to confirm if FMO1 participates in abiotic stress. FMO1 showed identical expression profiles under the drought/PEG and salt/NaCl treatments. It was rapidly downregulated at 0.25 h and showed elevated transcripts for up to 1 h. Then, it reverted to a low expression level. Moreover, under the MV (methyl viologen; induced oxidative stress) treatment, the expression of FMO1 was rapidly downregulated at 0.25 h up to 6 h after treatment; elevated transcripts were obtained 12 h later (Fig. 2). These results illustrate that FMO1 responded to drought stress and was likely induced by salt and oxidation. Moreover, it appeared to be a negative regulator at the early stage of the abiotic stress response.

Inhibiting the expression of FMO1 regulates drought resistance, and overexpression made it increasingly sensitive
Ten FMO1-OE and 8 FMO1-RNAi lines were generated to investigate the function of FMO1. The qRT-PCR screening results showed two FMO1-OE lines (OE-8 and OE-10) and two FMO1-RNAi lines (Ri-1 and Ri-3) with significant differential expression compared with the wild-type (WT) lines (Fig. 3a). These transgenic lines were selected and grown on MS media by drought and salt stress treatments for functional analysis. Mannitol, a compatible solute accumulated in plants, is a substance that produces osmotic stress; it can be used as a substitute to simulate osmotic stress in drought stress experiments [36]. Under the control MS media, the growth phenotypes of WT, RNAi, and OE plants showed no significant differences ( Fig. 3c; Supplemental Fig. S1, see online supplementary material). However, under mannitol stress, the growth and root length of the FMO1-OE lines were smaller than those of the WT lines, but the growth of the FMO1-RNAi lines  [14]. c Phylogenetic analysis of FMO1. The phylogenetic tree was constructed by MEGA 7.0 with neighbor-joining method at 1000 bootstrap replicate. The scale indicates branch length, and bootstrap values represent the divergence of each branch. was much higher than that of WT. The root lengths of the FMO1-RNAi lines were larger than those of WT ( Fig. 3b and d). Similarly, when treated with 200 mM NaCl, the root length of the FMO1-OE lines was smaller than that of the WT lines, but the root length of the FMO1-RNAi lines was larger than that of WT ( Fig. 3b and e). These results reveal that inhibiting the expression of FMO1 could improve the drought and salt resistance of tomato.
To further verify the effect of FMO1 expression on the regulation of drought tolerance, we grew the WT and FMO1 transgenic lines in soil under normal watering conditions and applied drought treatment up to the five-leaf stage. As shown in Fig. 4a, before the drought treatment, the WT and transgenic lines grew well. After the 7-day drought treatment, the leaves of the WT and FMO1-OE lines began to wilt, and no clear change was observed in the FMO1-RNAi lines. After 12-day drought treatment, the WT and FMO1-OE lines wilted more seriously than the FMO1-RNAi lines did. After 17-day drought treatment followed by a 3-day recovery process, the WT and FMO1-OE lines were still wilted and could not survive, but most of the FMO1-RNAi lines resumed their growth. After 17-day drought treatment, the survival rate of the FMO1-RNAi lines was >80%, which was considerably higher than the survival rate of the WT lines. Meanwhile, the survival rate of FMO1-OE was less than 20% (Fig. 4b). Moreover, the water loss detected in the FMO1-RNAi lines was smaller than that in the WT and FMO1-OE lines (Fig. 4c). This result indicates that inhibiting the expression of FMO1 increased drought tolerance, whereas overexpression of FMO1 decreased drought tolerance and increased sensitivity.

Inhibited expression of FMO1 improved the drought resistance of tomato by accumulating abundant proline and regulating ROS homeostasis
The capability of plants to cope with drought stress is closely related to the accumulation of proline (Pro) in plants, and the concentration of Pro is usually higher in stress-tolerant plants than in stress-sensitive plants. Moreover, excessive ROS production under drought has been reported to increase the malondialdehyde (MDA) content, an indicator of oxidative damage. Thus, in this study, the Pro and MDA contents of WT and transgenic plants under normal watering and drought stress conditions were investigated. The Pro and MDA contents of the WT and transgenic plants showed no significant difference under normal watering. After drought treatment, the Pro contents of the FMO1-RNAi lines were higher than those of the WT and FMO1-OE lines. On the contrary, the MDA contents of the FMO1-RNAi lines were lower than those of the WT and FMO1-OE lines ( Fig. 4d and e).
ROS induced by abiotic stress directly causes plant cell damage. To reduce the effects of ROS, plants develop low-molecular-weight antioxidants and reactive oxygen scavenging enzymes, including peroxidase (POD) and catalase (CAT). Therefore, the activities of CAT and POD in the WT and transgenic plants were monitored before and after drought stress in the current study. The CAT and POD activities of the FMO1-RNAi lines were higher than those of the WT and FMO1-OE lines ( Fig. 4f and g). This finding is consistent with the observation of mature leaves via histochemical staining by 3,3 -diaminobenzidine (DAB) and nitroblue tetrazolium chloride (NBT) after drought stress. The distribution of H 2 O 2 and O 2− in situ in the leaves of the FMO1-RNAi lines was less intense than that in the WT and FMO1-OE lines, especially at the top of the tomato leaves (Fig. 4h). This finding indicates that inhibiting the expression of FMO1 enhanced tomato drought tolerance via increased accumulation of Pro, improvement of the activities of CAT and POD, and MDA content reduction.

FMO1 silencing improved tomato drought tolerance by regulating the ABA-dependent pathway and drought stress-related genes
To further study the molecular mechanism of FMO1 gene regulation during drought stress, RNA sequencing was performed to analyse the genes and pathways regulated by FMO1. A total of 12 cDNA libraries, including WT and FMO1 transgenic lines (OE and Ri), were present in DT and CK (DT: drought treatment and CK: control). The cDNA library of each sample produced about 39.67-46.11 million reads, and the correct event rate of sequencing reached 98% (Table S1, see online supplementary material). The sequences were aligned to the tomato-released genome ITAG3.0, as shown in Table S2 (see online supplementary  material). Approximately 38.57-44.95 million reads were mapped to the tomato genome, accounting for 97.1% of the total reads on the average. More than 95% of the clean reads were uniquely located in the tomato genome, indicating that the RNA-seq results were reliable.
Next, |log 2 (FoldChange)| > 1 and padj ≤0.05 were used as criteria to screen genes with significantly different expression levels and to perform statistical analysis. As shown in Fig. 5a, under CK watering conditions, fewer differentially expressed genes (DEGs) were present among WT, Ri, and OE compared with the DT conditions. The heat clusters of numerous DEG expression changes are presented in Fig. 5b. For WT, Ri, and OE under CK and DT, the colors of the heatmap expression levels differed significantly between DT and CK, indicating that the expression of numerous genes was regulated by FMO1, especially under drought conditions. Many genes responded to drought, and the expression of FMO1 could regulate many genes under drought stress.
The Venn diagram analysis revealed 619 and 1583 common DEGs in Ri, WT, and OE under DT intercomparison and DT versus CK, respectively (Fig. 5c). We classified these differential genes to determine the key genes and metabolic pathways that FMO1 regulated during drought. The expression levels of several PP2Cs increased considerably after drought treatment, and they were higher in the FMO1-OE plants than in the FMO1-RNAi and WT plants (Fig. 5d), indicating that FMO1 and PP2Cs were coexpressed. This result also implies that FMO1 participated in the ABA-dependent pathway by modulating PP2C transcription to improve plant drought resistance. Gene ontology (GO) analysis further revealed that the DEGs mainly assembled during ion transport, photosynthesis, and plant hormone signal transduction (Fig. S2, see online supplementary material). This finding is consistent with the KEGG pathway enrichment results (Fig. S3, see online supplementary material). The shared DEGs under drought stress were enriched in the MAPK signaling pathway, which mainly included the ABA receptor PYR/PYL (Fig. S4a, see online supplementary material), PP2Cs (Fig. 5d), photosynthesis, plant hormone signal transduction, and carbon metabolism (Fig. S3, see online supplementary material). In the plant hormone synthesis and signal transduction pathway, except for ABA sensing and the signaling core component PP2Cs, the DEGs were enriched in the ABA synthesis pathway, e.g. the ABA biosynthesis gene NCED1 (Table S3, see online supplementary material).
To confirm whether FMO1 is involved in the ABA pathway, the ABA contents of FMO1-OE, FMO1-RNAi, and WT under DT or CK were determined. As shown in Fig. 6, the ABA contents of the WT and transgenic plants showed no significant difference under normal watering. However, after drought treatment, the ABA contents of the FMO1-OE lines were significantly higher than those of the WT and FMO1-Ri lines. The ABA content of WT was  also significantly higher than that of the FMO1-Ri lines. According to these observations, the FMO1 gene possibly participated in ABA-dependent signaling and negatively regulated the drought resistance of tomato.
The enriched genes in the GO term regulation of transcription showed stress-related DEGs, including cytoplasmic ascorbate peroxidase 1, monodehydroascorbate reductase 1, peroxidase, catalase isozyme 2 isoform X1, superoxide dismutase, and stress-related proteins, and other oxidative stress-related genes, transporters, and transcription factors (Table S4, see online supplementary material). The WRKY transcription factors were co-expressed with FMO1 (Fig. S4b, see online supplementary material), and most of the co-expressed genes were negatively regulated by FMO1. However, the LEA hydrophilic protein genes, which have been implicated in the stress responses of many plants, downregulated the FMO1-RNAi line after drought stress treatment and were positively regulated by FMO1 (Fig. S4c, see online supplementary material). To further verify the reliability of the transcriptome results, we selected 12 DEGs, such as PP2C, that were related to the ABA biosynthesis gene NCED1 and abiotic stress genes for real-time PCR verification (Table S3, see online supplementary material). The quantitative results (Fig. S5, see online supplementary material) of this experiment were in good agreement with the RNA-seq data (Table S3, see online supplementary material). Our results strongly support the idea that tomato FMO1 negatively regulates drought tolerance by regulating the ABA synthesis and signal transduction pathway and drought stress-related genes.

FMO1 interacted with CAT2 to confer drought resistance
Y2H screening of interacting proteins was performed to further analyse how FMO1 regulated drought tolerance. FMO1 was ligated to pGBKT7 to produce a pGBKT 7 -FMO1 (pBD-FMO1) fusion vector. Then, we used pBD-FMO1 to screen the tomato cDNA yeast library matchmaker Gold Y2H system. The positive clones were stained with X-α-Gal. Approximately 50 positive clones were selected and sequenced, and some proteins that possibly interact with FMO1 were obtained, including SlCAT2 (Solyc12g094620) and stressrelated proteins (Table S5, see online supplementary material). The deduced amnio acid sequence of SlCAT2 shared 89%, 86%, and 86% similarity with AtCAT2 (AT4G35090), AtCAT1 (AT1G20630), and AtCAT3 (AT1G20620), respectively (Fig. S6, see online supplementary material). To verify the reliability of the results, we cloned the putative interacting protein SlCAT2 gene for point-to-point verification. When the yeast contained tomato FMO1 and CAT2 protein, it grew and turned blue on the medium of SD/−Ade/His/−Leu/−Trp (containing X-α-Gal) (Fig. 7a), indicating that FMO1 interacted with SlCAT2 in the yeast. Furthermore, we used co-immunoprecipitation (Co-IP) and bimolecular f luorescence complementation (BiFC) to verify the results. The Co-IP assays demonstrated the existence of FMO1/CAT2 protein complexes in Nicotiana benthamiana leaves (Fig. 7b), and the BiFC results showed that when FMO1: YFP N and CAT2: YFP C were co-expressed in the N. benthamiana leaves, the leaves showed yellow f luorescence, which was not observed in the empty vector (Fig. 7c). This result further proves that FMO1 and CAT2 interacted with each other.
To determine if CAT2 is an antioxidant enzyme and confers drought resistance, we performed transgenic function analysis by using CAT2 OE and Ri plants. As presented in Fig. 7d, CAT2-OE showed a dwarfing phenotype under normal conditions. After drought stress, the CAT2-Ri and WT plants wilted, but the CAT2-OE lines grew well. The DAB and NBT histochemical staining results of mature leaves after drought stress revealed that the distribution of H 2 O 2 and O 2− in situ in the leaves of the CAT2-OE lines was less intense than that in the leaves of the WT and CAT2-Ri lines (Fig. 7d). This result indicates that CAT2 positively regulated drought resistance by scavenging ROS.

Tomato abiotic response gene FMO1 is widespread and highly conserved in plants
We identified a novel FMO through tomato drought transcriptome data analysis. A CLUSTAL alignment of FMOs in plants, which was highly homologous with tomato FMO1 sequences, was present, indicating that the FMOs were widespread in the plants. These findings imply that FMO1 may have similar functions in different species. In Schizosaccharomyces pombe, FMO is composed of FAD binding its complex with NADPH and an enzyme-substrate complex [14]. The plant FMOs in the present study shared highly conserved FMO-identifying and FAD-binding motifs but no NADPH interaction domain ( Fig. 1a and b). These findings differ from those obtained for the pepper FMO gene Bs3, which contains an NADPH binding site in the middle of amino acid residues [37].
According to the current tissue expression analysis of FMO1 in tomato, the gene was highly expressed in the fruits, f lowers, and leaves. FMO1 was downregulated at the early stage of drought stress (Fig. 2). FMO1 probably has a crucial function in the process of drought resistance during tomato growth, and whether it also improves the f lower and fruit preservation rate in tomato after drought stress remains to be further studied. Moreover, we detected the expression of FMO1 response to different plant hormones through qRT-PCR. The expression of FMO1 was rapidly downregulated at 1 h under ABA treatment but constitutively upregulated by Eth and JA (Fig. 2). JA and ABA are phytohormones that activate the defense system to enhance plant resistance to abiotic stresses [35]. ABA and Eth production increases during drought stress in tomato plants in response to biotic and abiotic stresses [38]. Therefore, FMO1 regulates ABA, Eth, and JA signaling and is probably involved in hormone crosstalk and drought stress.

FMO1 serves as a negative regulator of drought stress by modulating antioxidant-related enzymes and osmotic substances
Drought stress seriously restricts the growth of plants and therefore severely affects the yield and quality of crops. As a result, improvement of drought resistance has become a long-term goal in crop breeding. Recently, several genes have been successfully applied to increase the drought resistance of tomato. SlbZIP1 modulates an ABA-mediated pathway that contributes to drought stress tolerance [39], and overexpression of SlMAPK3 results in increased tolerance to Cd 2+ and drought in tomato [40]. Overexpression of SlNAC2 in tobacco improves plant survival and recovery, which in turn increases drought resistance. SlWRKY81 attenuates stomatal closure and subsequent drought tolerance by reducing nitric oxide accumulation [28]. Increased SlBRI1 expression levels reduce tomato drought resistance due to increased brassinosteroid signaling intensity [25], and overexpression of SlERF.B1 results in tomato plants that are less tolerant to drought and mannitol stress [41]. Many FMOs have been investigated in yeast and mammals, and relevant research has shown that their main function is to add molecular oxygen to lipophilic compounds or to produce ROS [14,15]. However, until recently, the function of FMOs in plants has remained elusive.
In this study, we performed transgenic functional analysis to verify whether FMO1 participates in the abiotic resistance of tomato. By comparing the phenotypic differences among FMO1-OE, FMO1-Ri, and WT plants after drought stress, we found that the drought tolerance of the FMO1-Ri plants was higher than that of the FMO-OE and WT plants (Figs 3d and 4a). Thus, tomato FMO1 negatively regulates drought tolerance, and it is unlike the first plant-identified FMO protein YUCCAs, which positively regulate muti-abiotic stress tolerance, including drought, aluminum stress, heat stress, and shade [18]. YUCCAs share similar properties and typical structures with mammalian FMO, which contains two conserved 'GxGxxG' sequences for FAD and NADPH binding. However, in this study, the tomato amino acid of FMO1 had very low homology with YUCCAs and mammalian FMOs, and the homology search and sequence alignment results showed that FMO1 was specific to plants. Moreover, tomato FMO1 only possessed a FAD-binding motif near the N-terminus, and there was no NADPH binding motif in the middle region ( Fig. 1a and b). Our observations led us to conclude that tomato FMO1 is a novel FMO that is a negative regulator of drought tolerance.
Drought stress alters cellular redox homeostasis and generates ROS, including H 2 O 2 , hydroxyl radicals (OH − ), and superoxide radical (O 2− ) [42]. Thus, ROS scavenging can protect plants against drought stress [43]. To minimize drought-induced oxidative damage, plants have various physiological functions, including antioxidant metabolism and osmotic adjustment [44]. Therefore, we measured the relevant physiological indicators of the FMO1-OE, FMO1-Ri, and WT plants before and after drought stress. The Pro content of the FMO1-RNAi lines was higher than that of the WT and FMO1-OE lines under drought stress (Fig. 4d), but the MDA content in the WT and FMO1-OE lines increased (Fig. 4e). FMO1 silencing conferred tolerance to drought by adjusting the Pro content and reducing the level of membrane lipid peroxidation. Similarly, under drought stress treatment, we found that the activities of CAT and POD in the FMO-RNAi lines were higher than those in the WT plants ( Fig. 4f and g), and the leaves of the FMO-RNAi lines exhibited lower ROS accumulation than those of the WT and FMO1-OE lines after visualization by DAB and NBT histochemical staining (Fig. 4h). Similar trends have been observed for Arabidopsis, where AtFMO1 is related to ROS production, especially H 2 O 2 ; it is involved in DNA damage and root meristem [45]. Over-accumulation of ROS caused by drought stress results in membrane lipid peroxidation, but some enzymatic antioxidants, such as CAT, play an essential role in eliminating ROS [46]. Overall, we posit that FMO1 negatively regulates drought tolerance by accumulating abundant antioxidant-related enzymes and osmotic substances to reduce ROS damage.

Molecular mechanism of FMO1 regulating drought resistance of tomato
Although ROS is involved in many biological processes by acting as signaling molecules, plant cells can suffer from oxidative damage due to excessive ROS accumulation [47]. Catalases exist in almost all living organisms and are known as the most effective ROS scavenger because of their high catalytic decomposition capacity toward H 2 O 2 , which plays important roles in plant abiotic stresses [47]. In Arabidopsis, three CAT genes, namely, AtCAT1, AtCAT2, and AtCAT3, have been characterized [48]. AtCAT1 is mediated by MAPK cascades, and its expression is upregulated by ABA, which suppresses ROS accumulation. AtCAT2 activity is necessary for long-term heat tolerance, and CAT2 (AtCAT2 mutant) accumulates abundant H 2 O 2 by reducing leaf catalase activity. Under drought conditions, AtCAT3 plays a vital role in H 2 O 2 homeostasis and stomatal regulation mediated by ABA. In this study, tomato SlCAT2 was found to interact with FMO1 (Fig. 7), and the expression of FMO1 affected CAT and POD activities to reduce the membranous peroxidation damage (Fig. 4f and g). SlCAT2 also interacted with SlMSR B2, which is involved in drought resistance by promoting ROS scavenging [49]. The OE of SlCAT2 resulted in a dwarfing phenotype and enhanced drought resistance by scavenging ROS (Fig. 7d). Altogether, our results strongly support the notion that FMO1 is involved in tomato drought resistance by directly modulating SlCAT2 and regulating the expression of other genes related to ROS-scavenging enzymes' expression (Table S3, see online supplementary material) or by affecting enzymatic activities ( Fig. 4f and g). Thus, exploration of the detailed mechanism of whether and how tomato FMO1 catalyzes the oxygenation of CAT2 under drought stress is worthy of attention.
The RNA-seq results revealed that the number of DEGs in WT, FMO1-OE, and FMO1-Ri lines under drought conditions increased significantly compared with that under normal growth conditions (Fig. 5). The DEGs mainly assembled during ion transport, photosynthesis, plant hormone signal transduction, carbon metabolism, and MAPK signaling ( Fig. 5d; Figs. S2 and S3, see online supplementary material), which are crucial processes in plants coping with drought stress. The MAPK signaling pathway is the core of the cell signal transduction system [50]. This pathway is crucial for plant responses to abiotic stresses and hormone signal transduction [50]. Through RNA-seq analysis, we found that the shared DEGs among WT, Ri, and OE in the MAPK signaling pathway included ABA receptor, PP2Cs, and some protein kinases (Fig. S3, see online supplementary material). In the plant hormone signal transduction pathway, aside from PP2Cs, the key gene in ABA biosynthesis, NCED1, was induced in the FMO1-OE plants and inhibited in the FMO1-Ri plants. At least 12 PP2Cs were co-expressed with FMO1 under drought stress (Fig. 5d). PP2Cs are responsible for negative regulation of ABA signaling via dephosphorylation of SnRK2 [51]. Thus, PP2Cs are insensitive to ABA [52], and overexpression of PP2Cs results in sensitivity to drought stress [53]. SnRK2 phosphorylates various substrates, including AREB/ABFs, FBH3/AKS1, and SNS1 in the ABAdependent pathway, and modulates their activities [5]. Recent studies have suggested that TaWRKY1 in wheat plays a crucial role in drought tolerance and the ABA-dependent pathway [54]. In our study, the WRKY transcription factors were co-expressed with FMO1 (Fig. S4b, see online supplementary material), and most of these co-expressed genes were negatively regulated by FMO1. These findings, together with our result that FMO1 negatively regulates drought tolerance (Fig. 4), led us to conclude that FMO1 modulates the ABA signal by regulating PP2Cs and other stress-resistant genes, such as WRKY transcription factors, to cope with drought stress in the ABA-dependent signaling pathway.
NCED1 was downregulated in the FMO1 RNAi lines and upregulated in the FMO1 OE plants (Table S3, see online supplementary material). NCED encodes a key enzyme during ABA biosynthesis to control plant accumulation of ABA [55]. For instance, the NCED3 mutant shows decreased ABA content, which reduces drought resistance. On the contrary, overexpression of NCED3 increases the ABA content and drought resistance [56]. This finding is consistent with our observation that ABA was abundantly accumulated in the FMO1 OE plants and significantly decreased in the FMO1 Ri plants (Fig. 6). A low ABA concentration inhibits PP2C activity and promotes root growth and hydrotropism [57]. Therefore, we suggest that the low ABA content in the FMO RNAi plants (Fig. 6) resulted in enhanced root growth under mannitol and NaCl stresses (Fig. 3d and e) and conferred drought tolerance (Fig. 4) by downregulating the expression of PP2Cs (Fig. 5d). The low accumulation of ABA in the FMO RNAi plants might be due to the feedback of PP2C downregulation by activating the ABA signaling pathway. Thus, we speculate that FMO1 negatively regulates drought resistance primarily by affecting ABA signaling then increasing ABA synthesis. Moreover, the pepper FMO gene Bs3 triggers plant cell death, impairs yeast growth, and produces H 2 O 2 , thereby promoting ABA signaling via the inactivation of the negative regulator PP2Cs [37,58]. This finding helps explain why the FMO1 RNAi-inhibited plants in our study accumulated less H 2 O 2 and had a low expression of PP2Cs. Our results strongly support the idea that FMO1 regulates ABA signaling and CATmediated antioxidant pathways to cope with drought resistance in tomato. Many clues show that the two pathways are not independent. A previous study has reported that the content of H 2 O 2 decreases when ABA production is reduced, which is closely related to the activity of catalases [59]. H 2 O 2 plays a vital intermediary function in the ABA signal pathway, thereby inducing the expression of CAT1 [60]. A recent study also showed that the transcription factor CDF4 induces the accumulation of ABA, which downregulates the transcript of CAT2, leading to the inhibition of H 2 O 2 scavenging [61]. We posit that the levels of FMO1, ABA, and CAT2 interlinked by feedback loops can adapt to drought stress.
We elucidated the mechanisms through which FMO1 negatively regulated the drought tolerance of tomato. When the plants coped with drought, FMO1 silencing inhibited the expression of PP2Cs to induce the response of the ABA signaling pathway. Meanwhile, the ABA biosynthesis gene NCED1 downregulated and inhibited ABA biosynthesis. Thus, the ABA content decreased. FMO1 directly interacted with CAT2 to promote ROS scavenging, thereby conferring drought tolerance by improving the activities of CAT and POD to eliminate ROS and reduce membrane lipid peroxidation. Therefore, FMO1 downregulation triggered the ABAdependent signaling pathway, leading to the induction of stressresponsive genes and drought tolerance (Fig. 8). In summary, FMO1 knockdown in tomato provided enhanced drought tolerance by regulating the ABA signaling and CAT2-mediated pathways maintaining ROS homeostasis. Thus, FMO1 is a novel target gene in genetic engineering for the enhancement of plant drought tolerance.

Bioinformatics analysis and transgenic plant generation
The coding sequence of FMO1 was PCR-amplified from tomato (Solanum lycopersicum) cDNA. The tomato FMO1 and its homologous protein sequences were compared using Clustal W2. FMO1-OE and FMO1-RNAi vectors were constructed as previously described [31]. In brief, FMO1 was cloned into pMV to yield FMO1-OE (under the CaMV35S promoter). The pHellsgate 2 vector (Invitrogen, CA, USA) was used for FMO1-RNAi construction via a recombination reaction using a 390 bp fragment. After genetic transformation to tomato M82, the resulting positive FMO1-OE and FMO1-RNAi plants were analysed through qPCR. The transgenic plants with significantly increased (FMO1-OE) or decreased (FMO1-RNAi) transcripts were selected for further analysis. The SlCAT2 OE and RNAi plants were obtained from a previously study [49]. The primers are listed in Table S6 (see online  supplementary

Plant treatments and quantitative real-time PCR
The tissue and plant growth regulator treatments were established as described previously [62]. In brief, WT seedlings at the five-leaf stage were used for the treatments. The leaves were sprayed with ABA (100 μM), GA (100 μM), ethephon (100 μM; an Eth releaser), JA (100 μM), SA (100 μM), and MV (100 μM) for the plant growth regulator treatment. The soil was irrigated with PEG (100 mM) and NaCl (100 mM) for the drought and salt treatment, and the control was sprayed with distilled water.
RNA extraction and cDNA synthesis were performed as described previously [62]. qRT-PCR reactions were implemented with the SYBR Green Real-time PCR system (Bao Guang, China), and the PCR products were monitored using the CFX96 PCR system.

Abiotic stress assays
The FMO1 transgenic plants with significant differential expression were selected for drought stress assays. We used a mannitolsimulated drought and applied NaCl for high-salinity treatments on MS medium. The seeds were germinated in MS medium, and the uniformly sprouted seedlings were sub-cultured in MS media containing 200 mM mannitol and NaCl for the treatments. Nostress treatment was used as the control, with at least four replicates per treatment. Drought tolerance testing was performed on the soil-grown transgenic plants as previously described [30], and a water loss assay was performed as previously described [31]. The MDA and Pro contents and CAT and POD activities were measured as described in a previous study [63]. Histochemical staining by DAB and NBT after drought stress was conducted as described previously [63].

Transcriptome analysis
The leaves of WT, FMO1 OE, and Ri lines under 17-day drought and normal watering were used for RNA extraction. After reverse transcriptase, we found that cDNA libraries of WT, OE, and Ri were present in DT and CK (labeled WTCK, OECK, RiCK, WTDT, OEDT, and RiDT; two biological replicates per treatment). The resulting cDNA libraries were sequenced using the Illumina NovaSeq platform (Nova, China). After sequencing, the filtered reads were aligned to the tomato genome (ITAG3.0). The mapped reads, differential expression, and gene enrichment were analysed as described previously [64].

Yeast screening library and analysis
The SlFMO1 gene sequence was amplified from tomato cDNA and ligated to pGBKT7 by enzyme digestion to obtain the pGBKT 7 -FMO1 (pBD-FMO1) fusion vector. The pBD-FMO1 was used as bait to screen the tomato cDNA library by following the manufacturer's instructions (Clontech, USA). In brief, pBD-FMO1 and pGADT 7 recombinant tomato homogeneous cDNA library plasmids were co-transformed into yeast then transferred to SD/−Trp-Leu-His-Ade media for screening. Fifty randomly positive clones were randomly selected, sequenced, and analysed.

Determination of ABA content
The ABA levels were determined by Nanjing Convinced-test Technology Co., Ltd. The leaves of the WT, FMO1-OE, and FMO1-Ri plants were sampled at 17 days from the DT and CK plants and ground to powder form in liquid nitrogen. The samples (1 g each) were transferred to 15 mL screw-cap tubes. Then, 8 μL (1 μg/mL) of ABA internal standards (Sigma, USA) were added to each 15 mL tube. ABA extraction and HPLC-ESI-MS/MS analysis (Q-EXACTIVE LC-MS System, USA) were performed as previously described [65].
The BiFC assays was conducted as previously described [31]. In brief, N. benthamiana leaves were infiltrated with GV3101 harboring the resulting FMO1: YFP N and SlCAT2: YFP C constructs that were mixed at a ratio of 1:1. After 48 h of infiltration, the samples were observed under a confocal laser microscope (Zeiss, LSM780, Germany).