Apple SERRATE negatively mediates drought resistance by regulating MdMYB88 and MdMYB124 and microRNA biogenesis

The function of serrate (SE) in miRNA biogenesis in Arabidopsis is well elucidated, whereas its role in plant drought resistance is largely unknown. In this study, we report that MdSE acts as a negative regulator of apple (Malus × domestica) drought resistance by regulating the expression levels of MdMYB88 and MdMYB124 and miRNAs, including mdm-miR156, mdm-miR166, mdm-miR172, mdm-miR319, and mdm-miR399. MdSE interacts with MdMYB88 and MdMYB124, two positive regulators of apple drought resistance. MdSE decreases the transcript and protein levels of MdMYB88 and MdMYB124, which directly regulate the expression of MdNCED3, a key enzyme in abscisic acid (ABA) biosynthesis. Furthermore, MdSE is enriched in the same region of the MdNECD3 promoter where MdMYB88/MdMYB124 binds. Consistently, MdSE RNAi transgenic plants are more sensitive to ABA-induced stomatal closure, whereas MdSE OE plants are less sensitive. In addition, under drought stress, MdSE is responsible for the biogenesis of mdm-miR399, a negative regulator of drought resistance, and negatively regulates miRNAs, including mdm-miR156, mdm-miR166, mdm-miR172, and mdm-miR319, which are positive regulators of drought resistance. Taken together, by revealing the negative role of MdSE, our results broaden our understanding of the apple drought response and provide a candidate gene for apple drought improvement through molecular breeding.


Introduction
Serrate (SE) in Arabidopsis is a conserved eukaryotic RNA processing factor that was first reported to mediate the formation of early juvenile leaves and phase length 1 . Encoding a C 2 H 2 zinc-finger protein, SE is required for normal shoot development 2 . Moreover, SE influences the alternative splicing of pre-mRNAs that primarily affect the selection of alternative 5′ splice sites of first introns 3 . Other genes with alternative splicing affected by SE encode transcription factors, splicing factors, and stressrelated proteins 3 . SE also functions in intron splicing and the transcription of intronless genes by pausing and elongating polymerase II complexes to promote their association with these intronless target genes 4,5 . Moreover, label-free quantitative proteomic analysis has revealed that SE is regulated by abscisic acid (ABA) under flooding stress 6 . In addition to these functions, SE has a role in microRNA (miRNA) biogenesis 7,8 , and previous studies report that SE and Hyponastic Leaves1 (HYL1) form a complex with DICER1 to achieve efficient and precise processing of pri-miRNAs 9 .
Drought stress is a major limiting factor that affects the yield and quality of apple. Researchers have long sought to increase the drought resistance of apple trees using molecular tools, such as genetic transformation and QTL mapping of loci associated with water use efficiency [10][11][12][13][14] .
To date, a number of genes have been reported to play positive or negative roles in apple drought resistance. For example, MdMYB88 and MdMYB124 are proven to be two positive regulators of apple drought stress that influence xylem formation and secondary cell wall deposition 11 . In addition, MdMYB88 and MdMYB124 bind to gene promoters containing the cis-element AACCG 11,13 to regulate expression. miRNAs are~20-25 nucleotide (nt) endogenous small molecules involved in various plant processes, including development and environmental stresses [15][16][17] . For example, the overexpression of miR156 improves the drought tolerance of alfalfa (Medicago sativa) by silencing SPL13 18 , a squamosa promoter-binding-like protein that binds to a core GTAC sequence in the promoter region of dihydroflavonol-4-reductase (DFR) to induce anthocyanin biosynthesis in response to drought stress 18,19 . miR393 is involved in the rice response to stress by targeting auxin receptors and plays negative roles in drought and salt stress 20 . In addition, transgenic Arabidopsis plants overexpressing miR399 exhibit hypersensitivity to drought but enhanced tolerance to salt stress and exogenously applied ABA 21 . In the apple genome, 23 conserved, 10 less conserved, and 42 apple-specific miRNAs or families with distinct expression patterns have been identified; these miRNAs target various genes and represent a wide range of enzymatic and regulatory activities 22 . Genome-wide miRNA analysis has revealed that 61 and 35 miRNAs are differentially expressed in drought-tolerant and droughtsensitive apple hybrid progeny, respectively, under drought stress 23 . Among these mdm-miRNAs, mdm-miR156 and mdm-miRn249 are two positive regulators of apple osmotic stress 23 .
ABA is a drought-induced phytohormone that plays important roles in plant responses to environmental stresses. Upon drought stress, ABA accumulates rapidly to promote stomatal closure and avoid water loss 24,25 . Exogenous ABA treatment effectively and sufficiently upregulates many stress-marker proteins in wheat and maize that are indicated to enhance drought tolerance 26,27 . ABA also acts as a signaling molecule in response to drought stress. Rice (Oryza sativa) OsPM1 (PLASMA MEMBRANE PROTEIN1) encodes an ABA influx carrier that mediates the movement of ABA across the plasma membrane and plays important roles in drought responses 28 . Under drought conditions, elevated ABA induces the production of H 2 O 2 in guard cells, and subsequent H 2 O 2 -activated Ca 2+ channels mediate the influx of Ca 2+ in intact guard cells to close stomata 29,30 .
In the current study, we provide evidence that MdSE participates in the drought resistance of apple by negatively regulating MdMYB88-and MdMYB124-mediated ABA homeostasis. MdSE also regulates the expression of miRNAs that play critical roles in drought resistance in apple. Our results highlight the roles of MdSE in the drought tolerance of apple and thereby provide genetic determinants for apple breeding.

MdSE interacts with and reduces the transcript and protein levels of MdMYB88 and MdMYB124
When applying affinity-purified mass spectrometry analysis to ascertain the interacting partners of MdMYB88 and MdMYB124, the SERRATE protein, which usually participates in miRNA biogenesis, pri-miRNA, and pre-mRNA splicing in Arabidopsis, was identified 3 . The interaction between MdSE and MdMYB88 or MdMYB124 was confirmed by BiFC analysis (Fig. 1a), which was further verified by a Co-IP assay (Fig. 1b). However, MdSE did not interact with MdMYB88 in yeast, as demonstrated by a yeast twohybrid analysis (Fig. S1), indicating that the physical interaction of MdSE with MdMYB88 or MdMYB124 may require another component.
The expression of MdSE was examined in MdMYB88 and MdMYB124 transgenic plants, which were generated previously 13 . qRT-PCR analysis revealed no regulation of MdSE by MdMYB88 or MdMYB124 under control or dehydration conditions (Fig. S2). To assess whether MdMYB88 or MdMYB124 expression levels are regulated by MdSE, MdSE RNAi and MdSE OE plants were generated. The transgenic plants were verified at the DNA and RNA levels (Fig. S3). After air dehydration for 2 h, transcripts of MdMYB88 or MdMYB124 were reduced dramatically in MdSE OE plants but increased in MdSE RNAi plants (Fig. 1c, d). Western blot analysis confirmed the downregulation of MdMYB88 and MdMYB124 by MdSE under drought (Fig. 1e), indicating that under drought conditions, MdSE decreases levels of MdMYB88 and MdMYB124 proteins.
Since SE is responsible for the alternative splicing of pre-mRNAs in Arabidopsis, we then examined the transcripts of MdMYB88 and MdMYB124 in MdSE RNAi plants by a RT-PCR assay. We found that decreased MdSE levels did not affect splicing of MdMYB88 and MdMYB124 in apple under control or drought conditions (Fig. S4).

MdSE subcellular localization and expression pattern
Protein alignment demonstrated that MdSE shares 67.2% sequence similarity with Arabidopsis SE and is more closely related to SERRATE from Prunus persica (Fig. S5). Based on a transient expression assay, the YFP-MdSE fusion protein was present in the nucleus of tobacco cells (Fig. 2a), consistent with the nuclear localization of SE in Arabidopsis 31 . SE from M. prunifolia was found to be expressed predominantly in flowers, followed by stems, leaves, and roots (Fig. 2b). The MdSE expression level was reduced in response to drought stress (Fig. 2c).

MdSE is a negative regulator of drought tolerance
To understand the biological function of MdSE in the drought response of apple, the drought tolerance of 5-month-old MdSE transgenic plants and nontransgenic GL-3 plants was examined. After withholding water for 30 days, and then rewatering for 7 days, 50-70% of MdSE RNAi plants survived, whereas only 40% of the control plants were alive (Fig. 3a, b). Water loss from MdSE RNAi plants was much lower than loss from GL-3 plants (Fig. 3c). Compared with GL-3 plants under drought for 18 days, MdSE RNAi plants had higher photosynthesis rates and water use efficiency (Fig. 3d, e).
When exposed to drought for 24 days and rewatered for 7 days,~30% of 5-month-old MdSE OE plants survived; in contrast, 50% of GL-3 plants remained alive (Fig. 3f, g), which suggested that MdSE OE plants are more sensitive to drought stress. In addition, MdSE OE plants lost significantly more water under dehydration (Fig. 3h), and MdSE OE plants had significantly lower photosynthesis rates and water use efficiency than GL-3 plants under drought (Fig. 3i, j). Together, these data suggest that MdSE negatively regulates apple drought resistance.  RNAi, and MdSE OE plants under control or drought conditions. Data are means ± SD (n = 3). Student's t test was performed, and statistically significant differences are indicated by *P < 0.05, **P < 0.01, or ***P < 0.001. e Protein level of MdMYB88 or MdMYB124 in GL-3, MdSE RNAi, and MdSE OE plants under control or drought conditions significantly higher in MdSE RNAi plants but lower in MdSE OE plants than in GL-3 plants (Fig. 4a). The ABA content was then measured in MdSE transgenic and GL-3 plants under control and drought conditions. LC-MS analysis showed that MdSE RNAi plants contained significantly more ABA but that MdSE OE plants contained less ABA in response to drought than GL-3 plants (Fig. 4b). Consistently, MdSE RNAi plants were hypersensitive to ABA-induced stomatal closure, whereas MdSE OE plants were less sensitive ( Fig. 4c-f).

MdSE is enriched in the MdNCED3 promoter
Regulation of the main biosynthetic pathway of ABA is mediated by NCED3, which cleaves 9-cis-epoxycarotenoids and produces in the active isomer of ABA 32 . Because the ABA content of MdSE transgenic lines was affected under drought conditions, we investigated whether MdSE influences the MdNECD3 expression level by associating with the MdNCED3 promoter. Chromatin immunoprecipitation (ChIP) experiments were carried out using an SE-specific antibody, followed by qRT-PCR with the same primers used to assess MdMYB88 and MdMYB124 binding activity to MdNCED3. The ChIP-qPCR results showed enrichment of MdSE at the MdNCED3 promoter in the same region where MdMYB88 and MdMYB124 bind (Fig. 5a, b). A dual luciferase reporter assay was also used to detect the influence of MdSE on MdNCED3 expression. The results showed that MdMYB88 and MdMYB124 enhanced MdNCED3 expression under both normal and dehydration conditions, whereas MdSE reduced the expression of MdNCED3. When MdSE was present, the expression of MdNCED3 induced by MdMYB88 or MdMYB124 was attenuated under control and drought conditions ( Fig. 5c-f). These results suggest that the regulation of MdNCED3 by MdSE depends on its association with MdMYB88 and MdMYB124.

MdSE regulates the biogenesis of miRNAs in apple under drought
Arabidopsis SE is required for the biogenesis of miRNA 33,34 . Loss-of-function mutant plants of SE have upward curling and radialized leaves 35 ; secondary inflorescences lack an associated cauline leaf, and inflorescences often produce siliques that emerge from the same node on the stem 34 . Curly leaves on MdSE RNAi plants were not observed (Fig. 3), which may be due tõ 30-50% MdSE being maintained in MdSE RNAi plants (Fig. S3A). To understand whether MdSE has a similar function to Arabidopsis SE in miRNA biogenesis, we analyzed the expression of drought-responsive miRNAs, including mdm-miR156, mdm-miR166, mdm-miR172, mdm-miR319, and mdm-miR399, using stem-loop qPCR. Mdm-miR156, miR166, miR172, and miR319 are positive regulators of apple osmotic stress 36 and drought resistance in alfalfa 18 , Arabidopsis 37,38 , and creeping bentgrass 39 , though miR399 is a negative regulator of drought resistance 21 . The expression levels of these miRNAs were reduced in MdSE RNAi plants under control conditions (Fig. 6), suggesting a similar role for MdSE and SE in miRNA biogenesis. Under drought stress, compared with GL-3 plants, the expression levels of mdm-miR156, mdm-miR166, mdm-miR172, and mdm-miR319 increased in MdSE RNAi plants, whereas mdm-miR399 expression was reduced (Fig. 6), consistent with the drought tolerance phenotype of MdSE RNAi plants. These data indicate that MdSE has a similar function in miRNA biogenesis as Arabidopsis SE and negatively modulates drought through regulation of drought-responsible miRNAs.

Discussion
In this study, we characterized a protein interacting with MdMYB88 and MdMYB124, MdSE, in response to drought. MdSE plays a negative role in apple drought resistance by negatively modulating the expression of MdMYB88 and MdMYB124, leading to the downregulation of MdNCED3 and reduced ABA levels. Furthermore, MdSE regulates the expression of droughtresponsive miRNAs under drought stress, which may contribute to its negative role in drought resistance.
The interaction between MdSE and MdMYB88 or MdMYB124 was confirmed by BiFC and Co-IP analyses. However, an in vitro interaction was not verified by Y2H analysis, indicating the possibility of a bridge between MdSE and MdMYB88. For example, the interaction of Arabidopsis C-terminal domain phosphatase-like 1 . Student's t test was performed, and statistically significant differences are indicated by *P < 0.05, **P < 0.01, or ***P < 0.001 (CPL1) and HYL1 requires SE as a bridge 31 . MdHYL1 is a putative candidate for bridging MdSE and MdMYB88 or MdMYB124. Indeed, we found that MdHYL1 interacts with MdSE in Y2H analysis (Fig. S6). The interaction between SE and HYL1 was also observed in Arabidopsis 33 , indicating that the functions of SE and HYL1 in some plant processes are conserved among plant species. Factors other than MdHYL1 might act as a bridge between MdSE and MdMYB88 or MdMYB124, but further study is necessary. SE is responsible for alternative splicing of pre-mRNAs 3 . In addition to the regulation of MdMYB88 and MdMYB124 transcripts by MdSE (Fig. 1), we hypothesized that MdSE might modulate the alternative splicing of MdMYB88 and MdMYB124. However, RT-PCR results showed that reduced MdSE levels did not affect splicing of MdMYB88 and MdMYB124 in apple under control and drought conditions (Fig. S4). It is possible that MdSE can affect alternative splicing of other genes under control or drought conditions. The only challenge of . e, j Intrinsic water use efficiency (WUEi). Data are means ± SD (n = 15). Student's t test was performed, and statistically significant differences are indicated by *P < 0.05, **P < 0.01, or ***P < 0.001 alternative splicing detection in apple is that 30-50% of MdSE was still functional in MdSE RNAi plants (Fig.  S3A), which might affect the accuracy of analyses.
We demonstrated that MdSE is a negative regulator of drought resistance. First, survival ability analysis suggested that MdSE RNAi plants were more tolerant to drought stress but that MdSE OE plants were more sensitive (Fig. 3). Second, under dehydration conditions, MdSE OE plants lost water more quickly than MdSE RNAi plants (Fig. 3c, h). Third, under drought stress, the cell membranes of MdSE RNAi plants were less damaged, as indicated by ion leakage (Fig. 3). Fourth, MdSE negatively regulated ABA accumulation and some droughtpositive miRNAs (Figs. 4 and 6a-d). In addition, MdSE positively regulated drought-negative mdm-miR399. Fifth, Arabidopsis se-1 mutants were also more tolerant to drought stress (Fig. S7). All these data support that MdSE plays a negative role in apple drought resistance and that the role of SE under drought might be conserved among plant species. Data are the means ± SD; 5 leaves were used, and at least 80 stomatal apertures were measured for each treatment. Student's t test was performed, and statistically significant differences are indicated by *P < 0.05, **P < 0.01, or ***P < 0.001 ABA is a plant stress hormone that regulates stomatal closure within and outside guard cells through combinational mechanisms 40 . In Arabidopsis, maize, wheat, rice, and apple, elevated ABA content contribute to plant drought tolerance by inducing stomatal closure and stress-related signal transduction 21,23,24,26,27,41,42 . NCED3 is considered to be the key contributor to ABA production under water deficit conditions, and ZEP and AAO3 . c-f Relative luciferase activity from dual luciferase reporter assays in N. benthamiana leaves. Pro35S::REN was used as an internal control. Quantification was performed by normalizing firefly luciferase activity to that of Renilla luciferase. Leaves of N. benthamiana were coinfiltrated and grown for 72 h, and then leaves were collected or dehydrated for an additional 2 h. Error bars indicate standard deviation (n = 10). Student's t test was performed, and statistically significant differences are indicated by *P < 0.05 or **P < 0.01 play minor roles 43,44 . In Arabidopsis, short vegetative phase (SVP) is able to bind to the promoters of the ABA catabolism pathway genes CYP707A1, CYP707A3, and AtBG1, and thus contributes to ABA homeostasis 45 . Homeostasis of ABA was also regulated by reversible glycosylation mediated by ABA-UGTs (uridine diphosphate glucosyltransferases) to affect ABA bioactivity (Fig.  S8) 46,47 . In our study, more ABA accumulated in MdSE RNAi transgenic lines compared with MdSE OE and GL-3 plants (Fig. 4b). Such negative regulation of ABA content should contribute to the negative role of MdSE in drought resistance. We also found that MdSE was enriched at the promoter region of MdNCED3, which is the same region bound by MdMYB88 and MdMYB124 (Fig. 5). Considering the in vivo interaction between MdSE and MdMYB88 or MdMYB124, we conclude that the enrichment of MdSE at the MdNCED3 promoter was due to recruitment by MdMYB88 or MdMYB124 instead of direct binding.
In summary, our study elucidated the roles of MdSE in drought stress resistance. MdSE plays a negative role in drought resistance by affecting miRNA biogenesis and negatively regulating protein accumulation of MdMYB88 and MdMYB124, which results in negative regulation of ABA accumulation. Our study provides a deeper understanding of the complex mechanism of MdSE in response to drought stress and identifies a candidate gene for drought improvement through molecular breeding.

Plant materials and growth conditions
For gene cloning, "Golden delicious" (Malus × domestica) grown in a greenhouse was used for RNA extraction. GL-3, a genotype selected from seedlings of "Royal Gala" (Malus × domestica), was used for genetic transformation 53

RNA extraction and qRT-PCR analysis
Detailed methods for RNA extraction and qRT-PCR analysis are provided in ref. 13 . The primers used for qRT-PCR analysis are listed in Supplementary Table 1.

Generation of transgenic apple
To generate a construct for MdSE overexpression, the coding region (CDS) of MdSE was cloned into pGWB414 to produce MdSE-pGWB414. To knock down MdSE, a 292-bp fragment of MdSE was introduced into pK7WIWG2D, resulting in MdSE-pK7WIWG2D. Both plasmids were transformed into Agrobacterium strain EHA105. For genetic transformation, we used an Agrobacterium-mediated transformation method. Plant transformation was carried out according to Dai et al. 53 . Briefly, 4-week-old GL-3 leaves were cut into strips in liquid MS medium (4.43 g/L MS salts and 30 g/L sucrose, pH 5.2) with EHA105 (OD 600 = 0.6-0.9) carrying the relevant plasmid for 15 min. Then, the leaf strips were transferred into maintenance medium (4.43 g/L MS + 2 mg/L TDZ + 0.5 mg/L NAA + 100 μM acetosyringone +1 mM betaine +7.5 g/L agar +30 mg/L sugar, pH = 5.8). After 3 days, leaf strips were transferred into selection medium (4.43 g/L MS + 2 mg/L TDZ + 0.5 mg/L NAA + 250 mg/L cefotaxime +50 mg/L kanamycin +7.5 g/L agar +30 mg/L sugar, pH = 5.8) for 4 weeks in the dark, and then incubated for 6 weeks under light. The transgenic buds that stayed green on selection medium were grown for 4 weeks. DNA and RNA were extracted from the transgenic plants and GL-3 and used to detect transgene insertion and MdSE expression levels by PCR and RT-qPCR, respectively. Transgenic plants with transgene insertion, as well as altered expression levels of MdSE, were selected for further experiments. The primers used are listed in Supplementary Table 1.

Drought treatment
Drought treatment was carried out by withholding water for a certain number of days, and then rewatering for 7 days, followed by calculation of the survival rate. Specifically, 5-month-old MdSE OE and GL-3 plants were treated with drought for 24 days, and then rewatered for 7 days. MdSE RNAi and GL-3 plants were treated with drought for 30 days, and then rewatered for 7 days. To obtain photosynthesis data (the rate of photosynthesis and intrinsic water use efficiency), a LiCor-6400 portable photosynthesis system (LiCor) was used.
Detached leaves from 5-month-old MdSE OE, MdSE RNAi, and GL-3 were used for the water loss assay.

Stomatal aperture measurements
For stomatal aperture measurements, we used leaves of 2-month-old soil-grown transgenic apple and GL-3 plants. Leaves were cut off and plunged into stomatal opening solution (30 mM KCl, 0.1 mM CaCl 2 , and 10 mM MES-KOH, pH 6.15) under light (120 μmol m −2 s −1 ) for 2 h to induce stomatal opening as described 54 . Then, ABA was added to the stomatal opening solution to a final concentration of 5 μM. The leaf epidermis was observed for stomatal aperture with an EX30 microscope (SDPTOP) after ABA treatment for 1 or 2 h. Stomatal length and width were measured by ImageJ software, and stomatal aperture was then calculated.

Subcellular localization, BiFC, and Co-IP assays
To generate constructs for BiFC assays, we cloned the CDS of MdMYB88 and its paralog gene MdMYB124 into pSPYNE-35S; the CDS of MdSE was cloned into the pSPYCE-35S vector. For subcellular localization, the CDS of MdSE was cloned into the pEearleyGate104 vector. Transient expression assays were performed according to Xie et al. 13 . After 3 days, fluorescent signals in transformed tobacco leaves were then detected using a Nikon A1R/A1 confocal microscope (Nikon).
For Co-IP analysis, the CDS of MdMYB88 was amplified by PCR and cloned into pEarleyGate 101; the CDS of MdSE was cloned into pEarleyGate 203. Co-IP analyses were performed as described previously 55 .
The primers used are listed in Supplementary Table 1.

ChIP-qPCR
ChIP-qPCR assays were performed as described previously 13 . Tissue-cultured GL-3 was used for crosslinking, and the ChIP assay was performed with an anti-SE antibody (Agrisera, AS09 532A). Three regions of the MdNCED3 promoter were examined by qPCR, with no antibody ChIP samples serving as the reference. The primers used for ChIP-qPCR are listed in Supplementary  Table 1. ABA measurement ABA was extracted as described 56 . Frozen apple leaf samples (about 100 mg fresh weight) was ground in liquid nitrogen, and then extracted with 1 ml of cold extraction buffer (methanol:isopropanol:acetic acid = 20:79:1, v/v/v). After centrifugation at 4°C and 12,000 rpm for 10 min, the supernatant was transferred into a 2 mL tube, and 500 μL cold extraction buffer was added followed by vortexing for 5 min. The extraction process was repeated three times followed by centrifugation at 4°C and 12,000 rpm for 10 min. The supernatant was filtered through a 0.22 μm PTFE filter (Waters, Milford, MA, USA). GC-MS analysis was carried out using a QTRAP ® 5500 LC-MS/MS (AB SCIEX, Redwood City, USA).