SUPPRESSOR OF MAX2 LIKE 6, 7, and 8 Interact with DDB1 BINDING WD REPEAT DOMAIN HYPERSENSITIVE TO ABA DEFICIENT 1 to Regulate the Drought Tolerance and Target SUCROSE NONFERMENTING 1 RELATED PROTEIN KINASE 2.3 to Abscisic Acid Response in Arabidopsis

SUPPRESSOR OF MAX2-LIKE 6, 7, and 8 (SMXL6,7,8) function as repressors and transcription factors of the strigolactone (SL) signaling pathway, playing an important role in the development and stress tolerance in Arabidopsis thaliana. However, the molecular mechanism by which SMXL6,7,8 negatively regulate drought tolerance and ABA response remains largely unexplored. In the present study, the interacting protein and downstream target genes of SMXL6,7,8 were investigated. Our results showed that the substrate receptor for the CUL4-based E3 ligase DDB1-BINDING WD-REPEAT DOMAIN (DWD) HYPERSENSITIVE TO ABA DEFICIENT 1 (ABA1) (DWA1) physically interacted with SMXL6,7,8. The degradation of SMXL6,7,8 proteins were partially dependent on DWA1. Disruption of SMXL6,7,8 resulted in increased drought tolerance and could restore the drought-sensitive phenotype of the dwa1 mutant. In addition, SMXL6,7,8 could directly bind to the promoter of SUCROSE NONFERMENTING 1 (SNF1)-RELATED PROTEIN KINASE 2.3 (SnRK2.3) to repress its transcription. The mutations in SnRK2.2/2.3 significantly suppressed the hypersensitivity of smxl6/7/8 to ABA-mediated inhibition of seed germination. Conclusively, SMXL6,7,8 interact with DWA1 to negatively regulate drought tolerance and target ABA-response genes. These data provide insights into drought tolerance and ABA response in Arabidopsis via the SMXL6,7,8-mediated SL signaling pathway.

ABA and SLs are two carotenoid-derived phytohormones that positively regulate drought tolerance in plants.ABA hypersensitivity and stronger drought tolerance were observed in the smxl6/7/8 triple mutant, and drought-stress-inducible ABA-response genes were also up-regulated in smxl6/7/8, indicating that SMXL6,7,8 may participate in ABA-mediated drought response through transcriptional regulation of ABA response genes [31,32].Chromatin immunoprecipitation sequencing (ChIP-seq) of SMXL6-HA had been reported, and many potential downstream target genes of SMXL6 were found by Wang et al. [30].These data led us to ask whether SMXL6,7,8, as transcription factors, participate in drought stress and ABA response through binding to downstream target genes.However, the underlying mechanism is not completely clear, and it is challenging to explore a new regulatory pathway that contributes to drought tolerance and ABA response in plants.Are SMXL6,7,8 involved in the regulation of drought tolerance by interacting with other proteins?Do SMXL6,7,8, as transcription factors, regulate drought tolerance and ABA response by inhibiting the transcription of ABA/drought response genes?These issues are not clear.
In this study, we screened a protein (DWA1) interacting with SMXL6,7,8 that promoted the degradation of SMXL6,7,8 in a 26S proteasome-dependent pathway.Furthermore, SMXL6,7,8 directly bound to the promoter of SnRK2.3 to repress its transcription and then its response to ABA-mediated seed germination.Accordingly, this work further reveals a possible molecular mechanism by which SMXL6,7,8 regulate drought tolerance and ABA response in plants.

Drought Stress and Relative Water Content (RWC) Analysis
Drought stress was induced with modification [31].In brief, seedlings were grown on 1 /2 MS medium (containing 1% sucrose and 1% agar) for 7 days under a 16-hour-light/ 8-hour-dark photoperiod at 22 • C. A total of 16 seedlings were transplanted into a small pot containing the same weight of soil, and these continued to grow for 3 weeks.Then, water was withheld from the plants until a phenotype appeared and was photographed.The plants were photographed again after 3 days of rewatering.The survival rates of multiple drought experiments were calculated.For relative water content (RWC) analysis, 30 rosette leaves from 4-week-old plants were detached and weighed at the indicated times [17].The experiment was performed with at least three biological replicates and at least three times with similar results.

Yeast Two-Hybrid (Y2H) and Yeast One-Hybrid (Y1H) Assays
To generate constructs fused to the GAL4 DNA-binding domain (BD) and activation domain (AD), the full-length coding sequences were cloned into the EcoRI-BamHI sites of the pGBKT7 and pGADT7 vectors, respectively.The primers and cloning sites used for plasmid construction are given in Table S1.The yeast assays were performed as described previously with modifications [38].The yeast transformation and two-hybrid library screen were performed according to the manufacturer's instructions (Clontech, Mountain View, CA, USA, https://www.takarabio.com/(accessed on 5 November 2018)).Reagents used in yeast experiments were from Clontech and Solarbio.The cDNA library was from Clontech (Make Your Own "Mate & Plate" Library System, Code No. 630490).The Y2H library was screened via the yeast-mating method.Briefly, pGBKT7-SMXL6, pGBKT7-SMXL7, and pGBKT7-SMXL8 constructs were transformed into Y2H Gold as bait.The library strains were combined with bait strains after autoactivation and toxicity tests.Subsequently, the candidate interacting proteins of SMXL6, SMXL7, and SMXL8 were screened using yeast two-hybrid libraries, and then all positive clones were sequenced.

Bimolecular Fluorescence Complementation (BiFC) Assay
The coding regions of SMXL6,7,8 and DWA1 were cloned next to the N-terminal part of YFP (nYFP) in the pEarleygate202-YN vector and the C-terminal part of YFP (cYFP) in the pEarleygate201-YC vector.Different constructs (including empty vectors) were transformed into the Agrobacterium strain GV3101 and then infiltrated into Nicotiana benthamiana leaves for transient expression.Fluorescence was observed and photographed in leaf epidermal cells 2 days after infiltration using a Confocal laser scanning microscope (Nikon, A1R+Ti2-E, Tokyo, Japan).
For the fluorescence observation of the OE-SMXL6/7/8 and OE-SMXL6/7/8/dwa1 transgenic lines, these seeds were sown in the same 1  2 MS medium plate and subjected to a 16-hour-light/8-hour-dark photoperiod and 60% relative humidity at 22 • C in a greenhouse.
Fluorescence observation was performed after 5 days.The seedlings of two transgenic lines, such as OE-SMXL6 and OE-SMXL6/dwa1, were placed under the same slide for observing using the layer scanning method with a confocal laser scanning microscope (Nikon, A1R + Ti2-E, Tokyo, Japan).Then, multiple-layer fluorescent images were superimposed into a single image using a stacking technique in Photoshop.Finally, the overall fluorescence in the multiple roots of the two lines were counted.

Toluidine Blue Staining
Toluidine blue (TB) staining was conducted as described previously with modification [40].The fifth leaves of 4-week-old plants were detached, immersed into a solution of 0.05% (w/v) TB, and softly shaken for 3 h.The excessive TB was washed with water, and then the leaves were photographed.

Quantitative Real-Time PCR (qRT-PCR) Assay
qRT-PCR was performed as described previously with modifications [31].Briefly, the total RNA of 10-day-old wild type, smxl6/7/8, OE-SMXL6, OE-SMXL7, and OE-SMXL8 seedlings were prepared using a MiniBEST Plant RNA Extraction Kit (Takara, Beijing, China) according to the users' manual.RNA samples were reverse-transcribed using a PrimeScript TM II 1st Strand cDNA Synthesis Kit (Takara).The qRT-PCR experiments were finished using gene-specific primers (Table S1) on a real-time system (ABI Q5) in a total volume of 20 µL containing 8 µL diluted cDNA, 0.5 µM gene-specific primers, and 10 µL ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China).The Arabidopsis ACTIN2 gene was used as the internal control.All experiments were set up with three biological replicates and three technical replicates.

Transient Expression Assays and Dual-Luciferase Reporter Assay
Transient expression in the Arabidopsis mesophyll cells was performed accordingly [41].The coding region of SMXL6, SMXL7, and SMXL8 were each cloned to pGreenII 62-SK vectors containing the 35S promoter.The pGreenII 62-SK-GFP was used as a control.To generate the pGreenII0800-LUC-SnRK2.3/2.6, the 2000 bp sequence upstream of the ATG start codon was amplified via PCR using genomic DNA as the template and the pair of primers (Table S1).
The dual-luciferase reporter assay was performed as described previously [30].pGreenII 62-SK-GFP or pGreenII 62-SK-SMXL6,7,8 were co-transformed into Arabidopsis protoplasts with pGreenII 0800-LUC-SnRK2.3/2.6,respectively.After overnight incubation in the dark, the protoplasts were centrifuged.The transient expression vector pGreenII 0800-LUC construct used in our experiment contains a REN gene controlled by a 35S promoter to provide an estimate of the extent of transient expression.The Renilla luciferase gene (REN) was used as an internal control.A Dual-Luciferase Reporter Assay Kit (Promega, Madison, WI, USA) was used to detect firefly luciferase (LUC) activity and the Renilla luciferase gene (REN) using a GloMax ® 20/20 Luminometer (Promega).The relative luciferase activity was expressed as Firefly Luc/Renilla Luc.

Electrophoretic Mobility Shift Assay (EMSA)
EMSAs were performed as described previously with minor modifications [30].The sequences of the biotin-labeled probes are listed in Table S1.In brief, 1 µg of purified GST or GST-SMXL6,7,8 proteins were incubated together with biotin-labeled probes in 20 µL reaction mixtures, with purified GST protein used as a competitive cold probe.The binding reactions were allowed to proceed at room temperature for 30 min and were halted by adding the 5 µL 5× loading buffer, and then they were separated with 6% native polyacrylamide gel in pre-cooled 0.5× TBE (Tris-base, Boric Acid, EDTA) buffer.

Statistical Analysis
Three independent experiments were performed, and results were expressed as the mean ± the standard deviation (SD).Data were analyzed by using GraphPad Prism 7 software.Statistical analyses were performed in SPSS Statistics (IBM, Tulsa, OK, USA).We used a one-way analysis of variance (ANOVA) with Duncan's multiple range tests and a Student's t-test.

The Degradation of SMXL6,7,8 Proteins Partially Depend on DWA1
DWA1 acts as the substrate receptor of the CUL4-based E3 ligase complex, which functions in ubiquitination [12].Therefore, we tested whether the stability of the SMXL6,7,8 proteins were affected by DWA1.GST-tagged SMXL6,7,8 recombinant proteins were incubated with protein extracts prepared from the Col-0 and dwa1-1 in the presence of ATP and analyzed via immunoblotting.Cell-free degradation assays showed that the recombinant GST-SMXL6, GST-SMXL7, and GST-SMXL8 proteins were degraded in the wild-type extracts, and their degradation was partially blocked by MG132, a proteasomespecific inhibitor (Figure 2).However, the GST-SMXL6, GST-SMXL7, and GST-SMXL8 proteins were degraded more slowly in dwa1-1 than in the wild-type extracts (Figure 2), suggesting that the degradations of SMXL6,7,8 are partially dependent on DWA1.The full-length SMXL6,7,8 proteins were fused to the GAL4 binding domain (BD), and the DWA1 protein was fused to the GAL4 activation domain (AD).DDO: SD/-Leu/-Trp; QDO: SD/-Ade/-His/-Leu/-Trp; X-α-Gal: 5-Bromo-4-chloro-3-indoxyl-α-D-galactopyranoside; AbA: Aureobasidin A (a cyclic, depsipeptide antifungal antibiotic).(b) Pull-down assay for detecting the interaction between GST-SMXL6,7,8 and His-DWA1.GST, GST-SMXL6,7,8, and His-DWA1 were expressed in the Rosetta (DE3) E. coli strain.Purified proteins were used for the pull-down assay.GST and GST-SMXL6,7,8 were bound to GST beads as baits and His-DWA1 was used as prey.The His-DWA1 protein was detected with anti-His antibodies, and the GST and GST-SMXL6,7,8 proteins were detected with anti-GST antibodies.The GST protein was used as a control."+" indicates that a relative protein has been added in lane, "−" indicates that no protein has been added.(c) Bimolecular fluorescence (BiFC) assay of the interactions between SMXL6,7,8 and DWA1.N. benthamiana leaves co-expressing SMXL6,7,8-nYFP and DWA1-cYFP fusion proteins were observed 2 days after infiltration.Co-expression of SMXL6,7,8-nYFP and cYFP, nYFP, and DWA1-cYFP or SMXL3-nYFP and DWA1-cYFP served as negative controls.The YFP signals were observed using confocal microscopy, and green pseudo-colour was used to indicate the YFP fluorescence signal generated by the interaction of the two proteins.Scale bars = 20 µm.The experiments were repeated three times with similar results.

The Degradation of SMXL6,7,8 Proteins Partially Depend on DWA1
DWA1 acts as the substrate receptor of the CUL4-based E3 ligase complex, which functions in ubiquitination [12].Therefore, we tested whether the stability of the SMXL6,7,8 proteins were affected by DWA1.GST-tagged SMXL6,7,8 recombinant proteins were incubated with protein extracts prepared from the Col-0 and dwa1-1 in the presence of ATP and analyzed via immunoblotting.Cell-free degradation assays showed that the recombinant GST-SMXL6, GST-SMXL7, and GST-SMXL8 proteins were degraded in the wild-type extracts, and their degradation was partially blocked by MG132, a proteasomespecific inhibitor (Figure 2).However, the GST-SMXL6, GST-SMXL7, and GST-SMXL8 proteins were degraded more slowly in dwa1-1 than in the wild-type extracts (Figure 2), suggesting that the degradations of SMXL6,7,8 are partially dependent on DWA1.GST and GST-SMXL6,7,8 were bound to GST beads as baits and His-DWA1 was used as prey.The His-DWA1 protein was detected with anti-His antibodies, and the GST and GST-SMXL6,7,8 proteins were detected with anti-GST antibodies.The GST protein was used as a control."+" indicates that a relative protein has been added in lane, "−" indicates that no protein has been added.To further verify these data, we first generated OE-SMXL6, OE-SMXL7, and OE-SMXL8 transgenic plants in a wild-type background, and the expressions of SMXL6,7,8 were verified with a qRT-PCR assay (Figure 3a-c).The results of the Western blot showed that the levels of SMXL6, SMXL7, and SMXL8 proteins were higher in the OE-SMXL6-4 OE-SMXL7-1, and OE-SMXL8-2 lines, respectively (Figure S1), which was consistent with the results of the qRT-PCR.Next, OE-SMXL6-4, OE-SMXL7-1, and OE-SMXL8-2 lines with higher expression levels were introduced into the dwa1-1 background via genetic crossing Then, the SMXL6-GFP, SMXL7-GFP, and SMXL8-GFP fluorescence of the root tips were observed.The results showed that the fluorescence intensity in the OE-SMXL6/dwa1-1 OE-SMXL7/dwa1-1, and OE-SMXL8/dwa1-1 plants was higher than in the OE-SMXL6, OE-SMXL7, and OE-SMXL8 plants (Figure 3d-i), suggesting that the degradations of SMXL6,7,8 are at least partially dependent on the DWA1 protein.To further verify these data, we first generated OE-SMXL6, OE-SMXL7, and OE-SMXL8 transgenic plants in a wild-type background, and the expressions of SMXL6,7,8 were verified with a qRT-PCR assay (Figure 3a-c).The results of the Western blot showed that the levels of SMXL6, SMXL7, and SMXL8 proteins were higher in the OE-SMXL6-4, OE-SMXL7-1, and OE-SMXL8-2 lines, respectively (Figure S1), which was consistent with the results of the qRT-PCR.Next, OE-SMXL6-4, OE-SMXL7-1, and OE-SMXL8-2 lines with higher expression levels were introduced into the dwa1-1 background via genetic crossing.Then, the SMXL6-GFP, SMXL7-GFP, and SMXL8-GFP fluorescence of the root tips were observed.The results showed that the fluorescence intensity in the OE-SMXL6/dwa1-1, OE-SMXL7/dwa1-1, and OE-SMXL8/dwa1-1 plants was higher than in the OE-SMXL6, OE-SMXL7, and OE-SMXL8 plants (Figure 3d-i), suggesting that the degradations of SMXL6,7,8 are at least partially dependent on the DWA1 protein.
We then investigated whether SMXL6,7,8 and DWA1 function together in response to drought stress.Under drought stress conditions, the dwa1-1 and dwa1-2 plants showed drought sensitivity compared with the wild type (Figures 4f, S2c and S3), while the smxl6/7/8 triple mutant plants displayed an enhanced drought tolerance (Figures 4f and S3).dwa1 had a lower survival rate and smxl6/7/8 had a higher survival rate than the wild type when deprived of water for 16 days (Figures 4g and S3).Simultaneously, the OE-SMXL6,7,8 lines were sensitive to drought stress (Figure S4).This proved that DWA1 positively regulates plant drought tolerance, while SMXL6,7,8 negatively regulates plant drought tolerance.The drought tolerance of two smxl6/7/8/dwa1 quadruple mutants was similar to that of the smxl6/7/8 triple mutant plants when deprived of water for 16 days (Figures 4f and S3), while the drought tolerance of smxl6/7/8/dwa1-1 was slightly weaker than that of smxl6/7/8 when deprived of water for 18 days (Figure S3).In addition, the relative water content of the dwa1-1 mutant was lower than that of the wild type.In contrast, the relative water content of the smxl6/7/8 mutant was higher than that of the wild type, and one of the smxl6/7/8/dwa1 quadruple mutants was lower than that of smxl6/7/8 (Figure 4h).It is well known that the permeability of the cuticle is also an important indicator of drought susceptibility [42].TB staining was used to visualize the potential differences of cuticle permeability among the lines described above.Compared with that of the wild type, the TB staining of dwa1 was enhanced, while that of the smxl6/7/8 triple mutant was reduced, and that of the smxl6/7/8/dwa1 mutants was similar to that of smxl6/7/8, which was consistent with the drought phenotype (Figure S5).Taken together, these findings indicate that SMXL6,7,8 act downstream of DWA1 to negatively regulate drought tolerance.A recent study reported that SMXL6,7,8 function as transcription factors by binding to the promoters and inhibiting the transcription of genes encoding downstream transcription factors.Deep-clustering GO enrichment analysis of previously published chromatin immunoprecipitation sequencing (ChIP-seq) data revealed 729 genes targeted by SMXL6-HA [30], and the results showed that SMXL6-regulated candidate genes were enriched for the pathways such as 'response to the ABA', 'response to water deprivation', and 'response to osmotic stress' (Figure S6a).Here, we selected some interesting genes in 'response to the ABA' to verify whether they could function as target genes of SMXL6,7,8 (Figure S6b).We then found that SnRK2.3, a member of ABA signaling, may be a candidate target gene for SMXL6 (Figure S6b).
3.5.SnRK2.2/2.3Act Downstream of SMXL6,7,8 in Response to ABA Previous reports showed that SnRK2.2 and SnRK2.3 had functional redundancy and that the snrk2.2/2.3 mutant was more insensitive to ABA than the wild type, snrk2.2, and snrk2.3single mutant during seed germination, while smxl6/7/8 was more sensitive to ABA than the wild type [7,31].Given that SMXL6,7,8 bound the SnRK2.3promoter and repressed its expression (Figure 5), we then asked whether SMXL6,7,8 were genetically related to SnRK2.3 to mediate ABA-induced repression of seed germination.Subsequently, we generated the smxl6/7/8/snrk2.3 6).In the absence of ABA, all genotypes exhibited no obvious differences in seed germination, but two concentrations of ABA treatments (0.5 µM and 1 µM) significantly inhibited seed germination (but to varying degrees).Consistent with previous findings, we also found that the germination rate of smxl6/7/8 was much lower than that of the wild type, whereas that of snrk2.2/2.3 was higher than both the wild type and snrk2.3.As expected, the smxl6/7/8/snrk2.2/2.3 was also insensitive to ABA during seed germination compared to the wild type, which was similar to the snrk2.2/2.3 seeds.In addition, the germination rate of smxl6/7/8/snrk2.3 was higher than that of smxl6/7/8, which was similar to that of the snrk2.3seeds.That is, the mutation of SnRK2.2/2.3 repressed the effect of ABA on the seed germination in smxl6/7/8, indicating that SnRK2.2 and SnRK2.3 are functionally redundant and are located downstream of SMXL6,7,8 to participate in the ABA response.The statistics of seed germination rate corresponding to the treatment under (a).Seed germination was recorded for the indicated times after sterilization on a medium supplemented with 0.5 µM and 1 µM ABA, respectively.The experiments described above were performed at least three times.For each biological replicate, we examined the seeds (more than 100) from the same batch three times as technical replicates.

Discussion
It has been shown that SLs have positive regulatory roles in plant adaptation to drought stress [17].The smxl6/7/8 triple mutant exhibited stronger drought tolerance with high ABA levels and upregulated drought-stress-inducible ABA-response genes [31].In this study, the CUL4-based E3 ligase DWA1 physically interacted with SMXL6,7,8 in vivo and in vitro (Figure 1).DWA1 acts as a substrate receptor of CUL4-based E3 ligase that interacts with DWA2 and ABI5 to regulate ABA signaling by promoting the degradation of ABI5 [12].We also assayed whether SMXL6,7,8 could directly interact with DWA2 and ABI5.However, the Y2H results showed that SMXL6,7,8 could interact with neither DWA2 nor ABI5 in yeast (Figure S11), indicating that SMXL6,7,8 specifically interacted with DWA1.
Ubiquitination is one of the most important post-translational protein modifications in eukaryotes and plays a key role in several cellular processes, including 26S proteasomedependent protein degradation.In the CUL4-based E3 ligase complex, the substrate adapter DDB1 interacts with substrate receptors and binds to a large number of proteins to form the CUL4-DDB1 complex [46].In turn, the substrate receptors interact with specific substrates that are targeted for degradation.The CUL4-based E3 ligase targets are key regulators of ABA responses.Another CUL4-DDB1 complex containing ABAhypersensitive DCAF1 (ABD1) acts as a substrate receptor by directly binding to ABI5 and positively affecting its degradation [47].DET1-DDB1-ASSOCIATED 1 (DDA1), as part of the CDDD (COP10-DET1-DDB1-DDA1) substrate adaptor module, targets the ABA receptor PYL8 for degradation [48].As a DWD protein, RNA EXPORT FACTOR 1 (AtRAE1) may act as a substrate receptor of CUL4-DDB1 E3 ligase and participates in ABA signaling by regulating the degradation of the ABA receptor PYL9 [49].In this study, DWA1 physically interacted with SMXL6,7,8, and the mutation of DWA1 partially delayed the degradation of SMXL6,7,8 both in vitro and in OE-SMXL6,7,8/dwa1 transgenic plants (Figures 1-3).Although we have preliminarily demonstrated that SMXL6,7,8 are partially dependent on DWA1 to be degraded, how the degradations of SMXL6,7,8 are triggered by DWA1 needs to be explored next.Previous studies have shown that the degradation of SMXL6,7,8 are dependent on MAX2 and that the GRKT motif is essential for the MAX2mediated degradation of SMXL6,7,8 [23,24].DWA1 and MAX2 show similar functions in the degradation of SMXL6,7,8 proteins and plant drought tolerance.Is there a possibility that DWA1 is a component of the SCF MAX2 complex to regulate the stability of SMXL6,7,8?Does mutation of the GRKT motif affect DWA1-mediated degradation of SMXL6,7,8?There needs to be further study using genetic and biochemical experiments, which will provide insight into the effect of DWA1 on the MAX2-mediated degradation of SMXL6,7,8.
Ubiquitin-conjugating enzymes and E3 ligases can negatively or positively participate in drought stress response.The endoplasmic reticulum-associated degradation (ERAD) component UBIQUITIN-CONJUGATING ENZYME 32 (UBC32) positively regulates drought tolerance in Arabidopsis by targeting the aquaporins PLASMA MEMBRANE INTRINSIC PROTEIN 2;1 (PIP2;1) and PIP2;2 for degradation [50].The E3 ubiquitin ligases PUB22 and PUB23 negatively regulate drought tolerance by targeting the ABA receptor PYL9 for degradation in Arabidopsis [51].In this study, smxl6/7/8 and smxl6/7/8/dwa1 exhibited the drought tolerance phenotype after 16 days of drought stress treatment compared with the wild type, and when the drought treatment time was extended, the drought tolerance of smxl6/7/8/dwa1 was lower than that of smxl6/7/8 after 18 days of drought stress treatment (Figures 4 and S3).Together, these data suggest that SMXL6,7,8 are degraded by DWA1 through their interactions, thereby regulating drought tolerance in Arabidopsis.
A previous study reported that dwa1 showed no difference in drought tolerance compared with the wild type [12].In contrast, the dwa1 mutant showed less tolerance after 16 days of drought stress treatment in this study (Figures 4, S2c and S3).This phenotypic difference in the dwa1 mutant in response to drought stress may be the difference in drought treatment time, and 16 days of drought treatment time was critical in our working condition and shorter than in Lee's [12].We also found that DWA1 partially promoted the degradation of SMXL6,7,8, and loss of DWA1 led to the accumulation of SMXL6,7,8 protein in planta (Figures 2 and 3).In addition, the overexpression of SMXL6,7,8 reduced drought tolerance compared with the wild type because of the negative regulatory function of SMXL6,7,8 (Figure S4).Thus, increased SMXL6,7,8 protein levels in dwa1 may contribute to its drought-sensitive phenotype (Figures 4, S2 and S3).
In this study, SMXL6,7,8 could directly bind to the promoters of SnRK2.3 and SnRK2.6 (Figures 5 and S8), which positively regulate ABA signaling and the drought stress response [8,55].SMXL6,7,8, as transcription suppressors, inhibit the expression of downstream genes to respond to SL signaling by interacting with specific transcription factors [21,23].SMXL6,7,8 also act as transcription factors to directly repress downstream gene expression, including that of themselves [30].The overexpression of individual SMXL6, SMXL7, and SMXL8 inhibited the activity of LUC driven by the SnRK2.3 and SnRK2.6 promoters compared with the expression of GFP alone (Figures 5 and S8).This suggests that the ABA/drought-response genes SnRK2.3 and SnRK2.6 are also transcriptionally regulated by SMXL6,7,8.The overexpression of SnRK2.3 and SnRK2.6 enhance drought tolerance [8].The snrk2.2/2.3/2.6 mutant significantly reduce the drought tolerance of Arabidopsis, while the overexpression of TaSnRK2.3 in common wheat enhance the drought tolerance [10,11].Meanwhile, AtPP2-B11, an F-box protein that is part of an SKP1/Cullin/F-box E3 ubiquitin ligase complex, has been reported to negatively regulate ABA signaling and the abiotic stress response by interacting with and targeting SnRK2.3 degradation [9].It has been speculated that activation of SnRK2.3 may require other components in addition to the classic ABA signaling pathway [56,57].The inactivation of SnRK2s is equally important to attenuate the ABA-induced drought response [58].The relative expressions of SnRK2.3 and SnRK2.6 were up-regulated in the smxl6/7/8 triple mutant (Figure S10).Genetic evidence suggests that snrk2.3 and snrk2.2/2.3 restored the ABA-sensitive phenotype of smxl6/7/8 (Figure 6), indicating that SnRK2.2/2.3 were located downstream of SMXL6,7,8 and that ABA hypersensitivity of smxl6/7/8 required functional SnRK2.2/2.3 in seed germination.In addition, we will investigate whether SnRK2.6 is involved in the regulation of ABA response and drought tolerance downstream of SMXL6,7,8 in the future.It is well documented that the ATAACAA motif is essential for SMXL6 to directly bind to the promoter of SMXL7 and repress its transcription [30].However, in the present study, we did not observe that ATAACAA contributed to SMXL6,7,8 binding to the promoters of SnRK2.3 and SnRK2.6 (Figures 5 and S8), which needs to be further explored.

Conclusions
Overall, a schematic diagram of how SMXL6,7,8 regulate drought tolerance was proposed (Figure 7).On the one hand, CUL4-based E3 ligase DWA1 functions upstream of SMXL6,7,8 to regulate plant drought stress responses by changing the stability of SMXL6,7,8 proteins.On the other hand, the SMXL6,7,8 bind to the promoter of SnRK2.3 and repress its transcription, leading to an ABA response.Our study will prompt us to decipher whether homologous proteins of SMXL6,7,8 in rice, maize, and other crops can also play similar roles in response to drought stress tolerance and ABA, which will enhance a better understanding of the molecular mechanism of SL signaling in response to the drought stress tolerance of crops.

Figure 5 .
Figure 5. SMXL6,7,8 directly bind to the SnRK2.3promoter.(a) Schematic diagram of dual-luciferase assay constructs.The promoter of SnRK2.3 was fused to the REN gene controlled by a 35S promoter and LUC gene as a reporter, and the effector constructs contained a GFP, SMXL6, SMXL7, or SMXL8 gene driven by the CaMV 35S promoter.(b) Transcriptional activities of SMXL6,7,8 on the SnRK2.3promoter in Arabidopsis protoplasts.The experiment was independently repeated three times with three biological replicates in each experiment, and similar results were obtained.Data are the means of three biological replicates ± SD.Different letters represent statistical significance at p < 0.05 (according to Duncan's multiple range test).(c) Truncated regions of the SnRK2.3promoter and synthesized fragments of EMSA probes.(d) Yeast one-hybrid assays showing that SMXL6,7,8 displayed strong binding to the fragment 3-1-2 of the SnRK2.3promoter.An empty vector expressing the AD domain alone was used as the negative control.(e) EMSAs showing that the SMXL6,7,8 proteins specifically bound to fragments '3-1-2-a' of the SnRK2.3promoter in vitro.For the competition, 25× excess unlabeled probes (cold probes) were mixed with biotin-labeled probes.GST, glutathione-Stransferase.Data represent three independent experiments.

Figure 5 .
Figure 5. SMXL6,7,8 directly bind to the SnRK2.3promoter.(a) Schematic diagram of dual-luciferase assay constructs.The promoter of SnRK2.3 was fused to the REN gene controlled by a 35S promoter and LUC gene as a reporter, and the effector constructs contained a GFP, SMXL6, SMXL7, or SMXL8 gene driven by the CaMV 35S promoter.(b) Transcriptional activities of SMXL6,7,8 on the SnRK2.3promoter in Arabidopsis protoplasts.The experiment was independently repeated three times with three biological replicates in each experiment, and similar results were obtained.Data are the means of