Characterization of the Role of SPL9 in Drought Stress Tolerance in Medicago sativa

Extreme environmental conditions, such as drought, are expected to increase in frequency and severity due to climate change, leading to substantial deficiencies in crop yield and quality. Medicago sativa (alfalfa) is an important crop that is relied upon as a staple source of forage in ruminant feed. Despite its economic importance, alfalfa production is constrained by abiotic stress, including drought. In this report, we investigate the role of Squamosa Promoter Binding Protein-Like 9 (SPL9), a target of miR156, in drought tolerance. Transgenic alfalfa plants with RNAi-silenced MsSPL9 (SPL9-RNAi) were compared to wild-type (WT) alfalfa for phenotypic changes and drought tolerance indicators. In SPL9-RNAi plants, both stem thickness and plant height were reduced in two- and six-month-old alfalfa, respectively; however, yield was unaffected. SPL9-RNAi plants showed less leaf senescence and had augmented relative water content under drought conditions, indicating that SPL9-RNAi plants had greater drought tolerance potential than WT plants. Interestingly, SPL9-RNAi plants accumulated more stress-alleviating anthocyanin compared to WT under both drought and well-watered control conditions, suggesting that MsSPL9 may contribute to drought tolerance in alfalfa, at least in part, by regulating anthocyanin biosynthesis. The results suggest that targeting MsSPL9 is a suitable means for improving alfalfa resilience towards drought conditions.


Introduction
As the world population increases, so does the demand for resources to support it. For example, global crop demands are projected to double by 2050 [1]. There exists a number of strategies to meet crop demands including increasing farmable land and improving productivity of existing farmland [1,2]. Unfortunately, some strategies that create new agriculturally available land, such as clearing, also result in the release of greenhouse gases, contributing to climate change [1]. Climate change can also cause extreme environmental conditions, such as drought, that must be overcome to achieve maximum agricultural production [3]. Areas such as the Canadian Prairies have seen a decrease in annual precipitation [4], leading to drought conditions that result in the reduction of crop yields [5]. Investigating and enhancing existing predispositions for tolerance to water deficiency could thus provide tools for crop improvement to secure the global food supply.
Medicago sativa (alfalfa) is a perennial legume grown on 3.8 million ha in Canada [6]. Alfalfa is an important staple in livestock nutrition due to its large supply of protein [7,8]. Its status as an important forage crop in Western Canada prompted careful evaluation of the trends in alfalfa yield in response to environmental variations [9].
In order to minimize water loss and to increase water uptake, plants respond to drought through physiological changes that include increasing root length, closing stomata, and delaying leaf senescence [10][11][12]. Plants can also utilize soluble secondary metabolites and inorganic compounds, known as osmolytes, to maintain cell turgor and hydrostatic pressure under drought [13]. Drought-sensitive plants can ultimately succumb to oxidative stress due to the build-up of reactive oxygen species (ROS) that cause changes in signalling cascades and transcription factors, leading to cell death [14,15]. Enzymatic ROS scavengers like Catalase 1 (CAT1) can mitigate ROS build-up and can increase drought tolerance of a plant [14][15][16]. Some secondary metabolites, such as flavonoids, are also capable of nonenzymatic ROS scavenging to decrease oxidative stress [14]. The ascorbate-glutathione cycle is an example of nonenzymatic ROS scavenging and of gene expression and activity of glutathione biosynthesis-related proteins such as Glutathione Synthase (GSH) increase in response to abiotic stress [17,18].
AtSPL9 functions by interrupting the DFR transcription-activating complex, consisting of MYB proteins, bHLH factors, and WD40 proteins [29]. DFR is an enzyme involved in the biosynthesis of anthocyanins and proanthocyanidins [30,31]. Proanthocyanidins prevent pasture bloat in cattle, a common ailment in animals primarily feeding on alfalfa forage [30,32,33]. Anthocyanins contribute to pigmentation of some plant tissues and to scavenging of ROS [34][35][36].
This study examines the role of MsSPL9 in drought tolerance and whether MsSPL9 plays a similar role in alfalfa as it does in Arabidopsis with regard to anthocyanin biosynthesis and its involvement in the miR156-SPL9-DFR pathway.

Phenotypic Characterization of SPL9-RNAi Plants
Measurements of phenotypic traits were recorded in two-and six-month-old plants grown under favourable normal greenhouse conditions in order to evaluate phenotypes before and after flowering, respectively. Three RNAi-silenced MsSPL9 (SPL9-RNAi) transgenic plants, R1, R2, and R3, were generated, and transcript levels of MsSPL9 were evaluated in each ( Figure 1). The three plants showed a range of MsSPL9 silencing: from strong silencing in R3, low in R1, and moderate in R2. SPL9-RNAi plants were indistinguishable from wild-type (WT) plants in all phenotypic measurements made in two-month-old plants except for stem thickness (Table 1), where SPL9-RNAi had thinner stems compared to WT plants. After six months, all SPL9-RNAi plants showed significantly diminished plant height compared to WT (Table 2). However, this trait was the only phenotype that was consistent in all RNAi plants. R1 and R3 had decreased average internode length and increased lateral and total branching, whereas only R2 showed decreased main branching, decreased number of nodes, and delayed flowering time. The aforementioned changes to phenotype in the transgenic plants did not result in significant changes to overall yield in either shoots or roots compared to WT.

Response to Drought Stress in SPL9-RNAi Plants
To determine whether MsSPL9 is regulated in response to drought in alfalfa, the MsSPL9 transcript level was assessed in WT alfalfa plants exposed to 12 days of withholding water ( Figure 2). The transcript abundance of MsSPL9 was significantly decreased under drought compared to a well-watered control treatment. To understand the role of MsSPL9 in drought tolerance, further drought experiments were performed on SPL9-RNAi and WT alfalfa plants. After 12 days of drought treatment, most plants showed signs of drought stress in the form of wilting, brown, and curling leaves (Figure 3). At this time, both WT and transgenic plants under drought had significantly decreased adaxial and abaxial stomatal conductance compared to their well-watered counterparts except in the abaxial surface of R2 ( Figure 4A). SPL9-RNAi plants appeared to tolerate drought better than WT plants because, after 10 days of stress, wilting and leaf rolling was observed in WT plants but not in SPL9-RNAi plants. Evaluation of the extent of the observed tolerance to drought treatment was analysed by quantifying leaf senescence after 12 days of withholding water. While all plants exposed to drought had increased leaf senescence compared to their well-watered counterparts, drought-stressed R1, R2, and R3 had decreased leaf senescence compared to WT ( Figure 4C). To investigate possible differences in their ability to grow under 12 days of drought stress, the difference between plant height before and after drought was determined. Decreased plant growth due to drought was only observed in the lowest silencers of MsSPL9, WT, and R1 plants, whereas the strongest silencers of MsSPL9, R2, and R3 plants were capable of maintaining growth under drought conditions ( Figure 4B). However, maintenance in growth did not extend to overall dry weight (DW) biomass of aboveground tissues. Decreased aboveground dry biomass was observed in drought-stressed R1 and R2 plants in addition to well-watered R2 plants compared to WT in each treatment ( Figure 5A). Differences in aboveground dry biomass were not observed between treatments in any plants. WT and transgenic plants were also indistinguishable between and within conditions when examining root dry biomass and length ( Figure 5B and Figure S1).

Response to Drought Stress in SPL9-RNAi Plants
To determine whether MsSPL9 is regulated in response to drought in alfalfa, the MsSPL9 transcript level was assessed in WT alfalfa plants exposed to 12 days of withholding water ( Figure  2). The transcript abundance of MsSPL9 was significantly decreased under drought compared to a well-watered control treatment. To understand the role of MsSPL9 in drought tolerance, further drought experiments were performed on SPL9-RNAi and WT alfalfa plants. After 12 days of drought treatment, most plants showed signs of drought stress in the form of wilting, brown, and curling leaves (Figure 3). At this time, both WT and transgenic plants under drought had significantly decreased adaxial and abaxial stomatal conductance compared to their well-watered counterparts except in the abaxial surface of R2 ( Figure 4A). SPL9-RNAi plants appeared to tolerate drought better than WT plants because, after 10 days of stress, wilting and leaf rolling was observed in WT plants but not in SPL9-RNAi plants. Evaluation of the extent of the observed tolerance to drought treatment was analysed by quantifying leaf senescence after 12 days of withholding water. While all plants exposed to drought had increased leaf senescence compared to their well-watered counterparts, drought-stressed R1, R2, and R3 had decreased leaf senescence compared to WT ( Figure 4C). To investigate possible differences in their ability to grow under 12 days of drought stress, the difference between plant height before and after drought was determined. Decreased plant growth due to drought was only observed in the lowest silencers of MsSPL9, WT, and R1 plants, whereas the strongest silencers of MsSPL9, R2, and R3 plants were capable of maintaining growth under drought conditions ( Figure 4B). However, maintenance in growth did not extend to overall dry weight (DW) biomass of aboveground tissues. Decreased aboveground dry biomass was observed in droughtstressed R1 and R2 plants in addition to well-watered R2 plants compared to WT in each treatment ( Figure 5A). Differences in aboveground dry biomass were not observed between treatments in any plants. WT and transgenic plants were also indistinguishable between and within conditions when examining root dry biomass and length (Figures 5B and S1).    The asterisks indicate significant differences between treatments (p < 0.05, where n = 8, two-way ANOVA and Sidak test). (B) Plant height difference between initiation and end of trial (12 days): the asterisks indicate significant differences between treatments (p < 0.05, where n = 10, two-way ANOVA and Sidak test). (C) Percent of total leaves that were senesced after plants were exposed to 12 days of control and drought   The asterisks indicate significant differences between treatments (p < 0.05, where n = 8, two-way ANOVA and Sidak test). (B) Plant height difference between initiation and end of trial (12 days): the asterisks indicate significant differences between treatments (p < 0.05, where n = 10, two-way ANOVA and Sidak test). (C) Percent of total leaves that were senesced after plants were exposed to 12 days of control and drought The asterisks indicate significant differences between treatments (p < 0.05, where n = 8, two-way ANOVA and Sidak test). (B) Plant height difference between initiation and end of trial (12 days): the asterisks indicate significant differences between treatments (p < 0.05, where n = 10, two-way ANOVA and Sidak test). (C) Percent of total leaves that were senesced after plants were exposed to 12 days of control and drought treatments: the asterisks indicate significant differences within treatments (Dunnett's test), and the bars indicate significant differences between treatments (Sidak test) (p < 0.05, where n = 10, two-way ANOVA). treatments: the asterisks indicate significant differences within treatments (Dunnett's test), and the bars indicate significant differences between treatments (Sidak test) (p < 0.05, where n = 10, two-way ANOVA).

Evaluation of Plant Water Status under Drought Stress
Plant water status can be used to shed light on mechanisms behind plant response to stress, and thus, it was used to evaluate responses of SPL9-RNAi plants to drought. Fresh weight (FW) of shoots and roots was compared in control and drought-treated plants. In plants with the highest MsSPL9 transcript level, WT and R1, shoot and root FWs were decreased when comparing well-watered and drought conditions ( Figure 5). Also, SPL9-RNAi plants had decreased shoot FW compared to WT when well-watered. WT plants, which had the highest MsSPL9 transcript level, had decreased relative water content (RWC) between conditions whereas SPL9-RNAi plants could maintain RWC between conditions ( Figure 6A). When comparing plants under drought stress, R1 and R2 plants had increased RWC compared to WT. However, R1 had decreased midday leaf water potential between treatments and an increase in water loss after 120 min under drought ( Figure 6B,C). No changes in water loss or midday leaf water potential in response to drought were detected in the other genotypes.

Evaluation of Plant Water Status under Drought Stress
Plant water status can be used to shed light on mechanisms behind plant response to stress, and thus, it was used to evaluate responses of SPL9-RNAi plants to drought. Fresh weight (FW) of shoots and roots was compared in control and drought-treated plants. In plants with the highest MsSPL9 transcript level, WT and R1, shoot and root FWs were decreased when comparing well-watered and drought conditions ( Figure 5). Also, SPL9-RNAi plants had decreased shoot FW compared to WT when well-watered. WT plants, which had the highest MsSPL9 transcript level, had decreased relative water content (RWC) between conditions whereas SPL9-RNAi plants could maintain RWC between conditions ( Figure 6A). When comparing plants under drought stress, R1 and R2 plants had increased RWC compared to WT. However, R1 had decreased midday leaf water potential between treatments and an increase in water loss after 120 min under drought ( Figure 6B,C). No changes in water loss or midday leaf water potential in response to drought were detected in the other genotypes.  The asterisks indicate significant differences within treatments (Dunnett's test), and the bars indicate significant differences between treatments (Sidak test) in a two-way ANOVA where p < 0.05 and n = 10. (B) Leaf water potential of plants measured around midday: the asterisks indicate significant differences between treatments (p < 0.05, where n = 9, 10, two-way ANOVA, Sidak test). (C) Water loss assay performed by weighing aboveground tissue every half hour for 3 h after decapitating plants that were exposed to drought: the asterisks indicate significant differences from WT within the time point (p < 0.05, where n = 10, two-way ANOVA and Dunnett's test).

Effect of Drought on the Antioxidant-Mediated ROS Scavenging Capabilities of SPL9-RNAi Alfalfa
The transcript abundance of the antioxidant-related genes Catalase 1 (CAT1) and Glutathione Synthase (GSH) was examined to determine whether MsSPL9 serves to maintain the levels of these enzymes in alfalfa exposed to drought. WT alfalfa had increased abundance of CAT1 under drought compared to control plants, but CAT1 abundance did not change between treatments in SPL9-RNAi plants ( Figure 7A). In fact, R2 had higher CAT1 abundance than WT under control conditions. Similar maintenance of GSH transcript levels was also detected between drought and control treatments in SPL9-RNAi plants ( Figure 7B). GSH abundance decreased in WT plants in response to drought, but R3 had increased GSH abundance under drought conditions compared to WT. Figure 6. Effect of drought on plant water status in WT and SPL9-RNAi plants: (A) relative water content (RWC) of 10 leaves harvested from plants under both conditions. The asterisks indicate significant differences within treatments (Dunnett's test), and the bars indicate significant differences between treatments (Sidak test) in a two-way ANOVA where p < 0.05 and n = 10. (B) Leaf water potential of plants measured around midday: the asterisks indicate significant differences between treatments (p < 0.05, where n = 9, 10, two-way ANOVA, Sidak test). (C) Water loss assay performed by weighing aboveground tissue every half hour for 3 h after decapitating plants that were exposed to drought: the asterisks indicate significant differences from WT within the time point (p < 0.05, where n = 10, two-way ANOVA and Dunnett's test).

Effect of Drought on the Antioxidant-Mediated ROS Scavenging Capabilities of SPL9-RNAi Alfalfa
The transcript abundance of the antioxidant-related genes Catalase 1 (CAT1) and Glutathione Synthase (GSH) was examined to determine whether MsSPL9 serves to maintain the levels of these enzymes in alfalfa exposed to drought. WT alfalfa had increased abundance of CAT1 under drought compared to control plants, but CAT1 abundance did not change between treatments in SPL9-RNAi plants ( Figure 7A). In fact, R2 had higher CAT1 abundance than WT under control conditions. Similar maintenance of GSH transcript levels was also detected between drought and control treatments in SPL9-RNAi plants ( Figure 7B). GSH abundance decreased in WT plants in response to drought, but R3 had increased GSH abundance under drought conditions compared to WT.

MsSPL9 Regulates Anthocyanin Biosynthesis
The involvement of MsSPL9 in anthocyanin biosynthesis was first investigated by examining the anthocyanin content in SPL9-RNAi plants under drought and control treatments and by comparing them to WT. Under drought, anthocyanin content was visually increased in SPL9-RNAi plants ( Figure 8A). The characteristic red hue of anthocyanins was also visible to a greater extent in well-watered SPL9-RNAi stems than in WT stems. Thus, anthocyanins were extracted from plant stems to determine relative quantities. Increased anthocyanin content was found in drought-stressed SPL9-RNAi plants ( Figure 8B). R2 also had increased anthocyanin content under control conditions compared to WT. Next, we wanted to determine if MsSPL9 impacts the DFR transcript level, as DFR is one of the critical enzymes in the anthocyanin biosynthesis pathway [30,31]. WT and R3 plants showed a decrease in DFR abundance in response to drought, whereas R1 and R2 plants were able to maintain their levels of DFR ( Figure 8C). When considering the plants in the drought treatment only, R1 had an enhanced DFR transcript level compared to WT. stems to determine relative quantities. Increased anthocyanin content was found in drought-stressed SPL9-RNAi plants ( Figure 8B). R2 also had increased anthocyanin content under control conditions compared to WT. Next, we wanted to determine if MsSPL9 impacts the DFR transcript level, as DFR is one of the critical enzymes in the anthocyanin biosynthesis pathway [30,31]. WT and R3 plants showed a decrease in DFR abundance in response to drought, whereas R1 and R2 plants were able to maintain their levels of DFR ( Figure 8C). When considering the plants in the drought treatment only, R1 had an enhanced DFR transcript level compared to WT. The asterisks indicate significant differences within treatments (Dunnett's test), and the bars indicate significant differences between treatments (Sidak test) in a two-way ANOVA where p < 0.05.

Discussion
The role of miR156 in plant growth and development has been well documented [19,[24][25][26][27], but functional characterization of its target SPL genes is still lagging, especially in major crops, such as alfalfa. In this study, we found MsSPL9 to play a role in determining plant height and stem thickness in alfalfa. Reduced MsSPL9 transcript levels resulted in diminished stem thickness (Table 1) and plant height ( Table 2) compared to WT plants, but the degree to which these traits were impacted was not proportional to the level of MsSPL9. For example, SPL9-RNAi plants displayed a gradient of MsSPL9 silencing (Figure 1), but a similar gradient in the reduction of stem thickness and plant height was not observed. This could indicate that further regulation of these traits is not dependent on the level of MsSPL9 under a certain threshold. MsSPL9 may also play a partial role in developmental control over other phenotypic traits like branching, number of nodes, and flowering; however, no consistency between the SPL9-RNAi plants was observed. The lack of consistency between transgenic genotypes may be a result of other MsSPLs functioning in tandem with MsSPL9 to exert phenotypic The asterisks indicate significant differences within treatments (Dunnett's test), and the bars indicate significant differences between treatments (Sidak test) in a two-way ANOVA where p < 0.05.

Discussion
The role of miR156 in plant growth and development has been well documented [19,[24][25][26][27], but functional characterization of its target SPL genes is still lagging, especially in major crops, such as alfalfa. In this study, we found MsSPL9 to play a role in determining plant height and stem thickness in alfalfa. Reduced MsSPL9 transcript levels resulted in diminished stem thickness (Table 1) and plant height ( Table 2) compared to WT plants, but the degree to which these traits were impacted was not proportional to the level of MsSPL9. For example, SPL9-RNAi plants displayed a gradient of MsSPL9 silencing (Figure 1), but a similar gradient in the reduction of stem thickness and plant height was not observed. This could indicate that further regulation of these traits is not dependent on the level of MsSPL9 under a certain threshold. MsSPL9 may also play a partial role in developmental control over other phenotypic traits like branching, number of nodes, and flowering; however, no consistency between the SPL9-RNAi plants was observed. The lack of consistency between transgenic genotypes may be a result of other MsSPLs functioning in tandem with MsSPL9 to exert phenotypic control. In Arabidopsis, AtSPL9 is paralogous to AtSPL15, and Schwarz et al. (2008) [24] found that SPL9/spl15 Arabidopsis double mutants displayed phenotypes that were different from those of single SPL9 and spl15 mutants. The Arabidopsis SPL9/spl15 double mutants had a greater number of rosette leaves and side shoots as well as later bolting and flowering than the single mutants [24]. Therefore, MsSPL9 may be regulated in conjunction with yet to be discovered MsSPL paralogs in alfalfa to attain miR156-mediated developmental control.
Reducing MsSPL9 transcript levels in alfalfa resulted in plants with improved tolerance to drought. Specifically, R2 and R3, the genotypes with the lowest MsSPL9 transcript abundance, maintained plant growth ( Figure 4B). In addition, all three SPL9-RNAi had decreased leaf senescence ( Figure 4C) and maintained RWC ( Figure 6A) in response to drought. This suggests that, despite underexpression of MsSPL9 in R1, levels were not low enough to warrant the same tolerance against drought that R2 and R3 possessed. Support for the dependence of drought tolerance on the level of MsSPL9 silencing could also be seen in the FW biomass of shoots and roots ( Figure 5). Again, R2 and R3 were able to maintain FW between well-watered and drought-exposed treatments while WT and R1 plants had decreased FW. This is consistent with a similar finding reported for MsSPL13, where only SPL13-RNAi alfalfa plants with reduced MsSPL13 levels but over a certain threshold showed significant drought tolerance [22].
The drought tolerant phenotypes observed in these SPL9-RNAi plants were only a subset of those reported in SPL13-RNAi plants by   [18], indicating that these SPLs may regulate different responses to stress.   [18] demonstrated that MsSPL13 is targeted by miR156 to increase drought tolerance mainly through its involvement in regulating root architecture and water loss. SPL13-RNAi plants had roots that were thicker, denser, and longer than control alfalfa plants and lost less water in response to drought [18]. In this study, reduced MsSPL9 levels did not affect root growth under control conditions or in response to drought ( Figure 5B, Figure S2). Additionally, water loss was not mitigated in SPL9-RNAi but was actually enhanced in R1 plants ( Figure 6C). Interestingly, R1, like WT, was not able to maintain growth when exposed to water-deficit conditions. Considering that R1 also had decreased leaf water potential in response to drought, the less than ideal water status of R1 plants likely contributed to its greater susceptibility to drought compared to the other SPL9-RNAi plants. Therefore, the differences in responses between SPL13-RNAi reported by Arshad et al. [18] and the SPL9-RNAi alfalfa plants analysed in this study suggest existence of multiple MsSPLs that are regulated when alfalfa is exposed to drought.   [18] also provided evidence for the involvement of miR156 in the regulation of antioxidants to combat ROS produced as a by-product of drought stress, noting an increase in total antioxidant content in miR156OE plants exposed to drought. If miR156 targets MsSPL9 to impact the levels of antioxidants present in the plant, miR156OE and SPL9-RNAi plants should have similar responses. The levels of CAT1 and GSH were measured to ascertain if MsSPL9 is involved in the regulation of antioxidant activity. WT plants had enhanced CAT1 in response to drought, but despite initially higher CAT1 under control conditions in R2, all SPL9-RNAi plants maintained CAT1 in response to drought ( Figure 7A). It could be interpreted that WT plants are under greater stress and therefore upregulate CAT1 to respond to possible enhanced ROS production, whereas SPL9-RNAi plants were already more tolerant to the drought conditions and thus maintained CAT1 levels. GSH was also maintained in SPL9-RNAi plants; however, in contrast to CAT1 levels, WT plants had reduced GSH in response to drought and R3 had enhanced GSH under drought ( Figure 7B). These contradicting results require further investigation to determine if MsSPL9 is definitively involved in antioxidant activity.
Underexpression of MsSPL9 increased anthocyanin content under drought stress in SPL9-RNAi plants ( Figure 8B). This supports the conclusions made by Cui et al. (2014) [21], who found that AtSPL9 regulates anthocyanin biosynthesis to affect drought tolerance. Thus, further investigations to confirm the direct interaction of MsSPL9 and DFR was warranted. Unexpectedly, levels of DFR transcripts did not exactly correspond to changes in anthocyanin content ( Figure 8C). DFR was significantly reduced in WT plants in response to drought, but no changes in anthocyanin content of WT between conditions were found. Surprisingly, DFR decreased in response to drought in R3 despite elevated anthocyanin content compared to WT plants under drought. In R1, enhanced DFR did correlate with enhanced anthocyanin content under drought. R2, however, had enhanced anthocyanin content compared to WT in both control and drought treatments but DFR did not change. A potential explanation for this discrepancy is that DFR transcription in alfalfa may be regulated through a different MsSPL to impact anthocyanin biosynthesis. Feyissa et al. (2019) found enhanced levels of DFR expression in SPL13-RNAi alfalfa that was also tolerant to drought and demonstrated the direct interaction between MsSPL13 and the DFR promoter region [22]. Analysis of a direct interaction between MsSPL9 and DFR needs to be examined to test the possibility that, for example, DFR transcription is not regulated by MsSPL9. Another potential explanation for this discrepancy is that MsSPL9 does interact with DFR in a similar fashion as was discovered in Arabidopsis but has a yet-to-be-discovered paralog partner, as was previously discussed. In this case, if the paralog is not silenced along with MsSPL9, it could be acting as the sole repressor of DFR, essentially making up for the absence of MsSPL9. Although no MsSPL15 has yet been found in alfalfa, conflicting results regarding anthocyanin biosynthesis and DFR transcript abundance point towards the existence of similar acting genes. Regardless, further investigations into the role that MsSPL9 plays in anthocyanin biosynthesis is required as it has potential to become a tool for improving alfalfa performance under drought stress.
Thinner stems and decreased plant height in SPL9-RNAi alfalfa demonstrated that MsSPL9 is involved in regulating alfalfa physiology. Under drought stress conditions, SPL9-RNAi alfalfa also displayed increased anthocyanin biosynthesis, resulting in plants with decreased leaf senescence, increased RWC, and the ability to maintain growth. Therefore, MsSPL9 is capable of regulating traits related to drought tolerance in alfalfa. Future studies regarding the exact mechanism through which MsSPL9 achieves this regulation are needed, and resulting information can be applied to improve the ability of alfalfa to resist the detrimental effects of drought conditions.

Plant Material
The WT Medicago sativa L. (alfalfa) germplasm used was propagated from clone N4.4.2 [37] obtained from Daniel Brown (Agriculture and Agri-Food Canada, London, ON, Canada). Plant material was grown under greenhouse conditions (16-h light/8-h dark, 56% humidity, 23 • C) for the duration of all experiments. WT and transgenic alfalfa plants were propagated from stem cuttings placed in Oasis Rootcubes ® for four weeks. Rooted cuttings were transferred to BX Mycorrhizae (PRO-MIX ®® , Smithers-Oasis North America, Kent, OH, USA) soil in six-inch pots. Plants were watered twice per week and allowed to grow for a month before being used in any experiment.

SPL9-RNAi Design and Construction
Medicago sativa-specific primers (Table S1) were used to amplify a blunt-end fragment of MsSPL9 that was cloned into the pENTR vector (Invitrogen, Carlsbad, CA, USA) according to the pENTR/D-TOPO Cloning Kit (Invitrogen). This fragment was designed to be outside the conserved SBP domain to ensure SPL9-RNAi specificity ( Figure S3). Positive transformants were used in a LR reaction with pHELLSGATE12 according to the Gateway LR Clonase II Enzyme Mix protocol (Invitrogen). SPL9-RNAi transgenic alfalfa were created by transferring the construct to Agrobacterium tumefaciens (GV3101) by heat shock, and the resulting strain was used to transform alfalfa N4.4.2 germplasm [37]. Alfalfa transformation proceeded by soaking N4.4.2 germplasm in A. tumefaciens-inoculated Luria-Bertani (LB) media containing 20 µM acetosyringone and by inducing callus formation, embryo induction, and plant development as described by [38]. Positive alfalfa transformants were first selected by kanamycin antibiotic selection, potential transgenic plants were tested for the presence of the transgene by PCR analysis using primers nptII-F and nptII-R (Table S1), and MsSPL9 transcript abundance was determined by qRT-PCR using primers LA-MsSPL9-Fq1 and LA-MsSPL9-Rq1 (Table S1).

SPL9-RNAi Growing Conditions, Propagation, and Characterization
Phenotypic characterization was performed on two-and six-month-old WT and SPL9-RNAi alfalfa with approximately 10 biological replicates of each genotype. Main branches were considered those that emerged directly from the soil, while lateral branches were considered any that emerged from main branches. The tallest main branch was used to measure plant height, number of nodes, and average internode length. Plant height was measured starting at the soil level, and average internode length was considered as the ratio between the plant height and number of nodes. Approximately halfway between the second and third nodes was used to measure stem thickness using a digital caliper (Mitutoyo, Kawasaki, Japan). Aboveground tissue was determined by decapitating the plant approximately 1 cm above the soil line, and any tissue below this point was considered roots. Tissue fresh weight (FW) was measured at the time of harvest, and dry weight (DW) was determined after the tissue was baked at 65 • C for 5 days. Root length was considered the length from the top of the root crown to the tip of the longest root. Flowering time was measured from the day cuttings were made to the first emergence of flowers.

Evaluation of Drought Tolerance
Drought trials consisted of one-month-old WT and SPL9-RNAi alfalfa, subjected to either drought or well-watered control treatments in two trials of five biological replicates. At the beginning of each trial, to normalize the amount of water each plant received, all plants were watered to pot capacity (approximately 150 mL in 6-inch pots). Drought was imposed by withholding water, while the control treated plants were watered regularly. All measurements were performed 12 days after the start of the trial. Growth during the trial was determined from the difference between plant heights before and after the trial. Leaf senescence was determined from the ratio of senesced to total number of leaves. Tissue biomass and plant height were measured using the same procedure as for SPL9-RNAi characterization. Analysis of roots under drought stress (root length, fresh weight, and dry weight) were performed by growing rooted cuttings in containers (Greenhouse Megastore, Danville, IL, USA) for five weeks before imposing trial conditions. The longest root was measured for length, and aboveground tissue was separated from the roots at the soil line to measure FW and DW.

Evaluation of Water Status
Ten leaves were used to measure relative water content (RWC) according to the following equation: where FW is the weight of the tissue 12 days after water was withheld, DW is the weight of the tissue sampled at the end of each condition after five days at 65 • C, and FTW is the fresh turgid weight of the tissue after two days of incubation in deionized water [18,39,40]. Leaf water potential was measured around midday, 12 days after the trial's initiation, using a pressure vessel SAPS II Portable Plant Water Status Console (Soilmoisture Equipment Corp., Santa Barbara, CA, USA). Midday leaf water potential was recorded when a single drop of liquid was visible from the exposed petiole [41]. Rapid water loss was evaluated by measuring the FW of decapitated aboveground tissue every half hour for three hours 12 days after the trial's initiation. Rapid water loss was calculated using the following equation: where FW t 0 is the initial weight of the tissue at decapitation and FW t n is the weight of the tissue at each time point (n) [18].

Stomatal Conductance
Adaxial and abaxial surface leaf stomatal conductance were determined using a Leaf Porometer (Decagon Devices Inc., Pullman, WA, USA). Measurements were taken around midday 12 days after starting the trial. The stomatal conductance of both surfaces of two leaves in a single plant were measured. The average of the technical replicates was considered one biological replicate. Two biological replicates of each genotype were measured in four separate trials.

Extraction and Analysis of Anthocyanin and Antioxidants
Approximately 250 mg of stem tissue collected from the base of each plant was used for anthocyanin extraction as described in [42]. Absorbance of the extracts was measured at 657 nm and 530 nm using a Multiskan GO spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Relative anthocyanin content was calculated by subtracting the difference of absorbance at 657 nm from that at 530 nm with 60% extraction solution used as a blank. Anthocyanin content was then normalized by the weight of the starting tissue.

cDNA Synthesis and qRT-PCR Analysis
RNA was extracted from 100 mg of alfalfa tissue using the RNeasy Plant Mini-prep Kit (Qiagen, Hilden, Germany) according to the manufacturer's manual. Transcript analysis of MsSPL9, DFR, GSH, and CAT1 were performed by extracting RNA from apical bud, stem, and leaf tissue, respectively. Approximately 0.5 µg of Turbo DNase (Invitrogen)-treated RNA was used to synthesize cDNA using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA). Actin-depolymerizing protein 1 (ADF1) and elongation initiation factor 4A (elF4A) were used as reference genes ( Figure S2).

Statistical Analysis
Statistical analysis of the data presented was performed using GraphPad Prism software. Phenotypic trait evaluations were analysed using one-way ANOVA with Dunnett's test for multiple comparisons between WT and SPL9-RNAi plants. The MsSPL9 transcript level in WT alfalfa in response to drought was analysed using unpaired, two-tailed Student's t-test. Evaluation of SPL9-RNAi response to drought was analysed using two-way ANOVA with Sidak and Dunnett's test for multiple comparisons between and within conditions, respectively.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.