The Novel Cucurbitaceae miRNA ClmiR86 Is Involved in Grafting-Enhanced Phosphate Utilization and Phosphate Starvation Tolerance in Watermelon

Watermelon (Citrullus lanatus) is a globally important Cucurbitaceae crop in which grafting is commonly used to improve stress tolerance and enhance nutrient utilization. However, the mechanism underlying grafting-enhanced nutrient assimilation remains unclear. Here, we demonstrate the possible involvement of a novel Cucurbitaceae miRNA, ClmiR86, in grafting-enhanced phosphate-starvation tolerance via CALCINEURIN B-LIKE INTERACTING PROTEIN KINASE 5 (ClCIPK5) suppression in watermelon. Transcript analyses revealed that the induction of ClmiR86 expression was correlated with the downregulation of ClCIPK5 in squash-grafted watermelon under phosphate starvation. In addition, the differential expression of ClmiR86 in various watermelon genotypes was consistent with their phosphate utilization efficiency. Furthermore, ClmiR86 overexpression in Arabidopsis enhanced root growth and phosphate uptake under phosphate starvation and promoted inflorescence elongation under normal conditions. These results suggest that the ClmiR86–ClCIPK5 axis is involved in phosphate starvation response as well as grafting-enhanced growth vigor and phosphate assimilation. The present study provides valuable insights for investigating long-distance signaling and nutrient utilization in plants.


Introduction
Grafting-the unification of different parts of two or more plants that grow together and form a new plant [1]-was found to improve resistance to various biotic and tolerance of various abiotic stresses [2][3][4][5]. Watermelon (Citrullus lanatus), a popular fresh fruit, is an economically important cash crop grown globally. Grafted watermelons using squash (Cucurbita moschata) and bottle gourd (Lagenaria siceraria) as the rootstocks were first adopted in Japan in the late 1920s [6,7]. Since then, grafting has been commonly used in watermelon production to improve disease resistance, specifically against the soil-borne Fusarium wilt disease. In addition, squash grafting was found to increase nutrient uptake in watermelon [8][9][10][11][12].
Phosphorus is a crucial and limiting soil nutrient, and it is fundamental to plant growth, development, and propagation [13,14]. However, because of precipitation and mineralization, the only form of soil phosphorus that can be assimilated by plants (inorganic phosphate) for vegetative growth and development is usually scarce [13,15,16]. As a result, in self-rooted watermelon [9]. Using small RNA deep sequencing of the same grafted watermelons, we identified ClmiR86 as a candidate long-distance signal between the scion and rootstock, since it exhibited a reverse expression pattern in these tissues after grafting. A qRT-PCR analysis confirmed that squash grafting induced and decreased ClmiR86 levels in watermelon leaves and squash roots, respectively ( Figure 1A,C), indicating that ClmiR86 may be transported from the rootstock to scion. Moreover, phosphate starvation further enhanced ClmiR86 accumulation in squash-grafted watermelon leaves, but not in auto-grafted watermelon leaves ( Figure 1A). Interestingly, ClmiR86 was induced by phosphate starvation in non-grafted squash roots, but not in grafted rootstock (squash root; Figure 1C), and ClmiR86 precursor showed a very similar expression profile to mature ClmiR86 in grafted watermelon leaves and squash roots under both phosphate-sufficient and phosphate-deficient conditions ( Figure 1B,D). These findings indicate that ClmiR86 may act as a graft-transmissible signal (rootstock to scion) and be involved in the graftingtriggered phosphate starvation response.

ClmiR86 May Enhance PUE of Watermelon
Our previous study demonstrated that under phosphate deficiency, phosphate uptake and PUE were significantly enhanced in squash-grafted watermelon compared with those in self-rooted watermelon [9]. Using small RNA deep sequencing of the same grafted watermelons, we identified ClmiR86 as a candidate long-distance signal between the scion and rootstock, since it exhibited a reverse expression pattern in these tissues after grafting. A qRT-PCR analysis confirmed that squash grafting induced and decreased ClmiR86 levels in watermelon leaves and squash roots, respectively ( Figure 1A,C), indicating that ClmiR86 may be transported from the rootstock to scion. Moreover, phosphate starvation further enhanced ClmiR86 accumulation in squash-grafted watermelon leaves, but not in auto-grafted watermelon leaves ( Figure 1A). Interestingly, ClmiR86 was induced by phosphate starvation in non-grafted squash roots, but not in grafted rootstock (squash root; Figure 1C), and ClmiR86 precursor showed a very similar expression profile to mature ClmiR86 in grafted watermelon leaves and squash roots under both phosphatesufficient and phosphate-deficient conditions ( Figure 1B,D). These findings indicate that ClmiR86 may act as a graft-transmissible signal (rootstock to scion) and be involved in the grafting-triggered phosphate starvation response.

Figure 1.
Grafting induced expression of ClmiR86 and Pre-ClmiR86 in watermelon leaves. Expression of (A) ClmiR86 and (B) Pre-ClmiR86 in the leaves of auto-and squash-grafted watermelon with or without P deficient treatment for 6 h. "Wm/Wm" and "Wm/Sq" represent auto-, and squashgrafted watermelon (Zaojia), respectively. Expression of (C) ClmiR86 and (D) Pre-ClmiR86 in the roots of non-grafted and grafted squash roots. "Sq" and "Wm/Sq" represent non-grafted and grafted squash roots, respectively. Different lower-case letters denote a significant difference in relative expression level (p < 0.05, one-way ANOVA and then Tukey's test for multiple comparisons). Values are means ± SD (n = 3).
To investigate whether ClmiR86 accumulation is consistent with improved phosphate utilization, we tested ClmiR86 expression patterns in watermelon genotypes with different PUEs. In our previous study, ZJ, a low-PUE genotype, exhibited relatively less Pre-ClmiR86 in the leaves of auto-and squash-grafted watermelon with or without P deficient treatment for 6 h. "Wm/Wm" and "Wm/Sq" represent auto-, and squash-grafted watermelon (Zaojia), respectively. Expression of (C) ClmiR86 and (D) Pre-ClmiR86 in the roots of nongrafted and grafted squash roots. "Sq" and "Wm/Sq" represent non-grafted and grafted squash roots, respectively. Different lower-case letters denote a significant difference in relative expression level (p < 0.05, one-way ANOVA and then Tukey's test for multiple comparisons). Values are means ± SD (n = 3).
To investigate whether ClmiR86 accumulation is consistent with improved phosphate utilization, we tested ClmiR86 expression patterns in watermelon genotypes with different PUEs. In our previous study, ZJ, a low-PUE genotype, exhibited relatively less biomass under phosphate sufficiency and low PUE under phosphate deficiency [44]. A cyclic variation in ClmiR86 expression with photoperiod was observed in both genotypes. Phosphate deficiency slightly and transitorily enhanced ClmiR86 expression in ZJ (Figure 2A). Mean-while, in the high-PUE genotype XN8, ClmiR86 expression was significantly induced after 12 h of phosphate starvation ( Figure 2B). This genotype-dependent response of ClmiR86 expression to phosphate stress indicates its possible involvement in phosphate starvation response and utilization in watermelon. biomass under phosphate sufficiency and low PUE under phosphate deficiency [44]. A cyclic variation in ClmiR86 expression with photoperiod was observed in both genotypes. Phosphate deficiency slightly and transitorily enhanced ClmiR86 expression in ZJ ( Figure  2A). Meanwhile, in the high-PUE genotype XN8, ClmiR86 expression was significantly induced after 12 h of phosphate starvation ( Figure 2B). This genotype-dependent response of ClmiR86 expression to phosphate stress indicates its possible involvement in phosphate starvation response and utilization in watermelon.

Figure 2.
ClmiR86 expression varies in roots of watermelon genotypes with different P utilization efficiencies. Expression of ClmiR86 in the roots of (A) inefficient P utilization watermelon genotype "ZJ" and (B) efficient P utilization genotype "XN8" under different P concentrations. Different lower-case letters denote a significant difference in relative expression level (p < 0.05, one-way ANOVA and then Tukey's test for multiple comparisons). Values are means ± SD (n = 3).

ClmiR86 Suppresses ClCIPK5, AtCIPK5, and AtCIPK25 at the Post-Transcription Level
miRNAs regulate the expression of target genes via binding mRNA transcripts, leading to mRNA degradation and/or translational repression. To better understand the biological functions of ClmiR86 in watermelon, the effects of grafting and phosphate deficiency were examined on the expression of ClCIPK5, a putative target gene of ClmiR86. Squash-grafting suppressed ClCIPK5 expression in watermelon leaves compared with auto-grafting. Moreover, phosphate deficiency dramatically enhanced ClCIPK5 accumulation in auto-grafted watermelon leaves, but slightly inhibited its expression in squashgrafted watermelon leaves ( Figure 3A). A similar cyclic variation in ClCIPK5 expression was also observed in the two watermelon genotypes with different PUEs ( Figure 3B,C). The negative correlation between ClCIPK5 and ClmiR86 expression under both grafting and phosphate deficiency indicates that squash-grafting might promote the phosphate starvation response of watermelon via the ClmiR86-ClCIPK5 axis (Figures 1 and 3). ClmiR86 expression varies in roots of watermelon genotypes with different P utilization efficiencies. Expression of ClmiR86 in the roots of (A) inefficient P utilization watermelon genotype "ZJ" and (B) efficient P utilization genotype "XN8" under different P concentrations. Different lowercase letters denote a significant difference in relative expression level (p < 0.05, one-way ANOVA and then Tukey's test for multiple comparisons). Values are means ± SD (n = 3).

ClmiR86
Suppresses ClCIPK5, AtCIPK5, and AtCIPK25 at the Post-Transcription Level miRNAs regulate the expression of target genes via binding mRNA transcripts, leading to mRNA degradation and/or translational repression. To better understand the biological functions of ClmiR86 in watermelon, the effects of grafting and phosphate deficiency were examined on the expression of ClCIPK5, a putative target gene of ClmiR86. Squash-grafting suppressed ClCIPK5 expression in watermelon leaves compared with autografting. Moreover, phosphate deficiency dramatically enhanced ClCIPK5 accumulation in auto-grafted watermelon leaves, but slightly inhibited its expression in squash-grafted watermelon leaves ( Figure 3A). A similar cyclic variation in ClCIPK5 expression was also observed in the two watermelon genotypes with different PUEs ( Figure 3B,C). The negative correlation between ClCIPK5 and ClmiR86 expression under both grafting and phosphate deficiency indicates that squash-grafting might promote the phosphate starvation response of watermelon via the ClmiR86-ClCIPK5 axis (Figures 1 and 3).  To further verify the miRNA-target interaction, we also examined AtCIPK5 and AtCIPK25 expression in three independent transgenic lines (35S::ClmiR86#2, #3, and #5) overexpressing the ClmiR86 precursor in the Arabidopsis Col-0 background. AtCIPK5 and AtCIPK25 were identified as the orthologous genes of ClCIPK5 in Arabidopsis ( Figure 4A), and the putative ClmiR86 target site was also conserved in these two CIPKs ( Figure 4B). qRT-PCR revealed a reduction in AtCIPK5 and AtCIPK25 expression in all p35S::ClmiR86 lines ( Figure 4C). Interestingly, miR86 transcripts were detected in the wildtype, suggesting that this miRNA is present in Arabidopsis. These results indicate that miR86 functions may be conserved between watermelon and Arabidopsis via the post-transcriptional regulation of CIPKs. To further verify the miRNA-target interaction, we also examined AtCIPK5 and AtCIPK25 expression in three independent transgenic lines (35S::ClmiR86#2, #3, and #5) overexpressing the ClmiR86 precursor in the Arabidopsis Col-0 background. AtCIPK5 and AtCIPK25 were identified as the orthologous genes of ClCIPK5 in Arabidopsis ( Figure 4A), and the putative ClmiR86 target site was also conserved in these two CIPKs ( Figure 4B). qRT-PCR revealed a reduction in AtCIPK5 and AtCIPK25 expression in all p35S::ClmiR86 lines ( Figure 4C). Interestingly, miR86 transcripts were detected in the wildtype, suggesting that this miRNA is present in Arabidopsis. These results indicate that miR86 functions may be conserved between watermelon and Arabidopsis via the post-transcriptional regulation of CIPKs.   Values are means ± SD (n = 3). "*" was use to show the 30th and 50th cite of the sequence respectively.

Ectopic ClmiR86 Expression Enhanced Phosphate Starvation Tolerance
To gain further insight into ClmiR86 function in plants, wildtype and transgenic Arabidopsis lines overexpressing ClmiR86 were examined under phosphate starvation. No obvious phenotype was identified in transgenic plants grown under phosphate sufficiency ( Figure 5A). However, 35S::ClmiR86 seedlings showed significantly increased primary root length after 1 week of phosphate starvation compared with the wildtype seedlings ( Figure 5A,B). Furthermore, phosphate starvation significantly decreased lateral root number in the wildtype seedlings but did not decrease this number in 35S::ClmiR86 seedlings ( Figure 5A,C). Moreover, leaf growth in both wildtype and transgenic plants were dramatically inhibited by phosphate starvation, whereas ectopic ClmiR86 expression slightly rescued leaf area under stress ( Figure 5D,E). These results suggest that ClmiR86 overexpression enhances root tolerance to phosphate starvation, and this effect may be associated with its regulation of CIPKs. Interestingly, PHO2, a well-known phosphate starvation-responsive target of miR399, was also downregulated in 35S::ClmiR86 seedlings ( Figure S1A). Moreover, pho2 mutant seedlings showed a root response similar to 35S::ClmiR86 seedlings ( Figures 5 and S1).
To further verify the role of ClmiR86 in phosphate starvation response, we examined the root and inflorescence phenotypes of transgenic Arabidopsis plants under phosphate starvation at the adult stage. After 1 week, 35S::ClmiR86 plants grown under phosphate starvation showed significantly increased root tip number compared with plants grown under phosphate-sufficient conditions. Such differences were not observed in wildtype plants ( Figure 6A,D). Phosphate deficiency did not affect leaf area or primary root length in any line ( Figure 6B,C). When the phosphate starvation period was extended to 17 days, inflorescence length and height of wildtype plants were significantly reduced, although the flowering time remained similar among all lines. ClmiR86 overexpression evidently relieved the inhibitory effects of phosphate starvation on inflorescence elongation ( Figure 6E,F). Moreover, phosphate starvation significantly decreased the phosphate levels in wildtype plants, but not in transgenic plants ( Figure 6G). In addition, wildtype seedlings showed slightly lower PUE than 35S::ClmiR86 seedlings under phosphate sufficiency and phosphate deficiency slightly increased PUE ( Figure 6H). These results suggest a pivotal role of ClmiR86 in phosphate signaling during root and shoot development.    Values are means ± SD (n = 5).

Ectopic ClmiR86 Expression Enhanced Plant Growth under Normal Conditions
To test whether the ClmiR86-ClCIPK5 signaling axis is involved in grafting-enhanced growth and phosphate assimilation in watermelon, we examined the phenotypic differences between wildtype and transgenic Arabidopsis under normal conditions. Following germination on half-strength MS medium for 4 days, 35S::ClmiR86 seeds showed a higher hulling rate than their background ( Figure 7A,B), resulting in a larger leaf area in transgenic seedlings at the early stage (9 DAG) than in wildtype seedlings. However, nearly throughout the vegetative phase, no visible differences in leaf growth were observed between the transgenic and wildtype seedlings ( Figure 7C). Interestingly, 35S::ClmiR86 plants exhibited more vigorous growth than wildtype seedlings after they reached the generative phase (29-30 DAG), resulting in a larger photosynthetic area and longer inflorescence ( Figure 7C,D). To establish a link between ClmiR86-ClCIPK5 signaling and enhanced PUE, we analyzed the expression of 12 phosphate transporter genes (PHTs) using qRT-PCR. Transcripts of several of these genes were upregulated in ClmiR86-overexpressing lines compared with in wildtype seedlings ( Figure S2B). Three CBL genes, namely AtCBL1, AtCBL3, and AtCBL4, interact with CIPK5 [45]. AtCBL1 and AtCBL4 expression was clearly increased in ClmiR86-overexpressing lines, perhaps due to feedback regulation ( Figure S2C). These results indicate that ClmiR86 overexpression promoted plant growth under normal conditions, which is likely related to enhanced phosphate transport via altered CIPK-CBL signaling.

Discussion
Grafting has been adopted as one of the most effective techniques for improving plant stress tolerance and modifying ion accumulation in a number of plant species [2,12,46]. It is also commonly used in horticultural crops to increase nutrient uptake and utilization, which reduces not only crop production costs, but also global environmental pollution. Here, we demonstrated the possible role of ClmiR86, a novel Cucurbitaceae miRNA, as a positive regulator of phosphate deficiency tolerance and grafting-enhanced phosphate utilization in watermelon.

Squash Grafting Enhanced Watermelon Phosphate Utilization
Watermelon is one of the most important economic crops in the world. Squash is commonly used as a rootstock for watermelon grafting, leading to greater production and stronger resistance/tolerance to various stresses. In several studies, squash grafting was found to increase iron, nitrate, potassium, calcium, or magnesium ion concentration in watermelon [8,[10][11][12]. In addition, we previously demonstrated enhanced phosphate assimilation with squash grafting [9]. However, the effects of rootstock and scion on each other and their overall effects on nutrient absorption remain largely unknown. Typically, modification of the root architecture is a popular means of enhancing ion uptake, accumulation, and utilization efficiency. Indeed, the squash rootstock has a vigorous root system, with more lateral roots and root hair, higher total root length, and greater root surface area than watermelon. The root dry weight of pumpkin-grafted watermelon was 2.24 higher than that of self-rooted plants [47]. Some plant metabolites, including sugars, hormones, and miRNAs, may function as long-distance signaling molecules and regulate ion uptake and/or homeostasis by affecting the activity of ion transporters. Unfortunately, research on ion transporters and miRNA-regulated nutrient uptake has focused on model and non-grafted plants, and information on grafted horticultural crops is limited. In the present study, the link between squash grafting-enhanced phosphate utilization and ClmiR86 accumulation (Figure 1; [9]) provided an important clue for elucidating the mechanism of grafting-enhanced abiotic stress tolerance and developing improved nutrient-efficient rootstocks.

ClmiR86 Is Involved in Phosphate Starvation Response and Phosphate Utilization in Watermelon
The roles of miRNAs in nutrient acquisition, transport, and homeostasis have been summarized in the literature [27,48]. Some miRNAs (e.g., miR160, miR167, miR393, miR171, and miR3979) have been proposed to play roles in nitrate uptake, possibly by promoting or suppressing root functions (primary root elongation and lateral or adventitious root emergence) [49][50][51][52]. miR169, miR156, and miR826/miR5090 affect nitrate uptake, metabolism, and consumption [53][54][55]. In addition, miR444, which targets MADS23, may be involved in nitrate and potassium uptake or translocation from old to young leaves [27,56]. Recently, miR399 and miR827 were demonstrated to be important for the phosphate starvation response, as they inhibit genes involved in PHT1 and PHO1 suppression, contributing to their accumulation under stress [25][26][27]. PHT1 proteins can use energy to cotransport Pi and H + ions and are therefore involved in inorganic phosphate acquisition [57]. PHO1 is involved in the loading of the acquired inorganic phosphate into the xylem and facilitating the root-to-shoot transport of this macronutrient in plants [58,59]. The miR399-PHO2 and miR827-NLA pathways are likely coordinated to allow for fine-tuning of the expression of the high-affinity Pi transporter PHT1 for maintaining phosphate homeostasis [27,[60][61][62].
Here, we reported the novel watermelon miRNA ClmiR86, which was induced by grafting and phosphate deficiency (Figures 1 and 2). Ectopic expression of this miRNA in Arabidopsis enhanced phosphate deficiency tolerance based on its promoting effect on primary root elongation and lateral root emergence ( Figure 5). miR86 function may be conserved between watermelon and Arabidopsis, as mature miR86 was also detected in wildtype Arabidopsis plants ( Figure S2A). Interestingly, this root phenotype was similar to that of the phosphate-over-accumulating mutant pho2 grown under the same conditions ( Figure S1), suggesting that ClmiR86-triggered phosphate deficiency tolerance is due to enhanced phosphate assimilation. In addition, the expression of several phosphate acquisition-related genes was affected in transgenic Arabidopsis seedlings ( Figure S2B), similar to that in the pho2 mutant [62], suggesting that ClmiR86 is involved in phosphate signaling via the functions of these genes.
Squash grafting enhanced ClmiR86 accumulation in scion leaves compared with autografting ( Figure 1A), which is consistent with the results observed by Liu et al. [1] that grafting altered miRNA expression levels in the leaves of pumpkin-grafted watermelon seedlings. In addition, our findings implied that ClmiR86 might act as a graft-transmissible signal exported from the squash rootstock to watermelon leaves or reverse responses to both squash grafting and phosphate deficiency in leaves and roots (Figure 1). Similarly, Li et al. [63] showed that the expression of some miRNAs in the leaves and roots differed between pumpkin-and auto-grafted cucumbers as well as between seedlings grown under nitrate and phosphate deficiency. Among these, the expression levels of miR399 and csa-miR-n08 were increased in leaves, but decreased in the roots of pumpkin-grafted cucumber seedlings compared with those of auto-grafted seedlings. The altered miRNA levels then affected the expression of MYB transcription factors, including phosphate starvation regulator 1 (PHR1), phosphate starvation regulator 1-like (PHR1-LIKE1), and E3 (ubiquitin-protein ligase) in cucumber, which are essential for phosphate absorption [63]. Similar response patterns of ClmiR86 and csa-miR-n08 to both grafting and phosphate deficiency in scion and rootstock indicate shared or conserved regulatory network in squash-grafted watermelon and cucumber under phosphate starvation.

CIPK5 Works Downstream of ClmiR86 under Phosphate Starvation and Normal Conditions
In plants, the CBL-CIPK signal components form a complex signaling network, which allows for flexible but specific signal-response coupling during adaptation to environmental stresses and nutrient deficiency [64]. The CBL-CIPK pathways have been reported to function as nutrient transport and homeostasis regulatory networks in response to low potassium [65,66], sodium [41,42], magnesium [67], nitrate [43,68], and phosphate [36]. In the present study, watermelon CIPK5 was found to be involved in early phosphate signaling, and ClCIPK5 transcript levels were rapidly decreased by squash grafting (Figure 3). In addition, ClmiR86 overexpression significantly decreased AtCIPK5 and AtCIPK25 (orthologs of ClCIPK5) expression, perhaps due to enhanced plant tolerance to phosphate deficiency (Figures 4-6). Recently, Brassica napus CIPK (BnCIPK6) was also found to be involved in the phosphate starvation response. BnCIPK6 expression was strongly induced by phosphate deficiency, and transgenic Arabidopsis seedlings overexpressing BnCIPK6 grew better than the wildtype ones under phosphate starvation [36]. These results suggest a positive function of CIPK in phosphate deficiency tolerance, contrary to that of ClCIPK5. Orthologous genes of BnCIPK6 in Arabidopsis and chickpea (AtCIPK6 of CaCIPK6) were found to affect lateral root formation and salinity tolerance, possibly via their function in auxin transport and sensitivity, highlighting the role of CIPKs in root development [69]. This is partly consistent with our findings that reduced ClCIPK5 transcript levels in transgenic Arabidopsis may result in significant lateral root differences (Figures 5 and 6). A recent study indicated that the CIPK23-CBL9 complex could regulate primary nitrate responses via its phosphorylation function on the dual-affinity nitrate transporter CHL1 [43]. Of note, even though no functional analysis of CIPK5 or CIPK25 has been performed in Arabidopsis, CIPK5 was found to weakly interact with CBL1 in a yeast two-hybrid system [70]. This finding indicates a possible indirect interaction between CIPK5 and CIPK23 during complex formation as well as a cross-talk between nitrate and phosphate starvation responses. Unfortunately, the downstream substrates of CIPK6, CIPK5, or CIPK25, which might directly affect phosphate sensing or transport, have not been identified. Knowledge of these substrates will become a critical piece for solving the jigsaw puzzle of plant phosphate starvation responses. Thus, the CIPK5 downstream substrates mediating interaction with CBL must be identified to unveil the mechanism of grafting-triggered phosphate deficiency tolerance and phosphate assimilation.
In conclusion, we identified a novel grafting-induced or grafting-transportable miRNA in watermelon. ClmiR86 likely targets the transcripts encoding CIPK5 and CIPK25, which are involved in phosphorus sensing or stress signaling. Plants overexpressing ClmiR86 showed decreased expression of the target CIPK genes, enhanced lateral root growth under phosphate deficiency, and increased growth vigor under normal conditions. These findings provide valuable clues to further explore the mechanism of grafting-enhanced stress tolerance and nutrient utilization.

Plant Materials, Growth Conditions, and Treatments
The watermelon cultivar 'Zaojia' (ZJ) was grafted onto the squash cultivar 'Feichangfuzuo' (Wm/Sq) and watermelon (Wm/Wm). Self-rooted squash plants were used as controls for squash roots. Two watermelon genotypes, namely 'Xinong 8 Hao' (XN8) and ZJ, with different PUE were used to characterize the genotypic variations in ClmiR86 and ClCIPK5 expression [9]. Following seed germination, all watermelon and squash seedlings were grown in the Professional Growing Mix (Fafard ® 51 L Mix). Grafting was performed when the scion and rootstocks were at the cotyledon stage and one-leaf-stage, respectively. All plants were cultivated in a growth chamber under photosynthetically active radiation at 28 • C (day)/24 • C (night) under a 16 h light period at an intensity of 600 µmol·m −2 ·s −1 and 50-85% humidity. Phosphate starvation treatments were performed until the grafted plants and the two watermelon genotypes reached the two-true-leaf stage.
Under phosphate deficiency stress, all watermelon seedlings were transferred to hydroponic growth containers with full-strength Hoagland solution (0.3 mM phosphate; PH 6) and acclimated for 3 days. The seedlings were then subjected to phosphate starvation (0.01 mM) (Table S1). After 6 h of treatment, the grafted watermelon leaves and squash roots were collected. Root samples were also harvested from XN8 and ZJ at specific time intervals (6, 12, and 24 h). All samples were stored at −80 • C for further assays.
Seeds of Arabidopsis pho2 mutants were kindly provided by Professor Huixia Shou (Zhejiang university, China) [71]. Arabidopsis thaliana ecotype Columbia-0 (WT) was used to generate ClmiR86 overexpression lines (35S::ClmiR86). For growth and development studies, WT and 35S::ClmiR86 Arabidopsis seeds were sown on the Jiffy seedling culture substrate, stratified at 4 • C for 3 days and then grown in growth chambers (Sanyo; http://www.sanyobiomedical.co.uk, accessed on 3 September 2021) at 22 • C under a 16 h photoperiod at 200 µmol·m −2 ·s −1 light intensity and 60% relative humidity. Rosette leaf samples were collected and stored at −80 • C for further assays.
For the Arabidopsis phosphate starvation experiment at the seedling stage, WT and 35S::ClmiR86, pho2 mutant seeds were surface sterilized twice with 20% (v/v) bleach for 1 min. After washing three times with sterile distilled water, the seeds were cultured in phosphate-sufficient (P + ) medium containing half-strength MS medium (PhytoTechnology Laboratories), 1% (w/v) sucrose, and 1.2% (w/v) agar; vernalized at 4 • C for 3 days; and grown vertically in a growth chamber for 1 week (Sanyo). Thereafter, some 1-weekold plants were transferred onto a phosphate-deficient (P − ) medium, in which 1.25 mM KH 2 PO 4 in the P + medium was replaced with 0.65 mM K 2 SO 4 , while the remaining plants grown on P + and half-strength MS medium as the controls (Table S2). After 1 week of culture, leaf area, primary root length, and lateral root number were measured and photographed. Six seedlings were measured for each treatment, and three biological replicates were performed.
For Arabidopsis hydroponic culture at the adult stage, WT and 35S::ClmiR86 Arabidopsis adult-plants (22-day old) were transferred to hydroponic culture with half-strength Hoagland solution containing 0.3 mM phosphate for additional 7 or 17 days. Phosphate starvation was initiated by replacing 0.3 mM phosphate with 0.01 mM phosphate (Table S1).

Vector Construction
To generate 35S::ClmiR86 Arabidopsis lines, a 272 bp fragment containing the ClmiR86 precursor foldback structure was amplified using total RNA extracted from watermelon leaves with the primer pair Pre-miR86-F and Pre-miR86-R; cloned into pDONR221 (Invitrogen); and recombined into the GatewayTM vector pMDC83 binary expression vector under the control of the CaMV 35S promoter using the Gateway cloning system [72]. The primers used for vector construction are shown in Table S3.

Arabidopsis Transformation
The ClmiR86-pMDC83 construct was introduced into Agrobacterium tumefaciens strain GV3101. Wildtype Arabidopsis was transformed using the floral dip method [73]. To screen for transgenic plants, seeds harvested from transformed Arabidopsis plants were germinated on MS medium containing 50 µg·mL −1 hygromycin B (Roche, http://www.roche.com; accessed on 10 September 2021 ), vernalized at 4 • C for 3 days, and incubated in a growth chamber (Sanyo) for 10 days. The 10-day-old hygromycin B-resistant plants were then transferred to the Jiffy seedling culture substrate. PCR and quantitative real-time PCR (qRT-PCR) of ClmiR86 in candidate transgenic Arabidopsis plants were used to confirm the success of transformation ( Figures S1 and S2A). WT and T3 plants were used in the present study. All primers used for the confirmation of Arabidopsis transformation are listed in Table S3.

Target Gene Prediction
The ClmiR86 mature sequence was used as the query to search for putative targets against the watermelon (ICuGI; http://www.icugi.org/cgi-bin/ICuGI/index.cgi; accessed on 10 September 2021 ) and Arabidopsis (TAIR; http://www.arabidopsis.org; accessed on 10 September 2021 ) databases with TagetFinder 1.6. To detect whether the target sites were conserved in Arabidopsis, the coding sequences of ClCIPK5, AtCIPK5, and AtCIPK25 in the genomes of watermelon and Arabidopsis were retrieved from the ICuGI and TAIR databases, respectively. A multiple sequence alignment of the nucleotide sequences of CIPKs (ClCIPK5, AtCIPK5, and AtCIPK25) and ClmiR86 was performed using ClustalX 2.0 with default settings, as described by Thompson et al. [74].

Phylogenetic Tree Construction
To identify the orthologous genes of ClCIPK5 in Arabidopsis, the amino acid sequences of 26 CIPKs in the genome of Arabidopsis were retrieved from the TAIR database (http://www.arabidopsis.org; accessed on 10 September 2021 ). A neighbor-joining tree based on the protein sequence alignments was constructed using MEGA 6.0, with 1000 bootstrap replicates.

PUE Measurement
Shoots of the WT and 35S::ClmiR86 seedlings grown in either P + or P − medium or soil were sampled separately. Phosphate levels were measured using the vanadium-molybdenum method [75]. PUE was calculated as plant total dry weight divided by phosphate levels [76].

RNA Isolation, cDNA Synthesis, and qRT-PCR
Total RNA, including miRNA, was extracted using the mirVana TM miRNA Isolation Kit (Ambion). RNA quantity and quality were assessed spectrophotometrically using the Thermo 2000 Bioanalyzer with RNA NanoDrop (Thermo Scientific; http://www.thermo. com; accessed on 10 September 2021 ). Samples showing A260/A230 ratios of 2.0-2.2 and A260/A280 ratios of 1.8-2.0 were used for further analysis. For quantitative real-time PCR, 1 µg of total RNA was reverse transcribed to first-strand cDNA in a final reaction volume of 20 µL using the miScript II RT Kit (Qiagen).
The ClmiR86 forward primer was designed based on its mature sequence, and the universal reverse primer was provided with the miScript SYBR Green PCR Kit (Qiagen).
miR167c was used as a reference gene to normalize miR86 expression in watermelon and squash [77]. Primers specific to the ClmiR86 precursor were used to detect Pre-miR86 expression levels. Pre-miR86 and ClCIPK5 expression in watermelon was normalized to yellow-leaf-specific protein 8 (ClYLS8) expression. AtUBQ10 was selected as a reference gene for Arabidopsis gene expression [25]. All primer sequences are listed in Table S4. qRT-PCR was performed on the StepOnePlus™ Real-Time PCR System (ABI) using the miScript SYBR Green PCR Kit (Qiagen). PCR included pre-incubation at 95 • C for 15 min, followed by 40 cycles of denaturation at 94 • C for 15 s, annealing at 55 • C for 30 s, and extension at 70 • C for 30 s. Primer amplification specificity and efficiency were measured by melt curve and standard curve analysis, respectively. The relative expression levels were calculated using the 2 −∆∆CT method. Each experiment was performed in triplicate.

Statistical Analysis
For the statistical analysis, one-way ANOVA followed by Tukey post hoc test were performed using SPSS Statistics Version 22.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/plants10102133/s1, Figure S1. Root phenotypes of pho2 mutant under P starvation condition. Figure S2. Expression of CIPK, CBL and PHT genes in transgenic and wild type Arabidopsis. Figure S3. PCR confirmation of ClmiR86 transformation in candidate transgenic Arabidopsis. Table S1. Formula of Hoagland nutrient solution. Table S2. Formula of full-strength MS medium. Table S3. Primers used for vector construction and confirmation of Arabidopsis transformation. Table S4. Gene-specific primers for qRT-PCR.