An efficient root transformation system for CRISPR/Cas9-based analyses of shoot–root communication in cucurbit crops

Abstract Cucurbit crops are suitable models for studying long-distance signaling in horticultural plants. Although thousands of substances are graft transmissible in cucurbits, functional studies have been hampered by the lack of efficient genetic transformation systems. Here, we report a convenient and efficient root transformation method for several cucurbit crops that will facilitate studies of functional genes and shoot–root crosstalk. We obtained healthy plants with completely transformed roots and non-transgenic shoots within 6 weeks. Furthermore, we combined this root transformation method with grafting, which allowed for gene manipulation in the rootstock. We validated our system by exploring salt tolerance mechanisms using a cucumber (Cucumis sativus)/pumpkin (Cucurbita moschata Duch.) (scion/rootstock) graft in which the sodium transporter gene High-affinity K+ transporter1 (CmoHKT1;1) was edited in the pumpkin rootstock and by overexpressing the pumpkin tonoplast Na+/H+ antiporter gene Sodium hydrogen exchanger4 (CmoNHX4) in cucumber roots.


Introduction
Cucurbits are economically important crops because of their dietary and medicinal value [1]. Grafting of a high-yield scion onto a stress-resistant rootstock has been used to improve crop production for thousands of years, and grafting is also an ideal approach for exploring shoot-root crosstalk in plants [2][3][4]. Grafting among the different genera of Cucurbitaceae is straightforward, and their phloem and xylem saps can be easily obtained; therefore, cucurbit crops are excellent models for studying long-distance signaling and graft-transmissible signals [5][6][7]. However, the lack of an efficient transgene system in these crops has limited long-distance signaling studies. Although stable transformation systems have been established in watermelon (Citrullus lanatus) and cucumber (Cucumis sativus) [8,9], they are lacking in other cucurbit species, such as pumpkin (Cucurbita moschata Duch.), melon (Cucumis melo), bottle gourd (Lagenaria siceraria), and luffa gourd (Luffa acutangula). Agrobacterium rhizogenes-mediated root transformation is a rapid and convenient alternative to conventional stable transformation procedures for gene manipulation in vivo [10,11]. Moreover, the genetic modification of rootstock roots could offer a convenient way to investigate crosstalk between rootstocks and scions. Nonetheless, an efficient A. rhizogenes-mediated root transformation system has not been established for most cucurbit crops.
Grafting is a practical approach for overcoming soilborne diseases and increasing abiotic stress resistance in crops [12]. Pumpkin is the most widely used rootstock for cucurbit crops [7,12,13]. For instance, grafting onto pumpkin rootstock has been used to increase the salt tolerance of a cucumber scion. In that case, the mechanism of enhanced salt tolerance was related to the higher capacity of pumpkin roots to limit the longdistance transport of Na + to shoots compared to that of cucumber roots [13]. X-ray analysis revealed that much more Na + is sequestered in the cortex of pumpkin roots than in cucumber roots, suggesting that the transport of Na + to the stele is restricted; thus, its long-distance transport to shoots is limited [14]. The higher salt tolerance of pumpkin is also associated with a higher K + uptake capacity relative to that of cucumber [5]. However, the molecular mechanism and the genes responsible for these differences between pumpkin and cucumber are largely unknown. Here, we developed a convenient, efficient, and rapid root transformation system for pumpkin and several other cucurbit crops. Healthy plants with roots that had been completely transformed were obtained within 6 weeks. This root transformation system was combined with grafting to take advantage of gene manipulation in the rootstock. We demonstrated the feasibility of our method using the clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) genome editing technology to knock out the sodium transporter gene CmoHKT1;1 in the pumpkin rootstock of cucumber/pumpkin grafted plants and by overexpressing the pumpkin tonoplast Na + /H + antiporter gene CmoNHX4 in cucumber roots. Our results demonstrate that this robust root transformation system is efficient for functional analysis of genes in the roots of cucurbit crops and also provides a promising approach for studying shoot-root communication.

An efficient root transformation system for pumpkin and several other cucurbit crops
Constitutively expressed green fluorescent protein (GFP) was used as a reporter for transformation and to facilitate the selection of transgenic roots. The transformation was performed as shown in Supplementary Fig. S1a, with an optimized A. rhizogenes strain, concentration of Agrobacterium solution, age of seedlings for injection, and co-cultivation time ( Supplementary Fig. S1b-e). In brief, hypocotyls of pumpkin seedlings whose cotyledons had just unfolded were cut at angles approximately 2 to 3 cm below the cotyledons. These excised apical parts were used as explants for transformation (Fig. 1a). The cut end of the hypocotyl was immersed in 1 mL of A. rhizogenes strain K599 bacterial solution containing 35S:GFP for 30 min for infection (Fig. 1b). The infected explants were then cultivated in boxes containing sterilized vermiculite moistened with MS medium (Fig. 1c). After 4-d cultivation in the dark at 23 • C, GFP fluorescence was examined. We observed that A. rhizogenes successfully infected the cut ends of the pumpkin hypocotyls ( Fig. 1d) with an infection rate (percentage of hypocotyls with GFP fluorescence after cultivation) of approximately 95%. The explants were then transferred to trays containing sterilized vermiculite moistened with half-strength Hoagland solution. The trays were covered to maintain high humidity to promote the development of hairy roots. After 7 d, the lid was removed, roots without GFP fluorescence were excised under UV light, and the plants were returned to the vermiculite for another week. When the length of the GFP-fluorescent roots was greater than 2 cm, the plants were cultured hydroponically to more easily reveal the non-transgenic roots and allow their removal without damaging the plants. It was important to remove the non-transgenic roots at the early stages to promote the growth of transgenic roots. After the transgenic roots were fully developed, the growth of non-transgenic adventitious roots from the hypocotyl was greatly reduced. After approximately 2 weeks in hydroponic culture, healthy composite pumpkin plants with completely transgenic roots were obtained with a survival

The root transformation system can be combined with grafting
Cucurbit crops are useful for studying grafting and graft-transmissible signals [6], and grafting a rootstock with a scion from another species is a common way to characterize mobile signals [2]. We tested whether our root transformation system could be combined with grafting, reasoning that it could provide a much needed and convenient way to genetically modify rootstocks in grafted plants. We used the cucumber/pumpkin (scion/rootstock) graft combination for our study. Pumpkin seedlings whose cotyledons had just unfolded were used as the rootstock because this stage is ideal for our A. rhizogenes-mediated transformation method. The pumpkin seedlings were cut at the hypocotyls as described above.
Grafting was performed using the excised apical part with the hole insertion grafting method [15,16]. Freshly made grafts were used as explants for A. rhizogenesmediated transformation (Fig. 2a). The transformation and cultivation steps were performed as described above for pumpkin. After that, the cucumber/pumpkin grafts were transferred to trays containing sterilized vermiculite moistened with half-strength Hoagland solution and covered with transparent plastic lids to maintain high humidity. One week of growth in highhumidity conditions was essential, as it promoted both hairy root development and post-graft healing. For cucumber/pumpkin grafts, the infection rate of 96.1% was very similar to that of pumpkin (Fig. 2b), but the survival rate of 78.56% was lower (Fig. 2c). These results showed that the root transformation system could be combined with grafting without significantly reducing its infection and survival rates.
CmoHKT1;1 contributes to salinity tolerance of cucumber/pumpkin grafts by limiting long-distance Na + transport Grafting onto pumpkin has been reported to increase the salt tolerance of cucumber because the pumpkin rootstock limits the transport of Na + from the roots to the shoots and thus prevents the accumulation of toxic Na + in the cucumber scion [13]. The sodium transporter gene CmoHKT1;1 plays a role in the longdistance transport of Na + from roots to shoots [17]. However, the function of CmoHKT1;1 has not been tested in pumpkin owing to the lack of an efficient and stable transformation system. In this study, we generated mutations in CmoHKT1;1 in the pumpkin roots of cucumber/pumpkin grafts using CRISPR/Cas9 technology. The pBSE403G-sgRNA CmoHKT1;1 vector was constructed by placing a CmoHKT1;1-targeting sgRNA in pBSE403G, which was constructed by inserting 35S:GFP into pBSE401 [18]. 35S:GFP was used to verify transformation. Cucumber/pumpkin grafts were transformed using A. rhizogenes containing pBSE403G and pBSE403G-sgRNA CmoHKT1;1 vectors as described above. Four weeks after transformation, we used Hi-TOM sequencing to assess the gene-editing efficiency in the transgenic roots that were positive for GFP fluorescence [19]. The gene-editing efficiency of CmoHKT1;1 reached 80.14% (Fig. 3a). The role of CmoHKT1;1 was further evaluated with a 75 mM salt treatment, as described previously [17]. We determined that the CmoHKT1;1 CR plants (i.e. plants transformed with pBSE403G-sgRNA CmoHKT1;1 ) were more sensitive to salt stress than the control plants (i.e. plants transformed with the pBSE403G empty vector) (Fig. 3b). Moreover, when subjected to salt stress, the CmoHKT1;1 CR plants accumulated more Na + in the shoots and less Na + in the roots than the control plants ( Fig. 3c and d), suggesting disturbed Na + transport from roots to shoots in CmoHKT1;1 CR plants. Meanwhile, when plants were subjected to salt stress, the Na + content was higher in the xylem sap of the CmoHKT1;1 CR plants relative to the control plants ( Fig. 3e and f). These data support the conclusion that CmoHKT1;1 contributes to salinity tolerance by limiting the long-distance transport of Na + in cucumber/pumpkin grafts.

Pumpkin CmoNHX4 confers salinity tolerance to cucumber roots
We also tested whether our method could be used for cell biological analyses in the roots of cucurbit crops. The higher salt tolerance of pumpkin relative to cucumber is associated with the higher K + uptake capacity of pumpkin [5]. The tonoplast Na + /H + antiporter AtNHX1 mediates potassium uptake [20]. In this study, we observed that the expression of CmoNHX4-a homolog of AtNHX1-was induced by NaCl treatment, and it was the most highly expressed member of the CmoNHX gene family in pumpkin roots treated with NaCl ( Fig. 4a and Fig. S4). We therefore tested whether the overexpression of CmoNHX4 in cucumber could improve salt tolerance. Vectors were constructed by inserting the CmoNHX4 coding sequence with or without a GFP tag downstream of the 35S promoter (Fig. S5a,b). 35S:DsRed2 was used to monitor transformation [21]. The vectors were transformed as described above. The 35S:CmoNHX4-GFP vector was used for the subcellular localization of CmoNHX4-GFP. The expression of CmoNHX4-GFP in cucumber roots was low and inconsistent ( Fig. S6a,b), similar to 35S:AtNHX1-GFP expression in the roots of transgenic Arabidopsis [22]. According to a previous report [22], the GFP signal within the vacuolar lumen and endosomes in cells with bright GFP expression may be the result of 35S promoter-driven overexpression (Fig. S6a,b). However, the CmoNHX4-GFP signal was found on the vacuolar membrane in cells with lower GFP signal, as evidenced by nuclei stained with DAPI detected outside the membrane boundary (Fig. 4b). The expression pattern of 35S:CmoNHX4-GFP in pumpkin roots was similar to that in cucumber roots, but the fluorescence intensity in pumpkin roots was lower (Fig. S6c-e). The intensity of the GFP signal diminished after plasmolysis of root cells induced by sucrose, with the vacuolar membrane-located GFP signal barely detectable and only a GFP signal within the vacuolar lumen or endosomal bodies observed outside the nuclei (Fig. S6f). The tonoplast membrane marker 35S:AtTIP1-RFP was then transiently co-expressed with 35S:CmoNHX4-GFP in Nicotiana leaves [23]. The colocalization of CmoNHX4-GFP with AtTIP1-RFP provided additional evidence that CmoNHX4-GFP was localized on the vacuolar membrane (Fig. S6g). The 35S:CmoNHX4 vector (without GFP tag) was used to overexpress CmoNHX4, and the resulting cucumber seedlings with transformed roots were used for phenotypic analysis and CoroNa green staining. Real-time PCR analysis showed that CmoNHX4 expression increased significantly in roots transformed with the 35S:CmoNHX4 vector compared with the empty vector (Fig. S5c). The overexpression of CmoNHX4 in cucumber roots significantly improved the salt tolerance of cucumber (Fig. 4c). We measured the Na + and K + content in control and CmoNHX4-overexpressing roots and found that CmoNHX4 overexpression decreased the Na + content and increased the K + content of cucumber roots ( Fig. 4d and e). We used CoroNa staining [24] to assess the accumulation of Na + in the vacuoles of cucumber roots and observed that the CmoNHX4-overexpressing roots accumulated less Na + in their vacuoles than the control roots (Fig. 4f). These data indicate that CmoNHX4 conferred salinity tolerance to cucumber plants by increasing the K + content and decreasing the Na + content of cucumber roots.
In conclusion, we developed a robust root transformation system for cucurbit crops. We obtained composite plants with transgenic roots using the apical parts of seedlings as explants. A hydroponic culture procedure allowed us to readily identify non-transgenic roots and excise them without damaging the plants. By combining our transformation system with grafting, we provide a promising approach for studying shoot-root communication. Using this system, we further explored the mechanism of higher salt tolerance in pumpkin compared with cucumber by revealing the roles of CmoHKT1;1 and CmoNHX4.

Discussion
Stable genetic transformation mediated by Agrobacterium tumefaciens is time-consuming, labor-intensive, and a major constraint in the functional genomics of cucurbit crops. In addition, the recalcitrant nature of some cucurbits reduces their transformation efficiencies, and an alternative technique is required to address these issues [1,25]. Hairy root transformation has been successfully performed in hundreds of plant species [26][27][28], including legumes like soybean [11] and peanuts [29], vegetables like potato [30] and tomato [31], and even fruits such as avocado [32], grapevine [33], and Prunus [34]. The hairy root transformation system has been used in biological research for decades and has proven to be a viable tool for studying gene function.
In the current study, we successfully developed a hairy root transgenic system that may be deployed to investigate gene function in cucurbit crops. This approach highlights the basic need for genetic verification systems, particularly in pumpkin, melon, bottle gourd, and luffa gourd, which currently lack highly efficient stable transgenic systems. This root transformation system, together with stably transformed roots, could be used for gene functional studies in the roots of cucurbit crops using overexpression and knockout techniques, promoting research in the fields of root metabolites, nutrient acquisition, root development, and root resistance to soilborne disease and abiotic stress. It could also be used for cell biological analyses in roots, such as subcellular localization and cell staining. Moreover, this root transformation system was used in combination with grafting to explore root-shoot communication in grafted cucurbit plants. However, there were a few limitations, such as the inability to obtain homozygous mutations edited by CRISPR/Cas9 and the fact that the resulting mutations could not be transmitted to the next generation. Nevertheless, its convenience, speed, and high efficiency make this system a promising tool for the preliminary validation of candidate genes before performing stable transformation. Our method of studying scion-rootstock communication by integrating the root transformation system with a grafting technique could be used in other crops such as flowers and fruit trees.

Vector construction
The pBSE403G vector was constructed by inserting 35S:GFP:Terminator into the EcoRI site of pBSE401 [18]. The sgRNA targeting CmoHKT1;1 was designed using Geneious software as described previously [8]. The CmoHKT1;1-targeting sgRNA was cloned into pBSE403G as described previously [18]. For the overexpression of CmoNHX4, the intermediate vector pBSE401R was first constructed by inserting 35S:DsRed2:Terminator [21] into the EcoRI site of the pBSE401 vector. Then, CmoNHX4 or CmoNHX4-GFP was cloned into pBSE401R between the XbaI/SacI sites and downstream of the 35S promoter. The pBSE401R-CmoNHX4 vector was used for phenotypic analysis and CoroNa green staining. pBSE401R-CmoNHX4-GFP was used to study the subcellular localization of CmoNHX4. Maps of the two vectors are provided in Fig. S5.

A. rhizogenes-mediated root transformation system
Seeds were soaked in water at 55 • C for 5 h and then incubated in a Petri dish at 28 • C in the dark until germination. A. rhizogenes strain K599 from cultures with OD 600 values between 0.8 and 1.0 were collected and re-suspended in MS medium containing 200 μM AS for bacterial infection. The subsequent steps are described in detail in the supplementary data. The hole insertion grafting method was used for grafting [15,16]. In brief, we removed the apical meristem of the rootstock, leaving the cotyledons, and made a hole on the top portion with a grafting needle. Then, the cucumber scion was cut at angles on both sides of the hypocotyls and inserted into the hole made in the rootstock.

DNA isolation and Hi-TOM analysis
About 2 weeks after co-cultivation, healthy composite plants with completely transgenic roots were used to analyze the efficiency of gene editing. The GFPpositive transgenic roots were used for DNA extraction with the CTAB protocol. The fragment containing the target site was PCR amplified and analyzed by Hi-TOM sequencing [19].

Hydroponic cultivation and salt treatment
For hydroponic culture, plants were grown in halfstrength Hoagland solution in plastic boxes (380 mm × 245 mm × 100 mm) as described previously [17]. Foam boards were used to anchor and support the plants. The Hoagland solution was renewed once a week. For the salt treatment, healthy plants with completely transgenic GFP or DsRed2 fluorescent roots and consistent growth were selected and grown in 5 L half-strength Hoagland solution containing 0 or 75 mM NaCl for 5 days. Three biological replicates per treatment were used for the experiment.

Measurements of Na + and K + content
Three plants per treatment were harvested. The roots and shoots were separated and placed into Kraft paper bags, which were then placed at 105 • C for 15 min and 80 • C until dry. The dried samples were ground into powder and weighed. Then, 0.1 g of powder was digested in 5 mL H 2 SO 4 at 300 • C for 1 h, and a drop of H 2 O 2 was added every half hour until the solution became colorless and transparent. The digested solution was diluted and used to determine the concentrations of Na + and K + using an atomic absorption spectrophotometer (6300C, Shimadzu). For the determination of Na + concentration in xylem sap, 20 μL xylem sap was diluted in 5 mL ddH 2 O and directly measured without digestion. Three biological replicates per treatment were used for the experiment.

Gene expression analysis by qRT-PCR
Cucumber roots transformed with an empty vector or a pBSE401R-CmoNHX4 vector were harvested and immediately frozen in liquid nitrogen for gene expression analysis. CmoNHX4 expression was measured according to a previously described method [35]. Primer3Plus software was used to generate specific primer sequences as listed in supplementary Table S1.

Xylem sap collection
Cross sections of the salt-treated and control plants were cut 1 cm above the graft union. The wound surface was then wiped with filter paper two to three times, and the xylem fluid, which flowed out because of root pressure, was carefully removed with a pipette and placed into a 1.5-mL centrifuge tube for subsequent ion determination.

Light microscopy and CoroNa staining
Na + content in pumpkin root vacuoles was measured with the green fluorescent Na + dye CoroNa as described previously [24]. Transgenic roots positive for 35S:DsRed2 were used for this analysis. The DMSO-dissolved dye was diluted to 20 μM with the measuring buffer (10 mM KCl, 5 mM Ca 2+ -MES, pH 6.1). Excised root tip segments (∼1 cm) were immersed and incubated in the dark for 2.5 h for staining. The stained roots were washed three times with measuring buffer and then observed with a confocal laser scanning microscope (SP8; Leica Microsystems). Colocalization of CmoNHX4-GFP and AtTIP-RFP was evaluated in tobacco leaves as described previously [17]. The nuclei were stained with 10 μg/mL DAPI solution for 10 min.