Na+ Sensitivity of the KAT2-Like Channel Is a Common Feature of Cucurbits and Depends on the S5-P-S6 Segment

Inhibition of Shaker K + channel activity by external Na + was previously reported in the melon ( Cucumis melo L.) inwardly-rectifying K + channel MIRK and was hypothesized to contribute to salt tolerance. In this study, two inward Shaker K + channels, CsKAT2 from cucumber ( Cucumis sativus ) and ClKAT2 from watermelon ( Citrullus lanatus ) were identified and characterized in Xenopus oocytes. Both channels were inwardly-rectifying K + channels with higher permeability to potassium than other monovalent cations and more active when external pH was acidic. Similarly to MIRK, their activity displayed an inhibition by external Na + , thus suggesting a common feature in Cucurbitaceae ( Cucumis spp., Citrullus spp.). CsKAT2 and ClKAT2 are highly expressed in guard cells. After 24 h of plant treatment with 100 mM NaCl, the three KAT2 -like genes were significantly down-regulated in leaves and guard cells. Reciprocal chimeras were obtained between MIRK and Na + -insensitive AtKAT2 cDNAs. The chimera where the MIRK S5-P-S6 segment was replaced by that from AtKAT2 no longer showed Na + sensitivity, while the inverse chimera gained Na + sensitivity. These results provide evidence that the molecular basis of the channel blockage by Na + is located in the S5-P-S6 region. Comparison of the electrostatic property in the S5-P-S6 region in AtKAT2 and MIRK revealed four key amino-acid residues potentially governing Na + sensitivity. with Amber20.


Introduction
Salt stress hinders plant development and leads to wholesale reduction of agricultural productivity.
For plants, the sodium ion is harmful since high levels of Na + in the cytoplasm, in contrast to K + , can disrupt various enzymatic functions (Benito et al. 2014). Due to physicochemical similarities between Na + and K + , excess Na + will tend to substitute K + for Na + at binding sites and impair cellular biochemistry (Bhandal and Malik 1988;Serrano 1996;Tester and Davenport 2003;Maathuis 2006). Therefore, maintaining a low concentration of Na + or high K + /Na + ratio in the cytosol of plant cells through regulation of K + and/or Na + transporters or channels (Kronzucker and Britto, 2011;Benito et al. 2014;Demidchik 2014) is essential for the adaptation of plants to saline conditions (Asch et al. 2000;Tester and Davenport 2003).
Regarding K + channels, particularly in the Shaker family, regulation by saline conditions has mostly been studied at the transcriptional level. The abundance of Shaker family transcripts differs greatly in response to saline stress (Su et al. 2001;Golldack et al. 2003;Pilot et al. 2003a;Zhang et al. 2006;Obata et al. 2007;Escalante-Pérez et al. 2009). For instance, Arabidopsis AtKC1, a general modulator of inward Shaker channel activity (Jeanguenin et al. 2011), its transcript accumulation was strongly induced in leaves under salt stress, suggesting a key role of the protein in adaption to saline conditions (Pilot et al. 2003a). In barley, HvAKT1 in the elongation zone of leaves was also induced by sodium, contributing to the maintenance of K + concentration in mesophyll cells under salinity (Boscari et al. 2009). Also in the halophyte S. salsa, up-regulation of SsAKT1 expression was observed under saline conditions, which may contribute to mediating significant K + uptake in roots from the external medium (Duan et al. 2015). Such regulation was reported for other plant species, like BrKCT2 from Brassica rapa ssp. Pekinensis (Zhang et al. 2006), BnGORK from Brassica napus (Chakraborty et al. 2016), and PtKPT1 in the bark of grey poplar (Escalante-Pérez et al. 2009). In contrast, the transcript levels of McMKT1 from the common ice plant (Mesembryanthemum crystallinum), OsAKT1 from rice (Oryza sativa L. ssp. japonica cv. Nihonmasari), and AtAKT1 from Arabidopsis thaliana have been shown to be down-regulated in the presence of excessive Na + (Su et al. 2001;Fuchs et al. 2005;Kaddour et al. 2009). However, PutAKT1 transcript levels from Puccinellia tenuiflora seem to be unaffected by salt stress (Ardie et al. 2010).
The activity of plant K + channels is also regulated by protein kinases and phosphatases (Zhu 2003).
Little is known about the effect of Na + on the activity of plant K + channels . Electrophysiological analyses on guard-cell protoplasts from a salt-tolerant Aster species revealed that an increase in cytosolic Na + concentration in guard cells caused a delayed and dramatic deactivation of K + inward conductance (Véry et al. 1998). Also, an irreversible blockage/inactivation of inward rectifying K + currents by low external Na + was observed in Avena sativa mesophyll cells (Kourie & Goldsmith 1992). More recently, a direct effect of external Na + on the activity of the Shaker inward K + channel MIRK (belonging to KAT subgroup) from melon (Cucumis melo) has been reported . In Xenopus oocytes, external Na + downregulated both inward and outward MIRK currents in a voltage-independent manner, suggesting a blocking site in the external mouth of the channel. The addition of 100 mM Na + at low (submillimolar) external K + concentrations completely suppressed the MIRK-like K + channel activity. The molecular site(s) of the inhibition by Na + in the MIRK channel was speculated to be located in the S5-P-S6 (Transmembrane segment 5-Pore-Transmembrane segment 6) region . Three site-specific mutants (S232KQ-PRK, N241-Y and L252-R) within the S5-P-S6 region were obtained and studied using electrophysiological analyses to identify the molecular site(s) of Na + inhibition in MIRK. The S232KQ-PRK and N241-Y, but not L252-R, mutant channels had a reduced Na + inhibition rate. Among these three mutants, N241-Y had the strongest reduction of Na + inhibition rate. However, as the Na + effect was not completely abolished, it is likely that MIRK sensitivity to Na + is determined by more than one residue . The molecular basis of sodium blockage in MIRK thus needs to be further explored.
In the Cucurbitaceae family, besides melon, cucumber (Cucumis sativus) and watermelon (Citrullus lanatus) are two other common kinds of vegetable crops cultivated worldwide. To date, no K + channels from these two plants have ever been documented. In this study, two Shaker genes that are homologous to KAT2 were isolated from Cucumis sativus L. and Citrullus lanatus L., named CsKAT2 and ClKAT2 respectively. The electrical properties of these two K + channels were studied in Xenopus oocytes, in particular the regulation by Na + of these two Shaker K + channels, with the aim of determining whether inhibition by Na + is ubiquitous in the KAT-type Shaker channels from Cucurbitaceae. The question of the molecular location of the sodium blockage in MIRK channel was also further analyzed in this study.

Identification of the CsKAT2 and ClKAT2 Shaker channel genes
The Shaker Channel genes CsKAT2 (GenBank accession number MF447463) and ClKAT2 (GenBank accession number MF447464) were obtained by PCR. The open reading frame (ORF) of CsKAT2 was 2100 bp long encoding a 700 amino acid polypeptide, and ClKAT2 contained 2061 bp encoding a 687 amino acid polypeptide. Results of BLAST revealed a high degree of homology between amino acid sequences of CsKAT2, ClKAT2 and MIRK. CsKAT2 and ClKAT2 shared 91% and 90% amino acid identity with MIRK, respectively. These three channels shared, respectively, 62%, 61% and 62% amino acid identity with KAT2 from Arabidopsis. They all displayed high sequence conservation in the pore domain with the putative K + transport motif (TTLTTTGYGD), which is common to the other Shaker channel genes ( Fig S1). The phylogenetic tree of Shaker channel proteins confirmed that CsKAT2, ClKAT2 and MIRK proteins were genetically close and that all belong to the KAT-like channel subgroup (Fig 1).

Electrophysiological characterization of CsKAT2 and ClKAT2 channels in Xenopus oocytes
In contrast to Xenopus oocytes injected with H 2 O (Fig 2A), the oocytes expressing CsKAT2 and ClKAT2 elicited the classical inwardly-rectifying currents, from a voltage threshold around -105 mV (CsKAT2) or -120 mV (ClKAT2) in 10 mM external K + (Fig 2B, C). When K-glutamate concentration (1, 3, 10, or 100 mM, at pH 5.5) increased in the external solution, the magnitude of whole-oocyte steady-state CsKAT2 and ClKAT2 inward currents was increased at the same voltage. As expected for an inward K + channel, the increase of external K + concentration also positively shifted the reversal potential (E rev ) of currents. A 10-fold change in the external K + resulted in 50 mV and 53 mV differences in Erev for CsKAT2 and ClKAT2, respectively, which was close to the theoretical behavior of a pure K + channel (Fig 2D, E). Half-saturation of CsKAT2 and ClKAT2 conductance was observed at around 7 mM and 9 mM of external K + , respectively ( Fig 2F). The affinity to K + of both channels was thus close to that reported for MIRK channel (K M =11 mM; Zhang et al. 2011) and was in the lowest range of most other plant inward Shaker channels (Very et al. 2014). Analysis of CsKAT2 and ClKAT2 voltage sensitivity unveiled that their gating voltage-dependency was unaffected by external K + (Fig 2G). When successively replacing K + in the bath solution with the same concentration of Rb + , NH 4 + , Li + or Na + , current-voltage relationships indicated that CsKAT2 and ClKAT2 channels were Downloaded from https://academic.oup.com/pcp/advance-article/doi/10.1093/pcp/pcab170/6447658 by INRAE Institut National de Recherche pour l'Agriculture, l'Alimentation et l'Environnement user on 03 December 2021 mostly permeable to K + , and a lesser extent to Rb + and NH 4 + ( Fig 2H). P Rb /P K and P NH4 /P K of CsKAT2 calculated from the reversal potential analyses were 0.30 ± 0.02 (n=9) and 0.05 ± 0.003 (n=9), respectively. For ClKAT2, they were 0.25 ± 0.02 (n=11) and 0.04 ± 0.001 (n=11). Both channels are not significantly permeable to Li + or Na + (Fig 2H). Like other KAT2-type channels, CsKAT2 and ClKAT2 were regulated by the external pH (Pilot et al. 2003b;Véry et al. 2014). The acidification, specifically from pH 7.5 to 6, led to an increase of inward CsKAT2 and ClKAT2 currents (Fig 2I).

CsKAT2 and ClKAT2 currents were inhibited by external Na + , as MIRK currents
Similar to the MIRK currents in Xenopus oocytes, CsKAT2 and ClKAT2 currents recorded in the solution containing 1 mM K + and 99 mM Na + were clearly smaller than those in the solution with same concentration of K + but with 99 mM Li + (Fig 3A, B; upper panels). The "tail" currents (transient outward currents recorded upon channel deactivation when the membrane polarization was back to weak value) of both channels were inhibited as well in the presence of external Na + (Fig 3A, B; lower panels). Current inhibition by Na + in either inward or outward directions was not dependent on voltage.
Mean I-V relationship led to the same conclusion ( Fig 3D, E). Like in MIRK, both channels were no longer blocked by Na + when high K + concentration (≥10 mM) was present in the bath ( Fig S2).
AtKAT2 from Arabidopsis was expressed in oocytes in order to make the comparison under the same recording conditions. The AtKAT2 currents were not sensitive to external Na + (Fig 3C, F). Consistently, at -155 mV, the extent of Na + inhibition in CsKAT2, ClKAT2 and MIRK was high, which contrasted to KAT2 from Arabidopsis. No significant difference in the degree of inhibition was found among the three Cucurbitaceae-family channels ( Fig 3G).

KAT2-like channel genes in cucurbits were significantly downregulated under NaCl treatment
Quantitative RT-PCR results provided evidence that CsKAT2 and ClKAT2, like MIRK , are primarily expressed in leaves, particularly in guard cells ( Fig 4A). To determine whether the expression of KAT2-like genes was modulated by external Na + treatments, total RNA was collected from leaves and guard cells of plants exposed to salt stress for 24 h using 100 mM NaCl-supplemented growth solution. As shown in Fig 4B, KAT2-like genes were slightly affected by the salt stress in whole leaves or guard cells. Salt stress tended to down regulate the KAT2-type genes, the decrease being significant in all of three genes in the whole leaf and guard cells ( Fig 4B).

Inhibition by Na + was abolished in a chimeric MIRK channel harboring AtKAT2 S5-P-S6
domain, but re-appeared in the reciprocal chimera A molecular determinant of the inhibition by Na + in the MIRK channel could be suspected to be located in the S5-P-S6 region (which constitutes the channel pore), especially in the S5-P linker . The S5-P-S6 region harbored eighteen amino acid differences between AtKAT2 and MIRK (and the two other KAT-type channels from Cucurbitaceae; Fig 5 and S1). To confirm the hypothesis that the molecular site of Na + inhibition in MIRK is located in this area, we replaced the whole segment of the S5-P-S6 region of MIRK containing the sequence of the eighteen amino acids different from AtKAT2 (i.e. amino acids between the two inverted triangles in Fig 5A) by the corresponding position in AtKAT2 ( Fig 5A). The chimera in MIRK cDNA was named MIRK-By-△AtKAT2. The potassium sensitivity of MIRK-By-△AtKAT2 currents, tested at three different potassium concentrations (1, 10 and 100 mM) was very similar to that reported for MIRK (Fig S3 B; Zhang et al. 2011), which indicated that the affinity to K + was not significantly modified in the chimera. The reversal potential of MIRK-By-△AtKAT2 current shifted toward a more positive value as the K + concentration increased, with a 56 mV difference per 10-fold change in external K + concentration ( Fig   S3 C). This was very close to the theoretical behavior of a pure K + channel, as previously reported for MIRK . Thus, the above analyses proved that the substitution of amino acids in the region forming the channel pore with AtKAT2 did not change the basic properties of MIRK. Thereafter, MIRK-By-△AtKAT2 currents were investigated at 1 mM K + , either in the presence of Li + (99 mM) or Na + (99 mM), to test the sensitivity to Na + . Comparing the representative current traces between these two conditions, the current activation and deactivation magnitudes were similar ( Fig 5B). Statistical analyses confirmed that there was no significant difference in the inward currents between the two conditions. This observation was verified in three independent experiments ( Fig 5C). Therefore, we concluded that the inhibition by Na + was abolished in the MIRK-By-△AtKAT2 channel.

Expressed in
To confirm the above result, we created the reciprocal chimera, termed AtKAT2-By-△MIRK, by substituting the whole segment of the S5-P-S6 region of AtKAT2 for the corresponding residues from MIRK. The same solutions (1 mM K + , either in the presence of Li + (99 mM) or Na + (99 mM)) were used to test the Na + sensitivity. Expressed in oocytes, AtKAT2-By-△MIRK is capable of inducing the typical inwardly-rectifying currents in both solutions. The inward currents in K1Li99 condition were twice as large than that in K1Na99. The same observation was found at the "tail" currents ( Fig 5D).
Mean I/V plots ( Fig 5E) confirmed that the inhibition by external Na + re-appeared in AtKAT2-By-△MIRK channel. In summary, we conclude that the S5-P-S6 region is important for determining Na + sensitivity in the MIRK channel.

Identification of two novel K + Shaker channels from Cucurbits
The Cucurbitaceae family includes more than 700 species worldwide and many of them are very agronomically important (Jeffrey 1980). For example, watermelon and cucumber are extensively cultivated in China, accounting for 60% of the cultivated areas in the world (http://faostat.fao.org/, 2018). Studies over the last few decades have focused on the physiological importance of K + nutrition on the growth, yield and fruit quality of these two vegetable crops (Gislerød and Adams 1983;Altunlu et al. 1997;Kaya et al. 2001), but few concerned the molecular identity of K + acquisition and transport pathways. In this study, two more Shaker channel genes, CsKAT2 and ClKAT2 from cucumber and watermelon, respectively, were cloned and studied, reinforcing the information on the Shaker family from the plant kingdom.
Based on the phylogenetic analyses, the closest homolog of CsKAT2 and ClKAT2 among reported Shaker channels is MIRK, also from the Cucurbitaceae family, melon (Fig 1). The three channels have more than 90% conserved amino acid identity. Electrophysiological results obtained in the heterologous system Xenopus oocytes showed that CsKAT2 and ClKAT2 shared a number of functional properties with MIRK as well as other homologs from the KAT subfamily, however there were a few differences in quantitative parameters (Fig 2). For instance, CsKAT2 and ClKAT2, like MIRK and other KAT2-type channels, share similar gating properties. However, among CsKAT2, ClKAT2, MIRK and AtKAT2, MIRK shows the highest equivalent gating charge (z=3, 2-folds higher than that of CsKAT2) and also the most negative half activation potential (> 30 mV more negative than that of AtKAT2) (Fig 2G, Pilot et al. 2003b;Zhang et al. 2011). The affinity of CsKAT2, ClKAT2 and MIRK to external K + are very similar (Fig 2F; Zhang et al. 2011), but more than 2-fold higher than that of AtKAT2 from Arabidopsis (Jeanguenin et al. 2011) and about 4-fold higher than that of ZmK2.1 from maize (Su et al. 2005). Comparison of CsKAT2, ClKAT2 and MIRK cationic selectivity to different monovalent cations resulted in a similar sequence: for conductances, K + > Rb + ≈ NH 4 + >> Li + ≈ Na + (Fig 2H, Zhang et al. 2011). Like most other channels from the KAT subfamily (Hoth and Hedrich 1999;Zhang et al. 2011;, CsKAT2 and ClKAT2 were activated by acidic external pH (Fig 2I).

KAT2-type channels inhibited by external Na + : a common feature in cucurbits?
MIRK from melon was the first cloned plant K + channel reported whose activity was directly inhibited by external Na + . In this study, we reported two other Shaker K + channels from cucumber and watermelon whose activities were also blocked by external Na + (Fig 3). All three channels are from the gourd species, belonging to Cucurbitaceae. Is this inhibition of channel activity by external Na + a common feature among KAT2-type channels from the whole Cucurbitaceae family?
Cloning other KAT2-type channels from more distant species in the Cucurbitaceae family and study of their electrical properties with respect to external Na + effect would be very informative regarding the above question.
Detailed analyses of the Na + effect on CsKAT2 and ClKAT2 activity revealed three common features with MIRK: i) the inhibition of inward currents by external Na + was immediately observed when Na + was added into the bath and quickly reversed upon Na + removal; ii) the inhibition by external Na + presented no voltage dependency; iii) Na + was not permeant through these channels ( Fig 2H and 3 D, E). Given the above consistency, it is highly likely that the inhibitory mechanism of CsKAT2 and ClKAT2 channels by Na + and the molecular regions underpinning this phenomenon are the same as for MIRK.
The inhibitory effect of Na + on MIRK inward activity was completely abolished by replacing the S5-P-S6 segment of MIRK with the corresponding sequence from Arabidopsis AtKAT2 and was recovered in the reciprocal chimera (Fig 5), indicating that S5-P-S6 is crucial for MIRK sensitivity to Na + . The three channels from Cucurbitaceae display the same level of inhibition by Na + (Fig 3G) and share 18 amino acid differences with ATKAT2 in this region (Fig S1). Among these amino acids, 2 are located at the base of the S5 segment and 13 in the external S5-P linker, known to constitute the external mouth of the channel pore. Since the level of inhibition by external Na + is not dependent on voltage and is immediately reversed upon Na + removal, it is likely that the molecular site of the inhibition within the S5-P-S6 region is in the S5-P linker . To identify key amino acid residues, we used MIRK for homology modelling and compared it to AtKAT2.

Homology modeling and electrostatic comparison identify four key amino-acid residues in MIRK that may account for Na + sensitivity
Based on the Cryo-EM structure of AtKAT1 (Clark et al. 2020), we built homology models of AtKAT2 and MIRK. Similar to AtKAT1, both AtKAT2 and MIRK assemble as a tetrameric channel with a 4-fold symmetry and with the pore domain (in Red) at the center ( Fig 6A). Electrostatic interactions in the MRIK and AtKAT2 models were mapped. As seen, the pore entrances are predominantly negatively charged (white arrow; Fig 6B). The amino acid residues from the end part of S5 and S5-P linker (green line surrounded, Fig 6B) were located around the mouth of the pore entrance.
In this region, MIRK had less positively charged amino acids than AtKAT2 (Fig S1), globally providing a more negative charge environment (Fig 6B, C). In this case, there is a higher possibility of favoring binding of sodium ions in MIRK than in AtKAT2. Further comparison of the electrostatic potential distribution on surface for residues N222 to S260 in MIRK and the corresponding residues from A216 to S254 in AtKAT2 was carried out (Fig S4). Of 15 residues, four amino-acid residues, P230, N241, E247 and N251 in MIRK had more negative surface charge than those corresponding residues, H224, Y235, T241 and S245 in AtKAT2 (Fig 6D and Fig S4), which may account for Na + sensitivity. Of these, the N241-Y mutation has been shown to induce a lower Na + inhibition rate ), and we also tested MIRK mutated at 6 sites together (SKQ232-PRK, N241-Y, L252-R and D283-F). The sensitivity to Na + in this sextuple mutant was significantly reduced but still not completely abolished (Fig S5). Given the modeling analysis and the oocyte results, MIRK sensitivity to Na + seems to be governed by several residues. Of note, glutamic acid (Glu; E) in MIRK and CsKAT2 or aspartic acid (Asp; D) in ClKAT2 (FigS1; Green triangle) is negatively charged and the ability for Na + binding is prevalent (Handling, et al. 2018). Further analyses on the channels including D/E mutation would be helpful to better understand the common Na + sensitivity.

Possible role of KAT2 type channels from cucurbits in response to salt stress
K + channels in plants play important roles in K + absorption and transport, particularly in regulating cellular homeostasis under abiotic stress (Pilot et al. 2003a;Maathuis 2005). Shaker K + channels from several plant species were found at the transcriptional level to be affected (down-or up-regulated) by salt stress (Su et al. 2001;Pilot et al. 2003a;Fuchs et al. 2005;Zhang et al. 2006;Escalante-Pérez et al. 2009;Kaddour et al. 2009;Jeanguenin et al. 2011;Chakraborty et al. 2012;Duan et al. 2015). In this study, the expression of CsKAT2, ClKAT2 and MIRK was down regulated in response to 100 mM NaCl (Fig 4), as observed for that of the Shaker AKT subgroup (Véry et al. 2014) McMKT1 in Mesembryanthemum crystallinum (Su et al. 2001) and OsAKT1 (Golldack et al. 2003;Fuchs et al. 2005). The transcriptional downregulation of the three cucurbit KAT-type channels is consistent with the inhibition of these channels by external Na + (Fig 3G), suggesting that Na + regulation has an important role in response to salt stress in these species.
The Shaker KAT group has been characterized by expression of its members in guard cells (Véry et al., 2014). Quantitative RT-PCR showed that all three KAT2 type channels from Cucurbitaceae family were mostly expressed in guard cells (Fig 4), supporting the fact that all three channels belong to the KAT group (Pilot et al. 2003b).
The inhibition of KAT2-like channel activities by external Na + occurred when K + was in the range 0.1-1 mM and Na + in the range 10-100 mM (Fig 3 and Zhang et al., 2011). This can be predicted to be physiologically relevant in guard cells upon salt stress. Very substantial reduction of inward K + transport across guard cells is expected with the combined transcriptional and channel activity inhibitions, in the presence of 100 mM Na + and ≤1 mM K + in the guard cell apoplast. It should also be noted that at a low K + concentration like 0.1 mM, an even higher suppression of channel activity occurred in the presence of 100 mM Na + , which reached 86% and 94% for CsKAT2 and ClKAT2 respectively (data not shown). Inhibiting the expression and activity of guard cell inward channels likely contributes to reduced stomatal aperture (Lebaudy et al. 2008), an adaptive response to the osmotic stress and to excess salt arrival to the shoot caused by salt stress (Robinson et al. 1997;Munns and Tester 2008). Our previous experiments showed high correlation between stomatal movement and MIRK transcription level (Wang et al. 2013). Thus, our results suggest that KAT-like K + channels of cucurbits (gourds at least) have special characteristics (intrinsic functional properties and expression regulation), which may contribute to salt stress adaptation.

Construction of the K + channel gene chimera MIRK-By-△AtKAT2 and AtKAT2-By-△MIRK
A chimeric cDNA of MIRK channel subunit in which AtKAT2 S5-P-S6 region replaced the corresponding MIRK sequence, MIRK-By-△AtKAT2 was obtained by a series of PCR programs. Based on sequence analysis of the S5-P-S6 region of the two Shaker channels AtKAT2 and MIRK (Fig. S1), three pairs of primers were designed (Table S1). The first PCR product MIRK-S1 segment from ATG to mid S5 was amplified with the primers MIRK-F(BamHI) and MIRK-S1-R, using MIRK cDNA as the template. The second PCR was run again with the same matrix while using primers MIRK-S2-F and MIRK-R(NotI), to obtain the cDNA segment MIRK-S2 from mid S6 to end. The third PCR product KAT2-S5PS6 segment from S5 to S6 was obtained with the AtKAT2 plasmid as the template and with the primers AtKAT2-S5PS6-F and AtKAT2-S5PS6-R. Finally, the chimera MIRK mutant gene (MIRK-By-△AtKAT2) was obtained by PCR, using three cDNA segments MIRK-S1, KAT2-S5PS6 and MIRK-S2 mixture as the template, with the forward primer MIRK-F(BamHI) and the reverse primer MIRK-R(NotI). Thereafter, MIRK-By-△AtKAT2 was inserted to the expression vector pGEM-Xho (at the enzyme restriction sites BamHI and NotI). Using the same method, the chimera gene AtKAT2-By-△MIRK was obtained and cloned in pGEM-Xho at the enzyme restriction sites BamHI and NotI.

Expression in Xenopus oocytes
CsKAT2 and ClKAT2 cDNA were integrated into the pGEMXho vector (adapted from pGEMDG vector, D. Becker, Würzburg, Germany) under the control of the T7 promoter between the 5' and 3' untranslated regions of the Xenopus beta-globin gene, using the enzyme restriction sites BamHI and NotI for CsKAT2, and BamHI and EcoRI for ClKAT2.

Two-electrode voltage clamp experiments
Two to four days after injection, whole-cell currents were measured using the two electrode voltage-clamp technique using a GeneClamp 500B amplifier (Axon Instruments, Foster City, CA, USA). Voltage-pulse protocols, data acquisition and data analyses were performed using pClamp 9 (Axon Instruments, Foster City, CA, USA) and Sigmaplot 8 software (Jandel Scientific, Erkrath, Germany). All electrodes were filled with 3M KCl. Both actual imposed membrane potential and current were recorded. A correction was made for the voltage drop occurring through the series resistance of the bath and the reference electrode by using two external voltage-measuring electrodes MgCl 2 and either 10 mM Mes / Tris when the pH was 5 or 6, or 10 mM Hepes-Tris when the pH was 7.5.

Hydroponic cultivation of cucurbit plants
The seeds of melon (Cucumis melo L.), water melon (Citrullus lanatus) and cucumber (Cucumis sativus) were firstly sanitized and spread on filter papers. After the emerging of radicles and plumules, the seedlings were transferred in plug containing a mixture of peat, perlite and slag (1:1:  (Table S1). The quality of the PCR reactions was estimated based on melting curves. Data were analyzed using CFX Manager software. Each qRT-PCR was based on three biological replicates and three technical replicates.

Homology modeling
The protein sequences of the monomers from AtKAT2 (NP_193563.3, Arabidopsis thaliana) and MIRK (AAZ66349.2, Cucumis melo) were modelled using the I-TASSER server (http://zhanglab.ccmb.med.umich.edu/I-TASSER/)  with default settings. Based on the confidence score (C-score) assigned to the top 5 models by I-TASSER, the best scored model was used for downstream analysis. alignment using the MatchMaker tool in UCSF Chimera (Pettersen et al., 2004). Briefly, each of the four chains in the 7CAL structure were aligned to a modelled monomer of the query protein to construct a tetrameric pore using the BLOSUM-62 matrix. The electrostatic potential calculated with Coulomb's law was computed and visualized using Coulombic in UCSF Chimera with Amber20. Color scale was set to ±3 kcal/(mol·e).

Data Availability Statements
The data underlying this article are available in the article and in its online supplementary material.

Funding
This work was supported by the National Natural Science Foundation of China (31372079), State Scholarship Fund of China Scholarship Council (No. 201208310510).

Conflicts of interest
No conflicts of interest declared.

Figure 3. External Na + blocked CsKAT2 and ClKAT2 currents but not AtKAT2 currents
A-C. Representative CsKAT2 (A), ClKAT2 (B) and AtKAT2 (C) current traces recorded in 1 mM K + and 99 mM of either Li + (-Na) or Na + (+Na) from +30 to -165 mV (A and B) and +40 to -155 mV (C) in -15 mV increments (top panels). The corresponding deactivated currents (dashed circle) were zoomed and showed in the lower panels.
D-F. Mean of steady-state CsKAT2 (D), ClKAT2 (E) and AtKAT2 (F) inward currents against the membrane voltage in the presence of Li + or Na + at the external concentration of 1 mM K + (K1). Data concern time-activated currents and are means ± SE, n=5 in D, E and F.
G. Comparison of the extent of blockage by Na + on MIRK, CsKAT2, ClKAT2 and AtKAT2 inward currents at -155 mV in 1 mM K + and 99 mM Na + . Different letters at the right of the bars indicate significant differences (t test, P<0.05).

Figure 4. Expression profile of KAT2-like channels in Cucurbitaceae plants.
A. Expression patterns of KAT2-like channels in different tissues of plants. The RNAs were extracted from the roots, stems, leaves, mid rib of leaves and guard cells respectively. Actin was used as internal control. Data are presented as means ± SD from three biological repeats.
B. Expression profile of KAT2-like channel genes in Cucurbitaceae plants after 24 h under NaCl treatment. KAT2-like channel gene transcript levels were detected by qRT-PCR 24 h after plant treatment by addition of 100 mM NaCl. Data were shown as means ± SD from three biological repeats.
The significance level was defined as * (P < 0.05), ** (P < 0.01). A. Ribbon diagrams of AtKAT2 and MIRK channels composed from 4-subunits. Red sections of the ribbon diagram indicate the P domain-motif, and blue coloured sections represent the residues from the S5 to S5-P linker (from N222 to S260 in MIRK; corresponding to the residues A216 to S254 in AtKAT2).
B. Electrostatic potential distribution representation from the top view of AtKAT2 and MIRK.
Negatively charged surface is indicated in red and positively charged in blue. Blue sections in the ribbon diagram from A, were highlighted in green and arrows point to the pore structure. Electrostatic potential range is -3kT/e (red) to +3 kT/e (blue).