Japonica rice cultivar Nipponbare (O. sativa L.) was used as the wild type in this study, and also used for the generation of all transgenic plant lines. IRRI (International Rice Research Institute) hydroponic solution for rice was conducted as previous method (Li et al., 2014; Wang et al., 2021). The modification of Na⁺ and K⁺ concentrations as indicated in the figure legends.

The rice genomic cDNA sequences encoding OsHAK12 was amplified by PCR using the primer pairs listed in Supplementary Table 1. The PCR product was constructed into pYES2 vector (digested with HindIII and XbaI) to generate pYES2-NC-OsHAK12. This construct and the empty vector were transformed into yeast strain K⁺ uptake-deficient CY162 or high-Na⁺ sensitive AXT3K, respectively. The yeast complementation assay were performed as previous methods (Anderson et al., 1992; Quintero et al., 2002).

Total RNA was isolated from Nipponbare rice using the TRIzol reagent (Invitrogen). Real time qRT-PCR analyses were performed as described previously (Livak and Schmittgen, 2001; Wang et al., 2021). All primers used for real time qRT-PCR assay are listed in Supplementary Table 1.

The 2,000-bp fragment located upstream of the OsHAK12 initiation codon was amplified from Nipponbare rice genomic DNA. This amplified promoter fragment was digested with EcoRI and HindIII, then cloned into pCAMBIA1301-GUS vector. The genetic transformation and histochemical analysis of GUS staining in different tissues of rice as described previously (Upadhyaya et al., 2000; Ai et al., 2009; Wang et al., 2021). All primers used for the GUS assay are listed in Supplementary Table 1.

The full length cDNA of OsHAK12 without the stop codon was amplified, after sequence confirmation and digestion with XbaI, the amplified DNA fragment was cloned into the binary vector pCAMBIA1390 to generate the 35S:OsHAK12-GFP fusion construct. Transient expression of the fusion protein was examined by the confocal laser-scanning microscopy using the LSM880 instrument (Carl Zeiss) as previous methods (Li et al., 2009; Wang et al., 2021). The primers used for the subcellular localization assay are listed in Supplementary Table 1.

To generate OsHAK12 knockout plants, the CRISPR/Cas system for targeted genome modification of rice was used (Xie et al., 2014; Usman et al., 2020). A 20-bp sgRNA sequences (GAGAGCTGGACCTCCCTTGG) was cloned into the pOs-sgRNA vector, and then subcloned into the Cas9 vector pYLCRISPR/Cas9Pubi-H (Supplementary Figure 1). Transgenic plants were obtained and identified as following the procedure (Upadhyaya et al., 2000; Wang et al., 2021). Two T2 generation homozygous mutant lines Oshak12-1 and Oshak12-2 were used for further study. The primers used for this assay are listed in Supplementary Table 1.

Measurement of chlorophyll and ion content (Na⁺, K⁺) analysis as previous methods (Porra et al., 1989; Wang et al., 2021). The collected method, Na⁺ and K⁺ concentration in the xylem sap and phloem exudates were determined using inductively coupled plasma/optical emission spectrometry ICP-AES (Varian 715-ES) following the method reported by Tian et al. (2021). Briefly, 5-days-old rice seedlings were cultivated in the solutions for 14 days and then transferred to the hydroponic cultures containing 0 or 100 mM Na⁺ for 2 days. The shoots were cut and then the xylem sap exuding at the cut surfaces was collected for 1 h. The xylem sap exudates were discard at initial half hour and the xylem sap exudates was collected during the first to second hour and the third to fourth hour on the same plants exchanging cotton balls on the same stumps. Then xylem sap was got from the cotton balls by centrifugation. To collect phloem exudates, the shoots were excised from seedling and the cut extremities were immediately dipped in a 15 mM EDTA solution (pH 7.5, K₂-EDTA buffer was used for Na⁺ content assay and Na₂-EDTA buffer was used for K⁺ content assay) and incubated in dark for 8 h under 90% humidity condition.

To dissect the functions of rice HAK family members, we generated loss-of-function mutants for each OsHAK transporter using CRISPR-assisted genetic analysis in Nipponbare rice background. Here, we focused on the functional identification of OsHAK12. We generated two independent knockout mutants (Oshak12-1 and Oshak12-2) of OsHAK12. The Oshak12-1 and Oshak12-2 mutants had a 4-bp and 1-bp deletion in the third exon of LOC_Os08g10550, respectively, leading to a frameshift mutations at the 192th and 194th amino acids and premature translation termination at 211 and 213 amino acids separately (Supplementary Figure 1). No off-target cleavage was found using the web-based tool CRISPR-P (Liu et al., 2017)¹.

Previous studies showed that some high affinity K⁺ transporter (HAK) family members responsed to low-K⁺ stress or salt tolerance in plants (Yang et al., 2014; Chen et al., 2015; Shen et al., 2015; Feng et al., 2019; Wang et al., 2021). First, we detected the growth of the Oshak12 mutants (Oshak12-1, Oshak12-2) under different K⁺ concentration conditions. We found that the Oshak12 mutants and the wild-type plants Nipponbare (Nip) both grew well with no distinct differences under either K⁺-sufficient (10 mM K⁺) or K⁺-deficient (0.01 mM K⁺) hydroponic solutions (Supplementary Figures 2A–E). In addition to seedling height, fresh weight, we also measured K⁺ content in both roots and shoots and found no differences between wild type and mutants, suggesting that disruption of OsHAK12 does not affect K⁺ homeostasis in rice at seedling stage. We further found that the grain length, grain width, 1,000-grain weight of the mature grains between wild type and mutants displayed no significant differences (Supplementary Figures 3A,Ba–f). Scanning electron microscopy (SEM) of transverse sections of mature endosperm revealed that the endosperm of Oshak12 and wild type mature grains both filled with larger, regular, tightly packed starch grains (Supplementary Figure 3C). In addition, no significant differences of pollen viability were observed between the Oshak12 mutants and the wild type (Supplementary Figures 3D,Ea,b). The above results suggested that disruption of OsHAK12 does not affect K⁺ homeostasis in rice at reproductive stage.

We then examined the growth of the two independent Oshak12 mutants under salt stress conditions. We transferred 14-days-old plants of Oshak12 and wild type grown in hydroponic culture to the same solution plus 100 mM Na⁺ for 6 days and found that the shoots of the oshak12 mutants displayed more withered and chlorotic phenotype as compared to that of wild type plants. In addition, the Oshak12 mutants showed decreased shoots growth under salt stress (Figure 1A). The above datas indicated that the oshak12 mutants were more hypersensitive to salt stress than the wild type plants. To quantify the phenotypes, we further determined the length and fresh weight of roots and shoots separately and measured the chlorophyll contents of the leaves. While the mutants showed similar levels as the wild type in roots, the mutant shoots were much stunted under salt stress as compared to the wild type (Figures 1B–E). Additionally, the chlorophyll contents of Oshak12 mutants were also lower than that of the wild type plants after NaCl treatment (Figure 1F), consistent with their chlorotic phenotype.

The above results showed that disruption of OsHAK12 was responsible for the hypersensitivity to salinity stress.

To understand the physiological role of OsHAK12, we first performed the expression pattern analysis of OsHAK12 in rice plants. The qRT-PCR analysis showed that OsHAK12 was expressed strongly in the roots and its lower amounts transcripts were also detected in stems, leaves, anther and glumes (Figure 2A). The expression of OsHAK12 was up-regulated in root during salt stress (Figure 2B). To detect the expression pattern of OsHAK12 in more detail, the GUS activity staining of transgenic rice plants harboring the OsHAK12 promoter-GUS fusion construct was performed. Strong GUS signals were found in the roots of the transgenic rice plants (Figure 2Ci), which was consistent with the qRT-PCR results (Figure 2A). Cross sections of GUS-stained roots showed that OsHAK12 was expressed almost in all cell types such as root hair, exodermis, cortex and endodermis, especially strongly expressed in vascular tissues (Figure 2Cii). Furthermore, GUS activity was present in mesophyll cells (Figure 2Ciii).

Then, we carried out the subcellular localization of OsHAK12 in plant driven by the cauliflower mosaic virus 35S promoter. A green fluorescence protein (GFP) reporter construct was developed to express the OsHAK12-GFP fusion protein, and the same vector expressing GFP only was used as a control. Subsequently, the OsHAK12-GFP fusion construct and the GFP-only control were transformed into the protoplasts of the rice leaf sheaths cells, respectively. GFP-only signal was present mainly in the cytoplasm and nucleus as expected, whereas OsHAK12-GFP fusions was localized at the plasma membrane, as indicated by overlaps between GFP and signals from the known plasma membrane protein fused to red fluorescent protein (SP1-RFP) (Li et al., 2009; Figure 3). Based on these results, we concluded that OsHAK12 is localized to the plasma membrane in rice cells.

Salinity stress generates both osmotic stress and Na⁺ ionic toxicity in plants (Tester and Davenport, 2003; Shen et al., 2015; Zelm et al., 2020). As 100 mM NaCl could cause both osmotic stress and ionic toxicity in plants, we compared the mutant and wild type plants grown under 20% PEG6000 (polyethylene glycol with an average molecular weight of 6,000 Da) that imposed osmotic stress but not ionic stress. No remarkable differences was found between the Oshak12 mutants and wild type plants (Supplementary Figures 4A–C). These results showed that the salt hypersensitivity of the Oshak12 mutants probably due to Na⁺ ionic toxicity but not to osmotic damage.

We then examined the Na⁺ contents in both shoot and root tissues of the above different genotypes plants during different NaCl concentrations. Under control condition (0 mM Na⁺), we found no significant differences of Na⁺ contents in roots or shoots between the mutants and wild type plants. However, under saline condition (100 mM Na⁺), Oshak12 mutant plants contained significantly higher levels of Na⁺ in their shoots but lower levels of Na⁺ in their roots as compared with the wild type plants (Figures 4A,B). These above results suggested that knockout of OsHAK12 leads to excessive root-to-shoot Na⁺ translocation in rice plants, resulting in over accumulation shoot Na⁺. Meanwhile, Oshak12 mutant plants had significantly less shoot K⁺ and similar root K⁺ content compared with wild-type plants under saline condition (Figures 4C,D). As a result, the Oshak12 mutants showed higher Na⁺/K⁺ ionic content ratio in shoots and similar Na⁺/K⁺ ionic content ratio in roots compared to those ratios in wild type plants (Figures 4E,F), which indicate that disruption of OsHAK12 damaged the Na⁺/K⁺ ionic homeostasis in shoots during salt stress.

The expression analysis suggested that OsHAK12 showed strong expression in root vascular tissues including xylem parenchyma cells (Figure 2Cii). Direct Na⁺ measurements suggested that, under saline conditions, the Oshak12 mutants accumulated more Na⁺ in the shoot and less Na⁺ in the root than wild type plants (Figures 4A,B). These results indicate that OsHAK12 may be involved in Na⁺ retrieval from the xylem vessels to xylem parenchyma cells in root tissues to prevent root-to-shoot Na⁺ translocation.

To address the role of OsHAK12 in regulating Na⁺ retrieving from the xylem sap to xylem parenchyma cells, the Na⁺ content in the xylem sap from different plants were measured. We found that, under control condition, there was no significant difference on Na⁺ content in the xylem vessels between the Oshak12 mutants and wild type plants. However, under 100 mM NaCl, the Oshak12 mutant plants had a significantly higher Na⁺ content in the xylem vessels than the wild type (Figure 5A), indicating that Na⁺ retrieving from the xylem vessels was defective in Oshak12 mutants root tissues. Meanwhile, we observed reduced K⁺ content in the xylem sap, reduced Na⁺ content and similar K⁺ content in the phloem sap in the Oshak12 mutant plants as compared with wild type plants (Figures 5B–D), which suggest that knockout of OsHAK12 also affect Na⁺ loading into the phloem sap and K⁺ homeostasis in the xylem sap. Considering that OsHAK12 is barely expressed in the phloem tissues (Figure 2Ciii), the role of OsHAK12 in phloem needs further investigation.

Taken together, our data suggest that OsHAK12 mediates Na⁺ retrieving from the xylem vessels to xylem parenchyma cells, then decrease the Na⁺ content in the xylem sap, therefore reducing Na⁺ translocation from root to shoot, eventually promotes shoot Na⁺ exclusion under high salt conditions.

Many HAK transporters display K⁺-transporting activity, however, some of HAK members were permeable to Na⁺ (Benito et al., 2012; Zhang et al., 2019). To evaluate the ion transport properties of OsHAK12 and interpret its vivo function under salt stress, we expressed OsHAK12 in the K⁺ uptake-deficient yeast strain CY162 (MATα, Δtrk1, trk2:pCK64, his3, leu2, ura3, trp1, and ade2) (Anderson et al., 1992) and Na⁺ sensitive yeast strain AXT3K (Δena1:HIS3:ena4, Δnha1:LEU2, Δnhx1:KanMX4) (Quintero et al., 2002), respectively, and examined the effect of OsHAK12 expression on the growth of these yeast strains under different ionic conditions.

The yeast strain CY162 lacking the high-affinity K⁺-transporters Trk1/2, therefore defective in K⁺ uptake (Anderson et al., 1992). We expressed OsHAK12 in the yeast strain CY162 to determine whether OsHAK12 can mediate K⁺ transport. When grown on the arginine phosphate (AP) medium containing 10 mM K⁺, yeast strain CY162 grew well with or without OsHAK12 construct (Figure 6A). A similar growth was also observed under 3 mM K⁺ (Figure 6A). When the K⁺ concentration in the AP medium was lower to less than 3 mM, yeast mutant transformed with either the empty vector or OsHAK12 construct both failed to grow on AP medium (Figure 6A). The results indicated that OsHAK12 confers little K⁺-transporting activity. The yeast complementation data were consistent with the finding that disruption of OsHAK12 did not affect K⁺ homeostasis in rice plants under non-saline conditions (Supplementary Figures 2, 3).

The Saccharonmyces cerevisiae strain AXT3K lacking major Na⁺ transporters (Na⁺ efflux proteins ENA1-4 and NHA1, and the vacuolar Na⁺/H⁺ antiporter NAX1) crucial for high-Na⁺ tolerance of yeast, which revealed growth inhibition above 50 mM Na⁺ concentrations (Quintero et al., 2002), so we expressed OsHAK12 in the yeast strain AXT3K to determine whether OsHAK12 can mediate Na⁺ influx. When grown on AP medium without Na⁺, Both transformants harboring either the empty vector or OsHAK12 construct showed similar growth (Figure 6B). However, when the Na⁺ concentration in AP medium was increased to 10 mM, yeast mutant transformed with OsHAK12 construct showed a hypersensitive phenotype than that of the empty vector control, and this phenotype became more evident when the Na⁺ concentration in AP medium was increased to 50 mM (Figure 6B), indicating OsHAK12 confers Na⁺-transporting activity. To verify this result, we then examined Na⁺ contents in the transformed AXT3K yeast strains, the results showed that OsHAK12-expressing AXT3K cells accumulated more Na⁺ content than that in the controls when exposed to 50 mM NaCl (Figure 6C). These observations indicate that OsHAK12 can mediate Na⁺ transport. Considering its expression pattern (mainly in roots) and subcellular localization (in plasma membrane), its disruption was responsible for the hypersensitivity to salinity stress and functions in Na⁺ retrieving from the xylem vessels (Figures 1–3, 5A), we suggest that OsHAK12, may be as a Na⁺-permeable transporter mediating Na⁺ transport in rice roots.