A tonoplast-localized magnesium transporter is involved in stomatal opening in Arabidopsis

Plant stomata play an important role in CO 2 uptake for photosynthesis and transpiration, but the mechanisms underlying stomatal opening and closing are still not completely understood. Here, through large-scale screening, we identified an Arabidopsis mutant ( cst2 for closed stomata2 ) defective in stomatal opening under light condition. A map-based cloning combined with complementation test revealed that the mutant phenotype was caused by a nucleotide substitution of a gene, which domains show similarity to human Mg efflux transporter ACDP/CNNM. Functional analysis showed that CST2 encodes a tonoplast-localized transporter for Mg. This protein is constitutively and highly expressed in the guard cells. Furthermore, CST2 is phosphorylated by calcineurin B-like protein (CBL)-interacting protein kinases 26 (CIPK26) in vitro , which is probably required for its activation. Knockout of this gene resulted in stomatal closing and growth retardation under high Mg concentration conditions, while overexpression of this gene increased tolerance to high Mg. Our results indicate that CST2 plays an important role in maintaining Mg homeostasis in plant cells through sequestering Mg into vacuoles especially in guard cells and that this homeostasis is required for stomatal opening, which provide a novel insight into mechanism of stomatal opening in plants. determined by a weak light after dark treatment for 30 min. Steady-state fluorescence (Fs) in illuminated leaf was measured in actinic light at 200 µmol m 2 s -1 . Maximum fluorescence (F m ') in light was determined by a saturating light pulse under actinic light at 200 µmol m -2 s -1 . Quantum yield of electron transfer to PSII (ΦPSII) was calculated as (Fm' -Fs)/Fm.


Introduction
Stomatal pores, formed by pairs of guard cells in the plant epidermis, regulate gas exchange between plants and the atmosphere. Stomatal opening is required for CO2 uptake for photosynthesis and water transpiration for maintaining leaf temperature 1,2 . A number of studies have done on the mechanisms of stomatal opening and closing in different plant species especially in model plant Arabidopsis. Among them, light has been demonstrated to be a key environmental cue to induce stomatal opening under natural conditions. The stomatal responses to light are regulated by two distinct light-activated signaling pathways: a blue light pathway and a photosynthetically active radiation (PAR)dependent pathway [2][3][4] . Blue light acts directly on guard cells to induce stomatal signaling, while PAR acts directly and indirectly on guard cells. It has been suggested that guard cell photosynthesis generates ATP or other metabolic processes for stomatal opening [5][6][7] and mesophyll cell photosynthesis induces a decrease of CO2 in the leaves, which acts on guard cells to mediate the opening of stomata 2, 6, 8 . Stomatal opening is driven by the guard cell-swelling in response to blue light, suggesting that a single guard cell possesses all signaling components, from blue light-perception to cell-volume increase, for stomatal opening 2,9 . Blue light activates the plasma membrane H + -ATPase via photoreceptor phototropinmediated signaling 10,11 . The activated H + -ATPase hyperpolarizes the plasma membrane, followed by activation of inward-rectifying K + channels [12][13][14] . In addition, phototropin signaling distinct from the H + -ATPase pathway activates the K + channels via CBL-interacting protein kinase 23 (CIPK 23) 15,16 . Theses K + channel activation induces an influx of K + , resulting in the accumulation of K + and the counter anions of Cl -, nitrate (NO3 -) and malate 2in a guard cell. K + and Clare transported into the vacuole via the tonoplast-localized transporters NHX1 and NHX2, and channels ALMT9 and CLCc, respectively [17][18][19][20] . Accumulation of these ions decreases the guard cell-water potential, leading to water uptake into the vacuole and turgor increase, which finally induces stomatal opening 14,21 .
Apart from these ion transport events, starch degradation in guard cell chloroplasts is induced downstream of phototropin-activated H + -ATPase, which also contributes to stomatal opening without affecting the capacity of guard cellion transports 22,23 . However, these signaling and the regulatory mechanisms for each event are not completely elucidated and our understanding of stomatal opening is still not sufficient.
In this study, to identify novel factors involved in stomatal opening, we performed a large-scale screening of mutants defective in stomatal opening in Arabidopsis using infrared thermography. Through gene mapping and detailed functional analysis of one mutant (cst2 for closed stomata 2), we found that a tonoplast-localized transporter for Mg, CST2 is involved in stomatal opening in Arabidopsis.

Isolation and physiological characterization of a mutant defective in stomatal opening
To further understand the mechanisms involved in stomatal opening, we performed a large-scale genetic screening of Arabidopsis mutants using infrared thermography 24 . We obtained several mutants (designated closed stomata (cst)) showing higher leaf temperature than WT, due to the impairment of stomatal opening. We have functionally characterized the first mutant cst1 24 , and in this study we report the second mutant cst2-1, which also showed higher leaf temperature and retarded growth (Fig. 1a). Physiological analysis showed that cst2-1 exhibited smaller stomatal opening than the wild type (WT) under both light and dark conditions (Fig. 1b), but the difference in stomatal opening between cst2-1 and WT was larger under light condition. cst2-1 mutant showed reduced photosynthetic electron transfer to PSII compared to WT (Fig. 1c). The growth of cst2-1 was greatly decreased compared with the WT (Fig. 1d). Intriguingly, when grown under drought stress condition, the WT was wilted, but the mutant survived well (Fig. 1e), probably due to stomata closure to prevent water loss. All these observations indicate that the phenotype of cst2-1 is caused by defect of stomatal opening under light condition.

Gene mapping and complementation test
In order to map the gene responsible for the cst2-1 phenotype, we performed map-based cloning and sequence analysis using the next-generation sequencer.
As a result, we identified a single nucleotide substitution from guanine (G) to adenine (A) at the sixth intron splicing acceptor site of the AT4G14240 gene ( Supplementary Fig. 1a). Western blot analysis indicated that there was no CST2 protein detected ( Supplementary Fig. 1b). To confirm this mapping result, we performed a complementation test by introducing the wild-type CST2 gene into cst2-1 mutant. Analysis with three independent lines showed that CST2 completely restored growth defect and stomatal opening of cst2-1 mutant ( Supplementary Fig. 1c, d). Furthermore, we obtained a T-DNA insertion line (GABI_322H07), which is designated as cst2-2 ( Supplementary Fig. 1a). There was no expression of CST2 gene and protein in this line ( Supplementary Fig. 1b, e). Furthermore, cst2-2 showed similar phenotypes to cst2-1 in terms of growth, stomatal opening and photosynthesis (Fig. 1b, c, d), indicating that cst2-2 is also a knockout mutant of CST2.

Phylogenetic analysis of CST2
CST2 encodes an uncharacterized protein with a Domain of Unknown Function 21 (DUF21) and a cytosolic cystathionine-b-synthase (CBS)-pair domain ( Supplementary Fig. 2a). In Arabidopsis genome, there are six homologs of CST2 with similarity ranging from 43.6 to 82.4% ( Supplementary Fig. 2b). Blast search revealed that the domains of CST2 protein are found in archaea, bacteria, fungi, algae, plants, and animals ( Supplementary Fig. 2c). Among them, bacterial homologs, CorB and MpfA were reported to mediate Mg efflux [25][26][27] Fig. 2c) was reported to be associated with manganese (Mn) homeostasis and located to the tonoplast, but its exact role in Mg homeostasis is unknown 31 .

Transport activity and subcellular localization of CST2
To test whether CST2 was also able to transport Mg like its homologs in other organisms, we performed yeast assay using a yeast mutant, mam3. When yeast was grown in SOD medium containing 3 mM Mg, mam3 knockout mutant showed reduced Mg accumulation compared to WT (Fig. 2a We then investigated the subcellular localization of CST2 by transiently expressing CST2-GFP fusion with tonoplast marker (Tonoplast target signal (TTS)-mCherry) and plasma membrane marker (mCherry-AHA1) in Nicotiana benthamiana leaves (Fig. 2b). The result showed that the fluorescence signal from CST2-GFP was well merged with tonoplast marker. These results indicate that similar to MAM3, CST2 protein is also localized to the tonoplast.

Expression pattern of CST2
We examined the expression pattern of CST2 by qRT-PCR. The expression of CST2 was higher in the leaves than in the roots (Fig. 2c). The expression in both the roots and shoots did not respond to Mg-deficiency and -excess (Fig. 2c).
Western blot analysis showed that CST2 protein accumulation was not altered by high Mg (20 mM) up to 72 hours (Fig. 2c). Furthermore, accumulation of the CST2 protein was not changed in response to 5, 10, 20 mM Mg supply compared with 0 mM Mg (Fig. 2d). These results indicate that CST2 is constitutively expressed at both transcriptional and protein levels.
We further investigated expression patterns of CST2 using transgenic plants expressing the GUS reporter gene under the control of native CST2 promoter.
Consistent with qRT-PCR result (Fig. 2c), GUS signal was ubiquitously observed in both the roots and leaves with stronger signal in the leaves (Fig. 3a-c, f, g). In the leave, a strong signal was observed in the guard cells (Fig. 3d, e) and vascular bundle tissues (Fig. 3c, d, f, h, i). Expression of CST2 protein was also confirmed by western blot analysis in guard cells, mesophyll cells, leaves, and roots ( Fig.   3j).

Effect of different Mg concentrations on growth of cst2 mutants and overexpression lines
To investigate the role of CST2 in Mg homeostasis, we compared the growth between WT and two independent cst2 mutants at different Mg concentrations including 0.25, 1, and 5 mM. When WT and cst2 mutants were grown hydroponically for 4 weeks at 0.25 mM Mg, the growth of both the roots and shoots were almost similar (Fig. 4a-c). However, at 1 (a concentration for normal growth) and 5 mM Mg, the growth of both the roots and shoots of cst2 mutants were significantly decreased compared with the WT (Fig. 4a-c). We further tested whether this response is specific to high Mg by exposing the plants to different ions. The results showed that the growth difference between WT and cst2 mutants was only observed at high Mg concentration ( Supplementary Fig. 3).
These results indicate that knockout of CST2 specifically increased sensitivity to high Mg concentrations.
We also generated over-expression lines of CST2. The over-expression lines showed higher accumulation of CST2 proteins ( Supplementary Fig. 4a). Overexpression of CST2 significantly enhanced tolerance to high Mg concentrations compared with the WT (Supplementary Fig. 4b). At Mg concentrations above 35 mM, the overexpression lines were able to survive, whereas WT died.
We compared Mg concentration in the leaves between WT and cst2 mutants.
The results showed that at low Mg concentration (0.75 mM), the leaf Mg concentration was similar between WT and cst2 mutant (Fig. 4d). However, at high Mg concentration (5.75 mM), the mutants contained lower Mg concentration than the WT. This decrease in Mg accumulation could be attributed to disrupted Mg homeostasis in the cells. Higher Mg concentration in cytosol in the mutant may down-regulate transporter genes involved in Mg uptake although further investigation is required.

Mg homeostasis is associated with stomatal opening.
To link Mg homeostasis with stomatal opening, we investigated the light- We further investigated Mg-dependent stomatal opening in a time-dependent manner. The stomatal opening capacities of the mutants started to decrease at 24 hours after the plants were transferred from 0.2 to 5 mM Mg (Fig. 5b). By contrast, the stomatal opening capacity of WT was hardly changed over time under the same conditions. We also compared stomatal conductance and photosynthetic rate between the WT and cst2 mutants. Magnitude and rate of stomatal conductance were reduced in cst2 mutants compared to those in WT ( Fig. 5c). Photosynthetic rate of cst2 mutant leaves was also reduced (Fig. 5d).
These results suggest that Mg homeostasis is important for stomatal opening and photosynthesis.

Functional relationship between CST2 and CBL-CIPK complex
It was reported that the tonoplast-localized calcium sensors, the calcineurin Blike (CBL) 2 and CBL3 and their regulating partners, CBL-interacting protein kinase (CIPK) 3, CIPK9, CIPK23, and CIPK26 are involved in regulation of Mg homeostasis through Mg sequestration into vacuoles 32, 33 . However, it is unknown whether CST2 is regulated by the CBL-CIPK complexes. Because the cbl2 cbl3 double mutant also showed similar phenotype to cst2 mutants (Mg hypersensitive and low Mg accumulation) 32 , we therefore, examined the possible interaction between CST2 and CBL2/3. We first compared Mg sensitivity among cst2-1, cbl2 cbl3, and cst2-1 cbl2 cbl3 mutants (Fig. 6a). The cbl2 cbl3 mutant showed a weaker growth inhibition than the cst2-1 mutant in response to high Mg supply. The Mg-hypersensitive phenotype of cst2-1 cbl2 cbl3 triple mutant was similar to that of cst2-1 (Fig. 6b). These results indicate that CST2 and CBL2/3 work on the same signaling pathway. We further investigated the effect of CST2 overexpression on Mg sensitivity in the cbl2 cbl3 background. Overexpression of CST2-GFP largely recovered the hypersensitive phenotype of the cbl2 cbl3 mutant (Fig. 6b), indicating that CST2 functions downstream of CBL2/3 in the same signaling pathway.
We then examined the protein-protein interaction between CST2 and CIPK26 using an in vitro pull-down assay. Recombinant GST-CIPK26 and GST-SRK2E were used as baits and incubated with recombinant His-CST2 C-terminal fragment (CST2 C-ter). CIPK26 interacted with CST2 but SRK2E as a control did not in vitro (Fig. 6c). Bimolecular fluorescence complementation (BiFC) assay in Nicotiana benthamiana leaves further confirmed the interaction between CIPK26 and CST2 (Fig. 6d).
We finally performed in vitro phosphorylation assay using the recombinant GST-

Discussion
Stomatal opening in plant leaves is caused by the swelling of a pair of guard cells, which is achieved by increasing cell volume through ion accumulations followed by water influx 2,34 . Until now, it has been reported that accumulation of ions including K + , Cl -, and nitrate (NO3 -), and malate 2in guard cells, in particular, the accumulation of K + and Clin the vacuoles largely contributes to the cell volume increase of guard cells 20,34 . In this study, we identified a novel component Our results revealed that sequestration of Mg by CST2 into the vacuole is required for stomatal opening under high Mg concentration conditions. There are at least three possibilities for involvement of CST2 in stomatal opening. The first one is that CST2 mediates transport of Mg into the vacuoles of guard cells ( Fig. 7), which contributes to increase in the osmotic pressure for guard cellswelling. However, this possibility is unlikely because cst2 mutants also showed similar stomatal opening as WT at low Mg concentrations (Fig. 5a). The second possibility is that Mg homeostasis in the guard cells affects other components involved in stomatal opening (Fig. 7). Although Mg is an essential element for plant growth and development, excess Mg in the cytosol due to lack of vacuolar sequestration may cause toxicity, which affects other events involved in stomatal opening. In fact, it was reported that excess Mg inhibits photosynthetic activity through affecting the stroma pH 35,36 . Because the guard cell photosynthesis provides ATP and/or reducing equivalents, which are the fuel for stomatal Recently, it was demonstrated that the tonoplast-localized calcium sensors, the calcineurin B-like proteins (CBLs) and their regulating partners, CBL-interacting protein kinases (CIPKs) regulates Mg homeostasis through Mg sequestration into vacuoles 32, 33, 38 . Our results revealed that CST2 is one of the target transporters phosphorylated by CBLs-CIPKs (Fig. 6e). This is supported by the genetic, physiological, and physical evidence presented in this study ( Fig. 6-d ) .
Phosphorylation of CST2 is probably required for activating its transport activity although further investigation is required.
In summary, CST2 is a novel type transporter of Mg, which is involved in stomatal opening in Arabidopsis. Its localization at the tonoplast functions to sequester Mg into the vacuoles, which is important for maintaining Mg homeostasis in guard cells and subsequently for stomatal opening. Our work provides novel insights into regulation of stomatal opening in plants.

Screening of mutants defective in stomatal opening
Ethyl methanesulfonate-mutagenized M2 seeds purchased from Lehle Seeds (the trichome-less glabra1 background), were used for screening of mutants as described previously . Plants were grown in soil under a 14 h white light (50 mmol m -2 s -1 ) and 10 h dark cycle at 20-25°C at a relative humidity of 55-75% in a temperature-controlled growth room. At 21-25 days, leaf temperature was measured using an infrared thermograph (TVS-500EX; NEC Avio Infrared Technology). Plants were grown for 3 weeks and thermal images were taken at 3-4 h after the start of the light period. Thermal images were analyzed using the Avio Thermography Studio software (NEC Avio Infrared Technology) (Fig. 1a). A mutant showing high leaf temperature, designated as cst2-1 for closed stomata 2 was used in this study.

Physiological characterization of cst2 mutants
To characterize the phenotypes of cst2 mutants, leaf temperature, stomatal aperture, photosynthetic electron transfer, shoot fresh weight, drought tolerance, and Mg concentration were measured and compared with WT as described below.

Measurement of stomatal aperture and stomatal conductance
For stomatal aperture measurements, the seedlings (4-week-old) grown in soil or hydroponically were kept in the dark before measurement. Rosette leaves were harvested from the dark-adapted plants, and epidermal fragments were isolated using a Waring blender (Waring Commercial) under dim red light. The epidermal fragments were collected on a 58 μm nylon mesh and used for stomatal aperture measurement according to Inoue et al. (2008b) 11

Transport assay of Mg in yeast
To determine transport activity of CST2 for Mg, yeast mutant mam3 was used.
Yeast strain BY4741 as WT and mam3 mutant (4741, mam3 Δ:: KANR) were purchased (Thermo). MAM3 coding sequence and CST2 cDNA were amplified from yeast genome, and Arabidopsis cDNA, respectively using the primers 5'-  Olympus). The plasmids constructed as above were transformed into the Agrobacterium tumefaciens GV3101 strain. Agrobacterium was transformed into the cst2-1 mutant by the floral dip method 57 . Transgenic plants were selected by resistance against hygromycin or Kanamycin and used for analysis. Transgenic lines were used for phenotypic analysis and observation as described below.

Phenotypic analysis of transgenic lines
For analysis of complementation lines, three independent lines (Comp1 to Copm3) lines were grown on soil for 4 weeks and used for comparison of growth, stomatal opening, and photosynthetic electron transfer as described above.

Promoter-GUS analysis
Leaves and roots were sampled for 1 to 3-week-old seedling for promoter-GUS

Construction of phylogenetic tree
Amino acid sequences of CST2 homologs were aligned using MUSCLE 59 . The phylogenetic tree was constructed using the neighbor-joining method and bootstrapping of 1000 replications in MEGAX 60 . Full-length amino acid sequences of CST2 and homolog proteins were used to construct phylogenetic trees.

Generation of CST2 antibodies
Anti

In vitro pull-down assay
Full-length cDNAs of CIPK26 and SRK2E were amplified from the wild-type (Col-0) cDNAs using the primers:

BiFC assay
To test the interaction of CST2 with CIPK26 in plant cells, bimolecular fluorescent complementation (BiFC) assay was performed using N. benthamiana leaves.
Full-length cDNAs of CST2, CIPK26, and SRK2E were amplified by RT-PCR using the following oligonucleotide primers: pSPYNE and pSPYCE containing the indicated inserts was performed as previously described 16,62 except that the p19 silencing suppressor strain was mixed and simultaneously infiltrated. Detection of reconstituted YFP fluorescence was monitored 4 days after infiltration using a confocal microscope as described above.

In vitro kinase assay
To test whether the CIPK26 phosphorylates CST2, we performed in vitro phosphorylation assay. The plasmid constructs for recombinant proteins, GST-CIPK26 and His-CST2 C-ter as described above were used. To inactivate the kinase activity of CIPK26 for use as a negative control, a mutated protein with a single amino acid substitution CIPK26 D154N was used. The aspartic acid is a binding site of Mg 2+ -ATP in protein kinases and substitution of the aspartic acid with asparagine leads to inactivation of protein kinases 11,63 . Nucleotide substitution was introduced into the CIPK26 gene in the pGEX2T vector as templates for inverse PCR and self-ligation. Inverse PCR was performed using the oligonucleotide primers 5′-AATTTTGGATTGAGTGCGTTGTCC-3′ and 5′-AGAGACTTTCAGATTTCCTTGAGC-3′. The resulting vector was introduced into the E. coli strain BL21 (DE3), and used for protein expression.
The recombinant GST-and His-fused proteins were expressed and purified as

Statistical analysis
All experiments were independently repeated at least two times. Statistical analyses were performed by Tukey's test. Significance of differences were defined as *P < 0.05, **P < 0.01, ***P < 0.001, or by different letters (P < 0.01).