Arabidopsis thaliana ecotype Columbia-0 (Col-0) and the T-DNA insertion mutant cml20 (SALK_079974c) generated in a Col-0 background were obtained from the ABRC¹. The zygosity of the mutant individuals was validated by PCR using the cml20-specific primer pair cml20-LP/-RP and the T-DNA left border primer LBb1.3². The relevant primer sequences are given in Table 1. To grow plants, seeds were surface-sterilized by immersing in 75% v/v ethanol for 3 min, followed by 95% v/v ethanol for 1 min, then the ethanol was allowed to evaporate. The seeds were then plated on solidified (0.7% w/v agar) half strength Murashige and Skoog (1962) medium (1/2 MS) and held first for 2 days at 4°C and then kept in a growth chamber under long-day conditions (16 h light/8 h darkness, 22 ± 1/16 ± 4°C) with illumination by 100 μmol m^-2s^-1 light); the relative humidity was maintained at ∼70%.

The coding region of CML20 (At3g50360) was amplified from Col-0 cDNA by using the primer pair CML20-HisF/-HisR (Table 1), while CaM7 (At3g43810) was amplified by using CaM7-HisF/-HisR (Table 1) as positive control. The amplicons and the prokaryotic expression vector pET30a (+) were both digested with NdeI and XhoI, and then ligated using T4 DNA ligase (TransGen Biotech, China). For heterologous expression of CML20, E. coli BL21 (DE3) transformants harboring recombinant plasmid CML20 were induced with 0.4 mM isopropyl β-D-1-thiogalactopyranoside (IPTG); the recombinant protein was thereafter purified using the Ni-NTA Purification System (Invitrogen, Uinted States). Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) mobility shift assay for Ca²⁺ binding was performed by exposing denatured CML20 to either 15 mM CaCl₂ or 15 mM ethylene glycol tetraacetic acid (EGTA), as described by Garrigos et al. (1991). The structure of CML20 was predicted using InterPro software³.

To engineer plant expression vector for CML20 over-expression, the gene’s open reading frame was amplified using the primer pair CML20-OEF/-OER (Table 1), and the resulting amplicon was digested with XbaI and SacI, then ligated into the XbaI and SacI cloning sites of pSTART (+) which generated the p35S::CML20 construct. The CML20 open reading frame was amplified using the primer pair CML20-CF/-CR (Table 1) cloned into pDonor 221 vector and then introduced into pB2GW7.0 vector through gateway reaction (Invitrogen, United States) to generate pB2GW7.0-CML20. cml20 mutant plants were transformed with pB2GW7.0-CML20 to generate complemented lines, C-1 and C-2. The CML20 promoter sequence was amplified using the primer pair CML20-PF/-PR (Table 1), then ligated into the HindIII/BamHI cloning site of pCambia::UBI-GUS to generate the construct pCML20::GUS. All the constructs were transferred into Agrobacterium tumefaciens strain GV3101 and then introduced into Arabidopsis using the floral dip method (Clough and Bent, 1998). T₁ transgenic seedlings were selected by growing on 1/2 MS containing 30 mg/L kanamycin (p35S::CML20 lines), or hygromycin (pCML20::GUS lines). The zygosity of T₃ lines was deduced from the transgene’s segregation behavior, and the abundance of the transgene transcript was assessed by using PCR or quantitative real time PCR (qRT-PCR), and the selected seeds from T₃ homozygous lines were used for further analysis.

A p35S::CML20-GFP construct was generated by inserting the CML20 open reading frame (amplified using CML20-GFPF/-GFPR, see Table 1) into the BamHI/SalI cloning sites of a modified GFP expression vector (Lin et al., 2009). Both the p35S::CML20-GFP and the empty p35S::GFP plasmids were purified using a NucleoBond^® Xtra Midi Kit⁴ (Macherey–Nagel) before introducing into Arabidopsis mesophyll protoplasts following Sheen (2001). After 16 h incubation at 23°C in the dark, protoplasts were examined for GFP signal using confocal laser scanning microscopy (performed at The Microscopy Characterization Facility, Shandong University).

T₃ homozygous transgenic plants harboring pCML20::GUS were assayed for GUS activity as described by Jefferson et al. (1987) with slight modification. Briefly, transgenic plant tissues were soaked in a GUS staining solution (2 mM X-Gluc, 2 mM K₃Fe(CN)₆, 2 mM K₄Fe(CN)₆, 0.1% Triton X-100, 10 mM EDTA in 50 mM sodium phosphate buffer, pH7.2) and incubated overnight at 37°C. After staining, the samples were washed in 50, 70, and 100% ethanol for 5 min consecutively, and then shook in 100% ethanol for about 5 h to remove chlorophyll. Subsequently, the samples were examined under an optical microscope.

To apply drought stress, the protocol described by Sakamoto et al. (2004) was followed with minor modifications. Seedlings were grown under well-watered conditions for four weeks, and then deprived of water for 2–3 weeks. Then, the plants were re-watered for over three days and photographed. For water-loss assays, rosette leaves were collected from 4-week-old plants as test samples. The samples were weighed immediately on a piece of paper, and then placed on the laboratory bench (relative humidity: 40%–50%; 22°C–23°C). The weight lost by each sample at a set of pre-assigned time points was recorded.

Arabidopsis thaliana plants were grown under a 10 h photoperiod (100 μmol m^-2s^-1 light), where the light and dark period temperatures were 22 and 20°C, respectively. The stomatal aperture assay was performed using fully expanded young leaves of 4-week-old plants. To characterize stomatal opening, detached rosette leaves were first floated adaxial side up for 2.5 h in the dark in 10 mM KCl, 7.5 mM iminodiacetic acid, 10 mM Mes-KOH (pH 6.15). Subsequently, either 50 μM ABA or solvent control (ethanol) was added, then the treated leaves were illuminated for 2.5 h. For the stomatal closure assay, initially leaves were floated in the solution containing 20 mM KCl, 1 mM CaCl₂, 5 mM Mes-KOH (pH 6.15) for 2.5 h in the light, then either ABA (1, 10, or 50 μM) or solvent control (ethanol) was added. In both cases, abaxial epidermal strips were peeled off after 2.5 h incubation (with ABA/solvent control ethanol) and photographed immediately. Stomatal aperture widths were estimated from the captured digital images using ImageJ v1.47⁵ software.

Arabidopsis thaliana guard cell protoplasts were isolated following the Zhang et al. (2008) method. K⁺ currents were recorded following Li et al. (2016) with slight modifications. The bath solution contained 4 mM MgCl₂, 10 mM MES-Tris (pH 5.6), 1 mM CaCl₂, 10 mM K-Glu, and the osmolarity was adjusted to 500 mOSM with sorbitol for measuring whole-cell channel K⁺ currents. The pipette solution contained 20 mM KCl, 10 mM HEPES-Tris (pH 7.8), and 80 mM K-Glu, and the osmolarity was adjusted to 550 mOSM, and also fresh ATP (5 mM Mg-ATP) was added before use. Axopatch-200B amplifier (Axon Instruments, United States) connected to a computer via an interface (TL-1 DMA Interface; Axon Instruments) was used to achieve whole-cell configuration, and the holding potential was set as –60 mV. The whole-cell currents were recorded 5 min later. The test voltage steps were from –200 mV to –40 mV with +20 mV increments, and each test voltage has 4.9-s duration. For the ABA treatment, 50 μM ABA was added to the bath solution after the whole-cell configuration was achieved. The whole-cell currents were recorded 10 min later. Guard cell anion currents were recorded as described by (Acharya et al., 2013). Briefly, the bath solutions contained 2 mM MgCl₂, 30 mM CsCl, 10 mM MES-Tris (pH 5.6), and 1 mM CaCl₂. The pipette solutions contained 150 mM CsCl, 2 mM MgCl₂, 6.7 mM EDTA, 3.35 mM CaCl₂, and 10 mM HEPES-Tris (pH 7.5). Osmolarity of the solutions was adjusted to 480 and 500 mOSM for bath and pipette solutions, respectively, with sorbitol. ATP (10 mM Mg-ATP) and GTP (10 mM) were added to the pipette solution before use. The holding potential was +30 mV. The voltage steps were applied from -145 to +35 mV with +30 mV increments, and each test voltage has a 60-s duration. For the ABA treatment, guard cell protoplasts were treated with 50 μM ABA for at least 1 h before recording.

Total RNA was isolated using the TRIzol reagent (Roche, Switzerland), treated with RNase-free DNase I (Invitrogen, United States) and reverse transcribed using oligo-dT primers and SuperScript^TM III Reverse Transcriptase (Invitrogen, United States). The resulting cDNA was used as template in both RT- and qRT-PCRs. The former employed the CML20-specific primer pair CML20-RTF/-RTR (Table 1) and reference reactions were primed by ACTIN2-RTF/-RTR (Table 1), to amplify the gene ACTIN2 (At3g18780). For the qRT-PCRs, leaves and epidermal strips derived from four-week old plants were exposed to 50 μM ABA for 6 h. All reactions were performed in triplicates using FastStart Universal SYBR Green master mix (Roche, Switzerland), and ACTIN2 was used as the reference. The CML20-specific primer pair CML20-QF/-QR and the ACTIN2 pair ACTIN2-QF/-QR (Table 1) were used for the qRT-PCR assays. The abundance of transcript of a set of known stress-responsive genes was also assayed by qRT-PCR using cDNA synthesized from seedlings (two-week old) that were harvested 6 h after exposure to 50 μM ABA. The gene/primer pair combinations were RAB18 (At5g66400)/RAB18F/R, COR47 (At1g20440)/ COR47F/R, MYB2 (At2g47190)/MYB2F/R, ERD10 (At1g20450)/ERD10F/R, KIN1 (At5g15960)/KIN1F/R, RD29A (At5g52310)/RD29AF/R, KIN2 (At5g15970)/ KIN2F/R, and APX2 (At3g09640)/APX2F/R. All primer sequences are given in Table 1.

Reactive oxygen species production in guard cells was detected by loading abaxial epidermal strips with H₂DCF-DA (Zhang et al., 2011). Epidermal strips were peeled from detached leaves from plants, and floated in 10 mM KCl, 7.5 mM iminodiacetic acid, 10 mM Mes-KOH (pH 6.15) for 2 h to induce stomatal opening. Then, the epidermal strips were transferred to 10 mM Tris HCl, 50 mM KCl (pH 7.2) containing 50 mM H₂DCF-DA, and held in the dark for 10 min. Unincorporated H₂DCF-DA was removed by rinsing three times in double distilled water. The guard cells were then subjected to laser scanning confocal microscopy, and treated with either 50 μM ABA or solvent control before data collection. The fluorescence of H₂DCF was captured by imaging at 2.5 min intervals over 25 min. ZEN software⁶ (2012, blue edition) was used to quantify the data. The change in fluorescence intensity at each time point was calculated in the form of a percentage of its initial intensity.

The predicted CML20 product was a CAM-like protein consisting of 169 aa residues. The structure of CML20, according to InterPro⁷, contains four EF-hand domains (Figure 1B), as seen in numerous Ca²⁺ binding proteins (Figure 1A) such as CAM2, CAM7 and CML9 (Fabienne et al., 2008; Garrigos et al., 1991; McCormack et al., 2005). The SDS-PAGE mobility shift assay revealed that CML20 migrated faster in the presence of free Ca²⁺ than in the presence of Ca²⁺ chelator EGTA (Figure 1C). Like CML20, we also observed similar result for CAM7 (Garrigos et al., 1991). Our findings suggested that CML20 could bind Ca²⁺ in vitro and thus potentially acts as a Ca²⁺ sensor in plant.

As deduced from the sites of GUS production in transgenic plants harboring pCML20::GUS, expression of CML20 was observed in the root, leaf, inflorescence and silique (Figures 2A–G). CML20 expression was also induced in the guard cells of plants that were exposed to ABA (Figures 2F,G). qRT-PCR result also supported that CML20 was induced in response to ABA treatment in the leaf and epidermis, which contains a large number of guard cells (Figure 2H). All of these findings suggested the possible involvement of CML20 in the ABA-mediated regulation of stomatal movement. When mesophyll protoplasts were transformed with p35S::CML20-GFP, GFP signal was detected exclusively in the cytoplasm (Figure 2I), but mesophyll protoplasts transformed with p35S::GFP empty vector control showed GFP signal dispersed throughout the protoplast (Figure 2J). These findings indicate that CML20 is a cytosol-localized protein.

Transcriptional analysis of the cml20 mutant (Figure 3A) was carried out using both RT- PCR and qRT-PCR which confirmed the absence of CML20 transcript (Figures 3B,C). Given that CML20 was clearly up-regulated by ABA in the guard cells, the response of cml20 guard cells was of interest to study its role in ABA signaling. Stomatal aperture was indistinguishable between the mutant and WT when plants were not treated with ABA, but in the presence of ABA (1, 10, or 50 μM), the mutant stomata showed smaller aperture than the WT (Figures 3D,E). The results indicated that cml20 stomata were hypersensitive to both ABA-promoted stomatal closure as well as the ABA-inhibited stomatal opening, and thus to minimize the water loss to increase the plant drought tolerance. This hypothesis was supported by the observed results of both the detached leaf assay and the whole plant response to drought stress as well (Figures 3F,G).

qRT-PCR result revealed that CML20 over-expressing lines, OE-1 and -2 both produced more CML20 transcript than WT (Figure 4A). Stomatal movement in these lines was less affected by ABA than in the WT (Figures 4B,C). As predicted, the rate of water loss from detached leaves of the two OE lines was higher than WT leaves (Figure 4D), and the OE-1 and OE-2 plants were more sensitive to drought stress (Figure 4E). In addition, cml20/CML20 complementation lines (C-1 and C-2) showed similar phenotype for both ABA regulation of stomatal movement and drought stress resistance compared to WT (Supplementary Figure S1). These findings provided additional support that CML20 functions as a negative regulatory signaling component in response to both ABA and drought stress.

Stomatal closure relies on anion efflux via channels that are activated by ABA, which also simultaneously inhibits the influx of K⁺. The patch clamp assay was conducted to test whether ABA-modulated currents depended on CML20. Without ABA treatment, we did not observe any difference in either the K⁺ or the anion currents between WT, cml20 and the two OE lines. In contrast, in response to ABA treatment, the mutant’s guard cells showed hypersensitivity to ABA inhibition of K⁺ currents and ABA activation of anion currents, while the over-expression of CML20 impaired them both (Figures 5A–D). These results coincided with the performance of stomatal movement in response to ABA treatment. The conclusion was that CML20 functioned as a negative regulatory signaling component for ABA regulation of stomatal movement, partly via its regulatory effect on ion channels activity in guard cells.

Transcriptional (qRT-PCR) profiling showed that in cml20, certain ABA-inducible genes (ERD10, RAB18, COR47, and MYB2) were up-regulated by exposure to ABA for 6 h (Figure 6). The imposition of drought stress also induced the transcription of an ABA-independent gene RD29A in the cml20 mutant (Figure 6). The aforementioned genes are up-regulated in response to abiotic stresses also their transcript level could be correlated with sensitivity to a specific stress. These findings not only implicate that CML20 negatively regulates transcription of these above indicated genes but also suggest that CML20 is also a regulator of the ABA and drought stress signaling pathways.

The ABA treatment resulted in a greater increase of ROS production in the guard cells of cml20 than in those of WT (Figures 7A,B). A qRT-PCR based assay of the transcription of a gene known to be involved in ROS elimination showed that APX2 (encoding an H₂O₂ scavenger) was down-regulated in cml20 whether or not the plants were exposed to ABA (Figure 7C). These above findings suggest that CML20 may play a positive role in guard cell ROS removal; in its absence, the down-regulation of APX2 could have compromised the plant’s ability to control the accumulation of ROS.