Arabidopsis thaliana (Col-0 ecotype) was grown at 23°C in soil under 16-h light and 8-h dark conditions. For all assays, rosette leaves were collected from 3–4 week-old Arabidopsis plants grown under these conditions. sine1-1 (SALK_018239C) and sine2-1 (CS801355), were previously reported, and shown to have no full-length SINE1 or SINE2 mRNA accumulates, respectively (Zhou et al., 2014). 35Spro::GFP-LIFEACT in Col-0 (GFP-LIFEACT; Tolmie et al., 2017) was crossed with sine1-1 or sine2-1 and bred until homozygous sine1-1 and sine2-1 mutants expressing GFP-LIFEACT were obtained. Genotyping for the sine1-1 and sine2-1 insertion alleles were performed as described before (Zhou et al., 2014).

All primers used in cloning and construct generation are outlined in Supplementary Table 1. The guard cell-specific promoter GC1 was cloned from whole seedling genomic DNA (Yang et al., 2008). Restriction sites for enzymes SacI and SpeI were added to the 5′ and 3′ ends and the amplified fragment was digested with the appropriate restriction enzymes. The GC1 promoter fragment was isolated, purified with the QIAquik PCR Purification kit (Qiagen) and subsequently ligated into a pH2GW7 vector (Takagi et al., 2011).

The Yellow Cameleon 3.6 (YC3.6) Ca²⁺ sensor N-terminally tagged with the rabbit heat stable protein kinase inhibitor α (PKIα) nuclear export signal (NES) was cloned from UBQ10pro::NES-YC3.6 (Krebs et al., 2011). The R-GECO1 sensor, N-terminally tagged with the SV40 nuclear localization signal (NLS) was amplified from CMVpro::NLS-R-GECO1 (Zhao et al., 2011). NES-YC3.6 and NLS-R-GECO1 were cloned into pENTR/D-TOPO vectors (Life Technologies) and then moved to GC1pro::pH2GW7 via the Gateway^® LR reaction (Life Technologies).

Binary vectors containing Ca²⁺ sensors were transformed into either Agrobacterium tumefaciens strains ABI or GV3101 by triparental mating (Wise et al., 2006). The Agrobacterium-mediated floral dip method was used to transform Col-0 ecotype (WT), sine1-1, and sine2-1 backgrounds (Clough and Bent, 1998). Transgenic plants were isolated on Murashige Skoog (MS) plates supplemented with using 30 μg/ml hygromycin, and positive transformants were confirmed by confocal microscopy. One homozygous T2 transgenic line for each background was used for all assays described here.

Stomatal bioassays were performed as previously described (Biel et al., 2020a). Briefly, rosette leaves of 3–4 week-old plants were placed abaxial side up in opening buffer (OB) containing 10 mM MES, 20 μM CaCl₂, 50 mM KCl, and 1% sucrose at pH 6.15 for 2 h under constant light. Stomatal closing assays were performed immediately after the opening assays, in which leaves were transferred to closing buffer (10 mM MES at pH 6.15) with or without the following treatments, as indicated: 20 μM ABA, 2 mM or 10 mM CaCl₂ and/or 10 μM LatB (Kang et al., 2002; Jiang et al., 2014). NIS-Elements AR version 3.2 software was used for stomatal aperture measurements.

Arabidopsis expressing the NES-YC3.6 Ca²⁺ sensor driven by the GC1 promoter was imaged using a Nikon Eclipse C90i system. The NES-YC3.6 sensor was excited with 457 nm light, and fluorescence emission was detected between 465 and 505 nm (CFP) and between 530 nm and 570 (cpVenus). Ca²⁺ imaging of stomata was performed as previously described (Behera and Kudla, 2013). Briefly, epidermal peels were mounted on a microscope slide with medical adhesive (Hollister, Libertyville, IL, United States) and incubated in opening buffer (10 mM MES, 20 μM CaCl₂, 50 mM KCl, and 1% sucrose at pH 6.15) under light for 3 h. A perfusion chamber (Grace Bio-Labs CoverWell) was attached to the slide and the perfusion system was attached to input and output tubing. The fluorescence signal of individual guard cells was observed using confocal microscopy with a 20x oil immersion objective and a 2x digital zoom. To establish a pretreatment baseline, the sample was treated with opening buffer for 5 min and images were captured every 10 s. Next, the epidermal peels were treated with 20 μM ABA (Biel et al., 2020a), 10 mM Ca²⁺ (Jeon et al., 2019) and/or 10 μM LatB (Biel et al., 2020a) and imaged for 45 min. Immediately after the 45-min treatment, guard cells were treated with 100 mM CaCl₂ for 5 min as a positive control for sensor function. The software NIS-Elements was used to isolate individual guard cells and quantify cytoplasmic or nuclear fluorescence. For NES-YC3.6, the YFP/CFP ratio for each time point was calculated by dividing the mean fluorescence intensity of YFP by the mean fluorescence intensity of CFP. Ca²⁺ peaks were defined as a YFP/CFP ratio increase and decrease of 0.5 A.U. or higher from the baseline.

Arabidopsis expressing the NLS-R-GECO1 Ca²⁺ sensor driven by the GC1 promoter were imaged as described above. The NLS-R-GECO1 sensor was excited with 561 nm light, and its emission was detected between 620 and 650 nm. Stomata were opened using full opening buffer (10 mM MES, 20 μM CaCl₂, 50 mM KCl, and 1% sucrose at pH 6.15) for 3 h under constant light and epidermal peels were prepared. The epidermal peels were pretreated for 5 min with a modified opening buffer lacking CaCl₂ (10 mM MES, 50 mM KCl, and 1% sucrose at pH 6.15). This adjustment was required because pretreatment with full opening buffer generated nuclear Ca²⁺ fluctuations prior to ABA addition and thus did not provide an adequate baseline (Supplementary Figure 1A). Removal of the 20 μM CaCl₂ from the full opening buffer abolished these fluctuations and was thus used to generate the 5-min baseline (Supplementary Figure 1B). The epidermal peels were imaged for 45 min after treatment with 20 μM ABA and then imaged for 5 min after treatment with 100 mM CaCl₂. The focal plane was adjusted to maintain maximum nuclear fluorescence signal during the time series to minimize any sample drift in the Z direction. RFP fluorescence was normalized by calculating (F − F₀)/F₀, where F₀ represents the average of the baseline values (30 frames) for that experiment. Nuclear Ca²⁺ peaks were determined based on an increase and decrease of 0.5 A.U. or higher from the normalized baseline. Area under the curve (AUC) for each data point was calculated using the following formula: | Y 1 + Y 2 | / ( 2 × ( X 2 + X 1 ) ) , where X is time and Y is the normalized Ca²⁺ value. The AUC for each guard cell is the sum of all calculated AUC points for each experiment.

Confocal microscopy was performed using a Nikon Eclipse C90i system. Images were taken using intact leaves at room temperature with a Plan Flour 60x oil objective (numerical aperture of 1.4, excitation wavelength 488 nm, emission wavelength 516 nm). Z-stacks of 3–18 slices were collected of the cortical layer of guard cells (ensuring exclusion of any nuclear signal) and used for subsequent quantification (Eun et al., 2001; Hwang and Lee, 2001; Gao et al., 2008). The number of filaments, occupancy, and mean angular difference were quantified as described previously with a few modifications (Biel et al., 2020b). For occupancy, a maximum intensity projection was created from the acquired z-stacks, and then the maximum intensity projection of the stoma was separated into individual guard cells. To perform noise reduction, a 1–5 pixel-band-pass filter was applied using the LPX Filter2d plugin (filter = bandpass; bpmode = Gaussian; lo = 1; hi = 5). The image was then binarized by thresholding. For mean angular difference, the binarized image was skeletonized using the LPX Filter2d plugin (filter = filtersbilevelThin; bilevelMode = bilevelThin_). The default settings were used. The average theta for each guard cell was calculated using the LPX Filter2d plugin, using the filter “lineFilters” and the linemode “lineFeature.”

Acridine Orange (AO) staining of vacuoles was performed as described previously with slight modifications (Mathur et al., 2003). Briefly, Arabidopsis leaves with the lower epidermal strip peeled were incubated in 10 μM AO (Sigma-Aldrich Item# A8097) in opening buffer (OB; see above) for 10 to 15 min under dark on a shaker and then washed twice for 2 min with OB. Leaves were then transferred to OB for 60 min under light and then transferred to specified buffers. Confocal microscopy was performed as described above, with the following exceptions: Utilizing an excitation laser of 488 nm, images were captured using 516 nm wavelength and single images were taken mid-cell, where vacuolar morphology was most visible. All images were analyzed using NIS-Elements AR version 3.2 software.

The number of guard cells analyzed for each line, in all figures, is ≥20. Error bars represent the standard error of means. Letters denote statistical groups (p < 0.01), as calculated using one-way ANOVA, followed by a post hoc Tukey HSD test. Selected pairwise comparisons were tested using Student’s t-test, and asterisks denote statistical significance.

The T-DNA insertion lines sine1-1 and sine2-1 have been characterized previously and shown to have defects in ABA-induced stomatal closure (Biel et al., 2020a). Col-0 expressing 35S-promoter driven GFP-LIFEACT (hereafter called GFP-LIFEACT; Tolmie et al., 2017) was crossed with sine1-1 and sine2-1 to visualize the cortical actin cytoskeleton (see “Materials and Methods”) and F2 lines homozygous for sine1-1 or sine2-1 and homozygous or heterozygous for GFP-LIFEACT were used for confocal imaging. To confirm that the actin marker did not influence the previously reported stomatal dynamics (Biel et al., 2020a), stomatal apertures of WT, sine1-1, and sine2-1 GFP-LIFEACT transgenic plants were measured at 0, 60, and 120 min after application of ABA. GFP-LIFEACT-expressing WT closes normally when exposed to ABA while sine1-1 and sine2-1 exhibit closing defects in the same range as previously shown for the marker-free lines (Figure 1A; Biel et al., 2020a).

Next, cortical F-actin patterns were imaged at 0, 60, and 120 min after ABA exposure. Two representative images each of maximal intensity projections of the guard cell cortex are shown in Figure 1B. To quantify the actin dynamics during stomatal closure, we used the following actin classification to qualitatively score images, based on Li et al. (2014): Type 1, radially arranged F-actin, with filaments extending from the dorsal to ventral guard cells walls; type 2, diffuse or mixed (radial and longitudinal) filaments; type 3, longitudinally bundled actin cables. We expanded these categories here to account for the loss of filaments seen in sine1-1 by subclassifying type 2 as type 2a and 2b: type 2a, predominantly diffused signal with fragmented F-actin; and type 2b, predominately mixed filaments with little or no diffuse signal (Figure 1C).

Images were visually scored and sorted into these four groups (Figure 1D). In WT guard cells at 0 min, the most prominent actin filament type was type 1 (radial, 71%), transitioning to type 2a at 60 min (diffuse, 73%) and to type 3 (longitudinal bundles, 67%). In contrast, sine1-1 guard cells at 0 min had significantly less type 1 (38%), and instead a mix of all three actin types. By 60 min, sine1-1 guard cells exhibited type 2a (73%) with little change seen at 120 min (type 2a, 67%). Loss of SINE2 resulted in a slight reduction of type 1 at 0 min (61%). At 60 min, sine2-1 guard cells had transitioned to type 2b (mixed filaments, 64%) with little change in organization seen at 120 min (type 2b, 78%).

Overall, fully open WT guard cells had predominantly radially organized actin filaments, which, upon ABA treatment, underwent reorganization, going through an intermediary, disorganized state (Figure 1D, 60 min), and arriving at a pattern of predominantly thick, longitudinally arrayed bundles (Figure 1D, 120 min). Conversely, sine mutants underwent less actin reorganization following exposure to ABA. Loss of SINE1 resulted in a loss of the organized, radial organization of actin filaments at the onset of the assay (Figure 1D, 0 min), which was followed by a state of largely diffuse signal indicative of F-actin depolymerization (Figure 1D, 60 min), that did not recover into an organized pattern by the end of the assay (Figure 1D, 120 min). Loss of SINE2 led to marginally fewer radial filaments prior to ABA perception (Figure 1D, 0 min), but the effect was less pronounced than for the loss of SINE1. F-actin remained mostly intact after ABA treatment, but became less radial and more mixed, and thereafter remained in a mixed state (Figure 1D, 120 min).

To quantify the observed differences in actin organization during ABA-induced stomatal closure, we calculated filament number (Figure 1E), mean angular difference (Figure 1F), and occupancy (Figure 1G). These parameters have been used previously to quantify actin and microtubule (MT) changes in guard cells (Yu et al., 2001, 2020; Higaki et al., 2010; Eisinger et al., 2012a,b; Biel et al., 2020b). Filament number is defined as the number of fluorescent peaks along the mid-width line of the guard cell, as described by Li et al. (2014). The number of filaments in WT guard cells decreases during closure and this change in filament number is used as a proxy to monitor actin bundling (Li et al., 2014). WT guard cells had significantly more filaments than sine1-1 or sine2-1 at 0 min (Figure 1E). During ABA-induced stomatal closure, filament number decreased significantly in WT, while sine1-1 decreased less and was similar to WT, and sine2-1 showed no significant change.

Mean angular difference was used to quantify F-actin orientation (see section “Materials and Methods”, Higaki et al., 2017; Biel et al., 2020b). Briefly, Z-stacks of individual guard cells were vertically arrayed, isolated, and the cell medial axis was defined. The mean angular difference was measured between actin pixel pairs and the nearest segments of the cell medial axis in the processed images (representative images shown in Supplementary Figure 2; Higaki et al., 2010). An angle closer to 90° represents more transverse filaments, while an angle closer to 0° indicates more longitudinal filaments. In WT, the mean angular difference decreased significantly from 35° at 0 min to 7° at 120 min, consistent with the loss of a radial filament array as reported previously (Higaki et al., 2010; Figure 1F). In sine1-1 and sine2-1, mean angular difference was already significantly lower than WT at 0 min (21^o and 28^o, respectively) and decreased significantly less than WT at 120 min (12^o and 14^o, respectively). This reflects the less pronounced radial arrays at the beginning of the assay and the less pronounced re-organization in both mutants. Thus, consistent with actin type and filament number, this parameter—which reflects changes in actin filament orientation—indicates less change over time in sine1-1 and sine2-1.

Occupancy (displayed in %) measures the average signal distribution within a guard cell and reflects the amount of actin present in filamentous form (Higaki et al., 2010). In WT, occupancy decreased at 60 min, then increased at 120 min, consistent with the visually scored actin reorganization and the literature (Higaki et al., 2010). Consistent with the visual observation, occupancy values for sine1-1 and sine2-1 change little over the 120 min time course (Figure 1G).

Together, these results show that loss of SINE1 and SINE2 results in a compromised actin network in open guard cells and overall deficiencies in actin reorganization in response to ABA-induced stomatal closure. Both SINE1 and SINE2 are required for a dense radially arrayed F-actin organization in open guard cells. During guard cell closure, the two mutants affect actin reorganization dynamics at different stages, with sine2-1 apparently preventing an early depolymerization step (0 to 60 min), while sine1-1 mutants are arrested in the reorganization of filaments post-depolymerization (60–120 min).

Jasplakinolide (JK) stabilizes and polymerizes F-actin, inhibiting stomatal closure in WT plants (MacRobbie and Kurup, 2007; Li et al., 2014). We have shown previously that the addition of JK, with or without ABA, induces stomatal closure in sine1-1, while sine2-1 mutants remain open, like WT (Biel et al., 2020a). Therefore, we tested here the effect of JK on the actin patterns in WT and sine1-1 with and without ABA. Because JK had no effect on stomatal closing in sine2-1, this line was not tested here. First, we confirmed that the previously reported WT and sine1-1 closing phenotypes based on JK were not altered by the actin marker. As shown in WT and sine1-1 without the marker (Biel et al., 2020a), 10 μM JK prevented ABA-based stomatal movement in WT GFP-LIFEACT stomata, while 10 μM JK induced partial closure in sine1-1 GFP-LIFEACT stomata, with or without concurrent ABA exposure (Figure 2A). This confirms that GFP-LIFEACT does not disturb the JK response in WT or the sine mutants.

Next, cortical F-actin patterns were imaged at 0 and 120 min after JK or JK + ABA exposure. At 0 min, the actin filaments are radially arrayed in WT and are in a mixed orientation in sine1-1, consistent with Figure 1B (Figure 2B). After 120 min of JK or JK + ABA exposure, F-actin in WT remains radially oriented. In contrast to the loss of F-actin seen after 60 and 120 min of ABA (Figure 1B), sine1-1 after 120 min of JK or JK + ABA exposure adopts a mixed F-actin pattern, with both longitudinal and radial filaments seen, with some fragmented actin also seen with JK + ABA (Figure 2B). Categorizing actin patterning, as described for Figure 1C, supports this observation (Figure 2C). This suggests that JK prevents the reorganization of actin filaments induced by ABA in WT, likely leading to the suppression of guard cell closure. In sine1-1 however, JK counteracts the loss of filaments after ABA treatment and inability to re-create an F-actin network. Under these conditions, sine1-1 stomata can close.

Filament number in WT guard cells decreased slightly with exposure to JK and JK + ABA, but this effect was much reduced compared to ABA treatment (Figures 1E, 2D). Filament number also decreased slightly in sine1-1 with exposure to JK and JK + ABA (Figure 2D). The mean angular difference is higher in WT than in sine1-1 at time point 0, and this difference is maintained with JK and JK + ABA treatment, consistent with the higher proportion of type 1 actin in WT stomata (Figures 2C,E). Occupancy at time point 0 was similar between WT and sine1-1 and decreased slightly by both treatments (Figure 2F). Together, these data indicate that the stabilization of actin filaments by JK overrides the closing defect caused by sine1-1, suggesting that the prolonged depolymerized state in the sine1-1 mutant has an inhibitory effect.

It has been established that ABA signaling results in cytoplasmic Ca²⁺ increases to induce stomatal closure, and that Ca²⁺ alone can stimulate closure (Allen et al., 1999; Hamilton et al., 2000; Kim et al., 2010; An et al., 2016). We have previously shown that exogenous Ca²⁺ partially rescues the sine1-1 and sine2-1 defect in ABA-induced stomatal closure, suggesting that SINE1 and SINE2 act downstream of ABA, but upstream of a Ca²⁺-dependent step in the pathway (Biel et al., 2020a). Given that both SINE1 and SINE2 are required for WT-like reorganization of the actin cytoskeleton during stomatal closure, we tested if exogenous Ca²⁺ also partially rescues this defect.

Exogenous Ca²⁺ (2 mM) was sufficient to induce stomatal closure in WT GFP-LIFEACT, sine1-1 GFP-LIFEACT and sine2-1 GFP-LIFEACT, again indicating that the actin marker does not interfere with Ca²⁺-induced stomatal dynamics (Figure 3A). As seen in the maximal intensity projections in Figure 3B and the visual scoring of actin patterns in Figure 3C, exogenous Ca²⁺ drastically reduced the amount of radial F-actin in WT guard cells (similar to ABA treatment, compare to Figures 1B,D). After 120 min, the signature longitudinal actin bundle pattern appeared in about half of the scored guard cells, suggesting that Ca²⁺-induced actin reorganization was less complete, or slower, than that induced by ABA. Exogenous Ca²⁺ increased the formation of longitudinal filaments in both sine1-1 and sine2-1, which was not seen with ABA. Curiously, sine1-1 displayed a substantial amount of radial filaments at 60 min after Ca²⁺ addition whereas ABA exposure resulted in a largely diffuse actin pattern (Figure 3C, middle panel vs. Figure 1D, middle panel). While the significance of this stage is not known, 2 mM Ca²⁺ clearly did not cause the accumulation of Type 2a actin observed in sine1-1 after treatment with ABA. Similarly, Ca²⁺ prevented for sine2-1 the “arrest” of filament reorganization at a disorganized stage (Type 2b) seen after ABA treatment.

Applying the same quantitative parameters as above, guard cell filament number declined more in the sine mutants after exposure to 2 mM Ca²⁺ than after ABA exposure (Figures 3D vs. 1E). No significant change in mean angular difference was revealed after Ca²⁺ exposure in sine1-1, consistent with the overall mixed patterns of actin filaments observed (Figures 3C,E). Mean angular difference in sine2-1 decreased further at 60 min and 120 min following Ca²⁺ treatment, consistent with the overall increase of the type 3 filament pattern (Figures 3C,E). Together, exogenous Ca²⁺ treatment resulted in a similar, if somewhat reduced (or possibly delayed) reorganization of actin in WT, compared to ABA treatment (Figures 3C–F). Both sine1-1 and sine2-1 went through a larger degree of actin reorganization than after ABA treatment, resulting in a larger number of guard cells with type 3 pattern at the end of the assay, consistent with the partial induction of stomatal closing in the mutants.

ABA-induced stomatal closure results in oscillatory changes in cytoplasmic Ca²⁺ (McAnish et al., 1990; Allen et al., 1999, 2000; Staxen et al., 1999; Schroeder et al., 2001). Given that our previous data suggest that SINE1 and SINE2 act upstream of a Ca²⁺-dependent step in the ABA pathway, we tested if ABA-induced cytoplasmic Ca²⁺ oscillations are disrupted in sine mutants. Using the genetically encoded ratiometric Ca²⁺ sensor YC3.6 tagged with a nuclear export signal (NES-YC3.6), we observed ABA-induced cytoplasmic Ca²⁺ oscillations in WT, sine1-1, and sine2-1 (Figures 4A,B). Counting and averaging the number of Ca²⁺ peaks for 45 min post-ABA addition indicated no significant difference between WT and sine1-1, but a significant reduction in peak number for sine2-1 (Figures 4B,C). The distribution of peak number for WT and sine mutants showed an absence of cytoplasmic Ca²⁺ peaks after ABA addition in approximately 25% of sine2-1 guard cells, but not in either WT or sine1-1 (Figure 4D). We next tested whether addition of exogenous Ca²⁺ could override the dampening of the cytoplasmic Ca²⁺ response to ABA in sine2-1 (Figures 4E–G). Both 1 mM and 10 mM external CaCl₂ have been shown previously to induce comparable cytoplasmic Ca²⁺ oscillations (Allen et al., 2000; Jeon et al., 2019). Here, inducing stomatal closure with 10 mM CaCl₂ resulted in an average of 6 Ca²⁺ peaks in 45 min in sine2-1, similar to WT and sine1-1 (Figure 4F). The portion of sine2-1 guard cells with no observable Ca²⁺ oscillations was reduced to 10% (Figure 4G). This suggests that, consistent with the closing phenotype, exogenous Ca²⁺, but not ABA, can trigger cytoplasmic Ca²⁺ fluctuations in sine2-1.

We have previously shown that addition of the actin depolymerizing drug Latrunculin B (LatB), in conjunction with ABA, is able to induce closure in sine mutants (Biel et al., 2020a). Thus, we tested if disrupting actin organization could also induce WT-like cytoplasmic Ca²⁺ oscillations in sine2-1 (Figures 4H–J). As a control, GFP-LIFEACT plants were imaged before and after 10 μM LatB treatment in both whole leaves and epidermal peels (Supplementary Figure 3A). Before addition of LatB, actin filaments (F-actin) are seen in both guard cells and the surrounding pavement cells in both whole leaves and epidermal peels. After 60 min LatB treatment in whole leaves, the F-actin network appears more diffuse and by 120 min appears completely disrupted (Supplementary Figure 3A). In epidermal peels, the F-actin network begins to noticeably disintegrate 10 min after LatB treatment (Supplementary Figure 3B). Addition of LatB and ABA induced, on average, 5 peaks per 45 min for sine2-1, similar to WT (Figure 4I). No traces without any Ca²⁺ oscillations were observed (Figure 4J). Taken together, these data demonstrate that ABA-induced cytoplasmic Ca²⁺ fluctuations are dampened in sine2-1 and that both the addition of exogenous Ca²⁺ and the addition of LatB can restore them to near-WT levels. Thus, induced actin depolymerization can override the cytoplasmic Ca²⁺ oscillation defect, consistent with the depolymerization defect of sine2-1 shown in Figure 1.

Nuclear Ca²⁺ fluctuations have been reported during biotic and abiotic stress as well as root development in a variety of plant species (Capoen et al., 2011; Liang et al., 2014; Charpentier et al., 2016; Charpentier, 2018; Leitão et al., 2019). Here, we tested whether guard cells also undergo nuclear Ca²⁺ fluctuations by using the genetically encoded Ca²⁺ sensor R-GECO1 tagged with a nuclear localization signal (NLS-R-GECO1). When the samples were treated with opening buffer for 5 min to establish a baseline prior to ABA addition, nuclear Ca²⁺ oscillations were observed (Supplementary Figure 1). To determine if this was due to the 20 μM CaCl₂ in the opening buffer, we compared guard cells exposed for 45 min to the full opening buffer (10 mM MES, 1% sucrose, 50 mM KCl, 20 μM CaCl₂) or a modified opening buffer lacking Ca²⁺ (10 mM MES, 1% sucrose, 50 mM KCl; Supplementary Figures 1A,B). Nuclear Ca²⁺ fluctuations were observed in full opening buffer but were absent in modified opening buffer, suggesting that 20 μM exogenous CaCl₂ is sufficient to trigger nuclear Ca²⁺ fluctuations in guard cells. We also tested whether sine mutants exposed to full opening buffer exhibit nuclear Ca²⁺ fluctuations (Supplementary Figures 1C,D). Using peak number and peak categorization to compare we observed similar average peak numbers and trends between WT and sine mutants (Supplementary Figures 1C,D). Therefore, to establish a baseline needed for ABA-induced Ca²⁺ experiments, we used the modified opening buffer for 5 min prior to ABA addition.

Changes in nuclear Ca²⁺ fluctuations were observed after addition of ABA but appeared irregular and variable compared to the oscillatory nature of cytoplasmic Ca²⁺ (Figures 5A,B). Compared to WT, the average number of peaks in 45 min post ABA addition was significantly reduced in sine2-1, but not sine1-1, mirroring the results for the cytoplasmic Ca²⁺ oscillations (Figure 5C). This was also reflected in the distribution of peaks per trace (Figure 5D). Given the irregular nature of the nuclear Ca²⁺ fluctuations, we also calculated the area under the curve for WT and sine mutants, with qualitatively similar results (see section “Materials and Methods”; Figure 5E). Taken together, these data show evidence for SINE2-influenced nuclear Ca²⁺ fluctuations in guard cells during ABA-induced stomatal closure.

Stomata open and close with concomitant changes in vacuolar morphology and actin reorganization (Fricker and White, 1990; Higaki et al., 2007; MacRobbie and Kurup, 2007; Li et al., 2013; Zheng et al., 2014, 2019). Upon ABA-induced stomatal closure, fused vacuoles of open guard cells undergo fission or constriction (hereafter called fragmentation; Zheng et al., 2014; Song et al., 2018). We tested if SINE1 and SINE2 are involved in vacuolar fragmentation during ABA-induced stomatal closure and if this process depends on actin reorganization. Acridine orange (AO; Fricker and White, 1990; Gao et al., 2005; Higaki et al., 2007; MacRobbie and Kurup, 2007; Li et al., 2013; Zheng et al., 2014) was used to monitor changes in vacuolar morphology during stomatal closure. We defined vacuolar fragmentation by the number of separated AO-containing structures. At 0 min of ABA treatment WT, sine1-1, and sine2-1 exhibited large, fused vacuoles (Figure 6A). After 90 min of ABA exposure, vacuolar fragmentation was seen in WT (Figures 6A–C). Addition of LatB + ABA resulted in similar vacuolar fragmentation as ABA treatment alone (Figures 6A,D,E). Conversely, vacuolar fragmentation after ABA treatment was significantly reduced in sine1-1 and sine2-1 (Figures 6A–C). Addition of LatB to ABA treatment increased the number of fragmented vacuoles in sine1-1 and sine2-1 (Figures 6A,D,E), more so than ABA treatment alone (Supplementary Table 2, first row).

These data suggest that ABA-induced vacuolar fragmentation is compromised in sine1-1 and sine2-1. Depolymerizing F-actin partially overrides this defect, consistent with the rescue of ABA-induced stomatal closure by LatB (Biel et al., 2020a). This is consistent with the assumption that an aspect of the dynamic actin reorganization is required for vacuolar fragmentation, that this is disrupted in the mutants, and that the inhibitory effect caused by this disruption is released by F-actin depolymerization.

During stomatal closure and vacuolar fragmentation, there is a rise in the cytoplasmic Ca²⁺ concentration via Ca²⁺ influx and Ca²⁺ release from intracellular stores (Ward and Schroeder, 1994; MacRobbie, 2000; Ng et al., 2001; Xiao et al., 2004; MacRobbie and Kurup, 2007; Song et al., 2018). Therefore, we tested whether exogenous Ca²⁺ induces vacuolar fragmentation and if any response is compromised by the loss of SINE1 or SINE2. Stomatal closure assays were performed by applying 10 mM Ca²⁺ and monitoring vacuolar morphology using AO (Figure 7). As shown in Figure 6, WT, sine1-1, and sine2-1 pre-opened stomata at 0 min had large, fused vacuoles (Figure 7A). After 90 min of exposure to 10 mM Ca²⁺, vacuoles of WT, sine1-1, and sine2-1 guard cells were still predominantly large and fused, with little change (Figures 7A–C). Addition of LatB caused no change, WT, sine1-1, and sine2-1 guard cells still exhibited large, fused vacuoles (Figures 7A,D,E). Stomatal closure was normal for both 10 mM Ca²⁺ and 10 mM Ca²⁺ + LatB treated samples (Supplementary Figure 4; Supplementary Table 2). It was noted that exogenous Ca²⁺ in a lower concentration (2 mM) led to a small degree of vacuolar fragmentation in WT and sine2-1, significantly less so than ABA but significantly more than 10 mM Ca²⁺ (Supplementary Table 2; Supplementary Figure 5). Taken together, stomatal closure can occur in the absence of vacuolar fragmentation when induced by exogenous Ca²⁺ and, under these conditions, actin depolymerization does not impact vacuolar fragmentation.