Drought-dependent regulation of cell coupling in Arabidopsis leaf epidermis requires plasmodesmal protein NHL12 Authors: Anam Ayyoub, Xiuyuan Yu, Xingjian Zhang, Chen Gao, Jiazhou Li, Shijiao Yin, Shaolin Chen, Johannes Liesche Published: 2024 ABSTRACT AbstractThe cytoplasm of most plant cells is connected by membrane-lined cell wall channels, the plasmodesmata (PD). Dynamic regulation of sugar, hormone, and protein diffusion through PD is essential for plant development and stress responses. Understanding this regulation requires knowledge of factors and mechanisms that control PD permeability through the modulation of callose levels in the cell wall around PD openings. We investigated PD regulation in leaf epidermal cells in relation to drought stress in Arabidopsis. PD-mediated cell wall permeability was decreased by drought stress and the hormone abscisic acid (ABA), and we tested how this related to several PD-associated genes with drought-responsive expression. Mutants of NON-RACE SPECIFIC DISEASE RESISTANCE/HIN1 HAIRPIN-INDUCED-LIKE 12 (NHL12) showed relatively low PD permeability that was unaffected by drought or ABA treatment. Overexpression of NHL12 in Nicotiana benthamiana epidermal cells increased PD permeability. Moreover, we showed that NHL12 can potentially interact with the callose synthase regulator NHL3 and we explored the effect of NHL12 abundance and/or lower interface permeability on ABA signaling genes. Our results indicate that NHL12 is a drought-responsive negative regulator of PD callose levels and, thereby, interface permeability. Results are discussed in relation to PD function during drought stress and the regulation of intercellular transport. Introduction Plasmodesmata (PD) are cell wall channels that provide a degree of cytosolic continuity between most plant cells. PD structure and abundance influence which molecules can move intercellularly and at what rate. The PD-mediated transport capacity is critical for processes such as organ development (Otero et al., 2016; Godel-Jędrychowska and Kurczyńska, 2021), pathogen defense (Cheval and Faulkner, 2018; Li et al., 2021), environmental acclimation (Gao et al., 2020), or carbon allocation (Comtet et al., 2017; Liesche et al., 2019). Accordingly, manipulation of PD has been proposed as a strategy to optimize stress resistance (Liu et al., 2021) and crop productivity (Amari et al., 2021). The main pathway for molecular movement through PD is the cytoplasmic sleeve, the space between the plasma membrane and the central endoplasmic reticulum tube called the desmotubule. The thickness of the cytoplasmic sleeve poses a size-exclusion limit that typically prevents medium-sized and large proteins from passing through PD (Wu et al., 2018). It also affects the rate at which molecules of permitted size pass through PD (Rutschow et al., 2011; Deinum et al., 2019). Cell differentiation programs or specific cues can cause the cytoplasmic sleeve to widen, narrow or even close completely (Benitez-Alfonso et al., 2013). One type of cue that influences PD permeability is abiotic stress. Heavy metal stress led to callose-mediated PD closure in maize, wheat, duckweed, and soybean (Sivaguru et al., 2000; Piršelová et al., 2012; Samardakiewicz et al., 2012). Chilling stress reduced interface permeability in maize leaves (Bilska and Sowiński, 2010; Fromm et al., 2013). Heat stress affects callose deposition at PD in tobacco (Iglesias and Meins, 2000; Rinne et al., 2005), and salt stress decreased interface permeability in Arabidopsis cotyledons (Hunter et al., 2019). In all these reports callose levels at PD were identified as the decisive factor for PD permeability regulation. There is still relatively little knowledge on the regulation of the enzymes responsible for callose synthesis and breakdown, callose synthases and β-1,3-glucanases (Hunter et al., 2019; Iswanto et al., 2020). An interesting approach was recently taken by Kirk et al. (2022), who analysed transcriptomics data of PD-specific proteins in several plant species to identify potential regulatory proteins for callose-dependent PD permeability regulation. The transcript levels of these factors showed associations with salinity and drought stress (Kirk et al., 2022). Biotic stress signals were shown to be integrated by a complex of NON-RACE SPECIFIC DISEASE RESISTANCE/HIN1 HAIRPIN-INDUCED-LIKE protein 3 (NHL3) and PLASMODESMATA-LOCATED PROTEIN 5 (PDLP5) that activates CALLOSE SYNTHASE 1 (CALS1) (Tee et al., 2023). NHL proteins were originally identified as mediators of the hypersensitive response triggered by the hairpin proteins of invading bacteria (Gopalan et al., 1996; Century et al., 1997). Based on transcript-level analysis in Arabidopsis, it was hypothesized that several NHL proteins also play a role in abiotic stress responses (Bao et al., 2016b). In the present study, we investigated the effect of drought stress on PD function in Arabidopsis leaves and looked for proteins that were involved in the response. Upon finding PD-mediated interface permeability and callose levels to be influenced by NHL12 abundance, we explored its potential mechanism of action. Materials and methods Plant material and growth conditions Seeds of Arabidopsis plants were sterilized in 75% alcohol for 2 min and 10% (v/v) sodium hypochlorite for 10 min, washed five times with sterilized water and imbibed for 3 d in the dark at 4 °C. Half-strength Murashige and Skoog (1/2 MS) solid medium was prepared with 1% (w/v) sucrose and 1% (w/v) agar at pH 5.8. Seeds were then grown on 1/2 MS medium plates for about 5 d (22 °C, 16 h light–8 h dark) until germinated and then transferred to a growth chamber on moist soil in pots kept at 16 h light–8 h dark, 90 μmol photons m−2 s−1, with temperature at 22 °C during the day and 20 °C at night. Plants were watered daily, except in the drought treatment group, for which irrigation was stopped completely at 10 d after germination. The relative water content of soil (RWC) was calculated by: Weights were measured for plants in their pots. Dry weight was determined by weighing after oven-drying the pot for 48 h at 60 °C. For the abscisic acid (ABA) treatment, 2 µl ABA solution (100 µM) was dripped on the leaf tip and plants kept in the dark for 1 h to avoid hormone breakdown. This treatment was repeated for 7 d before plants were used in the different experiments. Arabidopsis ecotype Col-0 was used as wild type. The seeds of the nhl12, at4g16380, and at4g27450 mutants (in Col-0 background) were obtained from the Nottingham Arabidopsis Stock Centre (NASC) (nhl12-1: N867643, nhl12-2: N606288, nhl12-3: N541762, at4g16380: N656837, at4g27450: N452389). T-DNA insertion sites of the mutants are illustrated in Supplementary Fig. S1. The insertion sites of the nhl12 mutants were in the 3ʹ-UTR, while the T-DNA insertions for the other genes was in the exons (Supplementary Fig. S1). The mutant genotype was confirmed by PCR and mRNA levels measured by qPCR with primers specified in Supplementary Table S1. Mutants nhl12-1 and nhl12-3 had detectable levels of mRNA of below 10% of wild-type plants, while the other mutants did not (Supplementary Fig. S1). Measurement of plasmodesmata-mediated cell wall permeability To measure PD-mediated cell wall permeability, we adapted the photoactivation method described in detail by Gao et al. (2020). Three-week-old plants were used for photoactivation experiments. Cage500 (Abberior GmbH, Göttingen, Germany) was used as photoactivatable tracer. To load the tracer, about 0.5 mm of leaf was sliced off with a razor blade and the leaf immediately submerged in 50 µM tracer solution. After 1 h incubation at room temperature, the leaf was rinsed with phosphate buffered saline (PBS) (1×: 130 mM NaCl, 2.6 mM KCl, 7 mM Na2HPO4, 1.2 mM KH2PO4, pH 7.2), and the whole plant was mounted on a slide. Fluorescence was monitored before, during, and after photoactivation on a Leica SP8 confocal microscope (Leica Microsystems, Wetzlar, Germany). The 405 nm laser was set to full power for photoactivation, and laser excitation was limited to the target cell using the free-form region of interest (ROI)-scanning function. For imaging activated fluorescein, the 488 nm line of the argon laser was used with 25% preset laser power and 15% gain of the acousto-optical tunable filter. Using the spectrometric filter, the detection window for fluorescence emission was set from 495 to 550 nm. A ×40 water-immersion lens (NA 1.2, Leica Microsystems) was used in all experiments. The gain of the detector (photomultiplier tube) was set below saturation level after photoactivation, at about 850 V. Image analysis was carried out in ImageJ (Schindelin et al., 2012), where ROIs were drawn for the neighboring cell and wall-adjacent regions of target and neighboring cells. Total fluorescence (integrated density) was determined for the neighboring cell ROI and used to calculate the tracer flux F by dividing by the photoactivation time. From the wall-adjacent ROIs, mean fluorescence was determined to calculate the concentration potential ∆c. Interface permeability was calculated as P=F/∆c. The theoretical background and control experiments for the method are described in Gao et al. (2020). Aniline blue staining in Arabidopsis For staining and quantifying callose in the leaf abaxial epidermis, we adapted the method of Zavaliev and Epel (2015). The entire leaf was immersed in 96% ethanol for bleaching and rehydrated in double distilled water with 0.01% Tween-20 for 1 h. A 5 mm by 2 mm strip was cut from each leaf. The samples were immersed in staining solution for 10 min under vacuum in a centrifuge concentrator (Concentrator Plus, Eppendorf, Germany), followed by 1 h shaking (40 rpm) at room temperature. The staining solution was a mixture of 0.1% (w/v) aniline blue and 0.01 M K3PO4 with pH 10.5. After staining, samples were analysed using a Leica SP8 confocal laser-scanning microscope (Leica Microsystems) with 405 nm excitation and 475–525 nm as emission filter setting. The pinhole was set to 2 Airy Units and observations made with a ×40 objective lens. Image analysis was carried out in ImageJ (Schindelin et al., 2012). To quantify the PD-specific aniline blue intensity, round ROIs with 1.8 µm diameter were overlaid on aniline blue spots and mean fluorescence determined. Transmission electron microscopy Squares of about 5 mm width were cut from leaves and immediately fixed in 2.5% glutaraldehyde 0.1 M PBS (pH 7.2) for at least 12 h at 4 °C. Samples were washed four times in 0.1 M PBS (pH 7.2) for 15 min before incubation in 1% osmic acid for 4 h at 4 °C. Samples were then washed in phosphate buffer and dehydrated in an ethanol and acetone series. Samples were embedded in LR White resin and afterwards sectioned on an ultramicrotome (EM UC7, Leica Microsystems). Sections were consecutively stained in 2% uranyl acetate for 20 mins and lead citrate for 15 min. Next, sample preparations were observed and imaged using transmission electron microscopy (TEM; HT7800, HITACHI, Tokyo, Japan). PD neck diameter and density were measured with ImageJ. PD density D was determined from electron micrographs as described by Liesche et al. (2019) with the formula D=count per µm wall length/(T+1.5r) with section thickness T=70 nm and estimated average PD radius r=20 nm. Any significant portion of a PD visible in a micrograph (typically at least one-quarter of the PD) was included in the analysis. Heterologous expression and fluorescence imaging The NHL12 gene was cloned from Arabidopsis cDNA, and restriction enzyme sites were added using the adequate primers (Supplementary Table S1). NHL12 was inserted into the pBI121-mCherry vector’s multiple cloning site. The vector was transformed into Agrobacterium tumefaciens strain GV3101 (Tolobio, Shanghai, China) following the supplier’s instructions. The method for infecting Nicotiana benthamiana plants with Agrobacterium was described previously (Xu et al., 2020). To show the localization of both callose and NHL12–mCherry, about 1 cm wide square samples were cut from leaves and immersed in aniline blue staining solution (see above) for 10 min under vacuum in a Concentrator Plus (Eppendorf), followed by 10 min shaking (40 rpm) at room temperature. Samples were observed immediately after staining. Nicotiana benthamiana abaxial leaf epidermal cells were observed using a Leica SP8 laser-scanning confocal microscope (Leica Microsystems) with a ×20 objective lens. mCherry fluorescence was recorded with 581 nm excitation and 595–625 nm emission detection window. For FRET-FLIM, the genes of interest and an mT2 or yellow fluorescent protein (YFP) were inserted in the vector pGWB441 at its multiple cloning site. Abaxial leaf epidermal cells of N. benthamiana transiently expressing the constructs were imaged using a laser scanning confocal system (Stellaris 8; Leica Microsystems) with ×20/0.75 NA dry objective. The image format was set to 512 × 512 and a scan speed of 400 Hz resulted in a pixel dwell time of 3.16 μs. For the excitation of mT2 and YFP, the laser lines at 448 nm and 514 nm, respectively, were used. The photon arrival times were recorded with two HyD detectors, covering the mT2 emission at 467–520 nm and the YFP emission at 521–601 nm. For the lifetime image analysis, fluorescence intensity images were used to manually select the regions where both proteins were present. Photon arrival times in these regions in the mT2 channel were fitted using a n-exponential reconvolution fit in Leica LAS X software (Leica Microsystems) and average lifetimes (in nanoseconds) per image recorded. Quantification of plant growth Photographs of plants were taken daily, starting from 10 d after germination, for 14 d. Photos were taken shortly before the end of the light period. In ImageJ, after setting the scale with the help of a ruler included in the photographs, a circle was drawn to include the tips of all leaves of one plant. This circle’s diameter was used as principal indicator of plant growth. The experiment was repeated three times. Measurement of mRNA levels RNA isolation with Trizol (Thermo Fisher Scientific, Waltham, MA, USA) was conducted as described by Xu et al. (2018). Synthesizing the first-strand cDNA and removing gDNA were conducted with TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix kit (Transgen, Beijing, China). The mRNA level was measured on a CFX Connect Real Time System (Bio-Rad, Feldkirchen, Germany) using the PerfectStart Green qPCR SuperMix kit (Transgen). Each group of data was derived from three independent biological samples and consisted of three technical replicates. Relative expression levels were normalized to reference genes Actin12 and UBQ10. The primers used in PCR are listed in Supplementary Table S1. Statistics The statistical analysis related to calculating PD permeability was performed using Microsoft Excel software (Microsoft, Redmond, WA, USA) or Python 3.11 (Python Software Foundation, Wilmington, DE, USA, 2022). Number of replicates is specified on figures and in the experiment descriptions. Student’s two-sided t-test was used for pairwise comparisons between control conditions and treatment conditions or wild type and mutants. Differences were considered significant when P<0.05. Results Drought stressed plants show reduced interface permeability We assessed PD function by measuring cell wall permeability mediated by all PD in a cell–cell interface, as single PD are too small to be observed under a conventional light microscope. Photoactivation microscopy, in which fluorescence of a tracer is activated in a single cell and the spread to neighboring cells monitored, has been established as a reliable way to measure PD-mediated interface permeability (Liesche and Schulz, 2012; Gao et al., 2020). To assess the influence of drought stress on interface permeability, we subjected Arabidopsis plants that were not watered for 11 d to photoactivation microscopy experiments. The fluorescence of the small tracer molecule Cage500 was activated by laser illumination in one cell of the abaxial leaf epidermis and tracer flux to neighboring cells and concentrations quantified (Fig. 1). For comparison, well-watered plants and well-watered plants sprayed with the hormone ABA were included. Tracer flux in water-deficient plants was significantly reduced (Fig. 1B), which translated to a reduction of interface permeability by about 25% (Fig. 1C). Well-watered plants treated with ABA showed similar decreases in flux and interface permeability as water-deficient plants (Fig. 1B, C). Interface permeability in leaf epidermal cells under control conditions, drought-stress conditions, and after abscisic acid (ABA) application in 3-week-old wild-type Arabidopsis plants. (A) Bright field images and images showing fluorescent tracer distributions before and after activation as well as drawings indicating the target cell area (black) and two neighboring cells (blue and green). During activation, tracer becomes fluorescent in the target cell (T) and diffuses to neighboring cells (N). St: Stomata. Scale bars 50 µm. (B) Average flux over target–neighbor interfaces indicated in (A). Flux was measured as the neighbor cell fluorescence concentration divided by interface length and activation time. (C) Plasmodesmata-mediated interface permeability, calculated as tracer flux divided by the tracer concentration potential across the interface. Asterisk indicates significant difference from control according to Student’s t-test (P<0.05). Boxes represent quartiles, × represents mean, and bars indicate 95% confidence intervals. Photoactivation experiments were carried out on at least three leaves. n=40 (control), 25 (drought), 31 (ABA). Drought stress changes plasmodesmal structure Since PD transport rates typically depend on PD structure, especially the neck diameter, we investigated whether this was affected by drought stress. Fluorescence imaging after aniline blue staining indicated callose levels at PD of leaf abaxial epidermal cells. Aniline blue fluorescence levels were increased in the drought-stressed plants compared with the well-watered control plants (Fig. 2A, B). In addition, we fixed and cross-sectioned leaves for PD structure analysis by TEM. TEM images of abaxial epidermal cells showed the PD neck diameter to be significantly reduced in drought-stressed plants (Fig. 2C, D), while PD density remained at a similar level to control plants (Fig. 2E). Plasmodesmal structure in leaf epidermal cells under control and drought-stress conditions in 3-week-old wild-type plants. (A) Images showing aniline blue-stained callose at plasmodesmata. (B) Quantification of aniline blue fluorescence intensity at plasmodesmata. (C) Transmission electron micrographs of plasmodesmata. (D) Quantification of plasmodesmal neck diameter based on TEM images. (E) Plasmodesmal density per wall length. Asterisk indicates significant difference from control according to Student’s t-test (P<0.05). Boxes represent quartiles, × represents mean, and bars indicate 95% confidence intervals. n>120 (B), 36 (D), 5 (E). Scale bars 50 µm (A), 100 nm (C). Transcript levels of several plasmodesmal-associated genes respond to drought stress To gain insight on the molecular basis of the drought-associated change in PD structure and function, we used publicly available leaf transcriptomics data of drought stress experiments to check transcript levels of 61 genes whose products have been identified as PD-specific in proteomics approaches (Fernandez-Calvino et al., 2011; Kraner et al., 2017; Leijon et al., 2018). Transcript levels of seven genes showed strong, consistent changes between drought- and control-conditions (Table 1). We chose these genes as candidates for a potential role in the modification of drought-related interface permeability. Table 1.Plasmodesma-associated genes with drought-responsive mRNA levelsDescriptionGene symbolLocus tagGSE106707 d −WGSE1064310 d −WGSE2417710 d −WGSE6541410 d −WGSE4006114 d −WGSE4676014 d −WMeanHydroxyproline-rich glycoprotein family proteinPELPK1AT5G095304.21.00.50.113.41.01.7Aluminium-induced proteinAT4G274501.32.70.50.60.90.61.0Plasmodesmata-located protein 3PDLP3AT2G33330−2.4−4.3−2.3−1.6−3.9−0.4−2.5Plasmodesmata-located protein 2PDLP2AT1G04520−3.4−5.2−1.6−1.7−2.9−0.3−2.5NDR1/HIN1-like 12NHL12AT2G35960−1.6−2.6−2.1−0.9−2.3−0.7−1.7GlycoproteinAT5G19240−4.2−1.3−0.4−1.3−1.3−0.1−1.4Heavy metal transport/detoxification superfamily proteinAT4G16380−1.4−1.8−0.7−0.8−1.0−0.8−1.0Soil-grown plants, between 2 and 4 weeks old, were deprived of water for a certain number of days (indicated below the GEO accession ID) before rosette leaves were harvested and subjected to transcriptomics analysis. All values denote log2 fold change. Plasmodesma-associated genes with drought-responsive mRNA levels Soil-grown plants, between 2 and 4 weeks old, were deprived of water for a certain number of days (indicated below the GEO accession ID) before rosette leaves were harvested and subjected to transcriptomics analysis. All values denote log2 fold change. Interface permeability in nhl12 mutant plants is low and does not respond to drought stress We used photoactivation microscopy on epidermal pavement cells to evaluate if T-DNA-insertion mutants of the genes listed in Table 1 affected interface permeability. Not all mutants for the candidate genes were available to us. We determined the PD-mediated cell wall permeability for nhl12-1, at4g16380, and at4g27450 mutants as well as wild-type plants. The mutants responded to drought in the same way as the wild type, except nhl12-1 (Fig. 3). Flux to neighboring cells and interface permeability were not significantly changed in drought-stressed nhl12-1 mutants compared with well-watered mutant plants (Fig. 3B, C). Similarly, ABA treatment had no significant effect (Fig. 3A–C). We also found interface permeability in nhl12-1 mutant plants, as well as the genetically different nhl12-2 and nhl12-3 mutants (Supplementary Fig. S1), to be significantly lower than wild type under control conditions with mean values of 5.36 µm s−1 in wild type (Fig. 1C) and 3.96 µm s−1 (nhl12-1), 3.22 µm s−1 (nhl12-2), and 3.27 µm s−1 (nhl12-3) in the mutants (Fig. 3C). Interface permeability in the epidermis of at4g16380 and at4g27450 mutants was similar to wild type (Fig. 3D). Moreover, it was reduced by drought conditions in a manner similar to wild-type plants (Figs 1C, 3D). Of the candidate genes that we were able to test here, our experiments indicated a potential role in the modulation of PD function only for NHL12. Therefore, our subsequent analysis focused on this gene. Interface permeability in leaf epidermal cells of 3-week-old mutant plants under control and drought-stress conditions. (A) Example images of nhl12-1 leaf epidermal cells showing fluorescent tracer distributions before and after activation. Bright field images are used for orientation. Regions of interest drawing indicate cell outlines of the target cell (T), in which fluorescence of the tracer was activated, and two neighboring cells (N). (B) Flux (measured here as fluorescence concentration in neighbor cell after activation divided by time of activation and interface length) into two neighbor cells indicated in (A). (C) Interface permeability in three nhl12 mutants. (D) Interface permeability in at4g16380 and at4g27450 mutants. Asterisks indicate significant difference from control conditions according to Student’s t-test (P<0.05). Boxes represent quartiles, × represents mean, and bars indicate 95% confidence intervals. Experiments were conducted on three plants for each line, with several photoactivation series conducted per leaf; n>11. ABA, abscisic acid. NHL12 overexpression increases plasmodesmal permeability To test if NHL12 protein can influence PD function, we used the transient expression system in Nicotiana benthamiana abaxial epidermal cells. The Arabidopsis NHL12 shares 76% nucleotide sequence similarity and 65% protein sequence similarity with its N. benthamiana homologue. Expression of the Arabidopsis NHL12 fused to the fluorescent protein mCherry under control of the constitutive 35S promoter enabled visualization of its intracellular location. It showed an inhomogeneous distribution at the cell periphery (Fig. 4A). The NHL12–mCherry fluorescence overlapped partly with aniline blue stained callose accumulations, indicating that NHL12 might be present at most, but not all PD, as well as the plasma membrane (Fig. 4B). Remarkably, the aniline blue fluorescence intensity was reduced in NHL12-mCherry-expressing cells compared with cells only expressing mCherry (Fig. 4C). This indicates lower callose levels at PD. This agrees with measurement of interface permeability, which showed higher interface permeability in cells expressing NHL12-mCherry than in cells expressing only mCherry (Fig. 4D, E). Plasmodesmata in Nicotiana benthamiana epidermal cells expressing 35S:NHL12-mCherry. (A) Fluorescence image of cells expressing 35S:NHL12-mCherry. (B) Magnification of 35S:NHL12-mCherry (red), callose-specific dye aniline blue (blue), and an overlay of the two channels. (C) Quantification of aniline blue fluorescence at aggregates. (D) Bright field images and images showing fluorescent tracer distributions before and after activation. Scale bars 100 µm. (E) Plasmodesmata-mediated cell wall permeability. Asterisk indicates significant difference from control according to Student’s t-test (P<0.05). Boxes represent quartiles, × represents mean, and bars indicate 95% confidence intervals; n=20. Plasmodesmal structure in nhl12 mutant leaves We also tested the levels of aniline blue-stained callose and PD structural parameters in the nhl12 mutants. Under control conditions, aniline blue fluorescence was increased in nhl12 mutants compared with wild-type plants (Figs 2, 5). The mean intensity of the mutant lines under control conditions (87, 86, and 79, respectively; Fig. 5B) was between the mean intensity of the wild type under control conditions (68) and under drought conditions (98; Fig. 2B). Importantly, water deficiency did not lead to a significant increase in aniline blue fluorescence in any of the mutant lines (Fig. 5A, B). Neither did the PD neck diameter of nhl12-1, as seen on electron micrographs of epidermal cells, show significant changes (Fig. 5C, D). Moreover, like in wild-type plants, PD density was not affected by water deficiency in the nhl12-1 mutant (Fig. 5E). The mean neck diameter of PD in the nhl12-1 mutant (Fig. 5D) was similar to the neck diameter of PD in drought-stressed wild-type plants (Fig. 2D) (t-test: P=0.3). Plasmodesmal structure in leaf epidermal cells of 3-week-old nhl12 mutant plants. (A) Images of nhl12-1 showing aniline blue-stained callose at plasmodesmata. (B) Quantification of aniline blue fluorescence intensity at plasmodesmata. We measured 70 aniline blue dots in at least 10 cells in three plants. (C) Transmission electron micrographs of plasmodesmata. (D) Quantification of plasmodesmal neck diameter based on TEM images. In box plots, boxes represent quartiles, × represents mean, line indicates median, and bars indicate 95% confidence intervals. Significant differences between control and drought conditions were evaluated with Student’s t-test (P<0.05). n≥40 (B), n≥36 (D), n=5 (E). Scale bars 50 µm (A), 100 nm (E). NHL12 potentially interacts with the callose synthesis regulator NHL3 To gain insight into how NHL12 might influence PD callose levels, we tested protein–protein interaction potentials with candidate proteins. Dynamic regulation of the callose synthase CALS1 was recently shown to depend on its interaction with PDLP5, with the interaction being influenced by NHL3 (Tee et al., 2023). Moreover, CALS1 could potentially interact with PDLP8 (Jones et al., 2014). We used Förster resonance energy transfer coupled with fluorescence lifetime imaging microscopy (FRET-FLIM) to evaluate the potential of NHL12 to interact with PDLP5, PDLP8, and NHL3. FRET-FLIM indicates a localization of donor and acceptor proteins within about 5 nm distance by a shortening of the donor lifetime due to energy transfer to the acceptor (Bücherl et al., 2014). Positive FRET-FLIM of plant proteins typically means that they physically interact and results are more reliable than FRET analysis, which is influenced by relative protein abundances (Harter et al., 2012). Moreover, the spatial resolution of FRET-FLIM allowed us to detect interactions specifically at the sites of protein co-localization. Expression of NHL3, PDLP5, or PDLP8 linked to the fluorescence donor mTourquoise2 (mT2) under control of the 35S promoter in N. benthamiana epidermal cells yielded similar mT2 lifetimes of about 3.8 ns for each construct (Fig. 6). The fluorescence intensity distribution pattern resembles the patterns previously published for these proteins (Wang et al., 2020; Johnston, 2021). NHL3–mT2 showed a relatively uniform distribution at the cell periphery, while PDLP5–mT2 showed a typical PD-localization pattern (Fig. 6A, B). PDLP8–mT2 was distributed along the cell periphery, but showed some accumulation (Fig. 6C). Co-expression of NHL3–mT2 with PDLP5 linked to the fluorescence acceptor YFP shortened the lifetime of mT2 to about 2.5 ns at the cellular locations where both proteins are present, indicating FRET (Fig. 6D, H). A similar result was obtained when PDLP5 was linked to mT2 and NHL3 to YFP (Fig. 6H). Co-expression of NHL12–YFP shortened the lifetime of NHL3–mT2, but not PDLP5–mT2 (Fig. 6E–H). Neither did NHL12–YFP change the lifetime of PDLP8–mT2 (Fig. 6G, H). Analysis of NHL12 protein–protein interaction potential with Förster resonance energy transfer–fluorescence lifetime imaging microscopy (FRET-FLIM). All fluorescent protein fusion constructs were transiently expressed under control of the 35S promoter in N. benthamiana epidermal cells. (A–C) Images of control samples showing expression of single constructs via mTourqoise2 (mT2) fluorescence intensity in grayscale and mT2 lifetime color-coded. Scale bars 20 µm. (D–G) Images of samples expressing protein pairs labeled with mT2 and yellow fluorescent protein (YFP) showing fluorescence intensity (grayscale) or mT2 lifetime (color-coded). Interaction is indicated by a shortening of mT2 lifetime (here a shift from green towards blue), the result of energy transfer from mT2 to YFP when both fluorophores are in close proximity. Scale bars 10 µm. (H) Quantification of mT2 lifetimes. For cells expressing two constructs, lifetimes were measured at the sites of co-localization and averaged per image. Short lifetimes indicate protein–protein interaction. Significance of differences of two-construct-samples to the corresponding single-construct-sample was tested with Student’s t-test (P<0.05) and indicated by asterisk. n>5. The FRET-FLIM analysis confirms the interaction potential of NHL3 and PDLP5. Moreover, it indicates that NHL12 can interact with NHL3, but not PDLP5 or PDLP8. Evidences of enhanced abscisic acid signaling in nhl12 mutant The results presented above indicate that PD in pavement epidermal cells of the nhl12 mutant resemble those in the drought-stressed wild-type plants. To explore how this affects plant physiology, we compared the growth of nhl12 and wild-type plants under control and water-deficiency conditions. Water deficiency was imposed by stopping irrigation after seedlings were grown for 10 d (Fig. 7). Without irrigation, soil water potential decreased to about 50% in the first 6 d and continued to fall below 10% in the following 6 d (Fig. 7B). Under control conditions, nhl12 mutant plants grew slower than wild type, with significantly smaller rosettes starting from day 5 (Fig. 7C; t-test between wild type and nhl12-1: P=0.002). Under water-deficit conditions, wild-type plants showed strongly reduced growth (Fig. 7A, C). Drought-exposed wild-type plants were significantly smaller than control plants starting from day 8 (t-test: P=0.002). Drought-exposed nhl12 plants generally grew as well as nhl12 mutants under control conditions (Fig. 7C). Moreover, on days 13 and 14, there was no difference between the drought-exposed wild-type and nhl12 mutant plants (Fig. 7C). Characteristics of nhl12 mutant plants. (A) Images showing wild-type and nhl12 plants grown under control and drought-stress conditions for 11 d (22 d after germination). (B) Relative soil water content during the drought stress experiments. (C) Daily quantification of rosette diameters. Lines connect mean values. Shading indicates 95% confidence intervals. Three experiments were conducted and data combined. n=9 (B), 17 (C). To check if this phenotype might be related to a constitutive drought stress response, we examined the mRNA levels of characteristic drought response genes under control conditions. In Arabidopsis, drought stress is signaled via ABA-dependent and ABA-independent pathways (Yoshida et al., 2014). Transcript level of DRE-BINDING PROTEIN 2A (DREB2A), a key factor in the ABA-independent pathway, was significantly lower in nhl12 mutant leaves (Fig. 8). Transcripts of key genes of the ABA-dependent signaling pathway, ABSCISIC ACID INSENSITIVE 1 (ABI1) and HOMEOBOX-7 (HB7), were much more abundant in the mutant than in wild-type leaves (Fig. 8). Transcript levels of the enzyme NINE-CIS-EPOXYCAROTENOID DIOXYGENASE 3 (NCED3), which participates in drought-dependent ABA synthesis, and the stress-responsive GLUTATHIONE PEROXIDASE 6 (GPX6), were similar in nhl12 mutant and wild type (Fig. 8). Transcript levels of genes related to the drought stress response measured under control conditions in 3-week-old nhl12 mutant plants relative to the transcript levels in wild-type plants. Boxes represent quartiles, line represents median, and bars indicate 95% confidence intervals. Asterisks indicate significant difference between nhl12-1 and wild type (P<0.05) by t-test. n=4. ABI1, ABSCISIC ACID INSENSITIVE 1; DREB2A, DRE-BINDING PROTEIN 2A; GPX6, GLUTATHIONE PEROXIDASE 6; HB7, HOMEOBOX-7; NCED, NINE-CIS-EPOXYCAROTENOID DIOXYGENASE 3. Discussion The permeability of PD has been previously shown to react to different environmental stresses (e.g. Fromm et al., 2013; Hunter et al., 2019). Our results show that this is also the case for drought stress, at least for the PD in the abaxial leaf epidermal pavement cells (Fig. 1). A clarification of whether PD in different cell layers and tissues behave similarly needs further callose quantification in fixed tissue sections and photoactivation experiments at deeper layers. What could be the purpose of reduced PD permeability under drought stress conditions? One answer could be the modulation of water flows. It is known that drought conditions change intercellular water distribution within plant tissues (Barzana et al., 2021), for which PD, besides aquaporins, are a major pathway (Anisimov, 2021). Another reason for lower interface permeability might be the resulting limitation of intercellular diffusion of signal molecules. The intercellular movement of reactive oxygen species, peptides, and ABA shapes the drought response in leaves (Takahashi et al., 2020). The PD permeability in the epidermis could influence the movement from vasculature to guard cells of ABA, which has to pass the pavement epidermal cells (Kuromori et al., 2022). Restricting ABA diffusion through PD might increase the efficiency of directional transport mediated by ABA transporters in the plasma membrane. The modulation of callose levels at PD is assumed to be the prevalent mode to regulate their permeability (Knox and Benitez-Alfonso, 2014). The amount of callose at the PD neck regions is a direct function of callose synthase and hydrolase activity (Levy et al., 2007; Han et al., 2014). Our results further illustrate the importance of callose levels in the regulation of PD permeability (Fig. 2). Moreover, they suggest that NHL12 has a function in controlling callose synthesis or hydrolysis, maybe as part of a feedback mechanism. Our results are compatible with NHL12 fine-tuning PD permeability within a certain range (Figs 3, 4). Drought stress, ABA application, or knock-out of NHL12 all increase callose to similar levels without complete blocking of PDs (Figs 2, 5). In contrast, heterologous NHL12 overexpression reduces callose to some degree (Fig. 4). NHL12 could directly or indirectly reduce callose synthase activity or enhance callose hydrolysis carried out by β-1,3-glucanases. Considering the extracellular domain is only about 20 amino acids long (Jumper et al., 2021), direct callose binding, as it was reported for the potential PD callose sensor PDCB1 (Simpson et al., 2009), appears unlikely. An indirect mechanism could involve interaction with PDLPs or CYS-RICH RECEPTOR-LIKE KINASE 2 (CRK2), which act as positive regulators of callose levels (Lee et al., 2011; Caillaud et al., 2014; Cui and Lee, 2016), mainly through activation of callose synthase (Hunter et al., 2019; Tee et al., 2023). A potential to interact has been demonstrated for NHL1 and PDLP8 (Jones et al., 2014) and for NHL3 and PDLP5 (Tee et al., 2023). PDLP8 and PDLP5 could interact with callose synthase CALS1 (Jones et al., 2014; Tee et al., 2023). Our results indicating that NHL12 can interact with NHL3, but not PDLP5 (Fig. 6), mean that NHL12 could block the formation of the active PDLP5–NHL3–callose synthase complex (Fig. 9). Callose levels would be fine-tuned according to the balance of NHL3 and NHL12. The reduction of NHL12 abundance under drought conditions would shift the balance towards increased complex formation and callose synthesis. Illustrations of potential functions of NHL3 and NHL12 in callose synthesis. (A) Complex formation of NHL3, PDLP5, and callose synthase CALS1 enhances callose formation at plasmodesmal necks (blue shading), thereby reducing permeability. (B) Binding of NHL12 to NHL3 might prevent activation of CALS1 and, thereby, callose formation and a reduction in permeability. In this way, the balance of NHL3 and NHL12 could finetune callose synthesis at PD. Protein structures are shown as predicted by AlphaFold (Jumper et al., 2021). NHL, NON-RACE SPECIFIC DISEASE RESISTANCE/HIN1 HAIRPIN-INDUCED-LIKE protein; PDLP, PLASMODESMATA-LOCATED PROTEIN; CALS1, CALLOSE SYNTHASE 1. The nhl12 mutant showed lower growth than wild-type plants (Fig. 7). This might be a consequence of reduced PD permeability. Investigation of additional cell–cell interfaces and various developmental stages is necessary to determine the impact of NHL12 deficiency on plant development. Besides the mechanical aspect, an exciting question is the potential direct involvement of NHL12 in hormone signaling. Many NHL proteins have been previously implicated in ABA signaling (Bao et al., 2016a, b). Transcript levels of the central ABA-dependent transcription factors ABI1 (Pinheiro and Chaves, 2011) and ATHB7 (Valdés et al., 2012) were increased in the nhl12 mutant (Fig. 8). Finding out if NHL12 interacts with ABA-related signaling factors and if narrower PD by themselves influence ABA signaling will be important for a better understanding of the plant’s drought response. Future studies should also investigate the potential role of NHL12 in other stress responses, especially biotic stress, in which other NHL family members have been implicated (Varet et al., 2002; Maldonado et al., 2014). Supplementary data The following supplementary data are available at JXB online. Fig. S1. Information on Arabidopsis mutants used in this study. Table S1. List of primers used in this study.