Arabidopsis thaliana was used as experimental material. Ethyl methanesulphonate (EMS)-mutagenesis using phot2-1 was carried out previously (Lightner and Caspar, 1998), and M₂ population was used for screening. Columbia-0 (Col-0) was used as control plant for bsk1 (SAIL_140_C04) and DNA extraction. The T-DNA insertion was confirmed by genomic PCR using primer sets shown in Supplementary Table S1 . Landsberg erecta (Ler) was used for a map-based cloning experiment. Plants were grown on soil under fluorescent lamps in the growth room. Photon flux densities, day length, temperature, and relative humidity was approximately 50 µmol m⁻² s⁻¹, 16-h, 20–24°C, and 40–60%, respectively.

Previously developed immunohistochemical technique using leaves (Ando and Kinoshita, 2018) was employed as screening tool wherein phosphorylation of the penultimate residue of PM H⁺-ATPase (Thr) in guard cells was visualized. Leaves were collected from each M₂ plants, then they were illuminated with red light (150 µmol m⁻² s⁻¹) and blue light (50 µmol m⁻² s⁻¹) simultaneously for 15 min. Illuminated leaves were fixed and attached to a microscope slide as described previously. Liquid handling including tissue permeabilization, blocking, antibody application, and washing the material steps was automated by Insitu Pro VSi (Intavis). Observed phenotype was confirmed using the remained leaves from the same M₂ plants if possible, and validated in M₃ plants.

For measurement of stomatal density and index, leaves were fixed and cleared according to the previous study (Kang et al., 2009). Six images per leaf were obtained, then the density and index were calculated on each image. Arithmetic mean of the six images was calculated as a representative value for the corresponding leaf. Data represent arithmetic means of the representative values obtained from at least three leaves with standard deviations.

F₂ plants were obtained by crossing EMS mutant line and its progenitor and those exhibiting mutant phenotype were selected and DNA was extracted from them in a bulk with the High pure PCR template preparation kit (Roche) according to the manufacturer’s instructions. The bulk population genomic DNA was subjected to whole-genome sequencing and analyzed by Mitsucal for mapping and identification of a mutation as described previously (Suzuki et al., 2018). Reanalysis of the mutation by sanger sequencing was performed using a primer set shown in Supplementary Table S1 .

Genomic fragment of BSK1 was amplified by nested PCR from Col-0 using primer sets shown in Supplementary Table S1 . The fragment including −1,650 bp to +3,742 bp of the start codon was fused to pCAMBIA1300 digested with EcoRI using In-Fusion HD Cloning kit (Clontech). Agrobacterium tumefaciens (GV3101) was transformed with the construct, then used for generation of the transgenic plants.

Statistical comparison of means was conducted using R software (R Core Team, 2023). Student’s t test or Dunnett’s test using multcomp package (Hothorn et al., 2008) was carried out for two independent means or multiple means with single control, respectively. Segregation ratio was analyzed by chi-square goodness of fit test. P < 0.05 was considered statistically significant.

Previously, we developed a novel immunohistochemical technique for visualizing PM H⁺-ATPase and the phosphorylation of its penultimate residue, Thr, in guard cells using whole leaves (Ando and Kinoshita, 2018). Conventional techniques, such as the isolation of guard cell protoplasts (Ueno et al., 2005), require isolation of a sufficient amount of epidermis before the experiments, limiting their application based on plant amount, size, or both. To assess the versatility of the new immunohistochemical technique, we applied it to genetic screening, where plants must be analyzed individually, and thus only single, and sometimes small, leaves are available for the experiment. Utilizing a liquid-handling robot, we successfully semi-automated the process of immunohistochemical visualization of guard-cell PM H⁺-ATPase to prepare many specimens simultaneously. We screened the M₂ population of EMS-mutagenized phot2-1 (phot2-EMS; about 8,700 plants) and Col-0 (Col-EMS; about 3,200 plants) based on the phosphorylation status of guard-cell PM H⁺-ATPase in leaves illuminated with red (150 µmol m^–2 s^–1) and blue light (50 µmol m^–2 s^–1) for 15 min ( Figure 1A ). During the experiment, we identified a plant that exhibited increased stomatal density, which was named jozetsu (jzt) after a Japanese word meaning ‘talkative’ ( Figure 1B ). The successful isolation of plants with stomatal defects demonstrates that our immunohistochemical technique using a leaf works even when the plant material is limited.

To identify the responsible mutation in jzt plant, we conducted further investigations on jzt and characterized this mutant. Initially, we examined whether jzt exhibits morphological defects other than the stomatal density. As shown in Figure 2A , we observed that jzt plants displayed a dwarf phenotype compared to its progenitor phot2, suggesting that the putative responsible mutation is involved not only in the stomatal development but also in general plant development. Next, we analyzed the stomatal index and density to quantitatively assess the jzt phenotype. In phot2 leaves, the stomatal index was approximately 20% on average, whereas jzt leaves exhibited a 1.5-fold higher stomatal index ( Figure 2B ). Stomatal density in phot2 and jzt was around 150 and 480 stomata mm^–2, respectively. These results indicate that jzt may have a defect in guard-cell differentiation during the stomatal development.

We crossed jzt with Ler or phot2 and conducted map-based cloning or Next-generation sequencing, respectively, to identify the mutation responsible for the jzt phenotypes. Map-based cloning revealed that the responsible mutation is located between 16.6 and 16.9 Mb of the chromosome 4 ( Figure 3A ). Through this, we identified single nucleotide substitutions that cause missense mutations in two genes: At4g35150 and At4g35230 encoding O-methyltransferase family protein and BSK1, respectively ( Figures 3B, C ; Supplementary Figures S1A, B ). Since BSK1 is a signaling complex for BR and BR has been shown to be a negative regulator of stomatal development in leaves (Tang et al., 2008; Kim et al., 2012). Public microarray data indicated that BSK1 transcripts are detectable throughout the entire plant, including leaves, whereas those of At4g35150 were indicated to express only in the developing embryo ( Supplementary Figure S1C ). These results strongly suggested that the responsible mutation is the substitution (G2262 to A) in BSK1, causing an amino acid change (Glu395 to Lys) at the linker region between the kinase domain and the tetratricopeptide repeat (TPR; Figure 3C ).

To validate the above results, we transformed jzt with a wild-type genomic BSK1 fragment, including its putative promoter region, and investigated whether the transgene complements the phenotypes in jzt. Although the dwarf phenotype was not fully restored in the transgenic lines ( Figure 4A ), they exhibited reduced stomatal index and density compared to jzt ( Figures 4B–D ). Therefore, the defect in stomatal development in jzt is most likely to be caused by the novel mutation in BSK1.

BSKs exhibit functional redundancy, and only the bsk3-1 mutant shows insensitivity to exogeneous BR treatment (Tang et al., 2008; Sreeramulu et al., 2013). A recent study indicated that a double knock-out of BSK1 and its homolog BSK2 is required to cause the defects in stomatal development similar to those observed in jtz (Neu et al., 2019). Consistent with these studies, we could not observe the high stomatal density phenotype in the T-DNA insertion bsk1 mutant ( Supplementary Figure S2 ). These results suggest that the novel mutation in BSK1 identified in this study does not simply result in the functional loss of BSK1 protein. Then, we hypothesized that the mutation is a dominant allele rather than a recessive loss-of-function allele. To test this, we reanalyzed the F₂ population obtained by crossing jzt and phot2 to quantify the stomatal phenotype in each plant in the population. We confirmed that a quarter of the F₂ plants exhibited an extremely high stomatal density like jzt (> 400 stomata mm^–2). Another quarter of the population showed phot2-like phenotype, where the stomatal density was below 150 stomata mm^–2. Interestingly, we found that about half of the F₂ plants showed partially increased stomatal density (150–230 stomata mm^–2). The segregation ratio of “phot2-like”: “intermediate”: “jzt-like” was fitted to the well-known 1: 2: 1 autosomal semi-dominant mode of inheritance ( Figure 5 ). These results indicate that the novel mutation observed in BSK1 (hereafter referred to bsk1-4D) is a semi-dominant allele and thus capable of causing morphological defects by itself.

The genetic screening conducted in this study is characterized by the direct observation of stomata. Although similar experiments can be found in literature studying guard-cell differentiation (Pillitteri et al., 2007), most previous studies on stomatal physiology have examined phenotypes that indirectly represent plant transpiration through stomata. For example, changes in leaf surface temperature or leaf weight serves as an index for water loss through transpiration (Merlot et al., 2002; Hashimoto et al., 2006; Negi et al., 2008; Tsuzuki et al., 2011; Takemiya et al., 2013a; Tomiyama et al., 2014; Yamauchi et al., 2016). The use of these phenotypes in genetic screening enables the simultaneous handling of multiple samples. On the other hand, immunohistochemical experiment appears to be unsuitable for handling many samples at once, as it involved various treatments with chemical solution as well as washing materials. Recently, however, commercial liquid-handling robots have enabled the automation of immunohistochemistry or in situ hybridization experiments (Friml et al., 2003; Matsuzaki et al., 2010). Here, we also demonstrated that the immunohistochemical detection of guard-cell PM H⁺-ATPase can be semi-automated by the robot. Genetic screening, taking advantage of the liquid-handling robot, is ongoing to isolate mutants that exhibit defects in the light-induced phosphorylation of guard-cell PM H⁺-ATPase, which will be reported in a future paper.

Previous techniques for detecting guard-cell PM H⁺-ATPase relied on the isolation of epidermal tissues containing guard cells, and thus, the application is restricted by the availability of epidermis from the plants (Ueno et al., 2005; Hayashi et al., 2011). In this context, the immunohistochemical technique using leaves was expected to enable us to conduct experiments such as genetic screening, where we have to handle individual, and sometimes dwarf, leaves (Ando and Kinoshita, 2018). Successful phenotyping by the immunohistochemistry using a single leaf even in dwarf plants like jzt/bsk1-4D, demonstrates that our technique has a broad range of application. Thus, it would enable various genetic investigations of guard-cell PM H⁺-ATPase in plants with small and/or few leaves like early-flowering mutants (Kinoshita et al., 2011).

In this study, we identified a novel semi-dominant mutation in BSK1 that causes morphological defects, including the overproduction of stomata. Incomplete complementation of the dwarf phenotype in jzt/bsk1-4D by genomic BSK1 may support that the mutation is a semi-dominant allele ( Figure 4A ). The BSK family genes exhibit functional redundancy, and simultaneous knock-out of BSK genes is required to induce the morphological defects (Sreeramulu et al., 2013; Neu et al., 2019). BSK1 is considered one of the substrates of BRI1 in BR signaling (Tang et al., 2008). Crosstalk exists between leucin-rich-repeat receptor-like kinases (LRR RLKs), including BRI1, which regulates the plant growth, development, and innate immunity (Zhu et al., 2013). Previously, the bsk1-1 mutation was identified as a suppressor mutation for the powdery mildew resistance phenotype in ENHANCED DISEASE RESISTANCE2, and BSK1 was shown to associate with FLAGELLIN SENSING2 (FLS2), another LRR RLK that functions in the immune response (Shi et al., 2013). In contrast to the mutation in jzt/bsk1-4D, bsk1-1 is a recessive allele that causes a missense mutation in the TPR domain of BSK1 (Shi et al., 2013). TPR was originally identified in yeast and has been considered to be involved in protein–protein interaction (Hirano et al., 1990; Sikorski et al., 1990; Blatch and Lässle, 1999). As bsk1-1 does not affect the BSK1-FLS2 interaction itself, a functional defect other than the protein interaction may be caused by the bsk1-1 mutation (Shi et al., 2013). jzt/bsk1-4D is located in the C-terminal region of the linker between the kinase domain and TPR ( Figure 3C ). Previous reports and the present results imply that the C-terminal region of BSK1, including Glu395 might function in the regulation of or interaction with protein(s) that regulates guard-cell differentiation. The putative target of BSK1 may include other BSK family proteins. Sreeramulu et al. (2013) indicated that loss-of-function of BSK1 restores the insensitivity to 24-epibrassinolide in bsk3,4,6,7 mutant, suggesting an antagonistic interaction between BSK1 and other BSK family proteins. It would be interesting to investigate the stomatal phenotype in higher-order bsk mutants, including jzt/bsk1-4D mutation. The jzt/bsk1-4D mutation identified in this study would be a beneficial tool to uncover the protein function of BSK1 in the plant development.

As the phosphorylation of guard-cell PM H⁺-ATPase was detected in jzt/bsk1-4D ( Figure 1B ), BSK1 may not be necessarily required for light-induced phosphorylation of PM H⁺-ATPase in guard cells. Given that BR signaling has pleiotropic functions (Zhu et al., 2013); however, BSK1 might function in other stomatal response, i.e. immunity (Zeng et al., 2010). Stomatal closure is a part of the immune response to restrict bacterial invasion, whereas the pathogenic bacteria have a mechanism for stomatal reopening to achieve their infection (Melotto et al., 2006). Bacterial invasion also induces a systemic reduction of stomatal density in new leaves emerged after the inoculation (Dutton et al., 2019). Previous studies have revealed a significant role of PM H⁺-ATPase in the plant-pathogen interaction including stomatal movement, in which RPM1-interacting protein 4 (RIN4) function as PM H⁺-ATPase activator for stomatal reopening (Liu et al., 2009; Elmore and Coaker, 2011). RIN4 is a putative phosphorylation target of FLS2, which is likely to be regulated by BSK1 as described above (Shi et al., 2013 ; Ray et al., 2019). It is noteworthy that BR induces the phosphorylation of PM H⁺-ATPase in the hypocotyl of etiolated seedlings (Minami et al., 2019). These results imply a functional connection between BSK1 and the stomatal immune response. The jzt/bsk1-4D mutant would also be a useful genetic resource to investigate this hypothesis.