For plant growth on soil, plants were grown under 16-h fluorescent light (50 µmol m^-2 s^-1)/8-h dark cycle at 24°C in 55-70% humidity in a growth room. For plant growth on plate followed by on soil, seeds were surface-sterilized and sown on Murashige and Skoog-agar plate supplied with 1% (w/v) sucrose under 16-h fluorescent light (50 µmol m^-2 s^-1)/8-h dark cycle. Four-week-old plants were transferred to soil and grown under 16-h fluorescent light (50 µmol m^-2 s^-1)/8-h dark cycle at 24°C in 55-70% humidity in a growth room. gl1 [Columbia (Col-0), carrying gl1 mutation] is the background ecotype of an rtl2 mutant and used as the wild type (WT). We backcrossed rtl2 with gl1 three times. The mutants used in this study, tsb1-1 (SAIL_886_A01), tsb2-1 (SAIL_598_H05), trp3-1 (CS8331; Radwanski et al., 1996), wei2-1 wei7-1 (CS16399; Stepanova et al., 2005) were obtained from the Arabidopsis Biological Resource Center (ABRC) (Ohio State University, Columbus, OH, USA). Col-0 is the background ecotype of these mutants. For plant growth in hydroponic culture, seeds were surface-sterilized and sown on Murashige and Skoog-agar plate as described above. Ten-day-old seedlings were transferred to hydroponic culture system with a nutrient solution (Gibeaut et al., 1997). The solution was constantly replaced every 1 week.

Mutant screening based on stomatal aperture-dependent water loss was performed as previously described (Tsuzuki et al., 2011). Briefly, ethyl methanesulfonate (EMS)-treated gl1 M₂ seeds were germinated and grown on soil. We measured the fresh weight of a detached rosette leaf at 0 and 90 min from each 4-week-old M₂ plant and isolated some rapid transpiration in detached leaves (rtl) mutants (Tsuzuki et al., 2011). In this study, we investigated an rtl2 mutant, which shows rapid weight change compared to WT plants. Mapping populations were generated by crossing the rtl2 mutant with the Landsberg erecta (Ler) accession of Arabidopsis thaliana. RTL2 locus was identified by mapping using simple-sequence length polymorphism (SSLP) and cleaved amplified polymorphism (CAPS) markers from 910 F₂ plants showing dwarf phenotype with pale green leaves and dark veins.

Stomatal apertures in the isolated epidermis were measured according to the previous method (Inoue et al., 2008) with modifications. The epidermal fragments isolated from overnight dark-adapted 4- to 6-week-old plants or 4- to 6-week-old plants from illuminated growth condition at zeitgeber time (ZT) 4 to 7 were incubated in basal buffer (5 mM MES-BTP, pH 6.5, 50 mM KCl, and 0.1 mM CaCl₂). For investigations of light-induced stomatal opening and effect of ABA or auxinole, the epidermal fragments were incubated under light [blue light (Stick-B-32; EYELA, Tokyo, Japan) at 10 µmol m^-2 s^-1 superimposed on background red light (LED-R; EYELA, Tokyo, Japan) at 50 µmol m^-2 s^-1] at 24°C in the presence or absence of 20 µM ABA or 10 µM auxinole for 2.5 hr or kept in the dark at 24°C for 2.5 hr. For investigation of the effect of auxin, the epidermal tissues were incubated in the basal buffer for 1.5 hr in darkness to deplete endogenous auxin. Then, the pre-incubated epidermal tissues were treated with 10 µM IAA in the dark for 3 hr. Stomatal apertures were measured microscopically by focusing on the inner lips of stomata in the abaxial epidermis.

For determination of stomatal aperture from test-crossed plants under illuminated growth condition, we used intact leaves for measurements according to the previous method (Toh et al., 2018). Rosette leaves were detached from the plants at ZT 4 to 7. We cut out central areas of the leaves without the midrib were cut out and mounted leaf disc on microscope slides with the abaxial side attached to the cover glass. We obtained Images using an optical microscope (BX43; Olympus) with a charge-coupled device (CCD) camera (DP27; Olympus) with a x20 objective lens (UPlanFL N; Olympus). For getting extended focus imaging, we used cellSens standard software (Olympus) to maximize the number of analyzable focused stomata within each image. Stomatal apertures in the abaxial side were measured on the inner lips of stomata using ImageJ software (http://imagej.nih.gov/ij/).

Plants were grown on MS plate for 4 weeks and transferred to soil in each pot for 1 week. We measured leaf temperatures using a TVS-500EX infrared thermography instrument (NEC Avio Infrared Technologies Co., Ltd.) and analyzed images using the Avio Thermography Studio software.

Stomatal size and density were determined according to a previous method (Wang et al., 2011).

We purified total RNA from rosette leaves of 4- to 6-week-old plants using an RNeasy Plant Mini Kit (Qiagen). We synthesized 1st-strand cDNAs using a Takara PrimeScript II First Strand cDNA Synthesis Kit (Takara) with oligo(dT)_12–18 primer. For RT-PCR, we amplified cDNA fragments by PCR using specific primers ( Supplementary Table S2 ). Quantitative RT-PCR (qRT-PCR) was performed as previously described (Kinoshita et al., 2011) using specific primers ( Supplementary Table S2 ). We used TUB2 (AT5G62690) as an internal standard for PCRs.

Anti-TSB1 antibody was raised against the recombinant TSB1ΔN, TSB1 protein without chloroplast targeting signal peptide in its N-terminus, as an antigen in a rabbit (Medical & Biological Laboratories). We amplified the TSB1 DNA fragment from 1st-strand Arabidopsis cDNA by PCR using the specific primers 5′-CGGGATCCGACCCGGCCCTGTGGCAAC-3′ and 5′-CGGGATCCTCAAACATCAAGATATTTAGCCACTGTCTGAAC-3′. We cloned the TSB1 CDS of 202–1413 bp containing BamHI site at both ends into the BamHI site of the pGEX-2T vector (GE Healthcare) to express as a fusion protein with glutathione S-transferase (GST). The pGEX-2T-TSB1 was transformed into the E. coli BL21 strain. The fusion protein (GST-TSB1ΔN) was purified using the glutathione-Sepharose 4B (GE Healthcare). The TSB1ΔN protein was obtained by digestion with thrombin to cut off the GST and used for the immunization as antigen.

Immunoblot analysis was performed according to the methods described in Hayashi et al. (2010) with modifications. We grinded leaves from 5- to 6-week-old plants using a mortar and pestle in extraction buffer (50 mM MOPS–KOH, pH 7.5, 2.5 mM EDTA, 100 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 20 µM leupeptin, and 2 mM DTT). Proteins (20 µg) were loaded and separated by SDS-PAGE. We detected TSB1 protein using the polyclonal antibody raised against recombinant TSB1ΔN protein (Anti-TSB1 antibody) in rabbits. We detected the 14-3-3 proteins with the anti-14-3-3 protein (GF14 ø) antibody (Kinoshita and Shimazaki, 1999) as a control. We used the antibodies at a 3,000-fold dilution.

Detections of PM H⁺-ATPase protein and phosphorylation status of the penultimate residue, Thr, of PM H⁺-ATPase in guard cell protoplasts (GCPs) from Vicia faba were performed using anti-H⁺-ATPase antibody and anti-pThr antibody, respectively (Hayashi et al., 2010).

TSB activity was determined according to Greenberg and Galston (1959) with modifications. For measurement in leaf extracts, leaves from 5- to 6-week-old plants were ground in phosphate buffer (0.1 M Potassium phosphate-KOH, pH 8.2, 30% [w/v] Insoluble-Polyvinylpolypyrrolidone [Sigma-Aldrich]) with 212-300 µm glass beads (Sigma-Aldrich) using a mortar and pestle. Then the extract was sonicated and centrifuged at 4 °C (12,000 g for 15 min). The supernatant was used as the enzyme source. The supernatant (150–250 µg of protein) was suspended in reaction buffer (50 µl; 80 mM Potassium phosphate-KOH, pH 8.2, 60 mM L-Serine, 200 µM indole, 10 µg/ml pyridoxal phosphate) and reacted at 30 °C for 90 min. The reaction was stopped by the addition of 5 µl of 0.2 M NaOH. To assay for residual unreacted indole, 200 µl of toluene was added to the reaction mixture and then shaken and centrifuged at 1,500 g for 15 min. The resulting toluene layer was transferred and mixed with 4 times the volume of Ethanol and twice the volume of Ehrlich’s reagent (36 mg/ml p-dimethylaminobenzaldehyde, 2.13 M HCl dissolved in Ethanol). After incubation at 25 °C for 30 min, the absorbance at 540 nm was measured and differences of absorbance between before and after the reaction was converted to the disappearance of indole using the standard curve generated with dilution series of indole. For expression of recombinant TSB1ΔN protein, the TSB1 DNA fragments derived from WT and rtl2 were cloned into the vector as described above. The recombinant TSB1ΔN proteins were purified using the glutathione-Sepharose 4B (GE Healthcare) and used for measurement of TSB activity.

We isolated guard cell protoplasts (GCPs) enzymatically from lower epidermis of leaves from 4- to 6-week-old Vicia faba according to a previous method (Hayashi et al., 2021). GCPs in suspension buffer (5mM MES-NaOH [pH 6.0], 10mM KCl, 0.4 M mannitol and 1mM CaCl₂) were treated with auxins (IAA, 1-NAA, 2,4-D) and fusicoccin (FC) at 10 µM in the dark for 30 min. Proteins (20 µg) were loaded and separated by SDS-PAGE.

ASA1; AT5G05730, ASB1; AT1G25220, AUR3; AT4G37390, IAA1; AT4G14560, IAA24; AT4G14560, SAUR9; AT4G36110, TSA1; AT3G54640, TSB1; AT5G54810, TSB2; AT4G27070, TUB2; AT5G62690.

To elucidate the mechanism of stomatal movement, we have performed a mutant screening based on stomatal aperture-dependent weight loss of detached leaves via transpiration using a microbalance (Tsuzuki et al., 2011). In addition to a rapid transpiration in detached leaves 1 (rtl1) mutant (Tsuzuki et al., 2011), we isolated an rtl2 mutant. The rtl2 showed rapid weight loss of detached leaves under growth condition. The WT (Col-gl1; background plant of the screening) leaf weight decreased to 59% of initial weight for 90 min, whereas the rtl2 leaf weight decreased to 2% ( Figure 1A ). In addition, rtl2 mutant showed severe dwarf phenotype with pale green leaves and dark veins under soil grown condition ( Figure 1B ). It is worthy of note that stomatal size and index in rtl2 and tsb1-1 were almost comparable to those in background strains. In contrast, stomatal density in rtl2 and tsb1-1 were ~30-46% increase compared to that in background strains ( Table 1 ). Furthermore, color of dry seeds from rtl2 and tsb1-1 showed a lighter color than background strain seeds ( Supplementary Figure 1 ) (Last et al., 1991).

As shown in Figure 1C , stomata of rtl2 opened widely compared to WT under illuminated growth condition. We further examined stomatal responses of rtl2 in detail. The stomata in WT closed in the dark and opened in response to light, and ABA (20 µM) suppressed light-induced stomatal opening ( Figure 1D ). By contrast, the stomata in rtl2 opened widely even in the dark and 20 µM ABA had no effect on stomatal aperture. In accord with open-stomata phenotype in rlt2 mutant, rtl2 mutant exhibited clear low leaf temperature phenotype under illuminated growth condition ( Figure 1E ). Average leaf temperature of the rtl2 mutant was reduced over 3.0°C relative to WT.

To identify the RTL2 locus, we performed a map-based analysis and found strong linkage the CAPS marker K5F14-1-1 and MBG8-1 in rtl2 ( Figure 2A ). According to the Arabidopsis Information Resource (TAIR) database, Tryptophan synthase beta-subunit 1 (TSB1; AT5G54810) is a candidate gene. Because there is a TSB1 locus between K5F14-1-1 and MBG8-1 and known TSB1 mutants showed dwarf phenotype with pale green leaves and dark veins same as in rtl2 ( Figures 1B , 2B ) (Jing et al., 2009). Sequence analysis of TSB1 cDNA from rtl2 revealed a single nucleotide substitution from G485 to A, which induces a novel missense mutation from Gly162 to Glu ( Figure 2B ). Next, we investigated theTSB1 transcript level and protein amount of TSB1 in rosette leaves. As shown in Figure 2C, 2D , the level of TSB1 transcript was similar to that of the WT, whereas the amount of TSB1 protein was significantly decreased in rtl2. We further determined the TSB activity in leaf extracts by measuring disappearance of indole. The results showed that rtl2 has ~35% TSB catalytic activity compared to that of WT ( Figure 2E ). Reduction of TSB activity in rtl2 may be caused by not only lower amount of TSB1 protein but also less activity of rtl2 mutation (Gly162 to Glu in TSB1), because the recombinant TSB1-G162E has only ~7.2% TSB activity compared to that of WT-TSB1 ( Figure 2F ).

To determine whether rtl2 is an allele of tsb1, a T-DNA insertion mutant tsb1-1 (SAIL_886_A01) was obtained from ABRC ( Figure 2B ). The tsb1-1 showed rtl2-like dwarf phenotype with pale green leaves and dark veins ( Figure 3A ). RT-PCR analysis revealed that tsb1-1 is a knockout mutant, which results in significant decrease of protein amount in rosette leaves of tsb1-1 ( Figures 3B, C ). Stomata of tsb1-1 also opened widely even in the dark and could not close in response to ABA, suggesting that TSB1 is responsible for the rtl2 phenotype ( Figure 3D ). To confirm this, we have tried to complement the WT TSB1 genome to rtl2, however, we could not obtain the transgenic rtl2 plants carrying the WT TSB1 gene under its own promoter, probably due to growth defect of rtl2 plants. Therefore, we performed test-cross of rtl2 and tsb1-1 to show that these are allelic gene. As shown in Figure 4A , F₁-generation plants showed dwarf phenotype with pale green leaves and dark veins same as in rtl2 and tsb1-1, indicating that these are allelic gene. In addition, stomata of F₁-generation plants widely opened under illuminated growth condition ( Figure 4B ). From these results, we concluded that TSB1 is responsible for the rtl2 phenotype.

TSB1 encodes a predominantly expressed Trp synthase ß subunit in Trp biosynthetic pathway, in which Trp is synthesized from chorismate via six reaction steps as shown in Figure 5A (Radwanski and Last, 1995; Tzin and Galili, 2010). To investigate whether other Trp biosynthesis enzymes affect stomatal opening, we measured the stomatal aperture of the double mutant of AS, wei2-1 wei7-1, single substitution mutant of TSA1, trp3-1, and a T-DNA insertion knockout mutant of TSB1 paralog, TSB2 (tsb2-1; Supplementary Figure S2 ). Among the mutants, trp3-1 showed open stomata phenotype similar to tsb1-1, but not in wei2-1 wei7-1 and tsb2-1 ( Figure 5B ).

Indole-2-glycerol phosphate and Indole are substrates of TSA and TSB respectively. These are thought to be converted to indole-3-acetic-acid (IAA) via Trp-independent pathway (Ouyang et al., 2000). Jing et al. (2009) reported that free IAA levels in two tsb1 alleles, trp2-1 and trp2-8, were elevated and auxin responsive genes were up-regulated in these mutants. To clarify whether IAA contents in rtl2 is also elevated, we investigated the expression level of some auxin responsive genes, AUXIN UP REGULATED 3 (AUR3), SMALL AUXIN UPREGULATED RNA 9 (SAUR9), INDOLE-3-ACETIC ACID INDUCIBLE 1 (IAA1) and INDOLE-3-ACETIC ACID INDUCIBLE 24 (IAA24). As shown in Figure 6A , AUR3 was significantly up-regulated in both rtl2 and tsb1-1, though other auxin responsive genes were not. Given that up-regulation of AUR3 expression in tsb1 alleles, we speculated that open stomata phenotype in rtl2 is caused by elevated levels of auxin. To address the possibility, we first investigated the effect of auxin on stomatal aperture. Incubation of the epidermal tissues with 10 µM IAA in the dark for 3 hr had no effect on stomatal aperture ( Figure 6B ). Next, we investigated effect of auxins, IAA and synthetic auxins, 1-naphthylacetic acid (1-NAA) and 2,4-dichlorophenoxyacetic acid (2.4-D), on phosphorylation status of the penultimate residue, Thr, of PM H⁺-ATPase in guard cell protoplasts (GCPs) from Vicia faba, because auxin activates PM H⁺-ATPase via phosphorylation during auxin-induced hypocotyl elongation (Takahashi et al., 2012; Uchida et al., 2018). However, consistent with a recent report that IAA did not induce phosphorylation of PM H⁺-ATPase in guard cells of Arabidopsis thaliana (Akiyama et al., 2022), all auxins had no effect on phosphorylation status of PM H⁺-ATPase for 30 min, although a fungal toxin fusicoccin (FC) induced phosphorylation of PM H⁺-ATPase ( Figure 6C ). These results suggest that short term treatment of auxin (within 3 hr) has no effect on both stomatal aperture and phosphorylation status of PM H⁺-ATPase in guard cells. Furthermore, we found that auxin antagonist, auxinole (Hayashi et al., 2012), did not change the stomatal aperture in rtl2 and tsb1-1 ( Figure 6D ).

To verify whether the open stomata phenotype in tsb1 mutants is caused by Trp deficiency, L-Tryptophan (L-Trp) was exogenously supplied to tsb1-1. We used only tsb1-1 mutant in this experiment due to less growth of rtl2 in hydroponic culture. Ten-day-old tsb1-1 seedlings grown on MS plate were transferred to hydroponic system supplemented with/without 0.25 mM L-Trp for 2-4 weeks. Similar to the previous report (Jing et al., 2009), dwarf and pale green phenotype in tsb1-1 were partially rescued by exogenous application of L-Trp ( Figure 7A ). Interestingly, tsb1-1 showed normal stomatal aperture by application of L-Trp ( Figure 7B ), suggesting that open stomata phenotype in tsb1-1 is due to Trp deficiency. On the other hand, under hydroponic culture condition, stomatal density in tsb1-1 did not show significant differences compared to that in Col-0 and L-Trp application did not change the stomatal density in both Col-0 and tsb1-1 ( Supplementary Table S1 ). Therefore, Trp deficiency may have no effect on the stomatal density.