Oligonucleotides (Supplementary Table S1) were obtained from Eurofins MWG Operon (Ebersberg, Germany). Tools provided by the National Center for Biotechnology Information¹, the Arabidopsis Information Resource², the Plant Transcription Factor Database³, and PLAZA 3.0⁴ were used for computational analyses.

Arabidopsis thaliana (L.) Heynh. (Col-0) was used as the wild type; transgenic lines are based on this accession. For experiments at seedling stage, seeds were surface sterilized and sown on half-strength Murashige and Skoog (MS) agar medium containing 1% (w/v) sucrose. Plants were grown in soil (Einheitserde GS90; Gebrüder Patzer) under a 12-h day (120 mmol m^-2 s^-1; 22°C) : 12-h night regime (22°C). For experiments shown in Figures 4, 5 and Supplementary Figures S1, S2 plants were grown at 21°C under a 16-h day and 8-h dark regime. 35S:JUB1 (JUB1Ox), RD29A:JUB1, and jub1-1 (SALK_ID 036474) plants were described previously (Wu et al., 2012). AtHB13Ox plants and athb13 mutants were reported in (Cabello and Chan, 2012; Cabello et al., 2012).

Gene constructs were generated by polymerase chain reaction (PCR) and directional cloning. All PCR-generated amplicons were checked for correctness by DNA sequence analysis (Seqlab or LGC Genomics). Constructs were transformed into Arabidopsis via Agrobacterium tumefaciens-mediated transformation (floral dip method). To generate the 35S:AtHB13-GFP and AtHB13-GST constructs, the AtHB13 open reading frame (ORF) was PCR-amplified without the stop codon and cloned into pENTR/D-TOPO vector using the pENTR directional TOPO cloning kit (Invitrogen). The AtHB13 ORF was then transferred to the GATEWAY vectors pK7FWG2 (Ghent University) and pDEST24 (Invitrogen), respectively, through LR recombination. To generate the JUB1 promoter deletion constructs (Pro_JUB1:GUS), 1.0, 0.73, 0.68, 0.31, and 0.21 kb long segments of the JUB1 promoter were PCR-amplified and cloned into pENTR/D-TOPO vector using the pENTR directional TOPO cloning kit (Invitrogen). The promoter fragments were then transferred to the pKGWFS7 GATEWAY vector (Ghent University) by LR recombination. pTUY1H-JUB1-373: The selected 373-bp JUB1 promoter region was first cloned via TA cloning into pCR2.1 entry vector (Invitrogen), and then transferred to the pTUY1H yeast transformation vector (with LEU2 as selection marker) (Castrillo et al., 2011) by restriction enzyme-mediated cloning.

Two-week-old seedlings were carefully transferred from agar media plates to Erlenmeyer flasks containing liquid MS medium (1% [w/v] sucrose) in the absence or presence of 10 mM H₂O₂, and incubated overnight. For drought treatments, 4-week-old soil-grown plants were not watered for 6 days, and leaves were harvested. Histochemical GUS staining was performed overnight at 37°C in the dark. Chlorophyll was removed by clearing the samples with 70% (v/v) ethanol. Quantification of GUS signal was done as described (Béziat et al., 2017).

Total RNA was extracted using the RNeasy Plant Mini kit (Qiagen, Hilden, Germany). Synthesis of cDNA and quantitative real-time PCR (qRT-PCR) using SYBR Green were performed as reported (Balazadeh et al., 2008, 2010) with ACTIN2 (At3g18780) as the reference gene. Primer sequences are given in Supplementary Table S1. Primers were designed using the QuantPrime tool⁵ (Arvidsson et al., 2008).

The bait construct pTUY1H-JUB1-373 (LEU2 selection marker; JUB1 promoter fragment upstream of HIS3 reporter) was transformed into yeast strain Y187, mating type α. The mating-based Y1H screen was done using a library of approximately 1,200 Arabidopsis TFs, established in vector pDEST22 (TRP1 selection marker) in yeast strain YM4271 (mating type a) (Castrillo et al., 2011). Screening for interaction between TFs and the 373 bp long JUB1 promoter fragment was done on SD medium lacking the essential amino acids Leu, Trp, and His in the absence or presence of different concentrations of 3-amino-1,2,4-triazol (3AT) to prevent false positive interactions.

AtHB13-GST fusion protein was purified from Escherichia coli expression strain BL21 Star (DE3) pRARE, which was generated by transforming the pRARE plasmid isolated from Rosetta (DE3) pRARE cells (Merck) into E. coli BL21 Star (DE3) (Invitrogen). Recombinant GST-fusion protein was purified using GST-agarose beads following the manufacturer’s instructions (Sigma–Aldrich, Taufkirchen, Germany). Electrophoretic mobility shift assays (EMSA) was performed as described (Wu et al., 2012) using the Odyssey Infrared EMSA kit (LI-COR). Sequences of 5′-DY682-labeled fragments are given in Supplementary Table S1.

Chromatin immunoprecipitation (ChIP) was carried out on chromatin extracted from 35S:AtHB13-GFP plants grown (i) well-watered for 4 weeks, (ii) well-watered for 4 weeks and then drought stressed for 6 days (by withholding water), and (iii) well-watered for 50 days. WT plants grown in parallel served as controls in each experiment. Three independent ChIP experiments were performed. The qPCR primers for the JUB1 promoter were designed to flank the AtHB13 binding site (Supplementary Table S1). As negative controls, primers annealing to promoter regions of two Arabidopsis genes lacking an AtHB13 binding site, i.e., AT3G18040 (Neg 1) and AT2G22180 (Neg 2), were used. ChIP-qPCR data were analyzed as described (Kaufmann et al., 2010).

For ion leakage measurements, the first six leaves of the rosette were immersed in 10 ml deionized water and shaken at room temperature for 30 min. Electrical conductivity (σ1) was measured at 25°C, using a conductometer (Schott, Mainz, Germany). Then samples were boiled for 15 min, cooled down to 25°C, and conductivity (σ2) was measured again. Ion leakage was calculated through the expression σ1/σ2 × 100. Three independent experiments were performed.

Plant material (five leaves per genotype) was weighed (fresh weight, FW), then put in a Petri dish containing water and kept at room temperature for 3 h. Then, leaves were weighed again (turgid weight, TW). Relative water content (RWC) was calculated using the following formula: RWC [%] = (TW–FW)/(TW) × 100. Three independent experiments were performed.

Plants were grown at a 16-h light : 8-h dark cycle and well watered for 25 days. Thereafter, field capacity was maintained at 50% in all pots by adding the needed quantity of water. Field capacity was determined by weighting the pots. Before the experiment was started, pots were saturated and weighed (100% field capacity). During the experiment, pots were weighed each day and water was added in order to maintain the 50% of the field capacity. The amount of water added to each plant is shown in the figure panels.

Plants were grown at a 16-h light : 8-h dark cycle and drought experiments were started by stopping irrigation at day 20. After stopping irrigation, leaves were detached at the indicated times (see figures), weighed (W1), submerged in tap water for 3 h and weighed again (W2). Water loss [%] was calculated using the following formula: (W2–W1)/W2 × 100.

ACTIN2 (AT3G18780), JUB1 (AT2G43000), ATHB13 (AT1G69780).

The expression of JUB1 rapidly increases after treatment of Arabidopsis plants with hydrogen peroxide (H₂O₂) or in the presence of different abiotic stresses such as salinity and heat (Wu et al., 2012; Allu et al., 2014; Shahnejat-Bushehri et al., 2016). To test whether JUB1 is also induced by drought, we analyzed its expression in plants subjected to water shortage. To this end, 4-week-old wild-type Col-0 (WT) plants were subjected to water withholding for 6 days, and whole rosettes were harvested to quantify JUB1 expression by qRT-PCR. As shown in Supplementary Figure S1A, JUB1 expression was considerably higher in drought-stressed plants than in well-watered controls. We also established transgenic Arabidopsis plants expressing the GUS reporter gene from the 1-kb JUB1 promoter (Pro_JUB1:GUS plants). We grew Pro_JUB1:GUS plants for 4 weeks under well-watered condition and then subjected them to drought stress (by stopping irrigation) for 6, 9, and 12 days. Histochemical GUS staining revealed drought-induced JUB1 promoter activity (Supplementary Figure S1B).

To investigate the function of JUB1 for the response to drought in Arabidopsis, we analyzed the phenotype of JUB1 transgenic lines during drought stress. We previously reported that constitutive overexpression of JUB1 results in reduced growth, delayed flowering, and late leaf senescence (Wu et al., 2012; Shahnejat-Bushehri et al., 2016); we here therefore included transgenic lines expressing JUB1 under the control of the stress-inducible promoter RESPONSIVE TO DESICCATION 29A (RD29A) in our drought assays. The RD29A gene is highly responsive to drought, low temperature, high salt concentration, and desiccation (Yamaguchi-Shinozaki and Shinozaki, 1993, 1994) and only basal expression is observed in non-stressed plants (Yamaguchi-Shinozaki et al., 1999). However, as we reported previously, different RD29A:JUB1 lines displayed different JUB1 expression levels under control conditions (Wu et al., 2012). Here, we selected a RD29A:JUB1 line with lowest expression of JUB1 under non-stress condition for our analyses. The selected transgenic line displayed growth phenotypes similar to WT, in contrast to the plants strongly expressing JUB1 from the constitutive CaMV 35S promoter (JUB1Ox; Shahnejat-Bushehri et al., 2016). Four-week-old RD29A:JUB1, JUB1Ox and WT plants, as well as jub1-1 knockdown mutants were dehydrated for 18 days. As shown in Figure 1A, both, RD29A:JUB1 and JUB1Ox plants exhibited strong drought tolerance, while jub1-1 and WT plants showed severe wilting after 16 days of drought (16DD), and were strongly dehydrated after 18 days (18DD). None of the WT and jub1-1 plants recovered after 18 days of drought followed by re-watering for 6 days (18DD+6DRW) further demonstrating their sensitivity to drought stress. In contrast, JUB1Ox and RD29A:JUB1 plants recovered rapidly after the 18 days of drought stress and showed complete survival after re-watering. Membrane stability under water deficit conditions was assessed by measuring ion leakage in both, control and treated plants. When water was withheld for 12 days, jub1-1 and WT plants showed a higher ion leakage (∼30%) than JUB1Ox and RD29A:JUB1 plants (∼15%). After 16 days of water withholding the differences in electrolyte leakage increased further; while jub1-1 and WT plants showed an ion leakage of ∼70 and ∼58%, respectively, ion leakage was only ∼30% for both, RD29A:JUB1 and JUB1Ox plants upon stress (Figure 1B).

Furthermore, RWC in leaves was determined after 8, 12, and 16 days of withholding water (Figure 1C). RWC was not significantly different between the genotypes at the early stage of drought (8DD), while at later stages of drought (16DD) a significantly higher RWC was observed in RD29A:JUB1 and JUB1Ox plants than in WT and jub1-1 plants (Figure 1C). Collectively, our results reveal that JUB1, when expressed from the constitutive CaMV 35S promoter or from the abiotic stress-induced RD29A promoter, confers superior tolerance to drought stress in Arabidopsis.

We intended to identify upstream regulatory factors controlling the expression of JUB1 during abiotic stress. To this end, we performed a deletion analysis of the JUB1 promoter and then conducted a yeast one-hybrid (Y1H) screen to identify TFs binding to a functionally relevant promoter segment. Various 5′ deletions of the JUB1 promoter were transcriptionally fused to the Escherichia coli β-GLUCURONIDASE (GUS) reporter gene and the constructs were transformed into Arabidopsis (Figure 2A). The promoter-reporter lines (hereafter, Pro_JUB1:GUS deletions) were subjected to H₂O₂ and drought treatments to identify a promoter region relevant for the response of JUB1 to these stresses. To this end, (i) 2-week-old seedlings of Pro_JUB1:GUS deletion lines were treated with 10 mM H₂O₂ and (ii) 4-week-old Pro_JUB1:GUS plants were subjected to desiccation for 6 days (6DD) and leaves were harvested for analysis. Histochemical GUS staining revealed that the 1-kb JUB1 promoter as well as the 0.73-kb and 0.68-kb deletion variants, but not the 0.31-kb and the 0.21-kb promoters, confer stress-inducible activation of the GUS reporter gene, indicating that the region responsive to H₂O₂ and drought is located between positions -0.68 and -0.31 kb of the JUB1 promoter (Figures 2B,C and Supplementary Figure S1C).

To identify upstream transcriptional regulators of JUB1, we performed a yeast one-hybrid (Y1H) screen using the 0.37-kb promoter fragment (-0.68 and -0.31 kb upstream of the JUB1 start codon) involved in the H₂O₂- and drought responsiveness of the JUB1 gene as bait. In a screen with nearly 1,200 Arabidopsis TFs we identified HD-Zip I protein AtHB13 (AT1G69780) as a TF binding to the abiotic stress-responsive segment of the JUB1 promoter (Figure 2D). As AtHB13 has previously been reported to affect the response of A. thaliana to various biotic and abiotic stresses including drought (Cabello and Chan, 2012; Cabello et al., 2012; Gao et al., 2014), we investigated whether it is a genuine upstream regulator of JUB1.

Interestingly, the 373-bp JUB1 fragment employed in the Y1H assay harbors an HD-Zip class I binding site, namely CAATAAATG (Figure 2E). The motif is identical to the one reported by Sessa et al. (1993) for AtHB1 (another HD-Zip I protein from Arabidopsis), with the exception of the underlined A which is a T in the AtHB1 binding site. To test whether AtHB13 binds to the CAATAAATG sequence within the frame of the JUB1 promoter, we performed EMSAs using recombinant AtHB13-GST (glutathione S-transferase) fusion protein and a 5′-DY682-labeled 40-bp JUB1 promoter fragment harboring the HD-Zip class I binding site. As shown in Figure 2F, AtHB13 protein physically interacts in vitro with the respective JUB1 promoter fragment. Next, we checked expression of JUB1 in rosette leaves of transgenic plants overexpressing AtHB13 from the CaMV 35S promoter (AtHB13Ox lines; Cabello et al., 2012) and WT plants at different developmental stages, i.e., at 10, 20, and 50 days after sowing (DAS). Interestingly, expression of JUB1 was not altered in the AtHB13Ox plants at 10 and 20 DAS, but was significantly upregulated in AtHB13Ox plants at 50 DAS (Figure 3A). Thereafter, we analyzed the expression of JUB1 in AtHB13Ox overexpressors and athb13 knockout mutants (Cabello and Chan, 2012; Cabello et al., 2012) during drought stress. To this end, 4-week-old plants were subjected to 6 days of water withholding. As shown in Figure 3B, expression of JUB1 was enhanced in AtHB13Ox plants compared to WT, and strongly repressed in the athb13-1 mutant upon drought stress. Collectively, our data indicate that AtHB13 functions as a positive regulator of JUB1 expression at later stages of development and upon drought stress.

To confirm that AtHB13 interacts with the JUB1 promoter in planta, we performed ChIP assays using Arabidopsis plants stably expressing GFP-tagged AtHB13 protein (hereafter, 35S:AtHB13-GFP). Analysis by confocal microscopy revealed nuclear localization of AtHB13-GFP protein, in accordance with the biological function of AtHB13 as a transcriptional regulator (Figure 3C). Next, we harvested rosette leaves from 35S:AtHB13-GFP and WT plants for gene expression analysis and ChIP-qPCR assays. Plants were grown (i) for 4 weeks in well-watered condition (control); (ii) for 4 weeks in well-watered condition, followed by 6 days without watering (drought); (iii) for 50 days in well-watered condition (50 DAS). Under control condition, JUB1 expression was similar in 35S:AtHB13-GFP and WT plants, while its expression was significantly induced under drought stress in 35S:AtHB13-GFP compared to WT. JUB1 expression was also elevated, although less strongly, in well-watered 50 DAS 35S:AtHB13-GFP plants (Figure 3D). ChIP-qPCR revealed enrichment of the JUB1 promoter fragment harboring the AtHB13 binding site, in particular at 50 DAS control and in drought stress conditions (Figure 3E).

Expression of AtHB13 and of its sunflower homologue HaHB1 is induced by water deficit and overexpression of the two TFs in transgenic Arabidopsis plants improves drought tolerance (Cabello and Chan, 2012), similar to the overexpression of JUB1 reported here. Considering the fact that AtHB13 binds the JUB1 promoter, and regulates JUB1 in planta, the higher drought tolerance of AtHB13 overexpressors might be conveyed through an upregulation of JUB1 expression by the HD-Zip I TF. To test this hypothesis, we tested the drought tolerance of AtHB13Ox, JUB1Ox, athb13-1, athb13-2, jub1-1, and WT plants grown side by side. Plants were grown under well-watered condition and drought stress was started at day 20, before bolting occurred in all genotypes. Thereafter, irrigation was stopped, which gradually produced severe drought stress. Supplementary Figure S2A shows the phenotype of all plants after eight days of drought stress indicating that the two TF overexpressors (AtHB13Ox, JUB1Ox) are more tolerant than the WT, while the athb13-1, athb13-2, and jub1-1 mutants are less tolerant. We determined water loss in plants of all genotypes at different days after stopping irrigation (Supplementary Figure S2B). Our results show that water loss was significantly lower in JUB1Ox than WT plants, in accordance with the observed increase in drought tolerance of these lines, while jub1-1 and athb13 mutants performed slightly worse than the WT in the experiments performed (Supplementary Figures S2B,C). Similar conclusions can be drawn from water consumption experiments; as shown in Supplementary Figures S2D,E, overexpressors consumed less water than WT and mutant plants during mild drought stress, although the reduction in water consumption was much more prominent in JUB1Ox than AtHB13Ox plants, which may in part be due to the more compact growth phenotype of JUB1 overexpressors compared to AtHB13 overexpressors.

To test whether AtHB13 requires JUB1 for improved drought tolerance we crossed the AtHB13Ox plant with the jub1-1 mutant. As seen in Figure 4A, water loss in the AtHB13Ox/jub1-1 line was similar to that of the jub1-1 mutant and the WT, while water loss was much lower in the AtHB13Ox plant, strongly indicating that AtHB13 confers drought tolerance at least in part through transcriptional control of JUB1 (Figure 4B).