Arabidopsis thaliana seeds (Bu-5 or Col-0) were sown on agar (0.8% w/v) plates containing full-strength Murashige and Skoog (MS) salts with a vitamin mixture (10 mg L⁻¹ myoinositol, 200 μg L⁻¹ glycine, 50 μg L⁻¹ nicotinic acid, 50 μg L⁻¹ pyridoxine hydrochloride, 10 μg L⁻¹ thiamine hydrochloride, pH 5.7), and 1% w/w sucrose. Plates were sealed with surgical tape, the seeds were stratified at 4°C for 4–7 days and transferred to a growth chamber (80 μmol photons m² s⁻¹; 16/8-h light/dark cycle; 22°C) for germination and growth.

Bu-5 seeds were irradiated as described in (Uchida et al., 2022).

Seeds of the following Arabidopsis mutants in the Col-0 background were obtained from the Arabidopsis Biological Resource Center (Ohio State University): uvh6-1 (CS6375), xpf (SALK096156C), xpg (CS3820), and ercc1 (SALK077000C). Seeds of the sog1–1 mutant (Col-0 background) were kindly provided by Dr. Kaoru Yoshiyama of Tohoku University. To generate a uvh6–1 sog1–1 plant, we crossed a uvh6–1 plant with the sog1–1 mutant.

Seedlings were grown on nylon mesh (990 μm) on an MS agar plate. At 7 days of age, they were mesh-transferred to a plate supplemented with 100 mM NaCl for 7 days. The seedlings were then mesh-transferred to a plate supplemented with 750 mM sorbitol for 36–49 days. Aerial parts of six randomly chosen seedlings from each experimental group (with or without various stresses) were harvested and subsequently homogenized in cold acetone. Chlorophyll content was determined spectrophotometrically using the equations described in (Porra et al., 1989).

All abiotic stress assays were performed in a growth chamber at 22 °C under a 16/8-h light/dark cycle with a light intensity of 80 μmol photons m^-² s^-¹. Seedlings were grown on nylon mesh (990 μm) on an MS agar plate. At 10 days of age, they were mesh-transferred to a plate supplemented with 650 mM sorbitol for 31 days (osmotic-shock stress) or 5 μM paraquat for 14 days (oxidative stress). Chlorophyll content was determined as described in (Porra et al., 1989).

Total RNA extraction and qRT-PCR were performed as described in (Isono et al., 2023). ACTIN2 was used as an internal standard for qRT-PCR. The PCR primers are listed in Supplementary Table 2 .

We crossed the aod12 mutant with Pog-0, an accession that shows acquired osmotic stress tolerance, and selfed the resulting F₁ progeny to generate an F₂ population. Genomic DNA was prepared from individual F₂ plants with the recessive phenotype for use as PCR templates. We used the simple sequence-length polymorphism (SSLP) markers listed in Supplementary Table 2 for mapping. PCR conditions were as follows: initial denaturation at 94 °C for 2 min; 34 cycles at 94 °C for 20 s, 52–55 °C for 20 s, and 72 °C for 20 s; and final extension at 72 °C for 2 min. The microsatellites were fractionated in 5%–7% agarose gels, and the recombination frequencies (%) were calculated from the band pattern.

Mutations were detected in the whole-genome sequencing data of aod12 as described in (Uchida et al., 2022). RNA-seq analysis was performed with three biological replicates for each condition. Differentially expressed genes (DEGs) were identified based on the criteria for fold change and FDR. The read data and the RNA-seq data were submitted to the DNA Data Bank of Japan (DDBJ) Read Archive (acc. Nos. DRR641246 and DRR641293–DRR641304).

For complementation analysis, we amplified the genomic region of AOD12/UVH6 (2.0-kb upstream of the ATG initiation codon) by PCR with the pRI909 UVH6 F and pRI909 UVH6 R primers ( Supplementary Table 2 ) and cloned it into the pRI909 vector (Takara Bio Inc.). The construct was introduced into Agrobacterium tumefaciens strain GV3101. Plants were transformed with the agrobacterium using the floral dip method. Transgenic plants were selected on MS agar plates containing 200 µg mL⁻¹ claforan and 20 µg mL⁻¹ hygromycin. Ten-day-old seedlings (T₁ plants) were transferred into soil pots.

The DNA damage assay was performed under the same growth chamber conditions as the abiotic stress assays, which were 22 °C with a 16/8-h light/dark cycle and a light intensity of 80 μmol photons m^-² s^-¹. To determine single-strand break (SSB) frequency, DNA extracted from seedlings was denatured by adding alkaline solution (0.5 M NaOH, 10% glycerol, and 0.25% [w/v] bromocresol green) and incubating for 30 min at 37 °C. DNA molecules were separated according to their single-strand molecular lengths in 0.7% alkaline agarose gels using static field electrophoresis and biased sinusoidal field gel electrophoresis (Genofield; ATTO Co.) (Hidema and Kumagai, 1998).

To determine double-strand break (DSB) frequency, DNA extracted from seedlings was mixed with loading buffer (10% glycerol and 0.25% [w/v] bromocresol green in TE buffer). DNA molecules were separated according to their single-strand molecular lengths in TBE agarose gels using static field electrophoresis and biased sinusoidal field gel electrophoresis (Genofield) (Hidema and Kumagai, 1998). The molecular length markers were DNA from Hansenula wingei chromosomes (smallest, 1.05 Mb) (Bio-Rad), bacteriophage T4 (170 kb), bacteriophage λ (48.5 kb), and the HindIII digest of λ DNA (23.1, 9.4, 6.6, 4.3, and 2.3 kb).

The SSB or DSB frequency was determined using a DNA damage analysis system (Tohoku Electric Co.) as previously described (Hidema and Kumagai, 1998) from a molecular length standard curve. DNA at each migration position was quantified and the values were expressed as SSB or DSB per 10⁶ bp.

Cell death was detected by trypan blue staining as previously described (Tsukimoto et al., 2021).

Statistical significance analysis was performed using the Student’s t-test. The significance results are indicated by asterisks: *P<0.05, **P<0.01, and ***P<0.001. For multiple comparisons, one-way ANOVA with post hoc Tukey HSD test was conducted. The 0.05 level of probability was used as the criterion for significance.

UVH6 protein structure predictions were performed using AlphaFold3 (Abramson et al., 2024) via the AlphaFold Server (https://alphafoldserver.com/welcome). For the superimposition of Arabidopsis thaliana XPD/UVH6 with human XPD and for domain-specific coloring, PyMOL Molecular Graphics System, Version 3.0 (Schrödinger, LLC.) was utilized. The positions of the domains were determined by aligning UVH6 with XPD and referencing the positions indicated in Peissert et al., 2020 (Peissert et al., 2020).

To identify genes involved in acquired osmotolerance, we screened over 30,000 ion-beam-mutagenized M2 (second generation after mutagenesis) seed pools derived from the osmotolerant Bu-5 accession and isolated the mutant aod12. Under normal growth conditions, aod12 leaves were pale green and chlorophyll content was significantly lower than in the WT ( Figure 1A ). Therefore, the osmotolerance of their seedlings was evaluated on the basis of chlorophyll content relative to that under normal conditions. After exposure to 100 mM NaCl, acquired osmotolerance to 750 mM sorbitol was significantly lower in aod12 seedlings than in WT Bu-5 ( Figure 1A ). Under normal growth conditions, 17-day-old aod12 and WT seedlings exhibited similar fresh weights ( Figure 1B ). While grown in soil for 5 weeks, aod12 retained its characteristic pale-green rosette leaves, a phenotype consistent with earlier observations ( Figure 1B ).

To determine whether the aod12 phenotype was broadly tolerant to stresses, we assessed its tolerance to other abiotic stresses, namely osmo-shock, salt shock, oxidative stress (paraquat) ( Figure 2A ), short-term heat (S-heat) stress ( Figure 2B ), and long-term heat (L-heat) stress ( Figure 2C ). The aod12 mutant showed significantly reduced tolerance to osmotic shock, S-heat, and L-heat stresses compared to WT Bu-5, but no difference in oxidative stress tolerance.

Furthermore, to gain insight into the molecular basis of the observed heat sensitivity in aod12, we examined the expression of heat shock protein (HSP) genes under heat stress. Our analysis revealed that the expression levels of both HSP70 and HSP17.6 were significantly lower in aod12 compared to WT when subjected to 37 °C for 1 hour ( Figure 2D ). This attenuated HSP induction is consistent with previous findings of a genome-wide reduction in gene expression in aod12 under heat stress conditions (Bourguet et al., 2018). These results suggest that the hypersensitivity of aod12 to heat stress is likely due to the impaired induction of essential stress-responsive genes, including HSPs, which are critical for heat tolerance.

To identify the locus responsible for the osmosensitive phenotype of aod12, we crossed it with Pog-0, an accession with acquired osmotolerance (Ariga et al., 2017), and performed high-resolution chromosomal mapping. The locus was mapped to a 313-kbp region on chromosome 1 ( Figure 3A ). Whole-genome sequencing of aod12 and WT Bu-5 revealed a 3-bp deletion in At1g03190, causing a deletion of a single arginine in the UVH6 protein (296) in aod12 ( Figure 3A ). At1g03190 was previously identified as responsible for the uvh6 mutant, which is hypersensitive to UV. To confirm that UVH6 is the causal gene of aod12, we evaluated the acquired osmotolerance of the mutant uvh6-1 (Col-0 background strong allele); (Liu et al., 2003). Since Col-0 lacks acquired osmotolerance, we acclimated the seedlings with 100 mM NaCl and exposed them to 650 mM sorbitol. The uvh6–1 mutant had pale green leaves under normal conditions and an osmosensitive phenotype, as did aod12 ( Figure 3B ). Complementation of aod12 with UVH6 with its native promoter (aod12_UVH6) restored acquired osmotolerance and leaf color to Bu-5 WT levels ( Figure 3C , Supplementary Figure 1 ). Therefore, UVH6 is the causal gene for the osmosensitive and pale-green-leaf phenotype of aod12.

UVH6 is a component of the TFIIH complex, which has two main functions: facilitating transcription initiation by RNA polymerase II and participating in NER, a key DNA repair pathway (Kciuk et al., 2020). NER repairs DNA damage spanning multiple bases, which is often caused by cross-linked structures (Kciuk et al., 2020). To assess aod12 sensitivity to DNA damage stress, we used cisplatin, which induces cross-linked DNA structures and inhibits growth. The sensitivity to cisplatin was higher in the aod12 mutant than in WT ( Figure 4A ), suggesting reduced DNA repair in aod12. This interpretation is strongly supported by prior studies in Arabidopsis demonstrating that the NER pathway is crucial for repairing cisplatin-induced DNA damage. Specifically, in vitro repair synthesis assays have shown that the repair of cisplatin-damaged plasmid DNA is significantly reduced in plants with depleted AtRAD1/ERCC1 activity, a likely homolog of mammalian ERCC1, a key NER component (Li et al., 2002). To evaluate the effect of DNA damage on osmotic stress tolerance, we treated WT seedlings with mock or cisplatin for 6 h and then exposed them to osmotic stress. Sensitivity to osmotic stress was higher in cisplatin-treated plants than in mock-treated plants ( Figure 4B ), indicating that DNA repair plays a role in the osmotic stress response in Arabidopsis.

Next, we investigated whether osmotic stress causes double- or single-strand DNA breaks and whether DNA damage accumulation differs between WT and aod12. Salt-acclimated seedlings were exposed to 750 mM sorbitol for 10 days, and the accumulation of DNA breaks was monitored. Osmotic stress induced considerable DNA damage, but no significant differences were observed between WT and aod12 ( Figure 6 ). These findings suggest that, although the UVH6 mutation in aod12 impairs DNA repair, it does not significantly affect the repair of osmotic stress–induced DNA damage.

In the Arabidopsis NER pathway, UVH6, along with XPF, XPG, and ERCC1, plays a role in damage-induced cleavage at the 5′ or 3′ ends (Tuteja et al., 2009; Yoshiyama et al., 2009). To further investigate the relationship between NER components and osmotolerance, we examined the sensitivity of the NER-related mutants (Col-0 background) uvh6-1, xpf (T-DNA insertion), xpg (strong allele), and ercc1 (T-DNA insertion) to cisplatin-induced DNA damage stress. All mutants were hypersensitive to cisplatin, but sensitivity was greater in xpf, xpg, and ercc1 than in uvh6-1 ( Figure 7A ). However, uvh6–1 had an osmosensitive phenotype, whereas xpf, xpg, and ercc1 retained WT-like osmotolerance ( Figure 7B ). Among these mutants, only uvh6–1 was hypersensitive to L-heat stress ( Supplementary Figure 2 ). These results suggest that the decrease in osmotic and L-heat stress tolerance caused by the mutation in UVH6 is independent of its DNA repair function.

Given that UVH6 has been implicated in heat stress–induced transcriptional changes and the release of heterochromatin silencing in Arabidopsis (Bourguet et al., 2018), we performed RNA-seq analysis of WT and aod12 seedlings under normal and osmotic stress conditions, and identified differentially expressed genes (DEGs) ( Figure 7A and Supplementary Table 1 ). Gene Ontology (GO) enrichment analysis revealed that osmotic stress response genes were enriched in WT-specific DEGs ( Figure 7B ), suggesting a partial reduction in osmotic stress response at the transcriptional level in aod12. ORG1, WRKY54, and UGT71B6 belong to the GO term of osmotic stress response, and qRT-PCR confirmed that they were expressed at lower levels in aod12 than in WT under osmotic stress ( Figure 7D ). For aod12-specific DEGs, GO terms related to decreased oxygen levels and defense responses were enriched ( Figure 7C ). To determine whether the immune response is activated in aod12 plants, we analyzed the transcript levels of pathogenesis-related (PR) genes (PR1, PR2, and PR5) in aod12 under osmotic stress. To further illustrate the altered gene expression in aod12, heatmaps of genes belonging to the “response to stress” and “defense response” GO terms, as identified in Figure 7C , are presented in Supplementary Figures 3A, B , respectively. The PR transcript levels were significantly higher in aod12 than in WT under osmotic stress, in line with increased cell death ( Figures 7E, F ).

UVH6 is required for efficient transcription at numerous loci under heat stress, as uvh6 mutants exhibit reduced transcript accumulation at genome-wide loci, including heterochromatin regions, under heat stress (Bourguet et al., 2018). We compared relative transcript accumulation at the chromosomal level between aod12 and WT under normal and osmotic stress conditions. Transcript accumulation was slightly higher in aod12 than in WT across the genome (log2 fold change < 0.5) under normal conditions ( Supplementary Figure 4 ), and did not differ significantly under osmotic stress ( Supplementary Figure 4 ).

To determine whether the reduced osmotolerance of uvh6 mutants was due to an enhanced immune response or impaired DNA repair, we analyzed PR gene expression under osmotic stress in the NER-related mutants (Col-0 background) uvh6-1, xpf, xpg, and ercc1. Under osmotic stress, the PR transcript levels were upregulated only in the uvh6–1 mutant ( Figure 8A ) and the ORG1 and WRKY54 transcript levels increased or tended to increase in WT and all mutants except uvh6-1 (for ORG1) or except uvh6–1 and xpg (for WRKY54) ( Figure 8B ). These results suggest that the impaired osmotolerance in UVH6 mutants is associated with an enhanced immune response and reduced expression of stress-response genes rather than defects in DNA repair.

While the precise molecular mechanism by which UVH6 regulates transcriptional responses and immune activation under osmotic stress remains to be fully elucidated, we sought to investigate the involvement of key immune regulators in the UVH6-mediated immune response. Given that heightened immune responses mediated by EDS1/PAD4 are known to compromise osmotic tolerance (Ariga et al., 2017; Kanamori et al., 2023), we generated a pad4 aod12 double mutant using genome editing to test if PAD4 is involved in the immune activation observed in aod12 ( Supplementary Figure 5A ). Upon assessing the osmotic tolerance of the pad4 aod12 double mutant, we found that it displayed a hypersensitive phenotype to osmotic stress comparable to the aod12 single mutant, when compared to the wild type ( Supplementary Figure 5B ). This striking result suggests that the immune activation induced by the UVH6 mutation under osmotic stress is mediated by a signaling pathway that is independent of PAD4.

Given that UVH6 is an essential component of the TFIIH complex, crucial for plant viability and involved in transcription initiation, we sought to investigate the potential structural and functional impact of the aod12 and uvh6–1 mutations on UVH6 protein. We first performed a multiple sequence alignment and found that the R296 residue, which is deleted in the aod12 mutant, is highly conserved across various species, including mammals and other plants ( Supplementary Figure 6A ), suggesting its functional importance. Next, we utilized AlphaFold3 to predict the three-dimensional structures of wild-type, aod12 mutant, and uvh6–1 mutant UVH6 proteins. To validate AlphaFold3’s predictive capability for Arabidopsis XPD (UVH6), we first generated its three-dimensional structure and compared it with previously reported crystal structures of human XPD, which is highly conserved across species from humans to archaea (Kokic et al., 2019). The predicted Arabidopsis XPD structure showed good agreement with the known human XPD crystal structures, suggesting that AlphaFold3 can reliably predict the three-dimensional structure of XPD ( Supplementary Figure 6B, C ). Subsequently, we predicted the three-dimensional structures of wild-type Arabidopsis UVH6, aod12 mutant type, and uvh6–1 mutant type using AlphaFold3. Our predictions revealed that the aod12 mutant type, which has a one-amino acid deletion in its Arch domain, is predicted to adopt an aberrant tertiary structure in this domain compared to the wild type ( Supplementary Figure 6D ). Interestingly, although the uvh6–1 mutant type carries a one-amino acid substitution (glycine to glutamate) in the Helicase domain 2 (Bourguet et al., 2018), not the Arch domain, it too is predicted to exhibit an abnormal tertiary structure in the Arch domain, similar to the aod12 mutant type ( Supplementary Figure 6D ). The Arch domain is known to be a crucial interaction site with MAT1, a core component of TFIIH (Peissert et al., 2020). We hypothesize that the structural aberrations in the Arch domain in both aod12 and uvh6–1 mutants may hinder their proper interaction with MAT1, leading to an abnormal TFIIH complex structure and consequently impacting transcription.

UV RESISTANCE LOCUS 8 (UVR8) is an UV-B receptor; PR 1 and 5 proteins are induced more rapidly and to a higher extent in uvr8–1 mutant than in WT (Kliebenstein et al., 2002). To investigate whether the enhanced immune response under osmotic stress caused by the UVH6 mutation is mediated by UVR8, we examined the osmotolerance of the uvr8 mutant. Its osmotolerance was similar to that of WT, and it had no osmosensitive phenotype observed in uvh6-1 (Supplementary Figure 7A ). The L-heat tolerance of uvh6–1 was also similar to that of WT ( Supplementary Figure 7B ), suggesting that the immune response triggered by the UVH6 mutation under osmotic stress is independent of the UVR8 pathway.

Immune responses are activated by DNA damage via SUPPRESSOR OF GAMMA RESPONSE1 (SOG1) (Yoshiyama et al., 2020) Double mutations in SOG1 and XPF suppress the hypersensitivity to gamma rays but do not affect the UV sensitivity observed in XPF single mutants (Preuss and Britt, 2003). To examine the role of SOG1 in UVH6-dependent osmotolerance, we generated a uvh6–1 sog1–1 double mutant in the Col-0 background and compared its osmotolerance with that of WT, uvh6-1, and sog1-1. That of sog1–1 was similar to that of WT, while the double mutant was hypersensitive to osmotic stress, similar to uvh6-1 ( Supplementary Figure 7C ), suggesting that the immune response induced by the UVH6 mutation under osmotic stress is independent of SOG1-mediated immune regulation. To further confirm this, we investigated the expression of the immune-related genes PR1 and PR2 under osmotic stress. As shown in Supplementary Figure 7D , the expression of both PR1 and PR2 was higher in the uvh6–1 and uvh6–1 sog1–1 double mutant compared to WT. Notably, the expression levels in the double mutant were even higher than those in the uvh6–1 single mutant, indicating that the sog1–1 mutation does not suppress the osmotic stress-induced expression of these immune genes. Collectively, these results indicate that the immune response pathway induced by the UVH6 mutation under osmotic stress is independent of UVR8 and SOG1 signaling.