Arabidopsis mutant line1-12 (esd4-3) was obtained from Agrobacterium-mediated genetic transformation. esd4-4 and fmo1esd4 were generated by CRISPR-cas9 gene editing system (Wang, 2015). eds1-2 and fmo1 were previously described (McDowell et al., 2000; Mishina and Zeier, 2006). All the mutants used in this study are in Col-0 ecotype. Col-0 was used as wild-type control and Ler ecotype was used for generating a mapping population with line1-12. As esd4 mutants are sterile, the heterozygous esd4 plants were used for producing offspring. The detailed information of mutants used in this study are summarized in Supplementary Table S1. All the primers used for mutant identification are listed in Supplementary Table S2.

For soil-grown plants, seeds were vernalized at 4°C for 2 days, sown onto sterile soil, and transferred to plant growth rooms under 16-h day/8-h night cycle, 22°C, and 65 ± 10% humidity conditions or 16-h day/8-h night cycle, 28°C, and 65 ± 10% humidity conditions. For selecting T1 transgenic plants, sterilized T1 seeds were sown in culture dishes containing 1/2MS medium with hygromycin (40 μg/ml) in growth chamber under 16-h day/8-h night cycle, 22°C and 65 ± 10% humidity conditions.

For map-based cloning, line1-12 homozygote (in Col background) was crossed with Ler (Landsberg-erecta) to generate F1 seeds. Then F1 plants were self-fertilized and the resulting F2 population was used for crude mapping. 64 plants exhibiting dwarfed morphology were picked out from F2 progenies and used for linkage analysis. The markers used for crude mapping were designed using the Monsanto Arabidopsis polymorphism and Ler sequence collections and listed in Supplementary Table S2 (Jander et al., 2002). After the mutation was narrowed down to a region of about 4.51 Mb between markers T4C9 (6.5 Mb) and F9F13 (11.01 Mb) on Chromosome 4, whole-genome resequencing was performed to identify mutations within this region. 1 g tissue from 52 F2 dwarfed plants from a backcross population between line1-12 and Col-0 was collected for genomic DNA extraction following a previously described procedure (Huang et al., 2019). The purified DNA was sequenced using Illumina whole-genome resequencing and the indel mutation in AT4G15880 was confirmed by Sanger sequencing using flanking primers listed in Supplementary Table S2.

The CRISPR/Cas9 constructs for deleting ESD4 were made following the protocol described previously (Wang, 2015). Briefly, ESD4 deletion target primers were designed using the CRISPR-PLANT website.¹ The ESD4 CRISPR cassette was generated by PCR amplification using pCBC-DT1T2 as template with designed primer pairs. The PCR products were digested with BsaI and cloned into binary vector pHEE401E, yielding ESD4-pHEE401E. The ESD4-pHEE401E construct was transformed into Col-0 or fmo1 plants by Agrobacterium-mediated floral-dip transformation (Clough and Bent, 1998). Genotyping primers flanking each target site were used to select T1 plants that harbored deletions in ESD4. Sanger sequencing was performed to define the ESD4 deletion. All the primers used are listed in Supplementary Table S2.

For oomycete pathogen Hyaloperonospora arabidopsidis (H.a.) Noco2 infection assays, seeds of the indicated genotypes were planted on soil and grown in a growth room under a 16 h day/8 h night cycle, 22°C and 65 ± 10% humidity. Two-week-old seedlings (3 pots for each genotype and 12 plants in each pot) were spray-inoculated with freshly harvested H.a. Noco2 conidiospores at a concentration of 1×10⁵ spores per mL sterile water. Infected plants were covered with a plastic lid and kept in a growth chamber under 12-h light/12-h dark cycle, 18°C, and 80% humidity. Seven days later, 1 g plant tissue of wild-type or esd4 mutant from each pot was collected and suspended in 2 ml sterile water. After vortexing, the conidiospores were counted using a hemocytometer under microscope.

Trypan Blue staining was performed following a previously described procedure (Zhang et al., 2012). Specifically, true leaves were transferred into 1.5-ml microcentrifuge tubes containing 1 ml lactophenol Trypan Blue solution (10 mg Trypan Blue, 10 g phenol, 10 ml lactic acid, 10 ml glycerol, and 10 ml water) diluted 1:1 in ethanol. After boiling for 2 min, removed staining solution and added 2 ml chloralhydrate solution (2.5 g/ml) on an orbital shaker for 2 h. Samples were further de-stained overnight with new chloralhydrate solution before examination by a light microscope.

DAB staining procedure was performed according to previously described procedures with minor modifications (Daudi and O’Brien, 2012). Briefly, true leaves were transferred into 2.5 ml micro centrifuge tubes containing 2 ml DAB solution (1 mg/ml DAB, pH 3.8). After overnight shaking on an oscillator, DAB solution was removed. 2 ml de-staining solution (70% ethanol) was added for de-colorization. Photos were taken under a light microscope afterwards.

Total RNA was extracted from 3-week-old soil-grown plants (50 mg) using Eastep plant RNA extraction kit (Promega, Ls1040). 2 μg RNA was reverse-transcribed to cDNA using Go Script reverse transcriptase (Promega, A5001). 50 ng cDNA was added as template in a 10 μl reaction on an ABI Step one Real-Time system machine. Real-time PCR was performed using a SYBR-Green PCR HS kit (AG, AG11701). qRT-PCR was carried out as described previously (Zhang et al., 2003). ACTIN1 was used to normalize the expression value. The primers used are listed in Supplementary Table 2.

With the arrival of better CRISPR-Cas9 gene editing methods in different organisms, knocking out redundant genes has become easier than ever. Designing two guide RNAs (gRNAs) targeting the flanking genes in a tandem array would yield a deletion to knock out these duplicated genes for reverse genetic analysis. Here, we tried to use this system with two gRNAs to knock out a tandem array of 20 genes encoding redundant receptor-like kinases (RLKs) from At4g23130 to At4g23320 in the Col-0 ecotype (Wang, 2015). In the T1 generation, among 114 plants tested, we did not detect any deletion products by PCR using primers flanking the two gRNA sites. However, an early flowering dwarfed plant that exhibited abnormal leaf phenotype and sterility was recovered and designated as line1-12. Interestingly, line1-12 resembled mutants with autoimmune defects (Figure 1A). When the defense marker genes Pathogenesis-Related 1 (PR1) and PR2 were examined by quantitative real-time PCR (qRT-PCR) in line1-12, heightened expression was observed (Figures 1B,C). Furthermore, cell death and accumulated H₂O₂ were observed in line1-12 using trypan blue staining and 3,3′-diaminobenzidine staining (Figures 1D,E), respectively. As Col is susceptible to H.a. Noco2, this model pathogen on Arabidopsis was used to test for enhanced immunity. When challenged with virulent pathogen H.a. Noco2 (Figure 1F), line1-12 showed less oomycete spores growth compared with wild-type plants. All these data indicate the autoimmunity of line1-12. We predicted that the mutation in this mutant should reveal a novel negative regulator of immunity.

Using map-based cloning, we tested if the mutation responsible for the autoimmunity of line1-12 is linked to the targeted loci by CRISPR-Cas9. When line1-12 was crossed with Ler, 64 dwarfed plants were observed in total 261 F2 population, suggestive of a single recessive mutation. Indeed, the mutation was mapped to chromosome 4 (Figure 2A), where the originally targeted RLK genes are linked to. After failing to find point or small insertion/deletion (Indel) mutations in the target genes through Sanger sequencing, we reasoned that there could be complex rearrangements in the locus. Whole-genome sequencing (WGS) was therefore conducted on the ¼ co-segregants that were dwarfed and flowered earlier in the Col-backcrossed F2 generation. Sequence analysis identified only one mutation, which is an exonic deletion of 49 bp in At4g15880 in the mapped region (Figure 2A).

At4g15880 encodes EARLY IN SHORT DAYS 4 (ESD4), which was reported to exhibit dwarfism and early flowering with abnormal leaves and sterility (Reeves et al., 2002; Pineiro et al., 2003; Xu et al., 2007). In addition, esd4 plants also have increased levels of defense hormone salicylic acid (SA; Villajuana-Bonequi et al., 2014), a well-known defense hormone. Thus, we concluded that the mutant line1-12 is an allele of esd4. As two alleles of esd4 were reported in the past, we named line1-12 as esd4-3. When the flanking sequences of the deletion in ESD4 were examined carefully, we did not observe any similarity with the original gRNA sequences, suggesting that the deletion in ESD4 might be due to CRISPR-Cas9 independent causes, likely through Agrobacterium-mediated mutations.

To further confirm the autoimmune phenotypes of esd4-3, we generated another allele of esd4, named esd4-4, by CRIPSPR-Cas9. esd4-4 which contains a 2,772 bp deletion in ESD4 (Supplementary Figure S1A), is also much smaller than wild-type plants and exhibits infertility phenotype under our growth condition (Figure 2B). Semi-qRT-PCR showed that the expression PR1 and PR2 were much higher in esd4-4 compared with that in wild-type (Figure 2C). Similar to esd4-3, esd4-4 showed enhanced resistance to H.a. Noco2 (Figure 2D). Taken together, our data further demonstrated that loss function of ESD4 leads to autoimmunity, indicating a negative role of ESD4 in plant immunity.

Previous study showed esd4 mutants have increased level of defense hormone SA and SA biosynthesis mutant ics1 partially reduced the PR1 expression in esd4 plants (Villajuana-Bonequi et al., 2014), suggesting that the autoimmunity in esd4 partly relies on SA signaling and additional factors contribute to the autoimmune phenotypes in esd4. As plant signal molecules NHP and SA were reported to cooperatively regulate plant immunity (Sun et al., 2020), we examined whether NHP signaling pathway was involved in the autoimmunity of esd4. We generated esd4 fmo1 double mutant by transforming the CRISPR/Cas9 construct for ESD4 deletion into fmo1, a mutant defective in NHP biosynthesis. A new allele of esd4, harboring a deletion of 123 bp, was isolated in the fmo1 background in T1 generation (Supplementary Figure S1B). As shown in Figure 3A, the esd4 fmo1 mutant exhibits intermediate size compared to wild-type and esd4 plant. Upregulated PR1 and PR2 levels in esd4 are also largely suppressed by fmo1 (Figures 3B,C). In addition, more H.a. Noco2 conidiospores grew on the esd4 fmo1 double mutant compared with esd4 single mutant (Figure 3D). Taken together, these data showed that fmo1 partially suppresses the autoimmunity in esd4, suggesting that NHP signaling pathway also contributes to autoimmune responses activated in esd4 mutant.

To test how NHP signaling pathway contributes to the enhanced disease resistance in esd4 mutant, expressions of NHP biosynthetic genes were examined. As shown in Figures 4A–C, ALD1, SARD4, and FMO1 were all highly upregulated in esd4, suggestive of elevated NHP level in esd4. NHP biosynthetic genes are coordinately regulated by two master transcriptional regulators SARD1 and CBP60g in A. thaliana (Sun et al., 2015). Interestingly, both the expression of SARD1 and CBP60g are greatly induced in esd4 mutant (Figures 4D,E). SARD1 and CBP60g are also the main regulators of SA biosynthetic genes, including ICS1, EDS5, and PBS3 (Zhang et al., 2010; Wang et al., 2011; Sun et al., 2015). Previously, it was shown that esd4 accumulates much higher SA level compared to wild-type plant and ICS1 is upregulated in esd4 mutant under short-day condition (Villajuana-Bonequi et al., 2014). Agreeably, the expressions of other SA biosynthesis genes EDS5 and PBS3 are also upregulated in esd4 mutant (Figures 4F,G). Taken together, these data demonstrate that NHP and SA biosynthesis are upregulated in esd4, likely due to the elevated expression of SARD1 and CBP60g.

EDS1 is a central player downstream of many NLRs and upstream of SA and NHP biosynthesis (Zhang and Li, 2019; Huang et al., 2020). As ESD4 seems to act upstream of SA and NHP biosynthesis, we further examined its relationship with the upstream defense regulator EDS1. esd4-3 was crossed to loss-of-function mutant of EDS1, eds1-2 (Wagner et al., 2013) to generate eds1esd4 double mutant. As shown in Figure 5A, the dwarfism of esd4-3 was largely suppressed by eds1-2. Although autoimmunity often results in reduced fertility, eds1 only mildly restored esd4 infertility (Supplementary Figure S2). qRT-PCR showed that eds1-2 also fully blocked the upregulated PR1 and PR2 in esd4-3 (Figures 5B,C). In addition, analysis of resistance to H.a. Noco2 indicates that enhanced resistance to oomycete in esd4-3 was largely compromised by mutation in EDS1 (Figure 5D). Taken together, these data demonstrate that autoimmunity in esd4 largely relies on EDS1, indicating that loss function of ESD4 activates NLR-mediated immunity.

Because activation of NLRs can trigger SA and NHP biosynthesis, we then tested whether elevated SA and NHP biosynthesis in esd4 mutants is due to the activation of certain NLR(s). We measured expressions of SA and NHP biosynthetic-related genes in eds1esd4 double mutant. As shown in Figure 6, except for PBS3 and CBP60g, all the other highly expressed genes in esd4 are reduced to wild-type level by eds1, further verifying our hypothesis that esd4 activates NLR-mediated immunity.

Previous studies showed that the autoimmune phenotypes of many autoimmune mutants are temperature-dependent and high temperature often inhibits NLR-mediated signaling. To test whether the autoimmune phenotype observed in esd4 is temperature-dependent, we grew esd4 under 28°C conditions. As shown in Supplementary Figure S3, the dwarf morphology and the expression of defense marker genes PR1 and PR2 were largely suppressed at 28°C. This is consistent with our finding that activation of NLR(s) leads to autoimmunity in esd4.