The transgenic Col-gl (Col-0 containing the glabrous mutation) line R1Y4 expressing RPW8.1-YFP from Ma et al. (2014) was used for EMS mutagenesis (Zhao et al., 2015). The b3-17 mutant was isolated for its dwarf phenotype. The xct-5 mutant was derived from crossing b3-17 with Col-gl. All seeds were sowed on 1/2 (W/V) Murashige Skoog (MS) basal media containing appropriate antibiotics and treated at 4°C for 2 days. Seedlings were transplanted into peat soil (Pindstrup, Beijing) after appearance of true leaves and were put in a growth chamber under the conditions of 23°C, 75% relative humidity, short-day (10 h light, 14 h dark) for vegetative growth or long-day (14 h light, 10 h dark) for induction of flowering or after inoculation of powdery mildew.

In order to generate enough polymorphism for linkage analysis of the mutated locus in b3-17 that was in Col-gl background, we constructed an F₂ segregating population by crossing b3-17 with Landsberg erecta (Ler). Then the individuals displaying mutant phenotype were selected for gene mapping. The simple sequence length polymorphism (SSLP) markers and new SNP markers (Supplementary Table S1) were used for initial mapping. After the mutation was mapped between the markers F26H11 and F7O24, the candidate genes were sequenced leading to the identification of a G-to-A mutation at the 3′ splice site of the 8th intron of XCT (At2g21150) (Supplementary Figure S1).

For complementation test, we amplified the wild type XCT containing the 2090-bp sequences upstream of the start codon and the 522-bp sequences downstream of the stop codon with the primers EcoRXCT-F and KpnXCT-R. We chose this region as the wild type XCT because the upstream sequences contain a predicted transcriptional start site and may function as the native promoter of XCT as reported in a previous paper (Martin-Tryon and Harmer, 2008) and the downstream sequences contain the 3′-untranslated region (UTR). The amplified fragment was cloned into the binary vector pCAMBIA1300 at the EcoRI/KpnI site generating the plasmid pP_XCT:XCT. The plasmid was introduced into b3-17 and xct-5 via Agrobacterium-mediated floral dip according to a previous approach (Clough and Bent, 1998). Positive transformants were screened on 1/2 (W/V) MS basal media containing 35 mg/L of Hygromycin B (Roche, United States).

To make transgenic lines with reduced expression of XCT through artificial microRNA (amiRNA), XCT specific primers were designed with the WMD Web MicroRNA Designer¹. The construct expressing amiRXCT targeting the 3′-UTR of XCT was made according to Schwab et al. (2006) using template plasmid pRS300 and primers XCTmiR-sI, XCTmiR-aII, XCTmiR^∗sIII, and XCTmiR^∗aIV (Supplementary Table S1) (Schwab et al., 2006). The resulting amiRNA was cloned into pKANNIBAL-35S-RBS (courtesy of Yuelin Zhang) at EcoRI /BamHI site generating the plasmid pamiRXCT. To make transgenic lines over-expressing XCT, the coding sequences of XCT was amplified with the primers Kpn-oeXCT-F and Kpn-oeXCT-R (Supplementary Table S1), and cloned into the binary vector pCAMBIA1300-35S at KpnI site leading to the plasmid p35S::XCT. The constructs were transformed into the agrobacterium strain GV3101 together with the helper plasmid pSOUP and then introduced into R1Y4 and Col-gl, respectively, via Agro-mediated floral dip (Clough and Bent, 1998). Positive transformants were screened on 1/2 MS media containing 35 mg/L of Hygromycin B, and the expression level of XCT was tested by quantitative reverse transcription PCR (qRT-PCR).

Total RNA was extracted from 80 mg leaves with TRIzol^® Reagent (Invitrogen). For time course analysis, leaves were collected at 0, 3, 6, 12, 24 h after treatment with 2 μM of flg22. For qRT-PCR assay, 500 ng total RNA was used for cDNA synthesis using ReverTra Ace^® qPCR RT Master Mix with gDNA Remover (Toyobo). Quantitative RT-PCR was performed with gene-specific primers (Supplementary Table S1) and QuantiNova^TM SYBR^® Green PCR Reagent (Sigma) in Bio-Rad CFX96^TM Real-Time system. ACT2 was used as an internal reference. Quantification of fold change was calculated by the 2[-Delta Delta C(T)] Method (Livak and Schmittgen, 2001). Statistical analysis was performed by t-test or by a one-way ANOVA followed by post hoc Tukey HSD analysis. Quantitative data were processed using Sigma Plot 10.0.

Powdery mildew isolate Golovinomyces cichoracearum UCSC1 was maintained on live pad4-1 sid2-1 plants. Inoculation, visual scoring, and quantification of disease susceptibility were done as described previously (Xiao et al., 2003). For bacterial proliferation assay, leaves of 5-week-old plants were infiltrated with the virulent strain P. syringae DC3000 (OD₆₀₀ = 0.0005), and the non-pathogenic strain P. syringae DC3000(hrcC^-) (OD₆₀₀ = 0.002) according to a previous report (Li et al., 2017). Bacterial growth was determined by colony counting as previously described (Zipfel et al., 2004).

Dying cells in inoculated leaves were analyzed by trypan blue (TB) staining at 10 days post inoculation (dpi) of powdery mildew using a method modified from Xiao et al. (2003). Briefly, inoculated leaves were soaked in TB solution, and boiled for 2 min, then de-stained using Chloral Hydrate (Sigma). For examination of H₂O₂ production by 3,3′-diaminobenzidine (DAB) staining, inoculated leaves were cut at petiole, submerged in 1 mg /ml of acidic DAB solution (Sigma), vacuum-infiltrated for 3 min at 0.2 gk/cm², incubated overnight in dark, and destained using cleaning solution. Images were acquired under Canon EOS Rebel T2i and Zeiss Imager A2.

For microscopy of the accumulation of RPW8.1-YFP, leaves from 5-week-old plants were examined under a laser scanning confocal microscope (Nikon A1) according to a previous report (Huang et al., 2014). All confocal graphs presented in the manuscript were two-dimensional projections of 10–30 Z-stack images. Images were processed using Image J, NIS Elements C and/or Adobe Photoshop CS5.

Total protein was extracted from 200 mg fresh leaves in extraction buffer supplemented with 100 mM PMSF, 1% (V/V) Triton X-100 following a previous report (Wang et al., 2012). After separation by 10% SDS-PAGE, the denatured protein was transferred to Immobilon^®-P Transfer Membrane (Millipore), and subjected to blot analysis using polyclonal anti-sera of GFP (1:2000) (BBI) and ClarityTM Western ECL Substrate system (BIO RAD) to detect RPW8.1-YFP.

Previously, we found that ectopic expression of RPW8.1 leads to enhanced resistance against different pathogens via boosting pattern-triggered immunity (Ma et al., 2014; Li et al., 2017). In order to identify regulators for RPW8.1-mediated immunity, we performed a forward genetic screen and isolated the b3-17 mutant that exhibited compromised cell death and smaller plant stature in comparison with R1Y4 (Figure 1A). The b3-17 mutant also exhibited susceptibility to powdery mildew as indicated by the white fungal mass on the leaf surface (Figure 1B). Quantification of spores at 10 dpi confirmed the susceptibility of b3-17 to powdery mildew, because the sporulation on b3-17 was significantly higher than that on R1Y4, although lower than that on Col-gl (Figure 1C). These results demonstrated the impairment of RPW8.1-mediated resistance against powdery mildew in the b3-17 mutant.

The b3-17 mutation was initially mapped to the chromosome 2 between the two simple sequence length polymorphism markers F6F22 and F2G1 (Supplementary Figure S1A). By using additional markers, the mutation was mapped to a 10.7-kb region between the markers F26H11 and F7O24. Finally, sequencing candidate genes in this region identified a G-to-A mutation at nucleotide 2016 of XCT (At2g21150) in b3-17, which occurred at the 3′ splice site of intron 8 (Figure 2A and Supplementary Figure S1B). Further sequencing XCT cDNA from b3-17 revealed that this mutation resulted in a splice shift of eight nucleotides downstream of the original splice junction (Figure 2A and Supplementary Figure S1C), generating a mRNA encoding a truncated XCT lacking the C-terminal 79 amino acid residues that covers one-third of the conserved X-chromosome Associated Protein 5 (XAP5) domain (Figure 2B and Supplementary Figure S1D).

To confirm that the b3-17 mutant phenotypes were due to the point mutation in XCT, we introduced the wild type XCT gene containing the 2090-bp sequences upstream of the start codon and 522-bp sequences downstream of the stop codon into b3-17. This wild type XCT includes the native promoter and 3′-UTR (Martin-Tryon and Harmer, 2008). The obtained >20 transgenic plants were all restored to the phenotypes of R1Y4, including plant size and defense against powdery mildew (Figures 2C,D). These data indicate that the b3-17 contains a novel allele of XCT. Because there are four xct mutant alleles reported previously (Martin-Tryon and Harmer, 2008; Fang et al., 2015), we designated this allele xct-5 and renamed b3-17 as R1Y4/xct-5.

XCT is a highly conserved protein across eukaryotes, and XAP5 domain at the C-terminus is the key functional domain. Therefore, truncation of this domain in xct-5 mutation obviously affects its function. To examine how the xct-5 mutation influences the function of RPW8.1, we measured some defense responses. First, we checked the distribution of cell death and production of H₂O₂ upon powdery mildew infection. Trypan blue staining of the inoculated leaf showed that clusters of dead cells were formed upon powdery mildew infection. The clusters of dead cells in R1Y4/xct-5 were obviously less than those in R1Y4 at 10 dpi (Figure 3A). Consistently, DAB staining revealed that H₂O₂ accumulation in R1Y4/xct-5 was less than that in R1Y4 (Figure 3B).

Then, we examined the defense responses induced by flg22. The results indicated that flg22-induced callose deposition was remarkably decreased in R1Y4/xct-5 compared to that in R1Y4 (Figure 3C). We also examined the induced levels of defense-related genes, including FLG22-INDUCED RECEPTOR-LIKE KINASE 1 (FRK1, At2g19190) and the PATHOGENESIS - RELATED1 (PR1, At2g19990), by qRT-PCR. The level of FRK1 was induced in R1Y4 upon application of flg22 and reached a summit at 6 h post application (hpa), then dropped back to relative lower levels at 12 and 24 hpa (Figure 3D). Similarly, the level of FRK1 was also induced in R1Y4/xct-5 upon application of flg22, but the amplitudes were significantly lower than those in R1Y4 at 3, 6, and 12 hpa (Figure 3D). The expression of PR1 was induced and reached a summit at 24 hpa upon application of flg22 in both R1Y4 and R1Y4/xct-5, notably, its levels in R1Y4/xct-5 were lower than those in R1Y4 at 0, 3, and 24 hpa (Figure 3D). These data indicate that RPW8.1-mediated defense responses were compromised by xct-5 mutation.

Previously, we found that R1Y4 displayed enhanced resistance to the virulent bacterial strain P. syringae DC3000 and limited the proliferation of the non-pathogenic strain P. syringae DC3000 (hrcC^-) (Li et al., 2017). We therefore tested the response of R1Y4/xct-5 to these strains. Our data indicated that the proliferation of both P. syringae DC3000 and P. syringae DC3000 (hrcC^-) in R1Y4/xct-5 was significantly higher than that in R1Y4 and became comparable to that in Col-gl (Figure 3E). Together these data indicate that XCT is required for RPW8.1-mediated defense responses and resistance to different pathogens.

To confirm that xct-5 is a loss of function allele, we made a construct expressing an artificial microRNA that targets the 3′-UTR of XCT (amiRXCT, Figure 2A). The construct was introduced into R1Y4. Four lines in R1Y4 background exhibited remarkable reduction of XCT identified by qRT-PCR from 23 positive transformants (Figure 4A). The expression levels of XCT in the four lines were reduced to about 40% of that in R1Y4 (Figure 4A). Two of them, i.e., R1Y4/amiRXCT-3 and R1Y4/amiRXCT-7, were selected for further experiments. The leaf size of R1Y4/amiRXCT lines was comparable with that of R1Y4, however, the cell death lesions were obviously less severe in R1Y4/amiRXCT lines than those of R1Y4 (Figure 4B). After inoculation with powdery mildew pathogen, the white fungus mass in the R1Y4/amiRXCT lines were obviously more than that in R1Y4, whereas, less than that in Col-gl (Figure 4C). Quantification analysis on spore numbers showed that the number of spores in R1Y4/amiRXCT lines was increased by about 2∼3 folds of that in R1Y4, but significantly lower than that in Col-gl (Figure 4D). These data indicate that the expression level of XCT is important for RPW8.1-mediated resistance to powdery mildew. Next, we examined the pathogen-induced cell death by trypan blue staining at 10 dpi of powdery mildew pathogen. The cell death lesion on the whole infected leaves in R1Y4/amiRXCT lines was obviously decreased compared to that in R1Y4 (Figures 4E,F). Consistently, the production of H₂O₂ revealed by DAB-staining was less in the R1Y4/amiRXCT lines than in R1Y4 (Figures 4G,H). We also examined flg22-induced callose deposition and the expression of FRK1 and PR1 in R1Y4/amiRXCT lines. The results indicated that callose deposited in R1Y4/amiRXCT lines was significantly less than those in R1Y4 (Figure 5A). Whereas, the basal level of FRK1 in R1Y4 and the R1Y4/amiRXCT lines was significantly higher than that in Col-gl, and also significantly higher in one of the R1Y4/amiRXCT lines than that in R1Y4 at 0 hpa (Figure 5B). At 3 hpa, the level of FRK1 became comparable in R1Y4/amiRXCT lines and R1Y4, but lower in R1Y4/amiRXCT lines than in R1Y4 at 6 hpa (Figure 5B). At 12 hpa, however, FRK1 levels in the R1Y4/amiRXCT lines became higher than that in R1Y4 (Figure 5B), implying that in the R1Y4/amiRXCT lines the induction of FRK1 by flg22 was delayed. Consistently, the level of PR1 in R1Y4/amiRXCT lines was significantly lower than that in R1Y4 at 12 hpa, although in one of the R1Y4/amiRXCT lines was comparable with that in R1Y4 at 0, 6, and 24 hpa (Figure 5C). These results suggest that lower expression of XCT in R1Y4 compromised RPW8.1-mediated defense responses.

Now that mutation or down-regulation of XCT led to impairment of RPW8.1-mediated immunity, it was anticipated that over-expression of XCT should increase resistance. To this end, we made a construct overexpressing XCT from the constitutive 35S promoter (OEXCT). The construct was introduced into R1Y4 via Agrobacterium-mediated floral dip. More than 20 positive transformants were obtained and the expression level of XCT was examined by qRT-PCR (Figure 6A). Two lines (R1Y4/OEXCT-3 and R1Y4/OEXCT-5) were selected for further experiments. The overexpression lines displayed enhanced cell death in comparison with R1Y4 in the absence of pathogen (Figure 6B), indicating than XCT positively regulated RPW8.1-mediated cell death. This observation prompted us to check if the mildew resistance in these lines was enhanced. Five-week-old plants were inoculated with powdery mildew and the disease phenotype was recorded at 10 dpi. While R1Y4 showed resistance in comparison with the noticeable susceptibility of Col-gl, there was hardly mycelium observed in R1Y4/OEXCT-3 and R1Y4/OEXCT-5 (Figure 6C). By quantitative analysis, the amount of spores in overexpression lines was significantly reduced than that in R1Y4 and Col-gl, respectively (Figure 6D). We also checked the spreading of cell death and accumulation of H₂O₂ in the overexpression lines. The clusters of dead cells in inoculated leaves of overexpression lines were obviously more than that of R1Y4 (Figures 6E,F). Similar to the cell death, overexpression lines displayed much more H₂O₂ production than R1Y4 (Figures 6G,H). Moreover, flg22-induced callose deposition was enhanced in R1Y4/OEXCT-3 and R1Y4/OEXCT-5 (Figure 7A). Flg-22 induced levels of FRK1 were higher in R1Y4/OEXCT-3 and R1Y4/OEXCT-5 than that in R1Y4 at 3 hpa (Figure 7B). Flg-22 induced levels of PR1 were also higher in R1Y4/OEXCT lines than that in R1Y4 (Figure 7C). Consistently, the proliferation of Pst DC3000 was reduced significantly in R1Y4/OEXCT-3 and R1Y4/OEXCT-5 in comparison with that in R1Y4 (Figure 7D). Taken together, these results demonstrated that up-regulation of XCT enhanced RPW8.1-mediated disease resistance and defense responses.

Because mutation and down-regulation of XCT results in impairment, but overexpression of XCT leads to enhanced RPW8.1-mediated disease resistance, we asked whether XCT is directly involved in defense. To this end, we obtained xct-5 mutant in Col-gl background by crossing b3-17 to Col-gl and introduced the wild type XCT into xct-5 to obtain the complemented lines with restored phenotypes (Figure 8A). Then we examined their responses to powdery mildew and two bacterial strains. Intriguingly, all these lines exhibited similar disease phenotypes of powdery mildew (Figures 8B–D). Furthermore, we also made transgenic lines expressing artificial microRNA or a 35S::XCT construct in the Col-gl background, and surveyed for their defense responses. After examining the expression levels of XCT in these transgenic lines by qRT-PCR (Supplementary Figure S2), we selected two knocked-down lines (amiRXCT-54 and amiRXCT-58) and two overexpression lines (XCTOE-11 and XCTOE-14) for analysis on disease resistance. When inoculated with powdery mildew, the fungal mass in both knocked-down lines and overexpression lines was quite similar to that in Col-gl, indicating that these transgenic lines were as susceptible as Col-gl against powdery mildew (Figures 8C,D). Moreover, the proliferation of Pst DC3000 in xct-5 mutant and the transgenic lines showed no difference from Col-gl (Figure 8E). In contrast, the proliferation of Pst DC3000 (hrcC^-) was slightly higher in xct-5 mutant and the knocked-down lines than that in Col-gl and the overexpression lines, but the differences were not significant (Figure 8F). These data demonstrate that XCT alone may not be directly involved in defense against powdery mildew and P. syringae.

It seems that XCT specifically and positively regulates RPW8.1-mediated resistance. From the cell death phenotype we speculated that the expression of RPW8.1 may be altered by mutation or overexpression of XCT. To this end, we examined the mRNA level of RPW8.1. The transcription of RPW8.1 in R1Y4/xct-5 was reduced to about 30% of that in R1Y4 (Figure 9A). When the expression levels of XCT were down-regulated in R1Y4/amiRXCT-3 and R1Y4/amiRXCT-7 to about 20–40% of that in R1Y4, the levels of RPW8.1 were decreased to ∼50% and 30% of that in R1Y4, respectively (Figure 9B). On the contrary, when the expression of XCT was increased in R1Y4/OEXCT-3 and R1Y4/OEXCT-5 by about 19- and 4-fold of that in R1Y4, the transcription of RPW8.1 was increased to about 2- and 3-fold of that in R1Y4, respectively (Figure 9C). These results suggest that XCT positively regulate RPW8.1 expression at transcriptional level.

To test if the accumulation of RPW8.1-YFP was consistent with its relative mRNA levels examined by qRT-PCR, we examined the fluorescent intensity of RPW8.1-YFP by Laser Scanning Confocal Microscopy (LSCM). The data showed that the RPW8.1-YFP signal was rarely found in leaves from R1Y4/xct-5 and R1Y4/amiRXCT-7 plants, whereas, the RPW8.1-YFP signal was obviously more in R1Y4/OEXCT-5 than in R1Y4 (Figure 9D). Quantification analysis of the YFP signal intensity confirmed that the accumulation of RPW8.1-YFP in R1Y4/xct-5 and R1Y4/amiRXCT-7 was significantly lower than that in R1Y4, whereas the accumulation of RPW8.1-YFP in the overexpression line was significantly higher than that in R1Y4 (Figure 9E). We next checked the expression of RPW8.1-YFP by western blotting analysis using GFP polyclonal antibody that could also detect YFP. We found that RPW8.1-YFP expression was higher in OE lines by ∼2 folds compared to that in R1Y4, whereas the expression of RPW8.1-YFP was barely detected in R1Y4/xct-5 and reduced in R1Y4/amiRXCT-7 (Figure 9F). These results suggest that XCT positively regulate RPW8.1 expression.

Now that XCT can regulate the transcription of RPW8.1, we asked whether RPW8.1 can also affect the expression of XCT. To this end, we compared the expression of XCT in R1Y4 and Col-gl. Intriguingly, the expression level of XCT in R1Y4 was increased to ∼2-fold of that in Col-gl (Figure 9G), indicating that the expression of RPW8.1 can also up-regulate the expression of XCT.

Therefore, it appears that XCT and RPW8.1 can mutually enhance each other’s transcription.