Loss of the common immune coreceptor BAK1 leads to NLR-dependent cell death.

Plants utilize a two-tiered immune system consisting of pattern recognition receptor (PRR)-triggered immunity (PTI) and effector-triggered immunity (ETI) to defend themselves against pathogenic microbes. The receptor protein kinase BAK1 plays a central role in multiple PTI signaling pathways in Arabidopsis However, double mutants made by BAK1 and its closest paralog BKK1 exhibit autoimmune phenotypes, including cell death resembling a typical nucleotide-binding leucine-rich repeat protein (NLR)-mediated ETI response. The molecular mechanisms of the cell death caused by the depletion of BAK1 and BKK1 are poorly understood. Here, we show that the cell-death phenotype of bak1 bkk1 is suppressed when a group of NLRs, ADR1s, are mutated, indicating the cell-death of bak1 bkk1 is the consequence of NLR activation. Furthermore, introduction of a Pseudomonas syringae effector HopB1, which proteolytically cleaves activated BAK1 and its paralogs via either gene transformation or bacterium-delivery, results in a cell-death phenotype in an ADR1s-dependent manner. Our study thus pinpoints that BAK1 and its paralogs are likely guarded by NLRs.

Plant innate immunity is a two-tiered immune system composed of pattern recognition receptor (PRR)-triggered immunity (PTI) and effector-triggered immunity (ETI) (1). PTI confers plants basal defense that allows resistance to most invading pathogens, whereas ETI is more often associated with hypersensitive response (HR), a type of programmed cell death.

To activate PTI, the cell surface-localized PRRs interact with pathogen-associated molecular pattern (PAMP) and subsequently initiate intracellular immune responses (2, 3). Many PRRs identified so far are receptor protein kinases (RKs) (2, 4). For example, FLAGELLIN-SENSING 2 (FLS2) and EF-Tu RECEPTOR (EFR), two leucine-rich repeat RKs (LRR-RKs), recognize flg22, a 22-amino acid peptide conserved among bacterial flagellin, and elf18, an 18-amino acid peptide conserved in EF-Tu, respectively (5, 6). BRASSINOSTEROID INSENSITIVE 1 (BRI1)-ASSOCIATED RECEPTOR KINASE 1 (BAK1), a different LRR-RK, which was originally identified as a coreceptor of a brassinosteroid receptor BRI1 (7, 8), is able to interact with FLS2 or EFR when flg22 or elf18 is present (9, 10). BAK1, also named SOMATIC EMBRYOGENESIS RECEPTOR-LIKE KINASE 3 (SERK3), belongs to a SERK subfamily, which contains five members in Arabidopsis. Unlike FLS2 or EFR that possesses a large extracellular domain (ECD) for a direct PAMP association, BAK1 contains only five LRRs in its ECD and is able to only interact with the new surfaces formed via the interaction of PRRs and their corresponding PAMPs. BAK1, therefore, is also considered as a coreceptor for multiple LRR-type PRRs. Structural assays indicated the ECD of BAK1 can directly recognize FLS2 and the C terminus of FLS2-bound flg22 (11). flg22 thus serves as molecular glue connecting the ECDs of FLS2 and BAK1. Consistently, bak1 single mutants show significantly impaired flg22-mediated responses, indicating BAK1 is essential for PAMP perception (9, 10). In general, PAMPs induce the interaction and transphosphorylation of PRRs with their coreceptors, initiating the downstream PTI cascades (2, 4).

Some microbial strains can deliver specific proteins, effectors, into host cells to repress PTI via disrupting key components in PTI signaling (12, 13). Intriguingly, plants have evolved additional immune receptors, originally termed resistance (R) proteins, to specifically recognize effectors, triggering faster and stronger immune responses, ETI (14). The majority of R proteins are nucleotide-binding leucine-rich repeat proteins (NLRs), containing either a Toll-interleukin 1-like receptor (TIR) domain or a coiled-coil (CC) domain at their N termini (15). Although some NLRs can directly associate with effectors to trigger ETI, a majority of NLRs detect effectors via monitoring the effector-targeting substrates, known as guardees or decoys. Modification of a guardee/decoy by an effector is guarded by NLRs that subsequently activate ETI (16). From a functional point of view, NLRs are thought to be composed of sensor NLRs and helper NLRs. Sensor NLRs can recognize specific effectors or guardees/decoys. Helper NLRs, on the other hand, cannot directly recognize effectors or guardees/decoys but are required for their corresponding sensor NLRs to trigger ETI (17, 18). Recent reports suggest an ACTIVATED DISEASE RESISTANCE 1 (ADR1) and its paralogs function as helper NLRs for several sensor NLRs, such as RPS2, RPP2, RPP4, CHS3, SNC1, and RRS1/RPS4 (19–22). The N-terminal CC domains of the ADR1 family members resemble an NLR protein RESISTANCE TO POWDERY MILDEW 8 (RPW8), and ADR1s are therefore termed as CCRPW8(R)-NLR proteins, which may represent a separate class of NLRs (23). Another CCR-NLR group, NRG1s (N REQUIREMENT GENE 1) also function as helper NLRs, downstream of TIR-type NLRs SNC1 and CHS3 (21). The nrg adr1 sextuple mutant showed reduced disease resistance to Pseudomonas syringae pv. maculicola ES4326 compared with that of their parents, suggesting NRG1s and ADR1s play synergistic roles on basal defense (21). The detailed molecular mechanisms of helper NLRs in regulating immune responses are yet to be determined.

Our previous genetic studies revealed that BAK1 is involved in a cell-death control pathway (24). Knocking out both BAK1 and its closest paralog, BAK1-LIKE 1 (BKK1), led to a spontaneous cell-death phenotype even under a sterile culture condition (24–26). Although the role of BAK1 in regulating PTI is now well documented, the cell-death phenotype observed in bak1 bkk1 is unlikely caused by the disruption of PTI responses (27). Previous studies indicated that knocking out a BAK1-associated PRR usually does not result in a cell-death phenotype in Arabidopsis (28). In addition, bak1-5, a dominant-negative mutant bearing a point mutation in BAK1, shows reduced PTI response compared to a bak1-4 null mutant, but a bak1-5 bkk1-1 double mutant is completely viable, again suggesting the cell death caused by loss of BAK1 and BAK1-mediated PTI are largely independent (29). NLR-mediated ETI activation is often accompanied by HR. Furthermore, like snc1, a gain-of-function mutant of an NLR gene, the autoimmune phenotypes of bak1-3 bkk1-1 showed at 22 °C can be greatly suppressed by growing at 28 °C (). We therefore hypothesized the HR-like cell death observed in bak1 bkk1 is likely caused by the activation of NLR-mediated ETI rather than the reduction of PTI.

Here we report that ADR1s contribute to BAK1 depletion-triggered cell-death. The expression levels of ADR1s are dramatically up-regulated in bak1 bkk1. Knocking out ADR1s can significantly suppress the autoimmune responses, including cell death in bak1-3 bkk1-1, suggesting the cell-death phenotype of bak1-3 bkk1-1 requires NLRs. Moreover, the expression of HopB1, a protease effector that targets BAK1 and other SERKs, not only caused impaired flg22-mediated immune responses but also resulted in the cell-death phenotype similar to bak1 bkk1. Furthermore, the HopB1-triggered cell-death symptom is also dependent on ADR1s. We conclude that the absence of BAK1 leads to the activation of NLRs, suggesting BAK1 is guarded by NLRs.

Results

Defense-Related Genes Are Up-Regulated in bak1 bkk1.

To identify new components involved in cell death triggered upon BAK1 loss, we compared the global gene-expression profiles of the seedlings of WT Columbia-0 (Col-0) and bak1-4 bkk1-1, a double null mutant. bak1-4 bkk1-1 starts to show a cell-death symptom a week after germination and is ultimately lethal even grown in sterilized culture media (24). We analyzed the differentially expressed genes in 7-d-old WT and bak1-4 bkk1-1 plants by using an RNA-sequencing approach. Among 23,496 detected transcripts, we set a cutoff of change at twofold or greater with P ≤ 0.05, which allowed us to identify 3,829 differentially expressed genes, including 1,848 up-regulated and 1,981 down-regulated ones in bak1-4 bkk1-1 compared to WT ( and Dataset S1). Gene ontology enrichment analyses showed that the genes associated with systemic acquired resistance, salicylic acid (SA) biosynthesis and signaling, cell death and HR, pathogen responses, and mitogen-activated protein kinase (MAPK) signaling were highly enriched in bak1-4 bkk1-1 (). These results suggest that an autoimmune response is activated in bak1-4 bkk1-1.

To investigate whether the autoimmune phenotypes of bak1-4 bkk1-1 are related to NLR-mediated responses, we analyzed the expression patterns of NLR genes and found a number of NLRs were up-regulated in bak1-4 bkk1-1 (). Since activated NLRs sometimes lead to the elevated transcriptional levels of their genes through a feedback loop, highly expressed NLRs might suggest the activation of the corresponding NLRs (30). We noticed three NLR subfamilies in which almost all their coding genes were highly up-regulated in bak1-4 bkk1-1. One of them is the ADR1 subfamily that was previously reported to function in multiple ETI signaling pathways. The second one was not reported before, and we named it a UNR1 (Uncharacterized NLR 1) subfamily. The third one is an RPS5 subfamily. RPS5, the founding member in this subfamily, recognizes the P. syringae effector AvrPphB (31). SUMM2, another member of the RPS5 subfamily, is required for the autoimmune phenotypes of two MAPK mutants, mpk4 and mekk1 (32). We thus tested the potential roles of the NLRs from these three subfamilies for their possible contribution to the cell death of bak1 bkk1.

Up-Regulation of ADR1s Is the Key for the Cell Death Triggered upon BAK1 Loss.

We first confirmed the expression patterns of all aforementioned candidate NLRs in bak1-3 bkk1-1, in which the transcription level of BAK1 is significantly reduced and that of BKK1 is absent. bak1-3 bkk1-1, showing obvious autoimmune phenotypes including cell-death when grown in soil, is completely fertile, making it an ideal double mutant for genetic analyses (24, 33). qRT-PCR results confirmed that almost all gene members in these three NLR subfamilies were up-regulated in bak1-4 bkk1-1 and bak1-3 bkk1-1 (Fig. 1 and ). Next, we tried to reduce the expression of these NLRs in bak1-3 bkk1-1 by using an RNAi approach. For each subfamily, DNA fragments conserved among the gene members were cloned into an RNAi binary vector pBIB-BASTA-35S-GWRNAi and transformed into bak1-3 bkk1-1. qRT-PCR results indicated the expression levels of most gene members in the three subfamilies were dramatically decreased in corresponding RNAi transgenic plants compared to bak1-3 bkk1-1 ().

Fig. 1.

ADR1s are up-regulated in bak1 bkk1 double mutants. (A and B) qRT-PCR assays indicate the expression levels of ADR1, ADR1-L1, and ADR1-L2 are significantly increased in bak1-4 bkk1-1 (A) and bak1-3 bkk1-1 (B). qRT-PCR was performed by using the total RNA from 8-d-old bak1-4 bkk1-1 seedlings grown on 1/2 MS media or 2-wk-old bak1-3 bkk1-1 plants grown in soil. (C) Overexpression of ADR1s in Col-0 results in autoimmune phenotypes similar to bak1-3 bkk1-1. Three-week-old plants grown in soil are presented. (Scale bars, 1 cm.) (D and E) Trypan blue staining (D) and DAB staining (E) assays indicate overexpression of ADR1s in Col-0 causes cell-death symptoms (D) and H2O2 accumulation (E) similar to bak1-3 bkk1-1. (Scale bars, 100 μm.) (F and G) PR1 (F) and FMO1 (G) are expressed in higher levels in ADR1 overexpression lines and bak1-3 bkk1-1. qRT-PCR was performed by using the total RNA from 3-wk-old plants grown in soil. ACT7 was used to normalize the transcript levels. Arbitrary units were used to show the relative abundance of transcript levels of ADR1s, PR1, and FMO1 as compared to Col-0. Bars represent mean ± SD (n = 3). Different letters indicate a significant difference following one-way ANOVA with Tukey's multiple comparison test (P < 0.05).

The autoimmune phenotypes were significantly suppressed in ADR1 RNAi in bak1-3 bkk1-1 plants, whereas, UNR1 or RPS5 RNAi in bak1-3 bkk1-1 showed no obvious phenotypic difference from bak1-3 bkk1-1 (). Consistently, the transcription levels of a defense marker gene PR1 and a defense and cell-death marker gene FMO1 were strongly decreased in the ADR1 RNAi plants but not in UNR1 or RPS5 RNAi lines compared to bak1-3 bkk1-1 (). To further understand whether SUMM2 is involved in the cell-death of bak1-3 bkk1-1, we generated a summ2 bak1-3 bkk1-1 triple mutant. Our results indicated that although summ2 was able to partially suppress the cell-death phenotype of mekk1 or mkp4, it cannot suppress that of bak1-3 bkk1-1 (). Trypan blue and DAB staining assays also suggest that SUMM2 may not contribute to the autoimmune responses of bak1-3 bkk1-1 (). Consistently, the expression levels of PR1 and FMO1 were not decreased in summ2 bak1-3 bkk1-1 compared to bak1-3 bkk1-1 ().

ADR1-mediated ETI signaling requires both SA and EDS1 (34). Similarly, the cell-death phenotype of bak1 bkk1 was partially inhibited when endogenous SA was depleted or an EDS1 mutation was introduced (25, 26). Previous study indicated that the mutation of ADR1-L2 was able to suppress the cell-death of lsd1, a lesion-mimic mutant showing a runaway cell-death phenotype under the treatment of an SA analog benzothiadiazole (34). More importantly, we found that reduced expression of ADR1s could suppress the cell death in bak1-3 bkk1-1. We therefore set to investigate the potential roles of ADR1s in regulating the cell-death of bak1 bkk1.

The ADR1 family contains three members: ADR1, ADR1-LIKE 1 (ADR1-L1), and ADR1-LIKE 2 (ADR1-L2) (20, 35). Overexpression of ADR1, ADR1-L1, or ADR1-L2 in Col-0 resulted in a dwarfed phenotype with compacted and curved rosette leaves and cell-death (Fig. 1). Trypan blue and DAB staining assays also indicated the cell death and H2O2 accumulation were significantly triggered in the overexpression lines (Fig. 1 ). PR1 and FMO1 were highly expressed in these transgenic lines (Fig. 1 ). These results demonstrate that enhanced expression of ADR1s leads to an autoimmune phenotype similar to bak1-3 bkk1-1.

Knocking Out ADR1s Suppresses the Cell-Death Phenotype of bak1-3 bkk1-1.

We next isolated the previously reported T-DNA insertion lines for all three ADR1s (19). RT-PCR analyses confirmed that adr1, adr1-L1, and adr1-L2 are true null mutants (). adr1, adr1-L1, adr1-L2, and adr1 adr1-L1 adr1-L2 plants do not exhibit any defective phenotypes, similar to Col-0 (). Compared to bak1-3 bkk1-1, the autoimmune phenotypes, including cell-death, accumulation of H2O2, and increased expression levels of PR1 in adr1-L2 bak1-3 bkk1-1 and adr1 bak1-3 bkk1-1 were partially suppressed (Fig. 2 and ). adr1-L1 bak1-3 bkk1-1 showed an enhanced cell-death phenotype (Fig. 2). To understand why adr1-L1 bak1-3 bkk1-1 showed enhanced autoimmune phenotypes, we analyzed the expression levels of all ADR1s in three different adr1s bak1-3 bkk1-1 triple mutants. qRT-PCR result showed that loss-of-function of ADR1-L1 caused a compensatory increased expression of ADR1 and ADR1-L2 in adr1-L1 bak1-3 bkk1-1 (). These results are consistent with an earlier report showing that adr1 or adr1-L2 suppressed the autoimmune responses of snc1 (20). In contrast, snc1 adr1-L1 double mutants showed enhanced phenotypes compared to snc1 due to compensatory expression of ADR1 and ADR1-L2 (20). To verify the aforementioned phenotypes of the adr1s bak1-3 bkk1-1 triple mutants, genomic sequences of ADR1s were cloned into a binary vector (modified from pFAST-G01) and transformed into the corresponding adr1s bak1-3 bkk1-1 triple plants. The resulting transgenic lines showed the phenotypes similar to bak1-3 bkk1-1 ().

Fig. 2.

The autoimmune phenotypes of bak1-3 bkk1-1 require ADR1s. (A) Mutation in ADR1 or ADR1-L2 weakly suppresses the cell death of bak1-3 bkk1-1. Mutation in ADR1-L1 enhances the cell-death symptoms of bak1-3 bkk1-1. adr1s bak1-3 bkk1-1 quadruple mutants show additional cell-death suppression compared to adr1s bak1-3 bkk1-1 triple mutants. The adr1 triple mutant strongly suppresses the autoimmune phenotypes of bak1-3 bkk1-1. Three-week-old plants grown in soil are presented. (Scale bars, 1 cm.) (B and C) Trypan blue staining (B) and DAB staining (C) assays indicate the cell-death symptoms and H2O2 accumulation in the plants presented in A. Two-week-old plants grown in soil were analyzed. (Scale bars, 100 μm.)

The cell-death symptoms of three quadruple mutants, adr1 adr1-L1 bak1-3 bkk1-1, adr1-L1 adr1-L2 bak1-3 bkk1-1, and adr1 adr1-L2 bak1-3 bkk1-1, were further suppressed compared to the adr1s bak1-3 bkk1-1 triple mutants. We subsequently generated a quintuple mutant adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 in which the autoimmune phenotypes were dramatically suppressed to a WT-like level (Fig. 2). We next tested whether the rescued phenotypes of adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 are caused by increased BAK1 transcripts or elevated BAK1 protein abundance. qRT-PCR analyses failed to detect the increased expression of BAK1 in adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 compared to bak1-3 bkk1-1 (). Immunoblotting analyses using an α-BAK1 antibody showed that the BAK1 protein level was not altered in adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 compared to bak1-3 bkk1-1 (). In summary, our genetic results indicated the cell-death phenotype of bak1 bkk1 requires ADR1s.

flg22-Mediated PTI Response Is Impaired but ADR1s-Mediated ETI Is Enhanced in bak1-3 bkk1-1.

To further dissect the functions of BAK1 and ADR1s in disease resistance, 3-wk-old Col-0, adr1 adr1-L1 adr1-L2, bak1-3 bkk1-1, and adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 plants were challenged with various strains of a bacterial pathogen P. syringae pv. tomato (Pto) DC3000. The adr1 triple mutant showed slightly reduced resistance to WT Pto DC3000 compared to Col-0 (Fig. 3 ), consistent with the results from a previous report (19). bak1-3 bkk1-1 exhibited enhanced resistance to Pto DC3000, which is consistent with the enhanced defenses in this mutant (Fig. 3 ). adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 plants, however, are more susceptible to Pto DC3000 compared to bak1-3 bkk1-1, indicating that the autoimmunity of bak1-3 bkk1-1 was suppressed in the quintuple mutant (Fig. 3 ). ADR1s function as helper NLRs for the sensor NLR, RPS2, which recognizes the bacterial effector AvrRpt2 (19). ADR1s are required to initiate AvrRpt2-triggered ETI activation. When treated with Pto DC3000 (avrRpt2) followed by 1-d covering to keep humidity, Col-0 showed resistance to bacterial infection and the adr1 triple mutant was more susceptible (Fig. 3). bak1-3 bkk1-1 plants are resistant to Pto DC3000 (avrRpt2), similar to Col-0. Since both Col-0 and bak1-3 bkk1-1 showed resistance to Pto DC3000 (avrRpt2) under this covering condition, we cannot distinguish the difference of resistance between Col-0 and bak1-3 bkk1-1. It was reported that increasing humidity can enhance the susceptibility of plants to pathogens (36). We therefore increased the covering time from 1 d to 3 d after the plants were treated with Pto DC3000 (avrRpt2). Under the altered condition, Col-0 showed slightly enhanced susceptibility to Pto DC3000 (avrRpt2) but bak1-3 bkk1-1 was still resistant to Pto DC3000 (avrRpt2) (Fig. 3). We thus concluded that ETI mediated by ADR1s is likely activated in bak1-3 bkk1-1. Under both aforementioned covering conditions after bacterial treatments, adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 showed enhanced susceptibility to Pto DC3000 (avrRpt2) compared to bak1-3 bkk1-1 (Fig. 3 ), indicating ADR1s contribute to the resistance of bak1-3 bkk1-1 to Pto DC3000 (avrRpt2). Our results also showed that adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 is more resistant than the adr1 triple mutant to Pto DC3000 (avrRpt2). Considering that additional NLRs besides ADR1s could be activated in bak1-3 bkk1-1 and may partially contribute to the autoimmune phenotypes of bak1-3 bkk1-1, we speculate that the defense responses activated by additional NLRs other than ADR1s in adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 are responsible for its resistance to Pto DC3000 (avrRpt2).

Fig. 3.

bak1-3 bkk1-1 shows enhanced effector-triggered responses and impaired PAMP-triggered responses. (A and B) Three-week-old plants were treated with Pto DC3000 and covered for 1 d (A) or 3 d (B). Bacterial growth was assessed at 0- and 3-d postinoculation (dpi). The adr1 triple mutant is slightly susceptible while bak1-3 bkk1-1 shows enhanced resistance to Pto DC3000. The adr1 triple mutant partially restores the response of bak1-3 bkk1-1 to Pto DC3000. Bars represent mean ± SD (n = 6). (C) Three-week-old plants were treated with Pto DC3000 (avrRpt2) and covered for 1 d. Bacterial growth was assessed at 0 and 3 dpi. Col-0 and bak1-3 bkk1-1 show resistance to Pto DC3000 (avrRpt2). The adr1 triple mutant is susceptible to Pto DC3000 (avrRpt2). The adr1 triple mutant partially suppresses the resistance of bak1-3 bkk1-1 to Pto DC3000 (avrRpt2). Bars represent mean ± SD (n = 6). (D) Three-week-old plants were treated with Pto DC3000 (avrRpt2) and covered for 3 d. Bacterial growth was assessed at 0 and 3 dpi. Col-0 shows susceptibility but bak1-3 bkk1-1 shows resistance to Pto DC3000 (avrRpt2). The adr1 triple mutant partially suppresses the resistance of bak1-3 bkk1-1 to Pto DC3000 (avrRpt2). Bars represent mean ± SD (n = 5). (E) Oxidative burst upon flg22 treatment is reduced in bak1-3 bkk1-1 and adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 compared to Col-0 and the adr1 triple mutant. ROS production was measured as relative light units (RLU) in a luminol-based assay. Values are mean ± SD (n = 5). (F) Col-0 and the adr1 triple mutant show similar MAPK activation upon flg22 treatment. bak1-3 bkk1-1 and adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1 show similar MAPK activation upon flg22 treatment. MAPK activation was analyzed by immunoblotting with an α-pERK antibody. The control for protein loading is shown by Coomassie brilliant blue (CBB). (G) flg22-mediated signaling is intact in the adr1 triple mutant. FRK1 expression was determined 1 h after treatment with 1 μM flg22. qRT-PCR was performed by using the total RNA from 7-d-old seedling. ACT7 was used to normalize the transcript levels. Arbitrary units were used to show the relative abundance of FRK1 transcript levels as compared to Col-0. Bars represent mean ± SD (n = 3). Different letters indicate a significant difference following one-way ANOVA with Tukey's multiple comparison test (P < 0.05).

Both the burst of reactive oxygen species (ROS) and MPK3/6 activities can be quickly triggered in plants after PAMPs are recognized by PRRs. We next analyzed the PAMP-mediated responses in various plant lines by detecting ROS burst and MPK3/6 activities upon the treatment of flg22. bak1-3 bkk1-1 showed partially reduced ROS accumulation and MPK3/6 activation compared to Col-0. bak1-3 bkk1-1 exhibited ROS accumulation and MPK3/6 activities similar to adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1, and adr1 adr1-L1 adr1-L2 displayed ROS accumulation and MPK3/6 activities similar to Col-0 upon the treatment of flg22 (Fig. 3 ). In addition, we analyzed the expression of FRK1, a marker gene for PTI signaling, in the different genetic backgrounds upon flg22 treatment. qRT-PCR results indicated that the flg22-mediated PTI response in adr1 adr1-L1 adr1-L2 was similar to that in Col-0. The PTI responses in both bak1-3 bkk1-1 and adr1 adr1-L1 adr1-L2 bak1-3 bkk1-1, however, were dampened compared to Col-0 (Fig. 3). These results demonstrate that flg22-triggered PTI responses are partially impaired in bak1-3 bkk1-1, and ADR1s are not involved in flg22-mediated PTI responses. Although PTI responses are partially impaired, bak1-3 bkk1-1 still showed enhanced resistance to Pto DC3000 and Pto DC3000 (avrRpt2), suggesting the elevated disease resistance in bak1-3 bkk1-1 is most likely caused by the activation of ADR1s.

Expression of HopB1 Mimics the Autoimmune Responses of bak1-3 bkk1-1.

Given the fact that knocking out or significantly knocking down BAK1 and BKK1 leads to NLR-dependent immune responses, we next investigated the biological significance of NLR activation upon depletion of BAK1 and its paralogs, SERKs. To promote full pathogenicity in the host, microbial pathogens deliver effectors to plant cells to shut down PTI signaling by attacking key components in PTI. The effectors HopF2 and AvrPtoB were found to associate with and disrupt BAK1 (37, 38). In a previous report, we identified a Pto DC3000-derived protease HopB1 that specifically cleaves flg22-activated BAK1 and other SERKs (39). We generated transgenic plants harboring estrogen inducible HopB1-FLAG in Col-0 (Est-HopB1-FLAG in Col-0). Upon treatment with estradiol to induce the expression of HopB1 for 2 wk, Est-HopB1-FLAG in Col-0 exhibited a phenotype with slightly more compacted rosette leaves compared to Col-0 (Fig. 4 and ). Because HopB1 was found to only cleave flg22-activated BAK1 (39), we used flg22 to activate BAK1. When treated with both estradiol and flg22 for 2 wk, Est-HopB1-FLAG in Col-0 showed a striking cell-death symptom reminiscent of bak1 bkk1 (Fig. 4 and see ). As a control, the abundance of BAK1 in Col-0 was not noticeably affected by the treatments of estradiol alone, flg22 alone, or estradiol plus flg22 for 2 wk. In Est-HopB1-FLAG in Col-0, the abundance of BAK1 was not changed upon the treatments of estradiol or flg22 alone but was significantly decreased when both estradiol and flg22 were applied for 2 wk. BAK1 abundance was analyzed by using an α-BAK1 antibody and the induced HopB1 was detected by using an α-FLAG antibody (Fig. 4). The treatment with estradiol and flg22 for 2 wk not only resulted in cell death, but also H2O2 accumulation, and up-regulation of PR1 and FMO1 in Est-HopB1-FLAG in Col-0 (Fig. 4 and ). In addition, HopB1 expression further enhanced the cell-death symptom of bak1-3 bkk1-1 in the presence of flg22 (). These data suggest cleavage of BAK1 and other SERKs by HopB1 triggers cell death in plants, mimicking the bak1 bkk1 double mutant.

Fig. 4.

Estrogen-induced HopB1 expression causes ADR1s-dependent immune responses. (A) Estrogen-induced HopB1 expression in Col-0 leads to a striking cell-death phenotype upon flg22 treatment. HopB1-induced cell death is significantly inhibited in the adr1 triple mutant. Three-week-old plants grown on 1/2 MS media supplemented with or without estradiol and/or flg22 are presented. (Scale bars, 0.5 cm.) (B) The abundance of BAK1 and induced HopB1 is analyzed by immunoblotting. The abundance of BAK1 was detected using an α-BAK1 antibody. BAK1 protein is significantly reduced in HopB1 transgenic plants when treated with both estradiol and flg22 for 2 wk. The HopB1 protein was detected using an α-FLAG antibody. HopB1 protein is induced in HopB1 transgenic plants when estradiol is applied. Coomassie brilliant blue (CBB) staining for a duplicated SDS/PAGE gel was used to show equal loading. (C and D) HopB1 expression causes up-regulation of PR1 (C) and FMO1 (D) in Col-0, which is significantly suppressed in the adr1 triple mutant in the presence of flg22. qRT-PCR was performed by using the total RNA from 3-wk-old plants grown on 1/2 MS media supplemented with estradiol and/or flg22. ACT7 was used to normalize the transcript levels. Arbitrary units are used to show the relative abundance of PR1 and FMO1 transcript levels as compared to Col-0. Bars represent mean ± SD (n = 3). (E) HopB1 expression up-regulates ADR1s in the presence of flg22. qRT-PCR was performed by using the total RNA from 3-wk-old plants grown on 1/2 MS media supplemented with or without estradiol and flg22. ACT7 was used to normalize the transcript levels. Arbitrary units were used to show the relative abundance of transcript levels of PR1, FMO1, and ADR1s as compared to Col-0. Bars represent mean ± SD (n = 3). Different letters indicate a significant difference following one-way ANOVA with Tukey's multiple comparison test (P < 0.05). E, estradiol; F, flg22; M, mock. Three biological replicates were conducted and similar results were obtained. Here are the representative results.

We showed the cell-death phenotype of bak1-3 bkk1-1 requires ADR1s. We next investigated whether ADR1s also contribute to HopB1-induced immune responses. Transgenic plants harboring estrogen inducible HopB1 in adr1 adr1-L1 adr1-L2 (Est-HopB1-FLAG in the adr1 triple mutant) were generated. In comparison with those of Est-HopB1-FLAG in Col-0, the cell-death phenotype of Est-HopB1-FLAG in the adr1 triple mutant is significantly suppressed when treated with both estradiol and flg22 for 2 wk (Fig. 4 and ). Immunoblotting analyses indicated that in both Est-HopB1-FLAG in the adr1 triple mutant and Est-HopB1-FLAG in Col-0 plants, treatments with estradiol and flg22 for 2 wk all caused dramatic reduction of BAK1 to an equivalent level (Fig. 4). Those results indicated the cell death caused by HopB1-induced BAK1 cleavage is ADR1s-dependent. In addition, HopB1 expression in the presence of flg22 resulted in increased expression of PR1 and FMO1 in Col-0, which was largely inhibited in the adr1 triple mutant (Fig. 4 ). Moreover, in the presence of flg22, the induced expression of HopB1 in Col-0 caused elevated expression of ADR1s, suggesting HopB1-mediated cleavage of BAK1 and other SERKs leads to the activation of ADR1s (Fig. 4).

To examine whether HopB1 affects PTI responses in which BAK1 plays an essential role, we tested MPK3/6 activation after BAK1 was activated by flg22 and HopB1 was induced by estradiol. We pretreated Est-HopB1-FLAG in Col-0 and in the adr1 triple mutant with estradiol for 24 h to induce the expression of HopB1. We then applied flg22 to activate BAK1, which led to a dramatic decrease of BAK1 abundance in both types of transgenic plants after a 15-min treatment (). Accordingly, MPK3/6 activation upon flg22 treatment was significantly suppressed in Est-HopB1-FLAG in Col-0 and in the adr1 triple mutant, indicating flg22-mediated PTI signaling is repressed when HopB1 is induced (). Similarly, analyses of FRK1 expression also showed HopB1 expression caused reduced flg22-mediated PTI response in Col-0 and the adr1 triple mutant (). These results suggest HopB1 expression dampened PTI responses in which ADR1s are not involved.

Bacterium-Delivered HopB1 Triggers ETI Responses in an ADR1s-Dependent Manner.

To further exclude the interference of the effectors other than HopB1 in the bacterial strain Pto DC3000, we used a Pseudomonas fluorescens Pf0-1 strain that has no effectors and can only trigger PTI. When sprayed with P. fluorescens-EV (Pf0-1-EV), Col-0 and adr1 triple mutant showed similar resistance, while eds1 and pad4 exhibited increased susceptibility, indicating that PTI confers plant resistance in a manner dependent on EDS1 and PAD4 but independent of ADR1s (Fig. 5). When sprayed with P. fluorescens-HopB1 (Pf0-1-HopB1), the bacterium grew to higher levels in the adr1 triple mutant compared to Col-0, and the adr1 triple mutant was as susceptible as to eds1 and pad4, indicating an ADR1-dependent immunity triggered by HopB1 (Fig. 5). The Pf0-1-HopB1 strain grew to much higher levels than the Pf0-1 EV strain on the adr1 triple mutant, indicating a profound role of HopB1 in virulence (Fig. 5 ). On Col-0 plants, the Pf0-1-HopB1 strain grew only slightly more than the Pf0-1 EV strain, which reflect an outcome of combined effect of the HopB1-triggered susceptibility and a HopB1-triggered immunity in normal plants. Consistent with the aforementioned antibacterial resistance, PR1 and FMO1 showed a modest induction in Col-0 and adr1 triple mutant when inoculated with the Pf0-1 EV strain compared to mock treatment (Fig. 5 ). This induction is abolished in eds1 and pad4 plants. When inoculated with Pf0-1-HopB1, a strong induction of PR1 and FMO1 was observed only in Col-0, which was significantly reduced in the adr1 triple (Fig. 5 ). When injected with Pf0-1-HopB1, the leaves of Col-0 showed a clear cell-death symptom which was clearly reduced in adr1 triple (Fig. 5 ), indicating that the bacterially delivered HopB1 can indeed trigger cell death in a manner dependent on ADR1s.

Fig. 5.

Bacterium-delivered HopB1 triggers ADR1s-dependent ETI responses. (A and B) Three-week-old plants were sprayed with Pf0-1-EV (A) and Pf0-1-HopB1 (B). Bacterial growth was assessed at 0 and 3 dpi. Compared to Col-0, the adr1 triple mutant shows similar resistance to Pf0-1-EV, but is more susceptible to Pf0-1-HopB1. Bars represent mean ± SD (n = 6). (C and D) Pf0-1-EV treatment causes similar levels of up-regulation of PR1 (C) and FMO1 (D) in Col-0 and in the adr1 triple mutant. Compared to Col-0, PR1 (C) and FMO1 (D) expression is significantly suppressed in the adr1 triple mutant upon the treatment of Pf0-1-HopB1. qRT-PCR was performed by using the total RNA from 3-wk-old plants grown in soil after Pf0-1-EV or Pf0-1-HopB1 injected and incubated for 10 h. ACT7 was used to normalize the transcript levels. Arbitrary units are used to show the relative abundance of PR1 and FMO1 transcription levels as compared to Col-0. Bars represent mean ± SD (n = 3). (E) Injection of Pf0-1-HopB1 induces clear cell-death phenotype in Col-0 but clearly reduced in adr1 triple mutant. (F) Trypan blue staining assays indicate the cell-death phenotypes of Col-0 after Pf0-1-HopB1 injection are suppressed in the adr1 triple mutant. (Scale bars, 200 μm.) Three biological replicates were conducted and similar results were obtained. Here are the representative results. Different letters indicate a significant difference following one-way ANOVA with Tukey's multiple comparison test (P < 0.05).

Discussion

It has been more than a decade since we first reported that bak1 bkk1 exhibited a cell death phenotype (24). Significant efforts have been made to elucidate mechanisms leading to such an unexpected phenotype. Genetic analyses identified a number of proteins that are involved in the cell-death control of bak1 bkk1, including an SA biosynthetic enzyme (SID2), components regulating ETI signaling and SA biosynthesis (EDS1 and PAD4), a nucleoporin subunit protein (SBB1), a regulator mediating endoplasmic reticulum quality control (STT3a), and two calcium ion channels (CNGC19/20) (24, 40–42). But the interrelationships among these proteins are not well understood. Especially the early events leading to BAK1-depletion triggered cell death are not elucidated.

In this study, we demonstrate that BAK1 is likely guarded by an ADR1-dependent NLR. First, the cell-death of bak1 bkk1 resembles the phenotype of NLR-mediated autoimmune responses. Second, genetic analyses indicated that ADR1s are required for the autoimmune phenotypes of bak1 bkk1. Third, the increased disease resistance of bak1-3 bkk1-1 to Pto DC3000 (avrRpt2) relative to WT is ADR1-dependent. Fourth, cleavage of activated SERKs by either transgenic expression of a bacterial effector protein HopB1 or bacterium-delivered HopB1 led to a cell-death phenotype similar to bak1 bkk1, which is also ADR1-dependent. These results demonstrate both the cell death and increased disease resistance phenotypes of bak1-3 bkk1-1 rely on the activation of ADR1s.

It was previously proposed that many plant autoimmune responses are caused by inappropriate activation of NLRs (43). For example, MAP kinases MEKK1 and MPK4 are two downstream components of the FLS2-BAK1–mediated PTI signaling. The activities of MEKK1 and MPK4 are guarded by an NLR protein, SUMM2. The summ2 mutant can partially suppress the autoimmune phenotypes of mekk1 or mpk4, indicating a surveillance system guards the downstream components of PTI (32). Whether PRRs or their coreceptors are guarded by NLRs is largely unknown. Mutations in RK PRRs, such as FLS2 and EFR, do not show any autoimmune phenotypes, suggesting RK PRRs are unlikely guarded by NLRs. BAK1, as a shared coreceptor, plays a key role in multiple PTI pathways, making it an ideal target for microbial effectors. It is an efficient strategy for plants to trigger much stronger defense responses to eliminate microbes if BAK1 is attacked (Fig. 6).

Fig. 6.

A model to show BAK1 is directly or indirectly guarded by NLR-mediated signaling. BAK1 acts as a coreceptor for LRR-type PRRs, by sensing PAMPs, through MAPK kinase cascades to positively regulates PTI signaling. When pathogens infect plants, pathogen-delivered effectors can target PTI components to promote virulence. Plants have evolved NLR proteins to monitor the situation of corresponding targets, either the downstream MAP kinases or the coreceptor BAK1/SERKs. Different targets are usually attacked by their corresponding effectors, their specific NLRs are then activated, leading to cell-death phenotypes. In bak1 bkk1, both BAK1 and BKK1 are absent, similar to the depletion of BAK1/SERKs by effectors such as HopB1, NLRs (such as ADR1s) can be constitutively activated, leading to spontaneous cell-death phenotypes even under sterile growth conditions.

A recent study showed mutations in PEP-RECEPTORs (PEPRs), encoding the receptors of Pep peptides, could partially inhibit the autoimmune responses of bak1-3 bkk1-1 (44). Our results suggest ADR1s-mediated cell death in bak1-3 bkk1-1 may be independent of PEPR-mediated immune responses. For example, compared to pepr1 pepr2, the adr1 triple mutant showed greater reduction of cell death caused by bak1-3 bkk1-1 mutations (). It was reported that Pep2 treatment can significantly inhibit root growth (44). The root inhibition by Pep2 in adr1 triple mutant is similar to that in WT (). In addition, the expression levels of PROPEP2 and PROPEP3, encoding Pep proligands, are moderately increased in Col-0 and dramatically elevated in bak1-3 bkk1-1 upon Pep2 treatment (). The expression levels of PROPEPs in adr1 triple mutant were similar to those in WT regardless of the treatment of Pep2, demonstrating ADR1s are not involved in PEPR-mediated immune signaling.

bak1 bkk1 is not a naturally existing double mutant. To validate the biological significance of the cell death observed in the double mutant, we studied the consequence when BAK1 and other SERKs are attacked by effectors from bacteria. HopB1 is a protease effector derived from P. syringae. Previous analyses indicated that HopB1 can directly interact with FLS2 and cleave flg22-activated SERKs to promote virulence when plants are infected by Pto DC3000 (39). While a transient induction of HopB1 transgene expression leads to increased susceptibility to Pto hrcC− bacteria (39), a prolonged induction of the HopB1 transgene in the presence of flg22 for 2 wk was found to activate immune responses in this study. As the prolonged treatment of estradiol and flg22 is expected to cause greater depletion of BAK1 and other SERKs, it is possible that a threshold of BAK1 and other SERKs must be reached before the activation of defenses. We propose that the protein levels of SERKs are monitored by an unknown NLR and that this NLR is activated once the SERK protein levels of BAK1 and other SERKs are below certain threshold.

It should be noted that HopB1 naturally delivered from Pto DC3000 does not trigger measurable ETI or cell death, as the strain is fully virulent on Arabidopsis. One plausible explanation is that there may exist an effector that masks HopB1-triggered ETI. This scenario is well supported by our experiments with the Pf0-1-HopB1, which does not carry any other effectors (Fig. 5).

Another recent study showed that an effector AvrRps4 can be recognized by sensor NLRs, RPS4/RRS1, together with helper ADR1s (45). Can HopB1, as an effector, also be recognized by an unknown sensor NLR and helper ADR1s? Inducible expression of AvrRps4 did not trigger HR-like phenotypes. Without flg22 treatment, expression of HopB1 also did not induce autoimmune responses. Based on these results, we cannot exclude the possibility that ADR1s may contribute to HopB1 recognition. First, HopB1-triggered cell death is flg22-dependent, indicating the cell death triggered by HopB1 expression likely involves flg22-related components, such as BAK1. Second, we failed to detect the interaction between ADR1s and HopB1 (). Therefore, we conclude HopB1 is unlikely recognized by ADR1s.

To adapt to ever-changing environments and maximize their chances of survival, plants have evolved sophisticated mechanisms to coordinate growth and defense. PTI activation allows plants to protect themselves against most invading pathogens. NLR-mediated signaling pathways, the stronger and damage-causing immune responses, need to be repressed during normal growth and development. When BAK1 and other SERKs are attacked by effectors, however, the depletion of BAK1 and other SERKs is detected and NLR-mediated defense responses are initiated at the cost of reduced growth. BAK1 and other SERKs are activated upon PAMP perception and positively regulates PTI responses. BAK1 and other SERKs also serve as guardees by NLRs. Depletion of BAK1 and other SERKs result in the activation of an unknown sensor NLR (NLRs), which acts upstream of ADR1s to activate immune responses including cell death (Fig. 6).

Materials and Methods

The detailed information about plant materials, plant growth and treatment conditions, Trypan blue staining, DAB staining, gene-expression analyses, plasmid construction, generation of transgenic plants, pathogen infection assays, immunoblotting, and oxidative burst measurement are described in .