NLR-Associating Transcription Factor bHLH84 and Its Paralogs Function Redundantly in Plant Immunity

Summary

In plants and animals, NLR immune receptors are utilized to detect pathogen-derived molecules and activate immunity. However, the mechanisms of plant NLR activation remain unclear. Here, we report on bHLH84, which functions as a transcriptional activator. Simultaneously knocking out three closely related bHLH paralogs partially suppresses the autoimmunity of snc1 and compromises RPS4-mediated defense, while overexpression of these close paralogs renders strong autoimmunity, suggesting functional redundancy in the gene family. In planta co-immunoprecipitation revealed interactions between not only bHLH84 and SNC1, but also bHLH84 and RPS4. Therefore bHLH84 family transcription factors associate with these NLRs to activate defense responses, enabling potentially faster and more robust transcriptional reprogramming upon pathogen recognition.

Introduction

Plants have evolved a sophisticated immune system to fight against invading microbial pathogens that threaten their normal growth and development. Plant immunity is in part mediated by resistance (R) proteins that recognize pathogen proteins known as effectors [1]–[3]. The majority of R proteins are NLR receptors that contain leucine-rich repeats (LRRs) at the C-terminus, a central nucleotide-binding site (NBS) and either a Toll/Interleukin-1 receptor (TIR) or a coiled-coil (CC) domain at the N-terminus [4]. In Arabidopsis, genetically downstream of the R proteins are the EDS1 (ENHANCED DISEASE SUSCEPTIBILITY 1)/PAD4 (PHYTOALEXIN DEFICIENT 4)/SAG101 (SENESCENCE-ASSOCIATED GENE101) complex and NDR1 (NON-RACE-RESISTANCE 1), which mainly mediate TIR-NB-LRR or CC-NB-LRR triggered defense responses, respectively [5]–[8].

While the mechanisms underlying effector recognition by R proteins have been intensively studied, little is known about the post-recognition events leading to defense activation. Recently, it has been shown that the nuclear pool of certain R proteins, including MLA10 (MILDEW A LOCUS 10) in barley, N in tobacco, Pb1 (Panicle blast 1) in rice, and RPS4 (RESISTANT TO P.SYRINGAE 4), RRS1 (RESISTANT TO RALSTONIA SOLANACEARUM 1) and SNC1 (SUPPRESSOR OF NPR1-1, CONSTITUTIVE1) in Arabidopsis, is important for the activation of defense responses [9]–[14]. The latest discoveries on the interactions between some of these R proteins and their associating transcription factors (TFs) further shed light on the activation mechanism of nuclear R proteins. For example, MLA10 interacts with WRKY TFs to de-repress PAMP (PATHOGEN-ASSOCIATED MOLECULAR PATTERN) triggered basal defense [9]. The active state of MLA10 can also release MYB6 (MYB DOMAIN PROTEIN 6) from WRKY suppression and promote its binding to cis-elements to initiate defense responses [15]. CC-type NLR Pb1 in rice interacts with WRKY45 and this interaction is believed to protect the TF from proteasomal degradation in the nucleus [16]. In addition, SNC1 associates with transcriptional co-repressor TPR1 (TOPLESS RELATED 1) to negatively regulate the expression of known defense suppressors, thereby activating plant immunity [17]. Lately, studies on N in tobacco showed that it is able to associate with the TF SPL6 (SQUAMOSA PROMOTER BINDING PROTEIN-LIKE 6) upon effector recognition [18]. From these data, it has been hypothesized that some NLRs associate with TFs inside the nucleus to directly participate in transcriptional reprogramming to regulate downstream defense responses.

In Arabidopsis, the gain-of-function NLR mutant snc1 constitutively expresses PATHOGENESIS RELATED (PR) defense marker genes and exhibits enhanced disease resistance against virulent bacteria Pseudomonas syringae pv. maculicola (P.s.m.) ES4326 and oomycete Hyaloperonospora arabidopsidis (H.a.) Noco2 [19], [20]. As snc1 displays strong autoimmune phenotypes while remaining fully fertile, it has become a useful tool for dissecting NLR mediated resistance. Forward genetic screens designed to isolate positive regulators of immunity were conducted in the snc1 background and over a dozen Modifier of snc1 (MOS) genes have been identified. Characterizations of the MOS genes and their encoded protein products have revealed complicated regulatory events surrounding snc1 mediated autoimmunity, which include nucleocytoplasmic trafficking, RNA processing, protein modification and transcriptional regulation [21], [22]. However, genetic redundancy and lethality may have prevented some essential positive regulators from being discovered through forward genetic approaches. Here, we employed a targeted reverse genetic screen to search for candidate TFs participating in the regulation of snc1-mediated defense. One basic Helix-loop-Helix (bHLH) type TF, which is a putative transcriptional activator, was isolated from the screen and found to be able to associate with NLRs to activate immunity.

Results

A targeted reverse genetic screen

Previously, SNC1 was found to participate directly in transcriptional reprogramming with TPR/MOS10 repressor proteins that do not directly bind DNA [17]. We did not find a DNA-binding TF that functions together with SNC1 from the MOS forward genetic screens, suggesting that multiple TFs may function redundantly in snc1-mediated immunity. To search for novel TFs regulating plant immunity, a reverse genetic screen was employed. As UV irradiation has been shown to induce resistance to pathogens and to induce transcription of defense related genes [23]–[25], we selected 36 putative TFs which show >1.7-fold enhanced expression level upon UV treatment based on publically available microarray data from The Arabidopsis Information Resource (Table S1). The genomic sequences of these genes were cloned into a binary vector pCambia1305 containing C-terminus GFP and HA double tags. Using the floral dip method [26], overexpression transgenic plants in snc1 and Col-0 backgrounds were generated. From the primary screen, we searched for transformants either suppressing or enhancing the dwarf morphology of snc1 or causing dwarfism in Col-0 background. Transgenic plants exhibiting heritable altered morphology were subject to a secondary screen, where altered resistance was examined using a Hyaloperonospora arabidopsidis (H.a.) Noco2 infection assay. Screening data for these candidate TFs are summarized in Table S1.

From the screen, we identified several TFs that displayed phenotypes in only snc1 or Col-0 background, but not in both when overexpressed (Table S1). However, overexpression of three TFs, At2g31230, At2g14760 or At5g61590, resulted in stunted growth in both the snc1 and Col-0 backgrounds (Table S1, Figure 1A and 1B). We selected two TFs with the strongest phenotypes for further analysis. At2g14760 encodes bHLH84, a predicted basic helix-loop-helix TF, while At5g61590 encodes ERF107, which belongs to the ethylene-response-factor (ERF) TF family.

Characterization of the OXbHLH84-GFP-HA and OXERF107-GFP-HA lines

To further explore the functions of bHLH84 and ERF107 in plant immunity, we isolated homozygous overexpression transgenic lines in Col-0 background. As shown in Figure 1B, both OXbHLH84-GFP-HA and OXERF107-GFP-HA plants exhibited dwarf morphology compared with WT plants. We further examined defense marker PR gene expression in these transgenic plants using real-time PCR. As shown in Figure 1C, the expression of both PR1 and PR2 was significantly up-regulated, with about 100- and 35- fold changes, respectively, in OXbHLH84-GFP-HA, indicating that the defense responses were constitutively activated. In OXERF107-GFP-HA transgenic plants, both PR1 and PR2 were around 15-fold up-regulated. Consistent with PR gene expression, resistance against virulent pathogen H.a. Noco2 was enhanced in both OXbHLH84-GFP-HA and OXERF107-GFP-HA plants (Figure 1D). As OXbHLH84-GFP-HA plants displayed more severe immune phenotypes than OXERF107-GFP-HA plants, we chose to focus solely on the functional study of bHLH84. Consistent with its predicted TF function, bHLH84-GFP-HA fluorescence was detected in the nuclei when the OXbHLH84-GFP-HA seedlings were examined by confocal fluorescence microscopy (Figure 1E).

bHLH84 functions as a transcriptional activator

To further investigate how bHLH84 regulates plant immunity, we tested whether it is a bona fide transcription factor by conducting a previously established protoplast transcription activity transient assay [27]. In this assay, the β-glucuronidase (GUS) reporter gene is driven by 2×Gal4 DNA-binding sites (DBS). Co-transformation of bHLH84 fused with the Gal4 DNA-binding domain (DBD) together with the reporter constructs in Arabidopsis mesophyll protoplasts resulted in drastically enhanced GUS expression (Figure 2A) compared to the control transfection, suggesting that bHLH84 functions as a transcriptional activator.

Knocking out bHLH84 and its two close paralogs does not compromise basal immunity while attenuating RPS4-mediated defense response

bHLH TFs constitute one of the largest TF families in Arabidopsis, with 147 members including bHLH84 [28]. bHLH84 has three alternatively spliced variants according to available expressed sequence tag (EST) data (Figure 2B). Based on sequence analysis, At2g14760.2 encodes a truncated protein without the C-terminal bHLH DNA binding domain, while the other two variants encode full-length proteins [28]. However, when the coding region of bHLH84 was amplified from cDNA of WT plants and sequenced, only At2g14760.1 was observed, suggesting that At2g14760.1 is the dominantly expressed version.

To further investigate the contribution of bHLH84 in plant immunity, knock-out analysis of bHLH84 was carried out. A T-DNA allele of bHLH84 (SALK_064296) was obtained from the Arabidopsis Biological Resource Centre (ABRC). As shown in Figure 2B, the T-DNA inserts in the first exon of At2g14760.1. As a consequence, the expression of bHLH84 was abolished (Figure S1A). SALK_064296 was thus assigned as bhlh84. When bhlh84 leaves were challenged with virulent bacterial pathogen Pseudomonas syringae pv maculicola (P.s.m.) ES4326, they exhibited similar bacterial growth as WT (Figure 2C), indicating that the immune response is not compromised in the knock-out mutant.

To investigate whether genetic redundancy masks the function of bHLH84, we carried out a phylogenetic analysis of bHLH84 and its paralogs. As RSL2 (ROOT HAIR DEFECTIVE 6-LIKE 2) is the closest paralog of bHLH84 (Figure 2D; [29]), a T-DNA knock-out line for this gene, SALK_048849, was obtained from ABRC. As shown in Figure S1B, no expression of RSL2 was detectable in SALK_048849, which was named as rsl2. Double mutant bhlh84 rsl2 was created and subjected to pathogen infection experiments. As shown in Figure 2C, the bhlh84 rsl2 double mutant did not exhibit resistance defects either. As RSL4 (ROOT HAIR DEFECTIVE 6-LIKE 4) is functionally redundant with RSL2 in regulating root hair growth [29], we further created the triple mutant by crossing bhlh84 rsl2 with rsl4 rsl2, which was characterized by Yi et al., 2010 [29]. The triple mutant bhlh84 rsl2 rsl4 still did not exhibit obvious defects upon infection with P.s.m. ES4326 compared to WT plants (Figure 2C), indicating that knocking out bHLH84 and its two paralogs does not compromise basal defense responses. Since no good T-DNA mutant line was available for bHLH139, we were not able to test higher level of redundancy using knockout approach.

To further examine the contribution of these TFs in specific R protein mediated immunity, we challenged single, double and triple mutant plants with Pseudomonas syringae pv tomato (P.s.t.) carrying either avrRPS4 or hopA1, which are effectors recognized by TIR-NB-LRR proteins RPS4 and RPS6, respectively. As shown in Figure 2E, significantly more P.s.t. avrRPS4 growth was observed in bhlh84 rsl2rsl4 triple mutant plant, while no detectable difference was observed when the TF mutants were challenged with P.s.t. hopA1 (Figure S2), suggesting that these bHLH TFs contribute redundantly to RPS4-mediated immunity.

Simultaneously knocking out bHLH84, RSL2 and RSL4 partially suppresses the autoimmunity of snc1

To investigate the biological function of bHLH84 and its paralogs in snc1-mediated immunity, we crossed bhlh84 rsl2 with snc1 and isolated triple mutant snc1 bhlh84 rsl2. The dwarf phenotype of snc1 was not suppressed in the triple mutant (Figure 3A). We further crossed snc1 bhlh84 rsl2 with rsl4 rsl2 [29] and isolated quadruple mutant snc1 bhlh84 rsl2 rsl4 from the F2 generation by genotyping bhlh84, rsl4 and snc1 loci. The quadruple mutant plants were significantly larger than those of snc1 (Figure 3A). Consistent with the morphological suppression, the expression of PR1 and PR2 in the quadruple mutant was significantly decreased compared to snc1 plants while only slight reduction was observed in the triple mutant (Figure 3B). In addition, when the quadruple mutant seedlings were challenged with H.a. Noco2 and P.s.m. ES4326, more pathogen growth was observed compared to snc1, although the resistance was not restored to wild type levels (Figure 3C and 3D). Taken together, the bhlh84 rsl2 rsl4 triple mutant partially suppresses snc1, suggesting that bHLH84 and its paralogs are functionally redundant and required for the autoimmunity of snc1.

When we further isolated snc1 rsl2 rsl4 (Figure S3A and S3B), the triple mutant was slightly larger than snc1. Since snc1 bhlh84 rsl2 plants were indistinguishable from snc1 in size, it can thus be concluded that these three TFs are not equally redundant; RSL4 seems to play a slightly larger role than bHLH84 in snc1-mediated autoimmunity.

Overexpression of bHLH84, RSL2, or RSL4 exhibits extreme dwarfism likely due to autoimmunity

To further test the redundant roles of bHLH84 and its paralogs, we overexpressed bHLH84, RSL4 or RSL2 in Col-0 by transforming plants with the coding sequence of each gene without any epitope tags under the control of the 35S promoter. When screening T1 populations, multiple plants with extremely dwarf morphology were observed for each genotype (Figure 4). Intriguingly, plants of intermediate sizes were observed in the transgenic lines overexpressing bHLH84, while the majority of the plants overexpressing RSL4 or RSL2 were tiny and gradually perished, presumably as a result of extreme autoimmunity. The phenotypic similarity in these overexpression progeny further supports the functional redundancy among these three TFs in regulating plant immunity.

SNC1 contributes to the constitutive activation of defense responses in OXbHLH84-GFP-HA transgenic plants

As with snc1, the dwarf morphology of OXbHLH84-GFP-HA plants was largely suppressed when grown at 28°C (Figure S4) [30]. This observation led us to ask whether SNC1 is required for the autoimmunity of OXbHLH84-GFP-HA. As shown in Figure 5A, the snc1-r1 allele (a loss-of-function allele of SNC1 in which 8 bp of the first exon of SNC1 is deleted from fast neutron mutagenesis; [20]) could largely suppress the dwarf morphology of OXbHLH84-GFP-HA. Consistent with the observed morphological suppression, defense response phenotypes conferred by OXbHLH84-GFP-HA, including up-regulation of PR gene expression and resistance to P.s.m. ES4326 and H.a. Noco2, were significantly suppressed by snc1-r1 (Figures 5B, 5C and 5D), indicating that a functional SNC1 is indispensable for the effects of bHLH84 overexpression. As CPR1 (CONSTITUTIVE EXRPRESSER OF PR GENES 1) targets SNC1 for degradation [31], we crossed OXbHLH84-GFP-HA with plants overexpressing CPR1 (OXCPR1). The dwarf morphology and enhanced resistance of OXbHLH84-GFP-HA were largely suppressed (Figure 5), providing further support that SNC1 contributes to the autoimmune phenotypes associated with OXbHLH84-GFP-HA. In addition, the bHLH84-GFP-HA protein level in snc1-r1 or OXCPR1 background was not changed (Figure S5), suggesting that SNC1 does not affect bHLH84 protein accumulation.

Epistasis analysis reveals that constitutive activation of defense responses in OXbHLH84-GFP-HA is EDS1- and SID2- dependent and NDR1- independent

To further dissect the function of bHLH84 in plant defense pathways, OXbHLH84-GFP-HA was crossed with various mutants of key components in plant immunity, including eds1-2, sid2-2, and ndr1-1 [8], [32], [33]. As shown in Figure 5A, eds1-2 and sid2-2 could fully and partially suppress the morphology of OXbHLH84-GFP-HA in terms of leaf shape and plant size, respectively, while ndr1-1 had little effect. The enhanced PR gene expression and resistance to H.a. Noco2 and P.s.m. ES4326 were fully suppressed by eds1-2 and partially by sid2-2 (Figure 5B, 5C and 5D), indicating that EDS1 and SA are required for the autoimmunity in OXbHLH84-GFP-HA. In contrast, ndr1-1 was not able to suppress the enhanced PR gene expression, H.a. Noco2 and P.s.m. ES4326 resistance conferred by OXbHLH84-GFP-HA, indicating that the constitutive activation of defense responses in OXbHLH84-GFP-HA is NDR1-independent.

bHLH84 does not directly regulate SNC1 transcription

As SNC1 is required for the constitutive activation of the defense responses of OXbHLH84-GFP-HA plants, we asked whether bHLH84 could directly regulate SNC1 transcription. We observed that the transcription and protein levels of SNC1 in OXbHLH84-GFP-HA plants were slightly higher than in WT (Figure S6). However, this up-regulation of SNC1 is probably due to the positive feed-back effect resulting from the high SA in the autoimmune transgenic plants [34]. To avoid interference from the feed-back up-regulation of SNC1, we used OXbHLH84-GFP-HA eds1-2 plants to examine SNC1 transcription level. Real-time PCR showed that no significant change in SNC1 transcription was detected in OXbHLH84-GFP-HA eds1-2 compared to eds1-2 control plants (Figure 6A). As a consequence, the SNC1 protein level in OXbHLH84-GFP-HA eds1-2 was similar to that of eds1-2 (Figure 6B). In addition, we tested the transcript levels of selected R genes including RPS6, RPS4, RPP2, RPP4, RPS2, RPS5, and RPM1 in the OXbHLH84-GFP-HA eds1-2 background. Similar to SNC1, none of the tested R genes showed over 1.2-fold transcriptional changes when compared to eds1-2 (Figure S7A). In addition, no significant up-regulation of R genes was observed in OXbHLH84-GFP-HA snc1-r1 double mutant compared to snc1-r1 control plants (Figure S7B). Taken together, bHLH84 does not seem to participate in the direct transcriptional regulation of SNC1 or other tested R genes, unless bHLH84 recruits both EDS1 and SNC1 for this regulation.

bHLH84 interacts with SNC1 and RPS4 in planta

As the dependence of OXbHLH84-GFP-HA on a functional SNC1 and the partial suppression of snc1 by bhlh84 rsl2 rsl4 resembles the genetic interactions between SNC1 and TPR1/MOS10, and SNC1 interacts with TPR1 [17], we further tested whether bHLH84 associates with SNC1. We attempted a nuclear co-immunoprecipitation (co-IP) experiment using OXbHLH84-GFP-HA transgenic plants, which carry C-terminal GFP and HA double tags. Unfortunately, we were unable to detect the bait after immunoprecipitation in the elution, while all the proteins were found in the flow-through fraction (Figure S8A). As an alternate approach, we transformed Arabidopsis plants with a construct expressing bHLH84 under its native promoter and containing an N-terminal GFP tag. The protein produced was functional, as the transgenic plants resembled the original OXbHLH84-GFP-HA plants (Figure S8B). However, when they were used for co-IP with anti-GFP beads, the bait still could not be pulled down (Figure S8C). The inability of bHLH84 to be pulled down using immunoprecipitation could be due to unknown structural complexity of the protein. Since we were not able to carry out a co-IP experiment with bHLH84 as bait using epitope-tagged bHLH84 transgenic plants, we decided to examine the interaction between SNC1 and bHLH84 using the Nicotiana benthamiana transient expression system [35]. Interestingly, when both proteins were expressed in N. benthamiana leaves, we consistently observed a faster hypersensitive response (HR), which was obvious a few hours earlier compared to when SNC1-FLAG was expressed with the control vector (Figure S9A and S9B). This was further confirmed by the ion leakage analysis of the infiltrated leaves (Figure 7A). Both proteins were expressed efficiently in N. benthamiana (Figure 7B). When co-immunoprecipitation was carried out, SNC1-FLAG could specifically pull down bHLH84-HA, but not an unrelated nuclear protein MAC5A-HA (Figure 7C, [36]), indicating that bHLH84 can interact with SNC1 in planta.

As bHLH84 is able to interact with SNC1 in planta, we further examined the interaction specificity between bHLH84-HA and other R proteins by conducting co-IP of bHLH84-HA with either RPS4-FLAG, RPS2-FLAG or RPS6-FLAG. As shown in Figure 7D, RPS4-FLAG could also immunoprecipitate bHLH84-HA, although not as efficiently as SNC1-FLAG. However, RPS2-FLAG or RPS6-FLAG could not pull down bHLH84-HA (Figure S10). Taken together, bHLH84-HA can specifically interact with SNC1-FLAG or RPS4-FLAG in planta.

SNC1 was previously shown to interact with transcriptional co-repressor TPR1, which does not contain a DNA binding domain [17]. Additionally, the SNC1-dependent phenotypes observed upon overexpressing bHLH84 are similar to those observed when TPR1 is overexpressed. We therefore asked whether bHLH84 interacts with TPR1. As shown in Figure 7E, bHLH84-HA could not be pulled down by TPR1-FLAG, indicating that bHLH84 does not interact with TPR1 in planta. In addition, when we co-expressed SNC1-FLAG, bHLH84-HA and TPR1-HA in N. benthamiana, SNC1-FLAG was able to pull down both TPR1-HA and bHLH84-HA (Figure S11). The IP efficiency of TPR1-HA by SNC1-FLAG with all three proteins expressed was comparable to that with only TPR1-HA and SNC1-FLAG expressed. On the other hand, the IP efficiency of bHLH84-HA by SNC1-FLAG varied from trial to trial. Taken together, these data suggest that the interactions of SNC1-bHLH84 and SNC1-TPR1 in planta are independent, although whether there is competition between bHLH84 and TPR1 in associating with SNC1 is unclear.

To further investigate whether bHLH84 is able to directly interact with SNC1, we carried out yeast-two-hybrid experiment by co-transforming bHLH84 fused with AD and SNC1 fused with BD. Since we failed in making a full-length SNC1 construct, we made truncated SNC1 segments. As shown in Figure S12, yeast cells transformed with bHLH84-AD and different truncated SNC1 fused with BD were not able to grow on the selection plates, suggesting that bHLH84 does not directly interact with the truncated SNC1 segments in yeast. Moreover, the interaction between bHLH84 and SNC1 probably demands a properly folded full-length SNC1 or an intermediate partner. As EDS1 is required for the function of bHLH84 and EDS1 was shown to interact with SNC1 [37], we asked whether EDS1 or its interacting protein PAD4 [7] might be the intermediate partner. However, we did not detect interaction between bHLH84 and EDS1or bHLH84 and PAD4 (Figure S13), suggesting that EDS1or PAD4 is not likely mediating the interaction between SNC1 and bHLH84.

Discussion

From a targeted reverse genetic screen, we have identified a group of TFs, bHLH84 and its paralogs RSL2 and RSL4, which serve as transcriptional regulators for plant immunity. bHLH84 constitutively activates defense responses when overexpressed, and this activation is SNC1-dependent. bHLH84 was further demonstrated to be a transcriptional activator. In addition, the autoimmune phenotypes of snc1 can be partially suppressed by bhlh84 rsl2 rsl4 triple mutant, suggesting that bHLH84 and SNC1 are mutually dependent. bHLH84 does not seem to directly regulate the transcription of SNC1 or other tested R genes. However, the specific interaction between bHLH84 and NLRs including SNC1 and RPS4 in planta suggests that it associates with nuclear NLRs to mediate downstream transcriptional reprogramming. As we failed to observe association between bHLH84 and the repressor protein TPR1 which also interacts with SNC1, we propose that bHLH84 activates defense responses by forming a complex with SNC1 that functions in parallel with the SNC1-TPR1 complex to activate downstream positive regulators (Figure S14).

The targeted reverse genetic screen is a useful approach to identify new players in biological pathways

Previous work on MLA, N, RRS1 and SNC1 suggests that the interactions between some nuclear R proteins and their associating TFs are essential in regulating defense responses [9], [11], [12], [15], [17], [18]. Different approaches have been utilized to isolate TFs that are able to interact with nuclear R proteins. TPR1, which associates with SNC1 to repress negative regulators of immunity, was isolated from a forward genetic screen for suppressors of snc1 [17]. Yeast-two-hybrid screens have been successfully used to identify TFs in plant immunity. For example, SPL6 was initially identified from a yeast-two-hybrid screen and was further confirmed to interact with N in tobacco [18]. In addition, identified from yeast-two-hybrid screens, MYB6 and WRKY1 were shown to interact with MLA in barley to initiate disease resistance signaling in an antagonistic manner [15]. In this study, we used an alternative reverse genetic screen and successfully identified a group of novel TFs that play critical roles in plant immunity.

Our targeted reverse genetic approach has several advantages. Since plant defense to UV radiation is regulated by many of the same factors as pathogen resistance [23]–[25], while UV treatment datasets exclude a large number of genes that are manipulated by pathogen effectors which are not directly related to defense responses [38], the number of target genes we chose from the UV-induced database is more manageable for a reverse genetics screen. All the selected TFs were overexpressed in both Col-0 and snc1 backgrounds, facilitating rapid identification of both defense enhancers and suppressors (Table S1). Furthermore, the functional redundancy predicament often encountered in forward genetic screens can be effectively avoided by using the overexpression approach. Finally, our approach evades self-activation problems that are often associated with yeast-two-hybrid screens for transcriptional activators. Specifically, bHLH84 exhibits strong self-activation when fused with GAL4 binding domain in yeast (data not shown), thus cannot be identified from a yeast-two-hybrid screen. However, our screen does rely on the availability of high-quality microarray data, which may still overlook TFs with relatively low expression level changes.

bHLH84 functions as a transcriptional activator that is able to bind N1- or N2-boxes

As bHLH84 was shown to be a transcriptional activator, we attempted chromatin immunoprecipitation (ChIP) to identify target genes of bHLH84. However, as with our co-IP experiments (Figure S8), the bHLH84-GFP-HA protein could not be pulled down when subjected to ChIP (Figure S15). Thus we were unable to identify the target DNA of bHLH84 in planta. Using yeast-one-hybrid assay as an alternative approach, we attempted to identify the DNA-binding sequences of bHLH84. Many bHLH type TFs were shown to bind sequences containing a consensus core element E-box (5′-CANNTG-3′), with the palindromic G-box (5′-CACGTG-3′) being the most typical form [39]. Some bHLH proteins bind to non-E-box sequences (N-box), such as 5′-CACGc/aG-3′ and 5′-CGCGTG-3′ [40], [41]. As shown in Figure S16A and Figure S16B, compared with the bHLH84 alone or cis-element alone negative controls, the most enhanced yeast growth was observed on SD-Leu-Trp-His media when AD-bHLH84 was co-transformed with pHIS2-N1-box, while considerably enhanced growth was observed when AD-bHLH84 was co-transformed with pHIS2-N2-box. No enhanced yeast growth was observed in G-box or N3 box co-transformations. These data suggest that bHLH84 is able to bind N1- and N2-boxes, but not N3- or G-boxes. These data are consistent with the prediction that TFs in this bHLH subfamily are non E-box binders [42]. Although the potential binding sites of bHLH84 have been revealed, it is still difficult to predict its target genes. More sophisticated ChIP experiments designed in the future may be able to solve this problem.

bHLH84 and its paralogs are implicated in plant immunity

The bHLH-containing proteins constitute a large conserved TF family in eukaryotes [43], [44]. They have been studied intensively in yeast and humans, providing evidence for their regulatory functions in cell proliferation and cellular differentiation pathways [45]–[48].

While only a few bHLH proteins have been studied in detail in plants, they have been shown to serve regulatory functions in multiple biological pathways. For example, a group of bHLH TFs in Zea mays regulate the production of the purple anthocyanin pigments by interacting with R2R3-MYB TFs [49]. In Arabidopsis, GL3 (GLABRA3) regulates trichome development through its interaction with MYB-like TF GL1(GLABRA1) [50]. Another small subfamily of bHLH TFs, referred to as phytochrome-interacting factors (PIFs), have been shown to play diverse functions including regulating light signaling pathways, seed germination, seedling photomorphogenesis, and shade avoidance responses via their interactions with phytochromes [51]–[56]. In addition, JAM1 (ABA-INDUCIBLE BHLH-TYPE TRANSCRIPTION FACTOR/JA-ASSOCIATED MYC2-LIKE), acts as a transcriptional repressor and negatively regulates JA signaling [57]. bHLH84 and its paralogs have previously been shown to regulate root hair elongation [29], [58]. However, they are the first few bHLH TFs found to be involved in plant immunity. Since bHLH TFs form one of the largest TF families in plants, it is difficult to imagine that these three TFs are the only bHLHs involved in immune regulation. Lethality of the knockout mutants or redundancy could be the factors prohibiting others from being discovered. Future novel methods, such as our overexpression approach, may facilitate the functional studies of more TFs in large families.

bHLH84 and its paralogs function redundantly in NLR-mediated immunity

As one of the largest TF families in Arabidopsis with 147 members, bHLH TFs are further subdivided into 12 major subfamilies based on sequence similarity. bHLH84 and its paralogs belong to the VIIIc subgroup [28]. In this study, we have experimentally shown that bHLH84, RSL2 and RSL4 redundantly regulate defense responses. Overexpression of any of these proteins results in constitutive activation of defense responses (Figure 4). Their redundancy was further demonstrated using the triple mutant of bhlh84 rsl4 rsl2, which is able to partially suppress the autoimmune phenotypes of snc1 (Figure 3), and compromise RPS4-mediated defense responses (Figure 2E). It is possible that additional members of the VIIIc subfamily are also functionally redundant with bHLH84. Future construction of higher order bhlh mutants may provide insight into the additional redundant relationships among these family members.

Typically, the bHLH domain contains approximately 60 amino acids and is comprised of a stretch of hydrophilic and basic residues at the N terminus, followed by two amphipathic alpha-helices connected by an intervening loop [44]. The helix-loop-helix and the basic region of the bHLH are required for DNA-binding, whereas the helix-loop-helix region alone often enables homo- or heterodimerization with other bHLH proteins. Since the single mutants of bhlh84, rsl2 and rsl4 do not exhibit obvious phenotypes, we speculate that if dimerization occurs, it would most likely be homodimerization rather than heterodimerization. The dimerized bHLH84 or its paralogs may bind to the same DNA region, thus regulating immunity in a similar manner. In addition, bHLH TFs often associate with other types of TFs, including MYBs and bZIPs for transcriptional reprogramming [49], [56], thus we cannot exclude the possibility that there are more unknown TFs that are also involved in the bHLH84-SNC1 complex.

As the expression level of SNC1 is comparable in eds1-2 and OXbHLH84-GFP-HA eds1-2 backgrounds (Figure 6), bHLH84 does not seem to regulate SNC1 expression. In addition, we did not observe transcriptional up-regulation of tested R genes in OXbHLH84-GFP-HA snc1-r1 or OXbHLH84-GFP-HA eds1-2 plants (Figure S7), suggesting that bHLH84 does not directly regulate the transcription of R genes.

As we also detected attenuated immunity against P.s.t. avrRps4 in bhlh84 rsl2 rsl4 triple mutant (Figure 2E), and interaction between RPS4 and bHLH84 in N.benthamina (Figure 7D), bHLH84 and its paralogs seem to be not just specific to SNC1. As both RPS4's and SNC1's nuclear localizations are critical to their defense activation [10], [14], we speculate that these bHLH TFs may work together with selective nuclear TIR-NB-LRRs to trigger downstream immunity. More in-depth investigations on the interactions of other nuclear TIR-NB-LRR proteins with these TFs might reveal more R proteins working together with these bHLH proteins.

bHLH84 and TPR1 function in parallel to regulate SNC1-mediated resistance

Overexpression of either bHLH84 or TPR1 results in SNC1-dependent autoimmunity, indicating that both bHLH84 and TPR1 positively regulate SNC1-mediated defense responses. Both bHLH84 and TPR1 were shown to associate with SNC1, although no interaction was detected between bHLH84 and TPR1, suggesting that bHLH84-SNC1 and TPR1-SNC1 probably function in distinct complexes (Figure 7, S11 and S14). Their downstream target genes are probably different, as bHLH84 is a transcriptional activator while TPRs are repressors. Defense activation induced by SNC1 is likely achieved through a combination of activation of positive regulators and repression of negative regulators.

Materials and Methods

Construction of plasmids

The genomic sequences of selected TFs, excluding the stop codon and including approximately 1.5 kb sequence upstream of the start codon, were amplified by PCR with two different restriction enzyme sites separately introduced at the two primer ends. The chosen restriction enzyme sites were KpnI, SalI, SacI, XbaI or PstI. The amplified fragments were then digested and ligated to modified pCambia1305 vectors harboring C-terminal GFP and HA tags. These constructs were transformed into snc1 and Col-0 using the floral dip method [26].

For overexpression of bHLH84, RSL2 and RSL4, coding sequences of the genes were amplified by PCR with two different restriction enzyme sites separately introduced at the two primer ends. The primer sequences can be found in Table S2. The fragments were then digested and ligated to the pG229HAN vector with a 35S promoter.

For the pCambia1300-35S-SNC1-FLAG, pCambia1300-35S-RPS4-FLAG and pCambia1300-35S-RPS6-FLAG constructs used in the transient expression in N. benthamiana, the genomic region of SNC1, RPS4 or RPS6 without the stop codon, was cloned into the pCambia1300 vector with a 35S promoter and a C-terminus FLAG tag. For other pCambia1300 constructs used in the transient expression, the CDS regions of the genes were cloned into the corresponding vectors. The primer sequences can be found in Table S2

Transgenic screening

Approximately 0.4 g of T1 transgenic seeds for each construct were first plated on solid MS medium containing 30 µg/ml Hygromycin B. 48 one-week-old transformant seedlings per genotype were selected and subsequently transplanted on soil. Col-0 and snc1 seeds were planted on solid MS medium without any selection and transplanted on soil at the same time to serve as controls. Among the transgenic plants of each genotype, the transformants which showed varied sizes were kept, and T2 seeds from these plants were planted on Hygromycin B plates to analyze transgene copy number, check for the presence of the transgene and validate the background using primers specific to the SNC1 locus [20]. The transgenic plants with heritable phenotypes and with the correct backgrounds were then subjected to H.a. Noco2 infection to examine whether their altered morphology is correlated with altered resistance. Resistance was scored based on the degree of deviation from that observed in the control plants. More specifically, transgenic plants in Col-0 background showing similar sporulation as Col-0 were scored as no change (NC). Plants showing less sporulation than Col-0 were scored as showing enhanced resistance phenotype with “+”. Plants exhibiting a little sporulation were scored as having more enhanced resistance phenotype with “++”, while the ones showing no sporulation were scored as the most enhanced resistance phenotype as “+++”. For transgenic plants in the snc1 background, plants showing more sporulation than snc1 were scored as suppressing phenotype with “−”, while the ones showing less sporulation than snc1 were scored as enhancing phenotype with “+”.

Confocal microscopy

Leaves from one-week-old seedlings were soaked in 1 mg/mL (1∶1 [g/v]) propidium iodide (PI) for 3 minutes and rinsed briefly with water before visualization. Root tissues were submerged in 1 µg/ml (1∶1 [g/v]) PI for 10 seconds and mounted in water. For GFP and PI visualization, a Nikon ECLIPSE 80i Confocal microscope was used under 488 nm and 543 nm filter sets.

Transient protein expression and co-immunoprecipitation in N. benthamiana

Transient protein expression in N. benthamiana was carried out as previously described [35]. The IP protocol was modified from [59]. Briefly, Agrobacteria containing the binary vector pCambia1300 constructed with the target genes and tags were cultured in LB media with kanamycin selection at 28°C overnight. The bacteria were inoculated into a new culture media (10.5 g/L K₂HPO₄, 4.5 g/L KH₂PO₄, 1.0 g/L (NH₄)₂SO4, 0.5 g/L NaCitrate, 1 mM MgSO₄, 0.2% glucose, 0.5% glycerol, 50 µM acetosyringone, and 10 mM N-morpholino-ethanesulfonic acid (MES) (pH 5.6), 50 µg/mL Kanamycin) by 1∶50 dilution and cultured for a further 8–12 hours. The bacteria were then harvested by centrifugation at 4000 rpm for 10 minutes and resuspended in MS buffer (4.4 g/L MS, 10 mM MES, 150 µM acetosyringone) to a final concentration of OD₆₀₀ = 0.2 for infiltration into four-week-old N. benthamiana leaves.

For co-immunoprecipitation, 3 g of N. benthamiana leaves were collected at 36 hours post-infiltration and ground into fine powder in liquid nitrogen using a cold mortar and pestle. The powder was mixed with 6 ml extraction buffer (10% glycerol, 25 mM Tris pH 7.5, 1 mM EDTA, 150 mM NaCl, 10 mM DTT, 2% w/v PVPP, protease inhibitor cocktail) and homogenized by further grinding. All the following steps were carried out at 4°C. The samples were centrifuged at 15000 g for 10 minutes and the supernatants were transferred to new tubes. These two steps were repeated twice before NP40 (Nonidet P-40 Substitute) was added into each supernatant to a final concentration of 0.15%. 30 µl pre-washed protein A or protein G agarose beads were added into each supernatant and incubated for 30 minutes. The mixtures were centrifuged at 4000 rpm for 2 minutes to remove the beads. Each supernatant was incubated with 30 µl anti-FLAG beads or protein A agarose beads for 3 hours, and the beads were pelleted down by centrifuging at 8000 rpm for 1 minute and washed 8 times using extraction buffer containing 0.15% NP40. Proteins specifically bound to the beads were competitively eluted using 100 µl 250 µg/ml 3×FLAG peptides. All the samples were boiled in SDS loading buffer for 5 minutes before running on SDS-PAGE gel.

Arabidopsis protoplast transient assay for transcriptional activity

The isolation and transfection of Arabidopsis protoplasts and the reporter gene assay were previously described in [27]. Briefly, the Arabidopsis protoplasts were transfected with the reporter construct, the effector construct and the internal control construct as illustrated in Figure 2A. GUS expression was determined using MUG assay (Acros Organics from Fisher Scientific). Fluorescence was measured using a fluorescence spectrophotometer (360/460 nm). The internal LUC expression was examined using a Dual-Luciferase reporter assay system (Promega, E1910).

Ion leakage assay

The ion leakage assay was performed as previously described [60], with a few modifications. Briefly, twelve leaf discs (7 mm in diameter) per measurement were punched from the infiltrated area at 23 hr post infiltration and placed in a 60 mm petri dish containing 10 ml of ddH₂O. After 30 minutes, the water was removed and another 10 ml of ddH₂O was added into the petri dish containing the leaf discs. Conductivity was measured using a 545 Conductivity Multi-purpose Cell (VWR Scientific) at the indicated time points.

Yeast-one-hybrid and yeast-two-hybrid assays

For yeast-one-hybrid assay, the pHIS2 derivatives (harboring the N1-, N2-, N3- and G-box cis-elements) were co-transformed with the construct of pAD-bHLH84 into the yeast strain Y187. For each co-transformation of pAD-bHLH84 and pHIS2 derivatives, yeast cells co-transformed with pHIS2 empty vector (EV) and pAD-bHLH84 as well as yeast cells cotransformed with pAD EV and the pHIS2 derivatives were used as negative controls. The positive transformants were isolated from SD-Trp-Leu medium. The transformants were then analyzed on the SD-Trp-Leu-His medium supplemented with 60 mM and 100 mM 3-Amino-1,2,4-Triazole (3AT).

For yeast-two-hybrid assays, the pGBKT7 derivatives containing various truncated SNC1 fragments were co-transformed with pAD-bHLH84 into yeast strain Y1347. pGBKT7 EV cotransformed with pAD-bHLH84 was used as a negative control. The positive transformants were isolated from SD-Trp-Leu medium. The transformants were then analyzed on SD-Trp-Leu-His medium supplemented with 3 mM 3AT.

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6

Attachment 7

Attachment 8

Attachment 9

Attachment 10

Attachment 11

Attachment 12

Attachment 13

Attachment 14

Attachment 15

Attachment 16

Attachment 17

Attachment 18