Pseudomonas syringae type III effectors are known to suppress plant immunity to promote bacterial virulence. However, the activities and targets of these effectors are not well understood.
We used genetic, molecular, and cell biology methods to characterize the activities, localization, and target of the HopD1 type III effector in Arabidopsis.
HopD1 contributes to P. syringae virulence in Arabidopsis and reduces effector-triggered immunity (ETI) responses but not pathogen-associated molecular pattern-triggered immunity (PTI) responses. Plants expressing HopD1 supported increased growth of ETI-inducing P. syringae strains compared with wild-type Arabidopsis. We show that HopD1 interacts with the membrane-tethered Arabidopsis transcription factor NTL9 and demonstrate that this interaction occurs at the endoplasmic reticulum (ER). A P. syringae hopD1 mutant and ETI-inducing P. syringae strains exhibited enhanced growth on Arabidopsis ntl9 mutant plants. Conversely, growth of P. syringae strains was reduced in plants expressing a constitutively active NTL9 derivative, indicating that NTL9 is a positive regulator of plant immunity. Furthermore, HopD1 inhibited the induction of NTL9-regulated genes during ETI but not PTI.
HopD1 contributes to P. syringae virulence in part by targeting NTL9, resulting in the suppression of ETI responses but not PTI responses and the promotion of plant pathogenicity.
The Gram-negative bacterial pathogen Pseudomonas syringae pv. tomato DC3000 is the causative agent of bacterial speck disease on tomato and pathogenic on the model plant Arabidopsis thaliana. One of its primary virulence factors is its ability to inject type III effector (T3E) proteins into host cells using a type III protein secretion system. Once injected, T3Es target and disrupt various host processes in order to suppress plant immunity (Block & Alfano, 2011; Deslandes & Rivas, 2012; Feng & Zhou, 2012).
Plant innate immunity can be broadly separated into two branches based on how microorganism recognition is achieved. The first of these branches involves the recognition of pathogen-associated molecular patterns (PAMPs) by extracellular pattern recognition receptors (PRRs; Jones & Dangl, 2006). This recognition induces multiple responses, including signaling cascades involving mitogen-activated protein kinases and calcium-dependent protein kinases. These signaling events lead to a rapid burst of reactive oxygen species (ROS) production and gene expression changes (Monaghan & Zipfel, 2012). Later responses include cell wall modifications such as the deposition of callose. The combined response arising from this recognition event is termed PAMP-triggered immunity (PTI; Jones & Dangl, 2006; Monaghan & Zipfel, 2012).
Pseudomonas syringae can use several of its injected T3Es to suppress PTI and achieve high growth rates within its host plants. To combat PTI suppression by T3Es and other pathogen effectors, plants evolved a second layer of immunity based on the recognition of effectors or their activities by plant immune receptors, know as resistance (R) proteins. The response from this recognition event is termed effector-triggered immunity (ETI) and induces many of the same responses as PTI but often with different timing and amplitude (Tsuda & Katagiri, 2010). ETI is usually associated with a form of programmed cell death known as the hypersensitive response (HR). Thus, in order to maintain its pathogen status, P. syringae probably adapted/evolved T3Es to suppress PTI and/or ETI.
If a P. syringae strain lacks the ability to successfully suppress ETI, bacterial growth is restricted. This is classically referred to as an incompatible interaction between an avirulent strain and a resistant plant (Keen, 1990). If the P. syringae strain can suppress PTI and ETI, or does not elicit ETI but suppresses PTI, it can be pathogenic and achieve high growth rates in plant tissue. This is classically referred to as a compatible interaction between a virulent strain and susceptible plant. Many P. syringae T3Es have been shown to suppress PTI responses and several suppress both PTI and ETI (Block et al., 2008; Gimenez-Ibanez & Rathjen, 2010; Feng & Zhou, 2012). However, it is less clear whether P. syringae T3Es have evolved to suppress ETI, but not PTI.
In a previous study we screened DC3000 T3Es for their ability to suppress ETI responses induced by the T3E HopA1 in Nicotiana tabacum (Jamir et al., 2004; Guo et al., 2009). One of the strong ETI suppressors identified in this screen was HopD1. HopD1 is a homolog of AvrPphD from P. syringae pv. phaseolicola race 4 strain 1302A (Wood et al., 1994). A P. syringae pv. phaseolicola avrPphD mutant showed no difference in growth in either resistant or susceptible pea cultivars, indicating that its immune suppression/induction functions are probably redundant with other T3Es in this pathogen (Arnold et al., 2001). Bacterial mutants lacking HopD1 homologs have differing effects on the virulence of other bacterial pathogens (Guo et al., 2002; Schulze et al., 2012). Therefore, depending on the pathogen and its likely redundancy with other T3Es, loss of hopD1 can reduce the virulence of a pathogen.
HopD1 has no known mechanism of action or homology to proteins with a known enzymatic function. A recent study used a high-throughput yeast two-hybrid screen to define the interactomes of P. syringae T3Es and the effectors from the oomycete pathogen Hyaloperonospora arabidopsidis with Arabidopsis proteins (Mukhtar et al., 2011). One potential interaction identified in this screen was that of HopD1 and the Arabidopsis NTL9 (At4g35580) protein (Mukhtar et al., 2011). NTL9 belongs to the NAC with Transmembrane Motif 1 (NTM1)-like family of transcription factors that are tethered to intracellular membranes (Kim et al., 2006). NAC transcription factors play a wide variety of roles in developmental and stress-related signaling and can also be involved in regulating cross-talk between different stresses (Puranik et al., 2012).
Here we show that HopD1 is an important DC3000 virulence factor. We show that it is a strong ETI suppressor in Arabidopsis but does not significantly suppress PTI responses. Additionally, we confirm that HopD1 interacts with NTL9 and that both HopD1 and NTL9 localize to the plant endoplasmic reticulum (ER). NTL9 is important for the Arabidopsis ETI response as ntl9 knockout plants allow for increased growth of ETI-inducing P. syringae strains and rescue the growth of a DC3000 hopD1 mutant. In addition, plants expressing constitutively active NTL9 are more resistant to virulent and ETI-inducing P. syringae strains, further indicating that NTL9 is a positive regulator of plant immunity. We identify several genes whose expression is induced by active NTL9 and show that HopD1 suppresses the induction of these genes during ETI but not PTI. These data indicate that HopD1 functions as a specific suppressor of ETI, in part by inhibiting NTL9-mediated gene expression.
Materials and Methods
Cloning and expression constructs
Genes were cloned into the Gateway entry vector pENTR-D. The pENTR-D constructs and their respective cloning primers are as follows: hopD1 with a ribosome binding site (pLN3228) primers P0967 (5′-CACCGGGACAGCTGATAGAACAATGA ATCCTCTACGCTC-3′) and P2892(5′-GGGTGCGGGCTGCCGCGA-3′); hopD1 (pLN5056) primers P4373 (5′-CACCATGAATCCTCTACGATCTATTCAACAC-3′) and P2892; NTL9 (pLN5057) primers P4379 (5′-CACCATGGGTGCTGTATCGATGGAGTCG-3′) and P4393 (5′-TGAACTCACCAGTGTCCTCCACATCC-3′); and NTL91-330 (pLN5058) primers P4379 and P4382 (5′-GAAAGCCATGAAGTCGTTGAAAGCATCCTCTG-3′).
For the hopD1 bimolecular fluorescence complementation (BiFC) constructs, hopD1 with a 3′ XhoI restriction site was amplified with primers P4373 and P4384 (5′-GATCCTCGAGGGGTGCGGGCTG CCGCGACGTG-3′); cyfp with a 5′ XhoI restriction site was amplified with primers P4342 (5′-GATCCTCGAGATGGACAAGCAGAAGAACGGCATC-3′ – which contains an ATG start codon) and P4343 (5′-TCAGATAGATCTCTTGTACAGCTC-3′); and nyfp with a 5′ XhoI restriction site was amplified with primers P4340 (5′-GATCCTCGAGATGGTGAGCAAGGG CGAGGAG-3′) and P4341 (5′-TCAGGCCATGATATAGACGTTGTGG). The PCR products were then digested with XhoI and ligated. The ligation products were used as templates to clone hopD1-cyfp with primers P4373 and P4343 into pENTR-D, resulting in the construct pLN5066, and hopD1-nyfp with primers P4373 and P4341 into pENTR-D, resulting in the construct pLN5067. For the NTL9 BiFC constructs, NTL9 with a 5′ XhoI restriction site was amplified with primers P4403 (5′-GATCCTCGAGATGGTGCTGTATCGATGGAGTCG-3′) and P4393; cyfp with a 3′ XhoI restriction site was amplified with primers P4406 (5′-CACCATGGACAAGCAAGCAGAAGAACGGCATC-3′) and P4407 (5′-GATCCTCGAGGATAGATCTCTTGTACAGCTC-3′); and nyfp with a 3′ XhoI restriction site was amplified with primers P4404 (5′-CACCATGGTGAGCAAGGCGAGGAG-3′) and P4405 (5′-GATCCTCGAGGGCCTGATATAGACGTTGTGG-3′). The PCR products were then digested with XhoI and ligated. The ligation products were used as templates to clone nyfp-NTL9 with primers P4404 and P4393 into pENTR-D, resulting in the construct pLN5068, and cyfp-NTL9 with primers P4406 and P4393 into pENTR-D, resulting in the construct pLN5069.
The BiFC clones pLN5066, pLN5067, pLN5068 and pLN5069 were placed into the plant constitutive expression vector pLN462 (Jamir et al., 2004), resulting in constructs pLN5070, pLN5071, pLN5072 and pLN5073, respectively. For plant expression, hopD1 (pLN5056) was fused to a C-terminal hemagglutinin (HA) tag by recombining into pLN462 with gateway technologies to give pLN5060 and to a C-terminal green fluorescent protein (GFP) tag by recombining into pK7FWG2 (Karimi et al., 2002) to give pLN5061. For complementation of the hopD1 mutant (UNL104), hopD1 (pLN3228) was recombined into the Tn7 vector pLN2992 (Choi et al., 2005), resulting in construct pLN4908. This vector was then used to insert the resultant hopD1-ha into the chromosome of UNL104. To create an N-terminal GFP fusion to NTL9, pLN5057 was recombined with pK7WGF2 (Karimi et al., 2002) to give pLN5062. To make an estradiol-inducible NTL91-330-HA, pLN5058 was recombined with the gateway-compatible pER8 vector pLN604 (Zuo et al., 2000) to give pLN5063.
Pseudomonas syringae strains were grown overnight at 30°C on King's B (KB; King et al., 1954) media with the appropriate antibiotics and resuspended to an OD600 of 0.2 (2 × 108 cells ml−1) in 10 mM MgCl2. Cells were serially diluted in 10 mM MgCl2 to the appropriate cell density and infiltrated into the fully expanded leaves of 4-wk-old Arabidopsis thaliana Col-0 wild-type, mutant, and/or transgenic plants using a needleless syringe. Plants were kept at 100% humidity and 1 cm2 leaf disks were sampled at the indicated times, ground in 10 mM MgCl2, serially diluted and plated on KB agar plates. Plates were incubated for 48 h at 30°C and number of cells cm−2 was determined. Four to six leaf disks were sampled for each treatment and the statistical significance of the resulting data was analyzed by one-way ANOVA.
HR and ion leakage
Pseudomonas fluorescens (Pf)(pLN1965), Pf(pLN1965 + pavrRpm1) and Pf(pLN1965 + pavrRpm1 + phopD1-flag; Guo et al., 2009) were grown overnight at 30°C in KB media with appropriate antibiotics. Bacterial strains were resuspended at 1 × 108 cells ml−1 in 5 mM 2-(4-morpholino)-ethane sulfonic acid (MES), pH 5.6. Wild-type Arabidopsis (Col-0) and rpm1 mutant (Eitas et al., 2008) leaves were infiltrated with bacterial suspensions using a needleless syringe. Nine leaf disks were harvested for each strain in each plant type using a 0.7 cm2 cork borer. Three leaf disks were placed in each 15 ml polypropylene tube with 5 ml of water and incubated in a shaker for 30 min at room temperature. Water was removed and 5 ml of fresh water was added to each sample and tubes were placed in a shaker. Electrolyte leakage was monitored at the indicated time points using an electrical conductivity meter (Fisher Scientific, Pittsburgh, PA, USA). Conductivity was measured in µS cm–1. For macroscopic HR, Pf(pLN1965 + pavrRpm1) and Pf(pLN1965 + pavrRpm1 + phopD1-flag) strains were infiltrated into leaves of wild-type Arabidopsis (Col-0) and the rpm1 mutant at 1 × 108 cells ml−1 using a needleless syringe. Leaves were photographed 2 d after infiltration.
Wild-type (Col-0) and HopD1-HA transgenic lines were syringe-infiltrated with 10 μM flg22, or wild-type Arabidopsis leaves were syringe-inoculated with 1 × 106 cells ml−1 of Pf(pLN1965), Pf(pLN1965 + hopD1-flag), Pf(pLN1965 + pavrRpm1) or Pf(pLN1965 + pavrRpm1 + hopD1-flag). Sixteen hours later, leaves were harvested and cleared with 100% (v/v) ethanol at 37°C for 4 h and washed twice with 70% (v/v) ethanol and three times with water. The completely cleared leaves were stained with 0.1% (w/v) aniline blue in a solution of 150 mM K2HPO4, pH 9.5, for 30 min and the callose deposits were enumerated as in Block et al. (Block et al., 2010).
Oxidative burst measurement
Leaf disks of 0.5 cm2 were cut from wild-type (Col-0) and HopD1-HA transgenic lines and floated on 0.1 ml of water in wells of a 96-well plate for 16 h in the dark. The water was then removed and replaced with 0.5 mM L-012 (Wako, Japan) in 10 mM MES buffer, pH 7.4. For PAMP treatment, 1 μM of flg22, elf18 or chitin was added to the buffer. The rate of ROS production was determined by counting photons from L-012-mediated chemiluminescence using a luminometer (Asai et al., 2008). The rates of ROS production were calculated as the number of photons released in 30 min after treatment with photons counted once a minute.
Gateway entry constructs carrying hopD1 (pLN5056) and NTL9 (pLN5057) were recombined by an LR reaction to the appropriate yeast two-hybrid destination vectors to generate pDEST-DB::hopD1 (pLN4988) and pDEST-AD::NTL9 (pLN4970). A detailed protocol for the yeast-two hybrid analysis used is described in Dreze et al. (2010). The yeast strains Y8930 (MATα) and Y8800 (MATa) were transformed with expression plasmids pDEST-DB:;hopD1 and pDEST-AD:;NTL9, respectively. Empty expression vectors (EVs) were also transformed into the corresponding yeast strains. Yeast Y8930 (pDEST-DB::hopD1 or EV) was mated to Y8800 (pDEST-AD::NTL9 or EV) in a mating plate (yeast extract peptone dextrose). Growth was checked on Sc-Leu-Trp and interactions were tested by plating mated strains in Sc-Leu-Trp-His + 1 mM 3-amino-1,2,4-triazole (3AT).
Agroinfiltration and confocal microscopy
Agrobacterium strains were coinfiltrated into N. benthamiana as in Block et al. (2010). Leaves were imaged 48 h later using a Nikon A1 confocal mounted on an Eclipse 90i Nikon compound microscope using the following excitation (ex) and emission (em) wavelengths: GFP and yellow fluorescent protein (YFP), 488 nm (ex), 500–550 nm (em); red fluorescent protein (RFP), 561.5 nm (ex) and 570–620 nm (em); Chl, 641 nm (ex) and 662–737 nm (em). Dual color image acquisition was sequential.
Wild-type Arabidopsis (Col-0) and pER8 NTL91-330-HA stable Arabidopsis lines were syringe-infiltrated with 0.2 mM estradiol or a water control. Leaf tissue was harvested for RNA extraction 24 h after treatment. Wild-type Arabidopsis (Col-0) plants were syringe-inoculated with water or 1 × 107 cells ml−1 of Pf(pLN1965 + pavrRpm1) or Pf(pLN1965 + pavrRpm1 + phopD1-flag). Leaf tissue was harvested for RNA extraction 4 h after infiltration. Total RNA was purified using RNeasy mini Kit with on-column DNase treatment (Qiagen). The reverse transcription of RNA was carried out using RETROscript (Ambion) using oligo(dT) primers with heat denaturation of the RNA. actin2 (At3g18780) was used as a reference gene with primers P3774 (5′-GCACTTGTGTGTGACAAACTCTCTGG-3′) and P3775 (5′-GGCATCAATTCGATCACTCTAGAGC-3′). The following gene-specific primers were used to measure expression levels in quantitative (qRT-PCR): HAB1 (At1g72770) P4556 (5′-GCGGTGATTCGAGGGCGGTTT-3′) and P4557 (5′-GCCACGTTTGTGTGATGTGCATT-3′); NIP2 (At2g17730) P4558(5′-TCTTCAGGATTTCCAGCTCGGTGAA-3′) and P4559 (5′-CGGGCAAGAACCGTGTCTAAGGA-3′); ADF6 (At2g31200) P4560 (5′-TTGCTTGGTCTCCTTCGACCTCTGG-3′) and P4561 (5′-TCTCAGTTCGCTCGTTCGCGT-3′); PAL1 (At2 g37040) P3922 (5′-AGCAGCAAGAGCAGCCTACGATAA-3′) and P3923 (5′-TGTTCCAAGCTCTTCCCTCACGAA-3′); FRK1 (At2g19190) P4476 (5′-ACCCCGAGTACTATTCGACTCGCCA-3′) and P4477 (5′-TGAGCTTGCAATAGCAGGTTGGCCT-3′); and NHL10 (At2g35980) P4562 (5′-TCACTGTTCCTGTCCGTAACCCAA-3′) and P4563 (5′-TGGTACTAAACCGCTTTCCTCGT-3′). qRT-PCR was run using IQ™ SYBR® Green supermix (BioRad) on a BioRad iCycler. Gene expression relative to Col-0 was calculated using with actin2 as the reference gene and mock-treated Col-0 as the reference sample.
Additional information on the methods used for the data presented in the supporting information can be found in Supporting Information, Methods S1.
HopD1 contributes to P. syringae virulence and enhances the growth of ETI-inducing P. syringae strains
To determine the extent of HopD1's involvement in P. syringae virulence, we measured the in planta growth of a P. syringae pv. tomato DC3000 hopD1 mutant (UNL104; Jamir et al., 2004) in wild-type Arabidopsis (Col-0; Fig. 1a). Bacterial growth assays showed a slight but significant reduction in the growth of the hopD1 mutant in Arabidopsis when compared with wild-type DC3000. Wild-type growth rates were restored when hopD1 was reintroduced into the hopD1 mutant using a T7 expression system such that hopD1 was expressed in single copy from a type III promoter (Fig. 1a). These data show that HopD1 is required for the full virulence of P. syringae on Arabidopsis.
Next we examined the ability of HopD1 to promote bacterial growth in planta using transgenic Arabidopsis constitutively expressing HopD1 fused to a C-terminal HA tag (HopD1-HA). These plants displayed no obvious phenotypes and we confirmed that they produced HopD1-HA (Fig. S1). Wild-type Arabidopsis and HopD1-HA expressing Arabidopsis plants were syringe-inoculated with wild-type DC3000, the DC3000 hrcC mutant defective in type III secretion, or DC3000 strains carrying plasmids expressing the T3E genes avrRpm1 or avrRpt2, which encode ETI-inducing T3Es, rendering these strains avirulent. Growth of these strains was determined at 0 and 3 d postinoculation (Fig. 1b). In planta expression of HopD1-HA led to significantly increased growth of the two ETI-inducing avirulent strains but had no effect on the growth of wild-type DC3000 or the nonpathogenic hrcC mutant (Fig. 1b). These data show that HopD1 is a strong suppressor of ETI in Arabidopsis. The lack of an effect of HopD1 on the growth of the hrcC mutant, which cannot inject T3Es but carries PAMPs that induce PTI, suggests that HopD1 is unable to significantly suppress PTI.
Bacterially delivered HopD1 can suppress AvrRpm1-induced ETI in Arabidopsis
The increased growth of avirulent P. syringae strains on Arabidopsis expressing HopD1-HA indicates that HopD1 can suppress ETI induced by AvrRpm1 and AvrRpt2 in Arabidopsis. To further characterize the ability of HopD1 to suppress ETI, we used a nonpathogenic Pf strain carrying pLN1965, which encodes a functional P. syringae type III secretion system (Guo et al., 2009), and two additional plasmids expressing HopD1-FLAG and AvrRpm1. This allows the effect of HopD1 on ETI to be determined in the absence of other T3Es that may be redundant to HopD1.
To test the ability of HopD1 to suppress ETI, Pf(pLN1965 + pavrRpm1) carrying either a vector control or a plasmid constitutively expressing hopD1-flag were syringe-infiltrated into wild-type Arabidopsis leaves. The coexpression of hopD1-flag with avrRpm1 suppressed the macroscopic HR (Fig. 2a). AvrRpm1-induced ion leakage was also measured in wild-type Col-0 infiltrated with these strains. Bacterial delivery of HopD1-FLAG suppressed ion leakage, which is correlated with cell death, to such an extent that it resembled that of leaves infiltrated with Pf lacking the ETI inducer AvrRpm1 (Fig. 2b). Furthermore, the ETI-induced deposition of callose in the plant cell wall of wild-type Arabidopsis leaves infiltrated with Pf(pLN1965 + pavrRpm1 + phopD1-flag) was significantly lower than that in leaves infiltrated with Pf(pLN1965 + pavrRpm1; Fig. 2c). These data show that HopD1 can suppress ETI-induced HR and callose deposition.
To confirm that the HR and ion leakage suppression by HopD1 was the result of suppression of ETI, we included an Arabidopsis rpm1 knockout line that cannot recognize AvrRpm1 and, therefore, does not exhibit AvrRpm1-induced ETI responses. Using this Arabidopsis rpm1 mutant, there was no induction of an HR by either Pf(pLN1965 + pavrRpm1) or Pf(pLN1965 + pavrRpm1 + phopD1-flag) and there was no difference in ion leakage (Fig. S2). Collectively, these data confirm that HopD1 can suppress AvrRpm1-induced ETI responses in Arabidopsis.
HopD1 does not suppress PAMP-induced oxidative burst or callose deposition in Arabidopsis
To determine if HopD1 could suppress PTI, we evaluated its ability to inhibit the oxidative burst that occurs within minutes of PAMP recognition (Nanda et al., 2010). To do this, we measured the production of ROS in wild-type and HopD1-HA-expressing Arabidopsis after treatment with the PTI inducers flg22, elf18, and chitin. Flg22 is a 22-amino-acid peptide of the PAMP flagellin that is recognized by the plant receptor FLS2 and elf18 is an 18-amino-acid peptide of the PAMP EF-Tu that is recognized by the receptor EFR. The PAMP chitin is recognized by the receptor CERK1 (Monaghan & Zipfel, 2012). No difference was observed in PAMP-induced ROS production between wild-type and HopD1-HA-expressing Arabidopsis, indicating that HopD1 cannot suppress this early PTI response (Fig. 3a).
If HopD1 targets an intermediate stage in PTI signaling, it may suppress later PTI responses without altering ROS production. Therefore, we measured the ability of HopD1 to suppress PAMP-induced callose deposition, a late PTI response (Nicaise et al., 2009), in wild-type and HopD1-HA expressing Arabidopsis plants treated with flg22. No difference in flg22-induced callose deposition was observed between wild-type Arabidopsis and Arabidopsis expressing HopD1-HA (Fig. 3b). To confirm this lack of PTI suppression, we measured the extent that bacterially delivered HopD1 could suppress PAMP-induced callose deposition in response to recognition of Pf PAMPs. To do this, wild-type Arabidopsis was infiltrated with Pf(pLN1965) carrying a vector control or a plasmid constitutively expressing hopD1-flag. No difference in PAMP-induced callose deposition as a result of the bacterial delivery of HopD1 was observed (Fig. 3c). These data suggest that while HopD1 is a strong suppressor of ETI, it has no detectable effect on these PTI responses. Therefore, it is likely that HopD1 suppresses ETI-specific components of plant immunity.
HopD1 interacts with the membrane-tethered Arabidopsis transcription factor NTL9
A high-throughput yeast-two-hybrid screen that defined the P. syringae effector interactome with Arabidopsis proteins identified the Arabidopsis membrane-tethered NAC transcription factor NTL9 as a possible interactor for HopD1 (Mukhtar et al., 2011). Three splice variants have been identified for NTL9 owing to differential splicing of its fourth intron. Splice variant 1 is the only one with a C-terminal transmembrane domain. RT-PCR analyses with primers flanking the fourth intron of NTL9 were used to differentiate between splice variant 1 and splice variants 2 and 3 during biotic stress. We performed RT-PCR using these primers with RNA isolated from untreated Arabidopsis and Arabidopsis treated with Pf(pLN1965 + pavrRpm1) or Pf(pLN1965 + pavrRpm1 + phopD1-flag). In all treatments, splice variant 1 was the only form observed, indicating that it is the dominant form of this gene in these conditions (Fig. S3).
As the interaction between NTL9 and HopD1 was identified using a yeast two-hybrid screen, we first confirmed this interaction by cloning hopD1 and NTL9 into yeast two-hybrid system vectors. Yeast strains expressing both hopD1 and NTL9 grew on selective media while hopD1 and NTL9 with their respective empty vector controls did not, confirming the interaction between HopD1 and NTL9 in yeast (Fig. 4a).
The interaction of two proteins in yeast does not necessarily translate into an interaction in planta. To confirm their interaction in planta, BiFC was performed. Nicotiana benthamiana was co-agroinfiltrated with NTL9 fused at its N-terminus to the N-terminal half of YFP and HopD1 fused at its C-terminus to the C-terminal half of YFP. Infiltrated leaves were examined with confocal microscopy and yellow fluorescence was observed in cells of the infiltrated leaves (Fig. 4b), indicating that these proteins interacted in planta. The reciprocal experiment with HopD1 fused at its C-terminus to the N-terminal half of YFP and NTL9 fused at its N-terminus to the C-terminal half of YFP also produced yellow fluorescence (Fig 4b). No yellow fluorescence was observed in leaves infiltrated with BiFC HopD1 or NTL9 constructs and their corresponding nYFP or cYFP controls (Fig 4b). These data indicate that HopD1 interacts with NTL9 in planta.
HopD1 and NTL9 localize to the plant endoplasmic reticulum
The reticulate pattern in the cytoplasm accompanied by a distinct ring around the nucleus observed in the BiFC experiments involving HopD1 and NTL9 is characteristic of ER-localized proteins. To investigate whether HopD1 and/or NTL9 localized to the plant ER, we performed colocalization experiments with a known ER marker. DNA encoding the ER-localized RFP marker CD3-960 (Nelson et al., 2007) was agroinfiltrated into N. benthamiana with DNA encoding either HopD1 fused at its C-terminus to GFP or NTL9 fused at its N-terminus to GFP, both under the control of a constitutive CaMV 35S promoter. Confocal microscopy revealed colocalization of HopD1-GFP with ER-RFP (Fig. 5a) and GFP-NTL9 with ER-RFP (Fig. 5b). These data indicate that both HopD1 and NTL9 are localized to the ER in N. benthamiana. The most likely orientation for NTL9 is with its C-terminal transmembrane domain inserted in the ER membrane with a minor portion of NTL9 in the ER lumen and the majority of the protein in the cytoplasm. Membrane-tethered transcription factors are held in a ready state in the membrane and upon activation are cleaved at the cytoplasmic side of the transmembrane domain, releasing an active transcription factor that can enter the nucleus and regulate transcription (Kim et al., 2006). The protease that performs this cleavage for NTL9 has not been identified. The localization of HopD1 and NTL9 to the ER and their interaction at this site as indicated in BiFC experiments suggests that HopD1 is targeting NTL9 while it is anchored in the ER membrane.
NTL9 is involved in the Arabidopsis immune response to P. syringae
As HopD1 is a strong suppressor of ETI, it stands to reason that if it targets NTL9, NTL9 is likely a component of plant immunity. An emerging theme for NAC transcription factors is their multiple roles in biotic and abiotic stress, as well as an ability to coordinate multiple stress responses (Puranik et al., 2012). This has been seen for the membrane-tethered NAC transcription factor NTL6 that coordinates the response to cold and pathogen attack as it up-regulates pathogenesis-related genes (Seo et al., 2010). A loss-of-function line for NTL9 (SALK-065051, ntl9-1) was isolated that is compromised in osmotic stress responses and has enhanced susceptibility to avirulent Hyaloperonospora arabidopsidis (Yoon et al., 2008; Mukhtar et al., 2011), suggesting that NTL9 is involved in the response to both osmotic and pathogen stresses.
To investigate the role of NTL9 in the response of Arabidopsis to P. syringae, wild-type Arabidopsis and the Arabidopsis ntl9-1 mutant were syringe-inoculated with various P. syringae strains and bacterial growth was measured at 0 and 3 d postinoculation (Fig. 6a). DC3000 grew to equivalent levels in wild-type and ntl9-1 Arabidopsis, as did the hrcC mutant. Interestingly, the reduced growth of the hopD1 mutant observed in wild-type Arabidopsis plants was restored to DC3000 values in ntl9-1 mutant plants. These data support the hypothesis that NTL9 is a target of HopD1. As expected, the DC3000 hopD1 mutant complemented with hopD1 showed equivalent growth to DC3000 in both wild-type and ntl9-1 mutant plants. The avirulent P. syringae strains DC3000(pavrRpm1) and DC3000(pavrRpt2) showed enhanced growth in Arabidopsis ntl9-1 mutant plants compared with wild-type Arabidopsis, indicating that NTL9 is involved in ETI. These results are consistent with our earlier results (Figs 1b, 2), indicating that HopD1 targets components of immunity associated with ETI.
A truncated version of NTL9 that consisted of amino acids 1–330 (NTL91–330) is constitutively active (Yoon et al., 2008). We made transgenic Arabidopsis lines expressing NTL91-330-HA under the control of an estradiol-inducible promoter. Wild-type and NTL91-330-HA-expressing Arabidopsis were sprayed with estradiol to induce transgene expression. Twenty-four hours later the plants were syringe-inoculated with 2 × 105 cells ml−1 of various P. syringae strains and bacterial growth was measured at 0 and 3 d postinoculation (Fig. 6b). Growth of both DC3000 and the complemented DC3000 hopD1 mutant were reduced to that of the hopD1 mutant in plants expressing activated NTL9. In addition, growth of the avirulent P. syringae strains was also reduced in the NTL91-330-HA-expressing lines. These data show that active NTL9 up-regulates plant immunity resulting in the restriction of P. syringae growth in planta.
Identification of NTL9-induced genes involved in plant immunity
If HopD1 functions by inhibiting NTL9's activity during ETI, it would be likely that HopD1 suppresses NTL9-dependent gene expression during ETI. NTL9 has been shown to regulate several genes involved in abiotic stress (Yoon et al., 2008), yet its regulation of genes involved in plant immunity is not well understood. To determine if NTL9 could regulate the expression of a subset of genes associated with immunity, FRK1, PAL1, and NHL10 expression was measured using qRT-PCR in wild-type and NTL91-330-HA-inducible Arabidopsis with and without estradiol treatment (Fig. 7a). The expression of all three immunity-related genes was induced exclusively in the estradiol-treated NTL91-330-HA plants showing that they are up-regulated either directly or indirectly by active NTL9. Because of the extensive overlap between ETI and PTI (Tsuda & Katagiri, 2010), we do not know if the expression of these genes represents an ETI or a PTI response in this experiment.
In a previous study, Kim et al. used PCR-mediated random binding site selection to identify the sequence TTGCTTANNNNNNAAG as the DNA-binding site for NTL9 (Kim et al., 2007a). We used this binding site to search for additional NTL9-regulated Arabidopsis genes using RSA-tools-dna-pattern search (http://rsat.ulb.ac.be/genome-scale-dna-pattern_form.cgi; van Helden et al., 2000). In all, 175 Arabidopsis genes were identified with the binding site motif in their promoter within 1 kb of their predicted translational start site. We selected 20 of these 175 genes for further analysis based on their possible role in innate immunity. Their possible involvement in immunity was established via literature searches or as a result of induction of their expression by DC3000 treatment according to the microarray database Arabidopsis eFP Browser (http://www.bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi; Winter et al., 2007).
Expression of the 20 candidate genes was measured by qRT-PCR in wild-type Arabidopsis and NTL91-330-HA-inducible plants with and without estradiol treatment (Table S1). Owing to the limited number of base pairs in the predicted binding site, the false positive rate was high and many of the genes tested did not show NTL9-regulated expression (Table S1). Despite this high error rate, four of the predicted genes were strongly up-regulated in estradiol-treated NTL91-330-HA-inducible lines. These genes were ADF6 (At2g31200), NIP2 (At2g17730), HAB1 (At1g72770) (Fig. 7a) and BZIP9 (At5g24800) (Table S1).
HopD1 suppresses NTL9-regulated gene expression during ETI but not PTI
In order to investigate HopD1's ability to suppress NTL9's function in ETI, the expression of NTL9-regulated genes was measured in wild-type Arabidopsis plants infiltrated with a mock control, Pf(pLN1965 + pavrRpm1) or Pf(pLN1965 + pavrRpm1 + phopD1-flag). One of the genes tested (BZIP9) was repressed by Pf(pLN1965 + pavrRpm1) treatment (Fig. S4). The other six genes showed increased expression upon AvrRpm1-induced ETI and little if any increase in expression by AvrRpm1 in the presence of HopD1 (Fig. 7b). FRK1 is known primarily to be associated with PTI. However, it is induced during AvrRpm1-induced ETI (Fig. 7b). These data suggest that HopD1 inhibits ETI-induction of NTL9-regulated gene expression and, therefore, it interferes with NTL9 function during ETI.
Because HopD1 specifically blocks ETI and not PTI responses, we examined the ability of HopD1 to affect the expression of the six NTL9-regulated marker genes during PTI. To accomplish this, wild-type and HopD1-HA-expressing Arabidopsis were syringe-infiltrated with flg22 or a mock control. Leaf tissue was sampled 4 h later and gene expression was measured using qRT-PCR (Fig. 7c). Three of the six genes, FRK1, PAL1 and NHL10, showed induction in response to flg22 treatment. However, the presence of HopD1-HA had no effect on the PTI induction of these genes. Thus, taken together, these data indicate that NTL9 is specifically involved in the ETI induction of immunity-related gene expression and the induction of a subset of these genes during PTI is probably controlled by other transcription factors.
In this study we examined the role of the HopD1 T3E in the virulence of the bacterial pathogen P. syringae. Baltrus et al. (2011) found that HopD1 was a well-distributed T3E, which was found in all sequenced strains of group I, two-thirds of the sequenced group III strains, and none of the six sequenced group II strains. We found that HopD1 was necessary for the full virulence of P. syringae (Fig. 1a) and that it was a strong suppressor of ETI (Fig. 2), but apparently not of PTI. The extent to which transgenically expressed HopD1 was capable of PTI suppression was evaluated with three commonly used PTI assays: the in planta growth of a P. syringae hrcC mutant (Fig. 1b), defective in type III secretion, or a nonpathogenic P. fluorescens strain (Fig. 3c); PAMP-induced ROS production (Fig. 3a); and PAMP-induced callose deposition (Fig. 3b). However, we did not test the extent to which HopD1 allowed better P. syringae growth in plants pretreated with a PAMP, which would be an additional assay to evaluate PTI suppression. Nevertheless, based on these commonly used PTI assays, it seems likely that HopD1 is not an effective PTI suppressor.
We found that HopD1 interacts with the NAC transcription factor NTL9 and both localize to the ER of plants (Figs 4, 5). The growth of the P. syringae hopD1 mutant and ETI-inducing P. syringae strains was enhanced in an Arabidopsis ntl9 mutant (Fig. 6a), while expression of constitutively active NTL9 suppressed the growth of both virulent and ETI-inducing P. syringae strains (Fig. 6b). This indicates that NTL9 is utilized in the Arabidopsis innate immune response to P. syringae. In addition, we observed that HopD1 suppressed ETI-induced but not PAMP-induced expression of NTL9-regulated genes (Fig. 7). Collectively, these data led us to conclude that HopD1 suppresses ETI and promotes virulence of P. syringae in part by blocking the ability of NTL9 to regulate ETI-induced gene expression.
HopD1 was shown previously to suppress the HopA1-induced HR in tobacco (Jamir et al., 2004; Guo et al., 2009). The ability of HopD1 to suppress ETI in two different systems (tobacco and Arabidopsis) and in response to multiple recognized T3Es (HopA1, AvrRpm1 and AvrRpt2) indicates that one of its primary targets is a common component of ETI. Interestingly, we did not find any evidence that HopD1 suppressed PTI. The specific effect of HopD1 on ETI suggests that the reduced virulence of the hopD1 mutant exhibited in Arabidopsis is probably the result of an ‘unmasking’ of recognized P. syringae T3Es owing to the loss of HopD1's ETI suppression activity. Earlier reports provided pioneering genetic evidence suggesting that avr genes, encoding recognized T3Es, can reside in virulent P. syringae strains and can be ‘unmasked’ when other T3E genes, encoding ETI suppressors, are mutated (Jackson et al., 1999; Tsiamis et al., 2000). Moreover, Arabidopsis mutants with defective R protein complexes allow for enhanced growth of a virulent P. syringae strain (Zhang et al., 2010), consistent with ETI playing an immunity role against virulent P. syringae. Our research further supports that ETI contributes to immunity against virulent P. syringae and that HopD1 disables ETI as a virulence strategy.
The coevolution of plants and pathogens is elegantly illustrated in the zigzag model of the plant immune system (Jones & Dangl, 2006). In this coevolutionary model, PRRs recognize PAMPs inducing PTI. To retain pathogenicity the pathogen must acquire or evolve an effector that can suppress PTI, resulting in effector-triggered susceptibility (ETS). Therefore, the earliest T3Es used in P. syringae virulence are predicted to be PTI suppressors. In the next phase, the plant evolves an R protein to recognize a specific pathogen effector, thereby inducing ETI. This would put selection pressure on the pathogen to disable the gene encoding the recognized effector or acquire another effector to suppress the ETI evoked from this recognition event, allowing the return of ETS. Importantly, the ability for the pathogen to regain ETS is probably dependent on its inventory of effectors that can suppress ETI and the strength of the ETI response induced by the recognized effector. P. syringae T3Es have been investigated mostly for their effect on PTI (Feng & Zhou, 2012). However, there are several examples of P. syringae T3Es that suppress both PTI and ETI (Abramovitch et al., 2003; Gohre et al., 2008; Gimenez-Ibanez et al., 2009; Wang et al., 2010; Wilton et al., 2010). This is perhaps not surprising given the highly overlapped nature of PTI and ETI (Tsuda et al., 2008). What is less common are examples of P. syringae T3Es that suppress ETI but not PTI. These effectors would be predicted to be acquired later in the coevolution of the pathogen–plant interaction after establishment of the R protein immune receptor surveillance system. HopD1 appears to be one such effector.
We clearly show by yeast two-hybrid and BiFC assays that HopD1 interacts with NTL9. GFP fusions of both of these proteins colocalize with an ER-targeted RFP fusion and BiFC assays show that HopD1 and NTL9 interact at the ER. NTL9 was previously reported to localize to the plasma membrane and nucleus of onion cells (Yoon et al., 2008). The large vacuole of onion cells that oppresses the ER against the plasma membrane, coupled with the difficulty in distinguishing nuclear and perinuclear localization with epifluorescence microscopy, may have led to these observed differences in localization.
NTL9 is one of over 110 NAC domain transcription factors in Arabidopsis (Ooka et al., 2003), of which at least 13 have C-terminal transmembrane domains (Kim et al., 2007b). There is precedent for pathogens to target NAC transcription factors, arguably the most well characterized of which is the capsid protein (CP) of Turnip crinkle virus (TCV) that interacts with the Arabidopsis NAC transcription factor TIP (Ren et al., 2000), blocking its ability to localize to the nucleus (Ren et al., 2005). We tested the extent to which HopD1 could affect the subcellular localization of NTL9 using Agrobacterium-mediated transient assays in N. benthamiana of GFP-NTL9 with or without HopD1-HA and were unable to discern any difference in GFP-NTL9 localization (Fig. S5). However, based on these results, DC3000 can induce the relocalization of NTL9-GFP to the nucleus in N. benthamiana. Additionally, we tested whether HopD1 was able to inhibit NTL9 transcription using a yeast one-hybrid system and were unable to detect a decrease in transcription in the presence of HopD1 (Fig. S6). Thus, HopD1 inhibits NTL9-dependent gene expression, but we do not yet know the mechanism of its inhibition.
NAC transcription factors have also been implicated in the plant immune response to bacterial pathogens. For example, the plasma membrane-bound Arabidopsis NAC transcription factor NTL6 has been implicated in the cold induction of innate immunity as a constitutively active form of NTL6 activates the expression of the pathogenesis-related genes PR1, PR2 and PR5 (Seo et al., 2010). Arabidopsis plants constitutively expressing active NTL6 are more resistant to P. syringae, and NTL6 RNAi lines are more susceptible to P. syringae after cold pretreatment (Seo et al., 2010). Another NAC transcription factor from Arabidopsis, ATAF1, is a negative regulator of immune responses against P. syringae (Wang et al., 2009).
Relatively recently a report showed that ntl9 knockout Arabidopsis plants are subtly more resistant to P. syringae (Kim et al., 2012). These data led the authors to suggest that NTL9 is a negative regulator of innate immunity. However, we did not observe any statistically significant difference in the growth of P. syringae in ntl9 knockout plants compared with wild-type plants and often observed a slight increase in P. syringae growth (Fig. 6a), even though these experiments were repeated many times. Importantly, Arabidopsis plants expressing constitutively active NTL9 were more resistant to P. syringae (Fig. 6b), which suggests that NTL9 acts as a positive regulator. We cannot account for the differences between our P. syringae growth phenotypes and that reported in Kim et al. (2012). However, their subtle P. syringae growth data and our own data suggest that the absence of NTL9 may be compensated for by other transcription factors in plant immunity.
Our study shows that NTL9 induces the classic immune marker genes FRK1 (He et al., 2006), PAL1 (Mishina & Zeier, 2007) and NHL10 (Zipfel et al., 2004), as well as at least three genes that have not been characterized as to their role in innate immunity. These genes are HAB1, NIP2 and ADF6. NIP2 has no known role in immunity. Although no direct role for ADF6 has been shown in the immune response to P. syringae infection, another actin-depolymerizing factor (ADF4) was shown to be specifically required for resistance triggered by the P. syringae effector AvrPphB (Tian et al., 2009). HAB1 is a negative regulator of ABA signaling. It is a 2C protein phosphatase that, in the absence of ABA, binds to SnRK2 kinases, inhibiting their activity (Soon et al., 2012). ABA has a negative role in plant immunity (Cao et al., 2011) and it would not be surprising, therefore, if the induction of a negative regulator of ABA signaling, such as HAB1, by NTL9 was important for immunity signaling. Other ETI-associated genes are also likely to be regulated by NTL9, but their identification will require extensive NTL9-dependent expression studies, which will be addressed in the future. Other future studies will examine HopD1's effect on NTL9 and other NAC transcription factors during pathogen stress, which will help to elucidate how this T3E inhibits the function of NTL9 as well as understanding the role NTL9 plays in plant immunity.
We would like to thank the Alfano group members for fruitful discussions and comments regarding this manuscript. We would also like to thank Dr Ed Cahoon for the vector CD3-96. The research was supported by grants to J.R.A. from the US Department of Agriculture (award no. 2007-35319-18336) and the National Institutes of Health (award no. 1R01AI069146-01A2).