•Little is known about how effectors from filamentous eukaryotic plant pathogens manipulate host defences. Recently, Phytophthora infestans RXLR effector AVR3a has been shown to target and stabilize host E3 ligase CMPG1, which is required for programmed cell death (PCD) triggered by INF1. We investigated the involvement of CMPG1 in PCD elicited by perception of diverse pathogen proteins, and assessed whether AVR3a could suppress each.
•The role of CMPG1 in PCD events was investigated using virus-induced gene silencing, and the ability of AVR3a to suppress each was determined by transient expression of natural forms (AVR3aKI and AVR3aEM) and a mutated form, AVR3aKI/Y147del, which is unable to interact with or stabilize CMPG1.
•PCD triggered at the host plasma membrane by Cf-9/Avr9, Cf-4/Avr4, Pto/AvrPto or the oomycete pathogen-associated molecular pattern (PAMP), cellulose-binding elicitor lectin (CBEL), required CMPG1 and was suppressed by AVR3a, but not by the AVR3aKI/Y147del mutant. Conversely, PCD triggered by nucleotide-binding site-leucine-rich repeat (NBS-LRR) proteins R3a, R2 and Rx was independent of CMPG1 and unaffected by AVR3a.
•CMPG1-dependent PCD follows perception of diverse pathogen elicitors externally or in association with the inner surface of the host plasma membrane. We argue that AVR3a targets CMPG1 to block initial signal transduction/regulatory processes following pathogen perception at the plasma membrane.
Plants are constantly threatened by microbial pathogens. However, the most common outcome of plant–microbe interactions is incompatibility (disease resistance), as the majority of plant species are resistant to infection by all genotypes of most microbial species. The ability of a microbe to breach preformed structural barriers or to detoxify constitutively produced antimicrobial compounds dictates the first challenge to plant infection. Having successfully overcome these obstacles, potential pathogens risk the recognition of exposed molecules by plant receptors that are analogous to the Toll- and Toll-like family receptors in mammalian innate immunity (Zipfel, 2008). This early warning system of pattern recognition receptors (PRRs) is typically localized in the plant extracellular matrix. PRRs have evolved to detect the most highly conserved, exposed molecules (pathogen- or microbe-associated molecular patterns: PAMPs or MAMPs) that are shared by various classes of microorganism, and are indispensable for fitness. The recognition of PAMPs or MAMPs activates signalling cascades involving the production of reactive oxygen intermediates (ROI) and mitogen-activated protein kinases (MAPK), leading to PAMP-triggered immunity (PTI), including the deposition of lignin and callose in the cell wall and the induction of pathogenesis-related (PR) gene expression (Chisholm et al., 2006; Jones & Dangl, 2006).
Successful plant pathogens overcome PTI by the deployment of virulence determinants, called effectors, that function in the apoplast or within living plant cells to suppress or manipulate host defences (effector-triggered susceptibility, ETS) (Chisholm et al., 2006; Jones & Dangl, 2006). Recent years have witnessed intense activity to determine the virulence functions of prokaryotic pathogen effector proteins, especially those ‘injected’ inside host cells via the bacterial type III secretion system (T3SS). T3SS effectors possess a range of enzymatic activities and can interact directly with host proteins to suppress defences at the levels of PAMP perception, signal transduction, defence gene expression and various post-translational modifications (Block et al., 2008), thus creating a susceptible environment for infection. However, effector proteins are themselves vulnerable to detection by resistance (R) proteins, a second line of inducible defence, that directly or indirectly perceive effectors or effector activities. Many defence responses induced by effector-triggered immunity (ETI) overlap with PTI, but appear to be amplified. Both responses can lead to rapid and localized programmed cell death (PCD) known as the hypersensitive response (HR) (Jones & Dangl, 2006).
Despite considerable advances in our understanding of bacterial pathogenesis, little is understood about how filamentous eukaryotic pathogens (fungi and oomycetes) cause disease. In recent years, it has been shown that the oomycete pathogens Phytophthora infestans, the cause of late blight on potato and tomato, and Phytophthora sojae, which infects soybean, secrete potentially hundreds of effectors containing a conserved amino acid motif RXLR (where R is arginine, X is any amino acid and L is leucine), which is required for the translocation of these proteins inside living plant cells (Whisson et al., 2007; Birch et al., 2008; Dou et al., 2008). All oomycete effectors that are recognized by host plants (avirulence; AVR proteins) have thus far been found to contain the RXLR motif (Hein et al., 2009), consistent with their detection by cytoplasmic nucleotide-binding site-leucine-rich repeat (NBS-LRR) R proteins.
Phytophthora infestans effector AVR3a is one of the best-studied oomycete RXLR effectors, and the first for which a potential virulence activity has been characterized (Bos et al., 2010). It exists as two forms in European P. infestans populations which differ in only two amino acids: AVR3aKI, but not AVR3aEM, is perceived by potato R protein R3a (Armstrong et al., 2005) to trigger an SGT1-dependent HR (Bos et al., 2006). AVR3aKI also functions to strongly suppress cell death induced by the oomycete PAMP, INF1, whereas AVR3aEM weakly suppresses INF1-triggered cell death (ICD) (Bos et al., 2006, 2009). Furthermore, deletion and substitution experiments have revealed that the C-terminal residue of AVR3a, tyrosine 147, is dispensable for R3a recognition, but essential for ICD suppression, allowing these two effector characteristics to be distinguished (Bos et al., 2009).
Recently, it has been demonstrated that both AVR3aKI and AVR3aEM interact with and stabilize host U-box E3 ligase CMPG1 in planta. By contrast, AVR3aKI/Y147del, a mutant form with a deleted C-terminal tyrosine residue that fails to suppress ICD, does not interact with or stabilize CMPG1 (Bos et al., 2010). The U-box domain of CMPG1 is essential for its activity as an E3 ligase, and this activity has been shown previously to be required for ICD (González-Lamothe et al., 2006). CMPG1 activity causes degradation of itself, and thus presumably of its substrates, in the plant cell via the 26S proteasome. In stabilizing CMPG1, AVR3aKI and, to a lesser extent, AVR3aEM thus alter its normal activity (Bos et al., 2010). The failure of AVR3aKI/Y147del to stabilize CMPG1 strengthens the link between the ability of AVR3a to stabilize this host protein and its ability to suppress ICD. Silencing AVR3a compromised the virulence of P. infestans. However, full virulence could be restored to silenced lines by in planta expression of AVR3aKI or AVR3aEM, but not the AVR3aKI/Y147del mutant form (Bos et al., 2010). As AVR3a is up-regulated during the biotrophic phase of infection, these data provide evidence that AVR3a is an essential virulence factor that targets and stabilizes CMPG1, preventing its normal activity, presumably to suppress CMPG1-mediated cell death during this phase of infection. By contrast, AVR3a is down-regulated and INF1 is up-regulated at the onset of the necrotrophic phase, withdrawing suppression, whilst triggering CMPG1-mediated cell death at a stage when it is of benefit to the infection cycle (Bos et al., 2010).
CMPG1 (ACRE74) was identified as one of a number of ubiquitination pathway-related genes rapidly induced after Cladosporum fulvum Avr9 elicitation in Cf-9 tobacco cells (Durrant et al., 2000; Rowland et al., 2005). Virus-induced gene silencing (VIGS) of NbCMPG1 in Nicotiana benthamiana using Tobacco rattle virus (TRV) demonstrated its essential role in the efficient production of Cf-9/Avr9-mediated HR (González-Lamothe et al., 2006). Furthermore, silencing CMPG1 in tomato by VIGS and RNA interference (RNAi) highlighted its function for full Cf-9-mediated resistance to C. fulvum. In addition to showing that both Cf-9 and ICD were dependent on CMPG1, González-Lamothe et al. (2006) demonstrated that the overexpression of a CMPG1C37A dominant negative mutant prevented HR triggered by Pto/AvrPto expression in tobacco. CMPG1 is thus involved in cell death triggered by perception of diverse molecules from fungal, oomycete and bacterial pathogens. A key objective in our work was to conduct a deeper investigation into the role of CMPG1 in pathogen-triggered cell death events. To that end, we determined its involvement in PCD triggered by a range of pathogen elicitors and effectors predicted to be perceived at either the host plasma membrane (PM) or within the cytoplasm. Surprisingly, the results revealed the involvement of CMPG1 in PCD triggered following perception of elicitors at PM, and no involvement in HRs following recognition of cytoplasmic effectors by the NBS-LRR R proteins R3a, R2 or Rx. In all cases, AVR3a exclusively suppressed CMPG1-dependent PCD, strongly supporting a hypothesis that this essential, cytoplasmic effector targets CMPG1 to block the signal transduction and/or regulatory processes following perception of pathogen molecules externally or associated with the inner surface of the host PM.
Materials and Methods
Microbial strains and growth conditions
pENTR1A constructs containing AVR3aEM, AVR3aKI and AVR3aKI/147-del (Bos et al., 2010) were recombined into pEarleygate 203; 5′ 35S-MYCtag-Gateway-OCS 3′ (Earley et al., 2006). These plasmids were transformed into Agrobacterium tumefaciens strain GV3101 and selected with kanamycin and rifampicin for transient expression in Nicotiana tabacum. Plasmid pGWB6 (Invitrogen) in GV3101, expressing green fluorescent protein (GFP), was used as a negative control. AVR3aKI, AVR3aEM and AVR3aKI-Y147del were also cloned into a Gateway version of the vector pGRAB mentioned in Bos et al. (2010) and transformed into A. tumefaciens strain AGL1, pSoup, pVirG cells by electroporation. An empty pGRAB vector in AGL1 was used as a control for these experiments. Positive transformants were subsequently grown in Luria–Bertani (LB) medium supplemented with rifampicin, chloroamphenicol, tetracycline and kanamycin for transient expression in N. benthamiana. AVR2 (PITG_08943) was initially amplified without the signal peptide using the primers 7F 5′-AAAGCAGGCTTCACCATGCTGCATGCAGCTCCAGGTG-3′ and 7R 5′-GAAAGCTGGGTCTTAACTCCTCTTGTCACCCTTAAT-3′. Full AttB recombination sites were added in a second PCR using the primers AttB1 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCACC-3′ and AttB2 5′-GGGGACCACT TTGTACAAGAAAGCTGGG-3′. The PCR product was run on an agarose gel, extracted with a Qiagen® Gel Extraction Kit and then recombined into pDONR221 (Invitrogen) with BP clonase (Invitrogen). After sequencing, AVR2 was recombined into pGRAB using LR clonase (Invitrogen) and transformed into AGL1 pVirG pSOUP and grown as above. Construct pKGW-R2 (Lokossou et al., 2009) was grown under the same antibiotic selections as pGRAB. GV3101, which expresses Rx (pB1:Rx-HA), and Potato virus X (PVX) coat protein (CP) (pBin61:CP-TK) were grown in LB medium supplemented with gentamycin, rifampicin and kanamycin (Bendahmane et al., 1999; Moffett et al., 2002). Constructs for the co-expression of Cf-4/Avr4, Cf-9/Avr9 (Van der Hoorn et al., 2000), 35S:Pto, 35S:AvrPto (Peart et al., 2002b) and pGR106:SP.CBEL (Gaulin et al., 2006) were grown in LB medium supplemented with rifampicin and kanamycin under Health and Safety Executive licence GM250.02.2. The bacterial apple and pear pathogen Erwinia amylovora 1430 (Eam) was prepared to 5 × 106 colony-forming units (CFU) ml−1 as described in Gilroy et al. (2007), but with the buffer mentioned below, for co-infiltration with pGRAB Avr3a constructs, or serially diluted to a range of concentrations. Eam hrpN::MuPR13 and hrcV::MuPR13 mutants were grown in LB medium supplemented with chloroamphenicol and serially diluted to a range of concentrations as mentioned in the next section.
Cell death assays
All A. tumefaciens cultures were grown at 27°C at 200 rpm for 2–3 d, spun at 4500 g and the pellet was resuspended in sterile 10 mM MES and 10 mM MgCl2 buffer with 200 μM acetosyringone, and subsequently mixed together where appropriate to a final optical density at 600 nm (OD600) of 0.5 for each construct, except with cellulose-binding elicitor lectin (CBEL) (OD600 = 1) and Eam in N. benthamiana (OD600 = 0.125). Cultures were infiltrated with a 1-ml syringe without a needle through the abaxial leaf surface superficially wounded with a needle. Three to four leaves on at least four plants were used for each biological replicate. Avr9 peptide infiltrate was prepared as stated in Hammond-Kosack et al. (1998) and then infiltrated at a dilution of 1/40 into Cf-9 transgenic N. tabacum (Rowland et al., 2005) 3 d after Agrobacterium delivery of pEARLEY::AVR3a constructs or pGWB6:GFP as a control. HRs were recorded and photographed between 1 and 8 d post-infiltration (dpi) depending on the elicitor, as described in the figure legends. An individual inoculation was counted as positive if > 50% of the inoculated area developed a clear PCD lesion. Data graphs present the mean percentage of total inoculations per plant developing a clear HR with error bars representing ± standard errors (SE) of combined data from at least three biological replicates. One-way ANOVA or t-tests were performed to identify statistically significant differences.
Plant material and VIGS constructs
Nicotiana benthamiana and tobacco were grown as described previously (Bos et al., 2010). VIGS experiments were conducted in containment glasshouses under Health and Safety Executive licence GM250.03.1. The adjacent regions of NbCMPG1 selected for silencing by VIGS in N. benthamiana spanned the equivalent 400-bp NtCMPGF8 region previously used for NtCMPG1 RNAi (González-Lamothe et al., 2006). Primer sequences used for cloning a 262-bp portion of NbCMPG1 into pBinTRV2b (Liu et al., 2002) in antisense orientation using EcoRI and HpaI restriction sites were: CMPG1 I For 5′-AAAAGAATTCTTTCAACATGGAGGAAAAGGA-3′ and CMPG1 I Rev 5′-TTAAGTTAACTGCTGTGATTTGGAATTGGT-3′. CMPG1 II is a 253-bp fragment cloned using CMPG1 II For 5′-AAAAGAATTCAAGGAAAATGATCCAACAATGG-3′ and CMPG1 II Rev 5′-AAAAGTTAACTGCAACAAATGCAGCTGATAA-3′, previously used for VIGS of NbCMPG1 in Bos et al. (2010). Silencing of NbCMPG1 using TRV constructs was confirmed by quantitative reverse transcription-polymerase chain reaction (qRT-PCR) using primers RTS1 5′-CCCATCACAAATCAAACATTG-3′ and RTS2 5′-TCCCATGATCCTTGTTCTCA-3′, employing the conditions described previously (Bos et al., 2010).
All plasmid constructs for fluorescence protein labelling, treatment with epoxomicin and conditions of confocal image preparation and handling were as described previously (Bos et al., 2010). The potato StCMPG1b-YFP (expressing full-length StCMPG1b fused, at the C-terminus, to yellow fluorescent protein (YFP)) construct was used for treatments as indicated above, and splitYFP constructs (again as described previously; Bos et al., 2010) used N-terminal fusions of YFP halves to full-length StCMPG1b. For nuclear labelling, a transgenic N benthamiana line expressing a histone H2B-mRFP (RFP, red fluorescent protein) fusion was used (Chakrabarty et al., 2007) and, for nucleolar labelling, a transgenic N. benthamiana line expressing a fibrillarin-mRFP fusion was employed (Goodin et al., 2007).
Cf-9-, Cf-4- and Pto-mediated cell deaths are dependent on CMPG1 and suppressed by AVR3aKI and AVR3aEM
As CMPG1 has been reported to be required for HR triggered by perception of the C. fulvum Avr9 protein by the tomato Cf-9 R protein (González-Lamothe et al., 2006), we initially endeavoured to reproduce this phenotype by silencing NbCMPG1 in N. benthamiana. To silence NbCMPG1, we used our previous TRV:CMPG1 VIGS construct (Bos et al., 2010) (renamed TRV:CMPG1 II) and, for independent verification, designed a second VIGS construct (TRV:CMPG1 I) containing an adjacent 263-bp DNA fragment. Nicotiana benthamiana plants expressing TRV:GFP, TRV:CMPG1 I and TRV:CMPG1 II were infiltrated with A. tumefaciens strains expressing Cf-9 and Avr9. As anticipated, we observed a significant, similar decrease in the visible HR in TRV:CMPG1 I and TRV:CMPG1 II plants compared with the TRV:GFP control (Fig. 1a).
As the Cf-9/Avr9 HR is dependent on CMPG1, we investigated whether A. tumefaciens-expressed AVR3a variants could suppress HR induced by Avr9 peptide infiltrated into transgenic tobacco expressing Cf-9 (Fig. 1b). We found that, indeed, AVR3aKI and AVR3aEM suppressed significantly the Cf-9/Avr9 HR compared with a control vector expressing GFP. One-way ANOVA indicated that their relative abilities to suppress this cell death were not significantly different from each other (Fig. 1b). The mutated form AVR3aKI-Y147del, however, did not suppress Cf-9-mediated HR.
Two ubiquitination-associated genes, ACRE189 (F-box E3 ligase) and ACRE276 (U-box E3 ligase), are essential for both the Cf-9/Avr9 HR and another tomato/C. fulvum recognition event, Cf-4/Avr4. Interestingly, ACRE189 is also essential for the Pto/AvrPto HR (Rowland et al., 2005; van den Burg et al., 2008). However, CMPG1 has not been reported previously to be involved in the development of Cf-4/Avr4 HR. We thus extended our studies to investigate whether CMPG1 was a requirement for this second C. fulvum recognition event. Agrobacterium tumefaciens strains delivering Cf-4 and Avr4 were inoculated into TRV:GFP, TRV:CMPG1 I and TRV:CMPG1 II N. benthamiana plants, resulting in a clear reduction in HR in CMPG1-silenced plants (Fig. 1c). Agrobacterium strains expressing Cf-4 and Avr4 were co-infiltrated into N. benthamiana with strains expressing each of the AVR3a variants. Both AVR3aKI and AVR3aEM suppressed significantly Cf-4/Avr4 cell death to similar levels. However, AVR3aKI-Y147del had no such effect compared with the empty pGRAB control (Fig. 1d). These results may indicate a general role for CMPG1 in Cf-mediated cell death.
Overexpression of a CMPG1C37A dominant negative mutant prevented HR triggered by Pto/AvrPto in tobacco (González-Lamothe et al., 2006). However, recently, this HR has been shown to be mediated by direct recognition of the phosphorylated C-terminal domain (CTD) of AvrPto, by unknown R protein Rpa, and not through interaction of the CD loop of AvrPto with Pto (Yeam et al., 2010). By contrast, co-expression of AvrPto and Pto in N. benthamiana leads to HR mediated by the Pto/Prf complex, which can be perturbed by VIGS of an endogenous NbPrf gene (Peart et al., 2002a,b; Lu et al., 2003; Kang et al., 2004). We therefore investigated the involvement of CMPG1 in AvrPto/Pto-mediated cell death in N. benthamiana by infiltrating A. tumefaciens strains co-expressing Pto and AvrPto into TRV:GFP, TRV:CMPG1 I and TRV:CMPG1 II plants. HR was compromised significantly where NbCMPG1 was silenced compared with the TRV:GFP control (Fig. 1e). When Pto/AvrPto cell death is compromised, it results in a characteristic yellow chlorosis. This may be a result of suppression of PTI by AvrPto in response to A. tumefaciens, which leads to increased growth of the bacterium (Hann & Rathjen, 2007).
As Pto/AvrPto PCD is dependent on CMPG1, we hypothesized that this HR should also be attenuated by AVR3a. As predicted, AVR3aKI could inhibit the Pto/AvrPto HR in N. benthamiana, whereas no such suppression was seen with the AVR3aKI-Y147del mutant compared with the empty vector control (Fig. 1f). Although there was a trend towards cell death suppression with AVR3aEM, this was not statistically significant using one-way ANOVA. Taken together, our results show that AVR3aKI and occasionally, to a lesser extent, AVR3aEM both compromise Cf- and Pto-mediated, CMPG1-dependent HR, whereas the AVR3aKI/Y147del mutant, which does not interact with or stabilize CMPG1 (Bos et al., 2010), fails to compromise these PCD events.
CMPG1 is required for necrosis induced by the oomycete PAMP, CBEL, and is suppressed by AVR3aKI
Cell death triggered by the P. infestans PAMP, INF1, has been shown to require CMPG1 (González-Lamothe et al., 2006; Bos et al., 2010) and to be suppressed by AVR3a (Bos et al., 2006, 2009). Moreover, CMPG1 is rapidly up-regulated in parsley cells after treatment with the oomycete and bacterial PAMPs Pep13 and Flg22, respectively, and after infection with a range of compatible and incompatible pathogens (Kirsch et al., 2001), suggesting that it may generally be involved in PTI. A further oomycete PAMP, CBEL, from Phytophthora parasitica var. nicotianae has been shown to elicit necrosis in N. benthamiana and Arabidopsis (Gaulin et al., 2006; Dumas et al., 2008).
We investigated whether CBEL-triggered cell death was dependent on CMPG1 by infiltrating TRV:GFP, TRV:CMPG1 I and TRV:CMPG1 II N. benthamiana with A. tumefaciens expressing pGR106:SP_CBEL. Despite being expressed by a PVX vector, which leads to systemic cell death (Supporting Information Fig. S1), necrosis is initially clearly localized to the zone of agroinfiltration. We observed a significant decrease in both local and systemic CBEL-induced necrosis in CMPG1-silenced plants compared with TRV:GFP controls (Figs 2a, S1), indicating a requirement for CMPG1. We then examined whether AVR3a could perturb this PTI-associated cell death response in N. benthamiana. AVR3aKI suppressed significantly the early, local CBEL-induced necrosis which was confined to the zone of agroinfiltration (Fig. 2b). However, no local cell death suppression was observed with the AVR3aKI/Y147del mutant compared with the empty vector control (Fig. 2b). Although there was a trend towards cell death suppression with AVR3aEM, this was not statistically significant in these assays (Fig. 2b).
HR mediated by R3a, R2 and Rx is independent of CMPG1 and not suppressed by AVR3a
All CMPG1-dependent pathogen perception events examined above are predicted to occur at the host PM. We extended our analyses to investigate the requirement for CMPG1 in the perception of cytoplasmic AVR effectors by NBS-LRR R proteins. Previously, it has been determined that the recognition of AVR3aKI and AVR3aKI/Y147del by the coiled coil (CC)-NBS-LRR protein R3a is independent of CMPG1 (Bos et al., 2010). We confirmed that the HR triggered by co-expression of AVR3aKI with R3a was not affected in N. benthamiana expressing the TRV:GFP, TRV:CMPG1 I or TRV:CMPG1 II construct (Fig. 3a). The next step was to determine whether this was specific to R3a-mediated cell death or also applied to other NBS-LRR R proteins recognizing cytoplasmic effectors.
Nicotiana benthamiana expressing each of the TRV constructs was inoculated with an alternative potato–P. infestans gene-for-gene combination, R2/AVR2 (Lokossou et al., 2009). HR developed to a similar extent in all TRV-infected plants, indicating that CMPG1 is not required for ETI signalling after recognition of the cytoplasmic RXLR effector AVR2 (Figs 3b, S2). Moreover, co-expression of R2 and AVR2 with AVR3aKI, AVR3aEM or AVR3aKI/Y147del had no effect on R2-mediated HR (Fig. 3c).
We also examined ETI triggered by the potato Rx (CC-NBS-LRR) protein which recognizes the CP of PVX (Bendahmane et al., 1999; Moffett et al., 2002). Again, no significant difference in HR was seen on TRV:GFP, TRV:CMPG1 I and TRV:CMPG1 II N. benthamiana following co-infiltration of A. tumefaciens strains expressing Rx and PVX-CP (Figs 3d, S2), indicating that CMPG1 is not required for Rx-mediated HR. As anticipated, the CMPG1-independent cell death elicited by Rx/PVX-CP was not suppressed by co-expression with AVR3aKI, AVR3aEM or AVR3aKI/Y147del (Fig. 3e).
The nonhost HR induced by Eam in Nicotiana spp. is dependent on CMPG1 and can be suppressed by AVR3aKI and AVR3aEM
Each of the cell death events examined above involved the expression of defined PAMP elicitors or AVR effectors in the model host N. benthamiana. We wished to extend our observations to a nonhost pathogen interaction, and selected Eam, a bacterial pathogen of apple and pear, which is known to induce nonhost HR in N. benthamiana (Gilroy et al., 2007; Oh et al., 2007) and tobacco, the latter of which is dependent on the Eam T3SS helper protein, HrpN, at low inoculum concentrations (5 × 106 CFU ml−1) (Sinn et al., 2008). Initially, we tested whether the Eam nonhost HR triggered on N. benthamiana was also dependent on HrpN at low concentration. We observed that, whereas a T3SS-disabled Eam hrcV− mutant caused no HR on N. benthamiana at high or low inoculation concentrations (108 or 5 × 106 CFU ml−1, respectively), as reported for tobacco, HR at the low concentration was dependent on HrpN, again in agreement with observations on tobacco (Sinn et al., 2008) (Fig. S3).
The expression of NbCMPG1 was examined over a 24-h time course after inoculation of 5 × 106 CFU ml−1Eam into N. benthamiana. Fig. 4(a) shows that NbCMPG1 was induced transiently at 1 h post-inoculation (hpi) and then again at 24 hpi, the point at which the onset of cell death is visibly detectable (Gilroy et al., 2007). Given that Eam induces NbCMPG1 expression, we investigated whether this nonhost HR requires CMPG1. Interestingly, silencing of NbCMPG1 compromised significantly Eam-induced cell death in N. benthamiana compared with the TRV:GFP control (Fig. 4b).
We tested the ability of AVR3a to suppress hrpN-dependent HR on both N. benthamiana and tobacco. We found that, consistent with the nonhost HR also being dependent on CMPG1, both AVR3aKI and AVR3aEM, but not the AVR3aKI/Y147del mutant, reduced significantly Eam-induced PCD in both N. benthamiana (Fig. 4c) and tobacco (Fig. 4d).
CMPG1 accumulates in the nucleolus when stabilized by AVR3a or epoxomicin treatment
Previously, CMPG1 has been shown to be continually degraded by the 26S proteasome as a result of its own E3 ligase activity (Bos et al., 2010). Co-expression of CMPG1 with AVR3aKI or AVR3aEM, but not AVR3aKI/Y147del, or treatment with the 26S proteasome inhibitor epoxomicin, resulted in the stabilization of CMPG1, which was detectable in western analyses and observed to accumulate in the nucleus using confocal microscopy (Bos et al., 2010). Given the association of CMPG1 activity with pathogen perception events at the host PM, the localization of CMPG1 was investigated further.
CMPG1-YFP was observed to accumulate, when expressed by the 35S promoter, in slow-moving vesicles, barely detectable in the cytoplasm of N. benthamiana cells (Fig. S4a). Following treatment with epoxomicin, or co-expression with AVR3aKI or AVR3aEM, the vesicles were still evident, but additional CMPG1-YFP fluorescence was detectable free in the cytoplasm, and significant fluorescence was observed in the nucleus (Fig. S4b–f). Interestingly, only the cytoplasmic and strong nuclear fluorescence were observed following bimolecular fluorescence complementation of CMPG1-YC with YN-AVR3aKI, suggesting that close proximity between CMPG1 and AVR3a does not occur in the vesicles (Fig. S4g). Co-expression of CMPG1-YFP with AVR3aKI in transgenic N. benthamiana expressing either the nuclear marker mRFP-histone2B (Fig. S4h,i) or the nucleolar marker mRFP-fibrillarin (Fig. S4j,k) revealed that CMPG1-YFP accumulated particularly in the nucleolus on stabilization.
We investigated the recognition of a range of pathogen-derived proteins in Nicotiana species to determine the role of CMPG1 in plant defence signalling, concomitantly confirming that CMPG1 is an important virulence target of the P. infestans RXLR effector AVR3a. We found that silencing of CMPG1 or Agrobacterium-mediated expression of AVR3a perturbed the cell death induced by pathogen elicitors and effectors predicted to be perceived at the host PM (Fig. 5). AVR3aKI was consistently the most efficient CMPG1-dependent cell death suppressor, AVR3aKI/Y147del mutant lacked suppression activity in all cases, and AVR3aEM showed a range of suppressor capabilities. HR following the recognition of cytoplasmic AVR effectors by the NBS-LRR R proteins under investigation was clearly not affected by silencing CMPG1 or by co-expression with AVR3a. Taken together, these data corroborate the correlation between the relative abilities of AVR3aKI and AVR3aEM to interact with and stabilize CMPG1 and their abilities to suppress CMPG1-mediated cell death events (Bos et al., 2010). In addition, the data strongly support a hypothesis that this essential effector targets CMPG1 to block signal transduction or regulatory processes following initial perception of pathogen molecules at the inner or outer surfaces of PM (Fig. 5). Each of these points, and the observation that the stabilization of CMPG1 by AVR3a results in its nuclear accumulation, is discussed below.
CMPG1 is a positive regulator of plant defence signalling from recognition events at PM
Silencing of CMPG1 in N. benthamiana demonstrated its requirement for efficient Cf-9/Avr9-mediated (Fig. 1; González-Lamothe et al., 2006) and Cf-4/Avr4-mediated (Fig. 1) HR. Both Cf proteins are type I membrane-anchored receptor-like proteins (RLPs) predicted to act at the host PM (Stergiopolis & de Wit, 2009). Cf-9 and Cf-4 recognize races of the leaf fungus C. fulvum that secrete the small, apoplastic cysteine-rich effectors Avr9 and Avr4, respectively. Avr9 is a cystine-knotted protein with structural homology to carboxypeptidase inhibitors (van den Hooven et al., 2001). Avr4 is a chitin-binding protein with a role in protecting C. fulvum against host perception of its cell walls (van den Burg et al., 2006). The predicted apoplastic localizations of both fungal proteins strengthen the assumption that they are detected at the host PM (Stergiopolis & de Wit, 2009).
Cf-4 and Cf-9 are similar RLPs that share identical cytoplasmic CTDs but differ in the extracellular LRR regions (Rowland et al., 2005). Consequently, many defence signalling components are shared between them, including the ubiquitination-associated genes SGT1 (suppressor of G2 allele of Skp1), ACRE189 (F-box E3 ligase) and ACRE276 (U-box E3 ligase) (Yang et al., 2006; van den Burg et al., 2008). Akin to PRRs, which require a cytoplasmic kinase as a signalling partner, Avr9/Cf-9 induced kinase 1 (ACIK1/ACRE264) is essential for the Cf-9-dependent HR (Rowland et al., 2005) and has since been shown to complex with Cf-9 through the CITRX thioredoxin that negatively regulates Cf-9-mediated HR signalling (Rivas et al., 2004). Interestingly, silencing ACIK1 in N. benthamiana also perturbs the Cf-4-dependent HR (Rowland et al., 2005), implying a similar role in signalling by this RLP. We have shown that CMPG1 is yet another shared component of Cf-4- and Cf-9-mediated HR (Fig. 5).
AvrPto is a T3SS-delivered effector from the bacterial pathogen Pseudomonas syringae which interacts with the tomato protein kinase Pto and, as N-myristoylation of both proteins is required for this HR (Wulf et al., 2004; de Vries et al., 2006), they are assumed to interact at the intracellular face of the host PM (Fig. 5). ETI signalling triggered by direct interaction of the CD loop of AvrPto with Pto is also dependent on a complex involving Pto and the NBS-LRR protein, Prf (Mucyn et al., 2006; Shan et al., 2008; Xiang et al., 2008). Interestingly, the CD loop of AvrPto also binds the kinase domains of the receptor-like kinase (RLK) BAK1 and associated RLKs (Shan et al., 2008; Xiang et al., 2008) to inhibit kinase activity and subsequent downstream signalling. BAK1 is linked to the perception of many PAMPs, such as flagellin, Ef-Tu, lipopolysaccharide (LPS), peptidoglycan, bacterial cold shock protein and INF1 (Lu et al., 2010).
AvrPto possesses three domains: in addition to the CD loop, there is an N-terminal domain required for translocation and PM localization, and a CTD. The CTD contributes to Pst virulence, but has no function in the suppression of BAK1-dependent PTI, and its phosphorylation by a plant kinase is responsible for the recognition of AvrPto in tobacco and Nicotiana sylvestris by an, as yet, unidentified R protein, Rpa (Yeam et al., 2010). Phosphorylation of CTD is not required for recognition by the Pto/Prf complex in tomato and N. benthamiana, indicative of two distinct, host-dependent recognition events and signalling pathways (Shan et al., 2000; Yeam et al., 2010). We demonstrate that silencing of CMPG1 perturbs the Pto-mediated recognition of AvrPto in N. benthamiana. Moreover, AVR3aKI strongly and AVR3aEM weakly suppress this HR. Therefore, we can again link CMPG1 dependence and AVR3a cell death suppression to a host PM recognition event (Fig. 5).
Cell death triggered by the P. infestans PAMP elicitin, INF1, is dependent on CMPG1 (González-Lamothe et al., 2006; Bos et al., 2010) and suppressed by AVR3a (Bos et al., 2009). INF1 interacts with a lectin-like receptor kinase (NbLRK1) with a single transmembrane domain which, when silenced, delays INF1-mediated HR, suggesting that it is a component of INF1 perception and signal transduction (Kanzaki et al., 2008). Furthermore, purified INF1 protein induces ICD when infiltrated into the intercellular spaces of tobacco leaves (Bos et al., 2010), endorsing the hypothesis that it is perceived extracellularly by nonhosts. Support for the view that INF1 elicitin can be regarded as a PAMP comes from the observation that ROS generation and cell death in N. benthamiana, following INF1 perception, are dependent on BAK1 (Heese et al., 2007).
We extended our analyses to cell death triggered by a second oomycete PAMP, CBEL, and again showed that this is dependent on CMPG1. CBEL is a glycoprotein containing two cellulose-binding domains (CBDs) which are necessary and sufficient for binding to the host cell wall and the elicitation of plant defences (Gaulin et al., 2006). Although surveillance of host cell wall integrity almost certainly involves a variety of sensory mechanisms, a PM-bound RLK, Theseus1 (THE1), has been implicated in cellulose sensing, and this, or a related RLK, has been postulated to form a link to CBEL perception (Dumas et al., 2008). Only AvR3aKI suppressed significantly CBEL-triggered cell death (Fig. 2). However, as first reported for suppression of ICD (Bos et al., 2006), AVR3aEM showed a trend towards the suppression of CBEL cell death which was not significant within the experimental replicates used in our assays. Nevertheless, the clear suppression by AVR3aKI provides further support for our hypothesis that AVR3a interaction with and stabilization of CMPG1 serve to suppress defence responses to perception events, including the recognition of oomycete PAMPs, which occur at PM (Fig. 5).
Our data reveal that CMPG1 is required for Cf-, Pto-, INF1- and CBEL-mediated PCD. Interestingly, ACRE189 (Avr9/Cf-9-induced F-box 1; ACIF1) is also required for HR induced by Cf-9/Avr9, Cf-4/Avr4, AvrPto/Pto and INF1, indicating that distinct ubiquitination-associated regulatory processes link these diverse perception events at the host PM.
CMPG1 is not required for ETI triggered by the NBS-LRR R proteins R3a, R2 and Rx
Silencing of NbCMPG1 was found not to affect the recognition of AVR3aKI by the CC-NBS-LRR protein R3a (Fig. 3; Bos et al., 2010). We investigated whether silencing of CMPG1 or transient expression of AVR3a would perturb the ETI responses triggered by NBS-LRR-mediated perception of other effectors. The potato R2 protein conveys resistance to isolates of P. infestans that deliver a different RXLR effector, AVR2 (Lokossou et al., 2009). In addition, Rx, which confers resistance to PVX by recognizing CP (Bendahmane et al., 1999; Moffett et al., 2002), was also examined. Intriguingly, these NBS-LRR-mediated recognition events are not dependent on CMPG1. This contrasts with another ubiquitination-associated protein, SGT1, a component of the Skp1-Cullin-F-box protein (SCF) ubiquitin ligases, which is essential for many HRs following perception of pathogen molecules, including AVR3a, PVX-CP, INF1, AvrPto, and C. fulvum Avr4 and Avr9 used in this work (e.g. Peart et al., 2002b; Bos et al., 2006; Gilroy et al., 2007).
In contrast with the observations above, the NBS-LRR protein Prf, in a complex with Pto at the host PM, responds to AvrPto with CMPG1-dependent HR, which is suppressed by AVR3a. Intriguingly, Pto kinase activity is required for the HR response to AvrPto (Balmuth & Rathjen, 2007), potentially indicating an additional signalling component that may differentiate this defence mechanism from the other NBS-LRRs. Pto has been shown to interact with and phosphorylate AvrPto-dependent Pto-interacting protein 3 (Adi3). Adi3 is an AGC kinase which, when activated by Pdk1, is a major negative regulator of host cell death that is conserved in both plants and animals (Devarenne et al., 2006). It has been proposed that Pto, interacting with AvrPto at the host PM, sequesters Adi3 at this location to prevent it from either interacting with the cytoplasmic Pdk1 protein, or phosphorylating its target proteins to suppress cell death (Devarenne & Martin, 2007). It would thus be interesting to investigate whether Adi3 and Pdk1 are targets for CMPG1-mediated degradation to promote host cell death.
The subcellular localization of CMPG1
Although it is not known where, within the cell, R3a-, R2- or Rx-mediated interaction with their cognate effectors occurs, all CMPG1-dependent cell death events reported previously (González-Lamothe et al., 2006), or in this work, have been shown, or can be predicted, to occur at the inner or outer surface of the host PM. Our data are thus suggestive of a specialized role for CMPG1 in cell death following recognition of elicitors at this subcellular location (Fig. 5). The location of CMPG1 within the plant cell thus warranted investigation.
Although CMPG1 activity mediates its own degradation by the 26S proteasome (Bos et al., 2010), CMPG1-YFP was nevertheless detectable in vesicles in the host cytoplasm (Fig. S4). Its stability within these vesicles implies that it is either inactive, or is sequestered away from additional components, such as the endogenous E2 conjugating enzyme, that mediate its activity as an E3 ligase. Indeed, the absence of fluorescence in the vesicles following split-YFP experiments between CMPG1-YC and YN-AVR3aKI suggests that AVR3a does not enter the vesicles to associate with CMPG1, supporting the potential inaccessibility of CMPG1 to additional factors. Co-expression of AVR3a, or treatment with the 26S proteasome inhibitor epoxomicin, both of which stabilize CMPG1, led to additional weak fluorescence free in the cytoplasm, and strong fluorescence in the nucleus, particularly in the nucleolus. We argue that the nucleus is likely to be a major site of 26S proteasome-mediated CMPG1 degradation. This is reminiscent of the transcriptional coactivator NPR1, a major regulator of salicylic acid (SA)-mediated defences, which is phosphorylated to promote NPR1 activity and entry into the nucleus, where it is degraded by the proteasome. This turnover of NPR1 is critical in modulating the transcription of its target genes (Spoel et al., 2009). Given that AVR3a suppresses CMPG1-mediated PCD, and strongly stabilizes it in the nucleus, this implies that, as with NPR1, the nucleus may be the major site of CMPG1 activity. Further work is required to determine how this potential site of CMPG1 activity relates to its regulatory role in promoting PCD triggered by perception of pathogen molecules at PM. However, the MAPK cascade components, which transduce pathogen perception events from PM, have been shown to accumulate rapidly in the nucleus after elicitor treatment (Lee et al., 2004).
Does CMPG1 dependence indicate that nonhost recognition of Eam occurs at PM?
Erwinia amylovora is the casual agent of fire blight in apple and pear. The recognition events responsible for the nonhost HR elicited by Eam in Nicotiana species are still being elucidated. Nevertheless, in both tobacco (Sinn et al., 2008) and N. benthamiana (this work), the nonhost HR triggered by low concentrations of bacterial cells [(2–5) × 106 CFU ml−1] is dependent on the T3SS protein hrpN. The harpin HrpN is a predicted T3SS ‘helper’ protein that elicits HR following its infiltration into the intercellular spaces of tobacco leaf tissue (Wei et al., 1992). It may thus be directly detected at the plant cell surface. However, it also aids the translocation of DspA/E, and presumably other T3SS effectors, into host cells and is required for full Eam virulence. Moreover, it is itself weakly translocated inside plant cells (Bocsanczy et al., 2008).
We have demonstrated previously that HR induced by inoculation of low concentrations of Eam in N. benthamiana is dependent on SGT1 and cathepsin B (Gilroy et al., 2007). The data presented here show that N. benthamiana HR triggered by Eam at these concentrations is also dependent on CMPG1. NbCMPG1 transcript accumulated rapidly and transiently in response to Eam, as described previously for a range of pathogen challenges (Kirsch et al., 2001), but then showed a second, distinct peak of induction at the onset of cell death symptoms (Fig. 4a). This is reminiscent of the dual peaks of induction observed during the compatible N. benthamiana–P. infestans interaction: one rapidly after spore inoculation, and a later peak at the onset of the necrotrophic phase (Bos et al., 2010).
We found that both AVR3aKI and AVR3aEM suppress the Eam nonhost HR in both N. benthamiana and tobacco, whereas AVR3aKI/Y147del does not, also in keeping with HR being dependent on CMPG1. From our model (Fig. 5), we predict that nonhost HR responses to Eam in these Nicotiana species, at low inoculum concentrations, probably involve recognition at the host PM. Such recognition could be a combination of extracellular perception of HrpN and/or intracellular perception of uncharacterized, HrpN-translocated effectors which, like AvrPto, may act at the inner surface of PM.
In conclusion, we report that the intracellular P. infestans effector AVR3a can exclusively suppress all of a wide range of CMPG1-dependent cell deaths, whereas the deletion mutant AVR3aKI/Y147del, which cannot interact with or stabilize CMPG1 (Bos et al., 2010), fails to suppress any CMPG1-dependent cell deaths. This corroborates our hypothesis that CMPG1 is an important virulence target of AVR3a, and the means by which it suppresses host cell death. AVR3aEM, which evades recognition by potato R protein R3a, was in many cases as efficient as AVR3aKI at PCD suppression. This is in agreement with our observation that both forms share an essential role in P. infestans virulence (Bos et al., 2010). We conclude that AVR3aEM has only partially compromised functional efficiency in its evolution to evade R3a recognition. We also revealed that CMPG1 has a critical role in the perception of pathogen molecules at the plant PM. Thus, the essential virulence role of AVR3a appears to be the prevention of signal transduction/regulatory events following the initial perception of diverse pathogen proteins at the inner or outer surface of the host cell.
We would like to thank Marie Anne Barny (INRA, Paris, France) for providing hrpN and hrcV Eam mutants. We thank Jake Harris and David Baulcombe (Cambridge University, UK) for pB1:Rx-HA and pBin61:CP-TK-expressing GV3101. John Rathjen (Australian National University, Canberra, Australia) kindly provided Agrobacterium containing 35S:AvrPto and 35S:Pto. Elodie Gaulin (Université Paul Sabatier, Toulouse, France) provided the pGR106:CBEL_SP construct. 35S::Cf-9 transgenic N. tabacum seeds were kindly provided by J. D. Jones (The Sainsbury Laboratory, Norwich, UK). We would also like to thank Anoma Lokossou (Wageningen University, the Netherlands) for the R2 construct. Thanks are also due to Ian Pitkethly (Scottish Crop Research Institute (SCRI), Dundee, UK) for assisting in the creation of our model in Fig. 5. We are very appreciative of Greg Martin’s open discussion and positive input. Many thanks go to Sophie Mantelin (SCRI), Jorunn Bos (SCRI), Greg Martin (Boyce-Thompson Institute, Ithaca, NY, USA) and Edgar Huitema (University of Dundee, UK) for helpful comments during writing. We are grateful to the Biotechnology and Biological Sciences Research Council, and Rural and Environmental Research and Analysis Directorate, UK, for funding this work.