Filamentous phytopathogens form sophisticated intracellular feeding structures called haustoria in plant cells. Pathogen effectors are likely to play a role in the establishment and maintenance of haustoria in addition to their better-characterized role in suppressing plant defence. However, the specific mechanisms by which these effectors promote virulence remain unclear. To address this question, we examined changes in subcellular architecture using live-cell imaging during the compatible interaction between the oomycete Hyaloperonospora arabidopsidis (Hpa) and its host Arabidopsis. We monitored host-cell restructuring of subcellular compartments within plant mesophyll cells during haustoria ontogenesis. Live-cell imaging highlighted rearrangements in plant cell membranes upon infection, in particular to the tonoplast, which was located close to the extra-haustorial membrane surrounding the haustorium. We also investigated the subcellular localization patterns of Hpa RxLR effector candidates (HaRxLs) in planta. We identified two major classes of HaRxL effector based on localization: nuclear-localized effectors and membrane-localized effectors. Further, we identified a single effector, HaRxL17, that associated with the tonoplast in uninfected cells and with membranes around haustoria, probably the extra-haustorial membrane, in infected cells. Functional analysis of selected effector candidates in planta revealed that HaRxL17 enhances plant susceptibility. The roles of subcellular changes and effector localization, with specific reference to the potential role of HaRxL17 in plant cell membrane trafficking, are discussed with respect to Hpa virulence.
Eukaryotic filamentous pathogens such as rust and powdery mildew fungi, and oomycetes including the downy mildew Hyaloperonospora arabidopsidis (Hpa) and Phytophthora species, form specialized feeding structures in their host cells that are called haustoria. Haustoria-producing pathogens include some of the most destructive plant parasites, causing huge economic losses in important agricultural industries and damage to natural ecosystems (Haverkort et al., 2008; Singh et al., 2011). Despite this, little is known about the molecular basis of pathogenicity in this important class of phytopathogens, mainly due to the fact that most of these organisms are difficult to culture and thus are not suitable for to molecular genetic approaches.
Haustoria are sophisticated but poorly understood structures. They are topologically distinct from the host-cell cytoplasm, and surrounded by an extra-haustorial membrane (EHM) of unknown origin. It has been shown that numerous host plasma membrane (PM) proteins are absent from the EHM of powdery mildew-infected Arabidopsis, suggesting that the EHM may originate from de novo membrane assembly or from the PM through a mechanism for excluding host proteins (Koh et al., 2005). Using in vivo live-cell imaging, the Arabidopsis syntaxin PEN1, which is involved in protein trafficking and disease resistance at the plant cell wall, has been localized to powdery mildew haustorial encasements (Meyer et al., 2009), whereas the broad-spectrum mildew resistance protein RPW8.2 from Arabidopsis has been reported to be localized to the EHM (Wang et al., 2009). During the formation and maturation of haustoria, callose and other plant cell-wall material is often deposited to form a collar around the haustorial neck, which can subsequently expand to encase the entire haustorium (Bushnell, 1972; Meyer et al., 2009). In contrast to subcellular dynamics at pathogen penetration sites (Howard and Valent, 1996; Takemoto et al., 2003, 2006; Genre et al., 2005), the rearrangement of plant cell compartments in cells hosting oomycete haustoria remains uncharacterized.
In contrast to fungal effectors, for which no conserved region has been yet identified, oomycete effectors usually contain a secretory signal peptide and a conserved domain featuring the motif RxLR, followed by a motif with a high frequency of acidic (D/E) residues, and a C-terminal domain(s) associated with virulence function (Rehmany et al., 2005; Birch et al., 2006; Win et al., 2007). To date, nine avirulent oomycete RxLR effectors have been identified, including ATR1 and ATR13 from Hpa (Allen et al., 2004; Rehmany et al., 2005). These intracellular effectors are presumed to be first secreted by the oomycetes (Whisson et al., 2007; Gilroy et al., 2011), and are then taken up by the host cell by an unknown mechanism involving the N-terminal RxLR motif (Kale et al., 2010). The sequencing of several oomycete genomes, including Hpa (Baxter et al., 2010), has allowed prediction of a large number of effector genes that share N-terminal sequences with the known effectors (Win et al., 2007; Baxter et al., 2010). The Hpa genome contains 134 high-confidence effector gene candidates (HaRxL genes), including the known effector genes ATR1 and ATR13, significantly fewer than Phytophthora genomes (Tyler et al., 2006; Jiang et al., 2008).
To assess the function of RxLR effector candidates in plant immunity, screening methods have recently been developed (Cabral et al., 2011; Fabro et al., 2011) Use of these methods has shown that the majority of the HaRxLs (approximately 70%) positively contributed to bacterial virulence, and this was positively correlated with their ability to suppress plant-triggered immunity (Cabral et al., 2011; Fabro et al., 2011). However, the mechanisms by which Hpa effectors promote virulence are still largely unknown.
The localization of pathogen effectors inside the plant cell gives an indication of their mode of action (Dowen et al., 2009; Schornack et al., 2010). Increasing evidence suggests that host protein secretion is an important process that mediates plant resistance against pathogens, and that pathogen virulence factors may target intracellular trafficking to suppress host immunity (Huckelhoven, 2007; Kalde et al., 2007; Robatzek, 2007; Nomura et al., 2011; Wu et al., 2011). In addition, bacterial effectors have been shown to localize to the nucleus of the plant cell where they may modulate plant gene expression (Gurlebeck et al., 2006; Poueymiro and Genin, 2009). In contrast with bacteria and necrotrophic fungi, many obligate biotroph pathogens are recalcitrant to stable transformation. In addition, when tranformants are generated, translocation of the effector in the host cell cannot usually be observed as the fluorescent signal is diluted in the plant cell cytoplasm (Whisson et al., 2007; Boevink et al., 2011). To circumvent this problem, fluorescent-tagged effector protein can be expressed directly in the plant cell (Kelley et al., 2010; Rafiqi et al., 2010; Schornack et al., 2010). Using this approach, the RxLR effector Avr3a was localized in planta during infection using Agrobacterium tumefaciens delivery of tagged effectors (Bos et al., 2010; Boevink et al., 2011). Thus, use of cell biology techniques to assess the virulence function of oomycete effectors is an emerging field. Further, Bozkurt et al. (2011) report focal recruitment of P infestans Avr-blb2:GFP fusions to haustorial membranes during the P. infestans/Nicotiana benthamiana interaction.
In order to study the molecular dialogue between Hpa and Arabidopsis, we developed an in vivo cell biology approach to analyse changes to subcellular compartments within the host cell and identify the compartments targeted by Hpa effectors. By transiently expressing 49 Hpa RxLR effector candidates (HaRxLs) fused to fluorescent tags in planta, we defined major classes of HaRxLs that accumulate in the plant cell nucleus and plant membranes. In planta functional analysis revealed that the membrane-localized effector HaRxL17 enhances plant susceptibility. This study allowed us to identify a role for plant cell membrane trafficking in modulation of Hpa virulence.
Dynamics of host plant cell compartments during Hpa infection, highlighting the close association between the plant cell nucleus and Hpa haustoria
We used a collection of Col-0 Arabidopsis fluorescent protein-tagged lines to unveil the organization of plant cell compartments in vivo in mesophyll cells hosting haustoria (Table 1). Ten-day-old seedlings of Arabidopsis marker lines were inoculated with virulent Hpa Noco2 isolate. Infected mesophyll tissues were then observed at an early stage of infection [4 days after infection (DAI)] by live-cell imaging. At this stage, young developing haustoria were observed near the growing tip of Hpa hyphae, and mature haustoria encased in callose-rich depositions were observed along the ageing hyphae.
Table 1. Summary of plant cell compartment rearrangements upon infection with Hpa in Arabidopsis mesophyll cells
We observed aggregation of ER membranes and accumulation of Golgi bodies around young Hpa haustoria, suggesting that production and secretion of plant materials are activated around the haustoria (n >20; Figure 1a and Figure S1a) as previously described by Chou (1970). In vivo Z-stack confocal imaging showed that actin microfilaments formed large bundles in cytoplasmic strands focused around the haustorium (asterisk) and the adjacent nucleus (arrow) (n =9, Figure 1b). Using free GFP as a nuclear/cytoplasmic marker, we confirmed the presence of cytoplasm around the haustorium (n >50, Figure 1c). Notably, we observed that the nucleus was always associated with developing haustoria in mesophyll cells (n >50, Figure 1b,c). This association between haustoria and nuclei in mesophyll cells was confirmed using the plant nuclear marker line H2A-GFP (n =15, Figure S1b). We observed that mitochondria accumulated around haustoria (n =15, Figure S1e), whereas chloroplasts and peroxisomes appeared to be free in the cytoplasm of mesophyll cells hosting haustoria (n =15, Figure S1d). This observation was surprising as the peroxisome-associated marker PEN2 localized around the haustorium during Hpa infection (n =5, Figure S1e).
Plasma membrane markers are excluded from the EHM, but the tonoplast is located close to the membrane surrounding the haustorium
It has been shown that numerous host PM proteins are absent from the EHM of fungi-infected Arabidopsis, suggesting that the EHM originates from de novo assembly of plant membranes or remodelling of the existing PM with exclusion of host proteins. Therefore, we investigated the subcellular localization of a PM marker in order to verify its exclusion from the EHM of oomycete-infected Arabidopsis. Ten-day-old Arabidopsis seedlings expressing the GFP-tagged aquaporin PIP1;4 were inoculated with the Hpa virulent isolates Noco2 and Waco9. Infected mesophyll tissues were then observed at an early stage of infection (4 DAI) by live-cell imaging. In non-infected cells (Figure 1d), GFP-tagged PIP1;4 localized to the rough ER and the PM, as previously reported (Geldner et al., 2009). In Hpa-infected cells, no GFP–PIP1;4 signal was observed at the EHM. However, a weak signal was detected in the cytoplasm around haustoria corresponding to the ER (n =15; Figure 1d).
We next determined the localization of the PEN1-GFP and RPW8.2-YFP fusion proteins under the control of native promoter (Np) during Hpa infection. In non-infected cells, PEN1–GFP localizes to the PM in mesophyll cells (Collins et al., 2003), but RPW8.2–YFP is not detectable (Wang et al., 2007). At an early stage of infection (3 DAI), PEN1–GFP is associated with the callose ring present at the neck of the developing haustorium (n >20, Figure S2a), but RPW8.2–YFP is localized to vesicle-like bodies surrounding haustoria (n =10, Figure S2b). No signal at the EHM was observed for either marker at early stages of haustorium formation. Later during infection, PEN1–GFP and RPW8.2–YFP localized to haustorium encasements (n >20, Figure S2c,d), as described for powdery mildew haustoria. Taken together, these results confirm the exclusion of PM markers from the EHM of Hpa haustoria and their accumulation in the callose-rich haustorium encasements as previously reported for fungal pathogens.
Previous ultrastructure studies of plant cells hosting haustoria described close proximity between the EHM and the tonoplast, resulting in a very thin layer of host-cell cytoplasm (approximately 0.1 μm between the EHM and the tonoplast) (Chou, 1970; Mims et al., 2004). Using in vivo confocal imaging, we detailed the behaviour of the tonoplast over a time course of Hpa infection. We showed that, in contrast to the exclusion of PM markers around haustoria, the tonoplast intrinsic protein γ-TIP–GFP strongly labelled the vicinity of the EHM in young haustoria (n >25; Figure 2a). At this stage of infection (4 DAI), the central vacuole of mesophyll cells hosting haustoria resembled the vacuole in a non-infected mesophyll cell (n >25, Figure 2a), as previously reported by Chou (1970). Later during infection (7 DAI), very bright γ-TIP–GFP membrane labelling was observed in the lumen space of the vacuole around mature haustoria (n >25; Figure 2b,c). The highly fluorescent structures observed (including lines, arcs and circles) are very similar to the double tonoplast cytoplasmic invaginations into vacuoles described in leaves and cotyledons of Arabidopsis (Saito et al., 2002), and are termed ‘bulbs’. Such bulbs were rarely observed in surrounding non-infected cells (n >25, Figure 2b). In vivo time-lapse imaging of γ-TIP–GFP-labelled tonoplast membrane revealed highly dynamic vacuole trafficking around mature haustoria in infected mesophyll cells. Bulbs accumulated around mature haustoria, but appeared to be unable to fuse with membranes surrounding haustoria (n =7; Figure 2b,c and Video Clips S1 and S2). Within the same infected cell, other bulbs trafficked freely, budding from and fusing to the tonoplast of the central vacuole.
Subcellular localization of the Hpa effector repertoire reveals that the host nucleus and membrane network are the main targeted compartments
As some plant compartments are highly dynamic upon Hpa infection, we investigated which host-cell compartments are targeted by the effectors of this obligate biotroph to promote its virulence. To address this question, we designed a fluorescence imaging-based screen in planta to assess the subcellular localization of a collection of Hpa RxLR effector candidates (HaRxLs) from the Hpa reference isolate Emoy2 (Fabro et al., 2011; Table S1). We used a recently developed transient assay (Dowen et al., 2009; Schornack et al., 2010) to express oomycete and fungal effectors in planta. We generated fluorescent protein-tagged versions of 49 HaRxLs downstream of their signal peptide cleavage site. Chimeric constructs were expressed transiently in Nicotiana benthamiana using A. tumefaciens infiltration (Table S1), and were imaged 24–48 h post-infiltration. The stability of the fusion protein was verified by Western blot analysis for each candidate (Figure S3).
Of the 49 HaRxLs tested, 16 (33%) localized to the plant cell nucleus and the cytoplasm, including the avirulent ATR1Emoy2 allele (Figure 3a,b, Figure S5 and Table S1). We identified three HaRxLs that strictly localized to the plant cell cytoplasm (Figure 3c, Figure S4 and Table S1) and one localized to the vacuole (Figure 3d). The majority (66%, merging the strictly nuclear-localized and the nuclear-cytoplasmic) of HaRxLs targeted the plant cell nucleus (Figure 3a,e–j, Figure S4, Figure S5 and Table S1), but in silico analysis showed that only 37.5% of the 16 HaRxLs that strictly localized to the nucleus carried a canonical nuclear localization signal (Table S1). Of the 16 HaRxLs that localized to the plant cell nucleus, we identified specific sub-nuclear localizations: GFP-HaRxL18 was excluded from the nucleolus (n = 15, Figure 3e); RFP-HaRxLL108 localized to the nucleoplasm and possibly labelled weakly the nucleolus (n = 20, Figure 3f); GFP-HaRxLL3b showed strong labelling of the nucleolus and fiber-like structures in the nucleoplasm (n = 10, Figure 3g); GFP-HaRxL36 showed strong nucleoplasmic labelling and nucleolar cap-like structures (n = 10, Figure 3h); GFP-ATR13Emoy2 concentrated into the cell nucleolus (n = 25, Figure 3i) and GFP-HaRxLL470b showed strong labelling of the nucleolus and speckles in the nucleoplasm (n = 15, Figure 3j). Of the 16 HaRxLs that localized to the plant cell nucleus, we found that 11 were nucleolar localized, including the virulent allele ATR13Emoy2.
In addition to the nuclear-localized HaRxLs, we identified a second class of effectors that targeted the plant membrane trafficking network. Thirteen HaRxLs (26%) localized to various plant membranes. Of these, nine localized to the ER network, including HaRxLL495b (Figure 3k, Figure S4 and Table S1). Surprisingly, only three candidates, including HaRxL77, localized to the plant PM (Figure 3l and Figure S4). The PM-localized proteins HaRxL77 and HaRxL146 are both predicted to be potentially post-translationally modified by N-myristoylation, whereas HaRxL47 is predicted to have two coiled-coil domains. Remarkably, one protein, HaRxL17, localized to the tonoplast (Figure 3m and Video Clip S3). HaRxLR17 sequence analysis revealed three coiled-coil motifs and a C-terminal W motif, a conserved domain that is found in avirulent effectors from Phytophtora species (Bos et al., 2006; Dou et al., 2008; van Poppel et al., 2008). In addition, these subcellular localizations did not correlate with predictions of putative trans-membrane domains. No localization to the chloroplast, peroxisome, mitochondria or cytoskeleton was found for any of the HaRxLs studied (Figures S4 and S5 and Table S1).
In order to validate the screening system in N. benthamiana, we next generated stable Arabidopsis transgenic lines expressing single GFP-tagged HaRxLs. We randomly selected 20 candidates that localized to different plant cell compartments (Table S1). For all of the GFP-tagged HaRxLs tested, we observed the same subcellular localization in both systems, thus validating the survey. Only 20% of the localizations observed were correctly predicted in silico.
HaRxLs polymorphisms reveal allelic differences in subcellular localization in planta
For some HaRxLs, evidence of polymorphism between eight Hpa isolates (Cala2, Emco5, Emoy2, Hiks1, Hind2, Maks9, Noco2 and Waco9) has been found (Fabro et al., 2011). We used transient assays in N. benthamiana to assess the subcellular localization of three HaRxLs that show allelic variation between Hpa isolates, ATR13, HaRxL45 and HaRxL9. Live-cell imaging showed that YFP–ATR13Maks9 and GFP–ATR13Emoy2 localized to the plant cell nucleolus, whereas GFP–ATR13Emco5 localized to the plant nucleus and cytoplasm (n >30, Figures 3i and 4a,b). This difference in localization can be explained by the presence of a predicted nuclear localization signal in both ATR13Maks9 and ATR13Emoy2 that is absent in ATR13Emco5 (Figure 4g). GFP–HaRxL45Emoy2 localized to the plant nucleoplasm and nucleolus (Figure 4c), but GFP–HaRxL45Hiks1 localized to the plant cytoplasm and the nucleus and was excluded from the nucleolus (Figure 4d). These differences in subcellular localizations were not correlated with polymorphism with regard to a predicted nuclear localization signal in the sequences of HaRxL45 alleles (Figure 4g). We also found that, while HaRxL9Emoy2 was localized to the nucleus and cytoplasm (Figure 4e), HaRxL9Noco2 localized to the plant PM (Figure 4f). Thus, for the tested polymorphic HaRxLs, amino acid polymorphisms cause a change in their subcellular localization in planta, which may influence their virulence functions.
HaRxL17 and HaRxL77 effectors confer enhanced plant susceptibility
The role of membrane-localized HaRxLs during Hpa infection in Arabidopsis was assessed by examining their effect on pathogen virulence. We selected three candidates that localized to different membranes, ER-localized HaRxLL495b (Figure 3k), PM-localized HaRxL77 (Figure 3l) and tonoplast-localized HaRxL17 (Figure 3m). We first verified that these candidates were expressed in planta during Hpa infection using RT-PCR (Figure S6). We next generated Arabidopsis Col-0 plants that stably expressed a single GFP-tagged HaRxL. We confirmed the presence of fusion proteins in the tested transgenic plants by Western blot analysis (Figure S7). These lines, and the GFP–ATR13 line as a control, were challenged with the virulent Hpa Waco9 isolate.
Two independent GFP–ATR13-expressing transgenic lines were more susceptible to Hpa than Col-0 (P >0.001; Figure 5a), consistent with the observation that ATR13 positively contributes to pathogen virulence inside host cells (Sohn et al., 2007). In contrast, the transgenic lines expressing HaRxLL495b did not show any difference in susceptibility to Hpa compared with Col-0 (Figure 5b). Independent transgenic lines expressing GFP–HaRxL17 and GFP–HaRxL77 were more susceptible to Hpa than wild-type plant (P >0.001; Figure 5c,d). Thus, the HaRxL17 and HaRxL77 effectors enhance plant susceptibility during a compatible interaction between Hpa and its host.
We next analysed whether HaRxL localization changed during infection with Hpa using live-cell imaging. We first confirmed that, in Arabidopsis transgenic lines, independently of the level of expression, GFP-tagged HaRxLs localized to the same plant cell compartment as observed in a transient assay in N. benthamiana (n >30, Figure S8). Using a biochemical approach, we next confirmed that GFP–HaRxL17 accumulates in the microsomal fraction in Arabidopsis transgenic line (Figure S9). Subcellular localization of GFP-tagged HaRxLs at 3 DAI with Hpa showed that GFP–ATR13 localized to the plant cell nucleolus in Hpa Waco9 haustorium-invaded mesophyll cells (n =20, Figure 5e). HaRxLL495b–GFP localized around the haustoria and in the plant cell nucleus (n =10, Figure 5f). GFP–HaRxL77 still localized at the PM in the infected cells, but also at the haustorial neck (n =20, Figure 5g). In addition, a weak fluorescent signal corresponding to GFP–HaRxL77 was observed around haustoria (Figure 5g). Thus, no re-localization of nucleolar-localized ATR13, PM-localized HaRxL77 or ER-localized HaRxLL495b was observed in Hpa Waco9 haustorium-invaded mesophyll cells. However, in transgenic lines expressing GFP-tagged HaRxL17, strong re-localization of this effector around the haustoria was observed in early stages of Hpa infection (n >20, Figure 5h). At this stage, no signal in the tonoplast or PM was observed in infected cells (n >20, Figure 5h). Thus, the Hpa effector HaRxL17, which suppressed plant immunity during a compatible interaction between Hpa and its host, re-localized around haustoria at early stages of infection.
HaRxL17 suppresses plant immunity during an incompatible interaction and localizes at the EHM
We tested whether any of the selected HaRxLs could compromise effector-triggered immunity. In Arabidopsis Col-0, RPP4 confers resistance to the Hpa Emoy2 isolate (van der Biezen et al., 2002). Transgenic lines stably expressing GFP–ATR13, HaRxLL495b–GFP, GFP–HaRxL77 or GFP–HaRxL17 were therefore challenged with Hpa Emoy2.
In wild-type Col-0 cotyledons infected with Hpa Emoy2, asexual or sexual reproduction of the avirulent pathogen was rarely observed (Figure 6a,b). Under our laboratory conditions, no more than two conidiophores or two oospores were usually seen (Figure 6a,b). No differences in resistance were observed in the transgenic lines expressing GFP–ATR13, HaRxLL495b–GFP or GFP–HaRxL77 (Figure 6a). However, a considerable increase in sexual reproduction of Hpa Emoy2 was observed in the transgenic lines expressing GFP– HaRxL17 (Figure 6a).
Differences in Hpa Emoy2 growth in Col-0 versus GFP–HaRxL17 lines were analysed in detail by trypan blue staining of infected cotyledons. Emoy2-infected Col-0 cotyledons showed development of few short hyphae in the epidermis and occassionally haustoria (Figure 6b). In contrast, GFP–HaRxL17 lines supported increased hyphal growth of Hpa Emoy2 (Figure 6c). This hyphae production may allow hyphae inter-twining, which is essential for sexual reproduction (Figure 6c). We also noted an increase in haustoria development (Figure 6c) in mesophyll cells, which resembled what was observed in the compatible interaction. To better understand the role of this effector during infection, we analysed haustoria formation in the GFP–HaRxL17 line infected with Hpa Emoy2 at the subcellular level. We identified encased haustoria at a late stage of the interaction (6 DAI) (Figure 6d). We then analysed the subcellular localization of HaRxL17 in Hpa Emoy2 haustorium-invaded mesophyll cells. In vivo confocal imaging revealed that HaRxL17 was strongly localized inside the encasement (Figure 6e), probably to the EHM. Thus, HaRxL17 is an effector that is strongly localized to the membrane around haustoria and increases the virulence of Hpa in both compatible and incompatible interactions.
The haustorium is a highly specialized feeding structure of phytopathogens of taxonomically distinct classes, including fungal powdery mildews, rusts and oomycetes. Here, we developed an in vivo cell biology approach in order to study the interaction between Hpa and Arabidopsis. We monitored changes to the subcellular architecture during Hpa infection in order to identify subcellular compartments that play significant roles in defence responses. Further, we attempted to correlate this information with the subcellular localization of Hpa effector candidates and their functional role in compatible and incompatible interactions. Significantly, by identifying an effector that associates with the host tonoplast in non-infected cells and with the EHM in infected cells, this study has highlighted a possible relationship between vacuolar and defence-induced membranes that plays a role in establishment of a compatible interaction.
In order to identify the plant cell compartments targeted by pathogen effectors, several cell biology approaches have been used (Boevink et al., 2011; Kemen et al., 2011). As transformation systems are not yet available for most biotrophic pathogens, functional analysis of tagged effector proteins using heterologous systems remains one of the most promising options (Kelley et al., 2010; Rafiqi et al., 2010; Schornack et al., 2010). In this study, we used fluorescent protein-tagged effectors that were either transiently or stably expressed in the plant cells. This approach allowed easy and rapid identification of promising pathogen effector candidates. However, we cannot exclude the possibility of any dominant-negative effects or interference by the GFP tag that could affect effector functions.
During the compatible interaction between Hpa and Arabidopsis, haustorium-invaded mesophyll cells undergo rearrangement. The plant cell nucleus stays close to the developing haustorium, presumably via movement through the actin cytoskeleton in infected mesophyll cells (Ketelaar et al., 2002; Iwabuchi et al., 2010). It is conceivable that the haustorium directly influences nuclear position in the cell to enhance delivery of effectors that compromise nuclear processes involved in defence. Alternatively, defence may be activated by hyphal penetration and haustorial development, although this is not sufficient to stop the growth of the pathogen. In this case, nuclear proximity to the haustorium maximizes the efficiency of delivery of anti-microbial transcripts and proteins to the intruding structure. The plant ER and Golgi surround developing haustoria, suggesting that production and secretion of plant materials are activated around the haustoria. We showed that, in contrast to the exclusion of PM markers around haustoria, the tonoplast is in close proximity to the EHM. In vivo time-lapse imaging of γ-TIP–GFP-labelled tonoplast membrane revealed the presence of numerous bulbs in Hpa-infected cells. Extensive examination by fluorescence and electron microscopy demonstrated that the bulbs are projections of cytoplasm, often consisting of two membranes (Saito et al., 2002). The outer and inner membranes, as well as the limiting vacuolar membrane, appear to be connected (Saito et al., 2002). Even though the role of these bulbs remains unclear (Saito et al., 2002, 2011; Escobar et al., 2003; Hicks et al., 2004; Beebo et al., 2009), it has been suggested that, if some vacuoles are not easily fused, double-membrane structures (bulbs) may be left behind for some time (Saito et al., 2002; Beebo et al., 2009). As bulbs accumulate in haustorium-invaded mesophyll cells, we speculate that turnover of the vacuolar membrane in mesophyll cells may be disturbed in the presence of the pathogen. The high-affinity phosphate transporter PT1 also localized to bulb structures (Escobar et al., 2003). Thus, it is plausible that bulbs enable re-mobilization of phosphate from the central vacuole to the haustoria.
Why do many Hpa RxLR effectors target the plant cell nucleus?
Increasing evidence suggests that the plant cell nucleus is one of the main targets for pathogen effectors. Recent studies that examined interactions between plants and bacterial pathogens found that effectors are probably targeting the host nucleus in order to re-program the plant cell transcriptional machinery, suppress the plant immune response and subsequently promote pathogen virulence (Kay and Bonas, 2009; Boch and Bonas, 2010). Recent advances in functional analysis of oomycete effectors from P. infestans confirmed that CRINKLER (CRN) effectors target the nucleus (Schornack et al., 2010). The authors showed that inhibition of CRN8 nuclear accumulation prevented cell death, and thus that CRN8 nuclear localization is required for cell-death induction. In the present study, the survey of the in planta subcellular localization of one third of the predicted HaRxLs revealed that a large majority (66%) targeted the plant cell nucleus. In contrast with effectors from bacterial pathogens, canonical nuclear localization signals are rarely predicted in the sequence of these oomycete effectors. Thus, oomycete effectors may have evolved other ways that allow them to target the plant cell nucleus, perhaps by association with endogenous host protein(s).
Twenty-one per cent of the HaRxLs tested (merging the strictly nucleolar and the nucleolar-cytoplasmic), including ATR13Maks9 and ATR13Emoy2, localized to the host nucleolus. Of all plant pathogens, only secreted proteins from cyst nematodes have been reported to target the plant nucleolus (Tytgat et al., 2005; Elling et al., 2009). The main functions of the nucleolus lie in transcription of ribosomal RNA gene units, processing and modification of precursor rRNA, and ribosomal subunit assembly (Brown et al., 2005; Raska et al., 2006). We hypothesize that Hpa hijacks the plant cell transcriptional machinery, acting on ribosome biogenesis and thus on protein translation in order to prevent de novo induction of defence responses. In addition, the nucleolus has been implicated in a variety of other functions, including biogenesis, transport, splicing of RNAs and transcriptional gene silencing (Pontes and Pikaard, 2008; Rodor et al., 2010). Thus, these additional processes could be the target of nucleolar-localized HaRxLs. We found specific sub-nuclear localizations for some HaRxLs that resemble the localization of already characterized Arabidopsis nucleolar proteins. For example, the nucleolar cap-like structures observed using GFP–HaRxL36 in planta resemble the localization of the Prp19-like U-box proteins, which are associated with the MOS4-associated complexes that are involved in plant innate immunity (Monaghan et al., 2009). Thus, detailed functional analysis of nucleolar-localized HaRxLs and identification of their plant targets may identify unknown processes that are essential for Hpa virulence.
Why do Hpa RxLR effectors target the plant plasma membrane?
The first level of microbe recognition in plants is performed by membrane proteins termed pattern recognition receptors (PRRs), which perceive molecular signatures characteristic of a whole class of microbes, termed pathogen-associated molecular patterns (PAMPs). To infect plants successfully, pathogens must overcome PAMP- triggered immunity (PTI). Several recent studies clearly showed that one strategy to do so is to directly target PRRs and their associated proteins by the action of virulence effectors (Robert-Seilaniantz et al., 2006; Thieme et al., 2007; Dowen et al., 2009; Wu et al., 2011). In contrast with bacterial effectors, nothing is known about membrane-localized effectors of eukaryotic phytopathogens. Recently, the receptor-like kinase SERK3/BAK1 was shown to be required for basal resistance against P. infestans in N. benthamiana (Chaparro-Garcia et al., 2011), indicating that some of the defence mechanisms employed by host cells in response to bacteria overlap with those induced in response to oomycetes. In addition, recent data showed that PAMP pre-elicitation impairs Hpa growth (and reproduction), indicating that Hpa must counteract PTI responses in order to promote virulence (Fabro et al., 2011). In the present study, we identified three Emoy2 HaRxLs and one Noco2 HaRxL that target the plant PM. Arabidopsis lines expressing the PM-localized HaRxL77 were more susceptible to Hpa than Col-0, indicating that HaRxL77 increases Hpa virulence during a compatible interaction. The localization of HaRxL77 at the PM suggests that this effector may act on PTI responses during pathogen growth. Recent evidence also suggested that vesicle trafficking is an important component of PTI (Frei dit Frey and Robatzek, 2009). Thus, we hypothesize that the PM-localized HaRxL77 may prevent exocytosis of defence components or endocytosis of PRR in order to dampen or regulate PTI responses. Bozkurt et al. (2011) showed that the P. infestans AVRblb2 effector interferes with secretion of host cysteine protease C14, thus reducing its contribution to defence.
How can the study of membrane-localized HaRxLs help to understand EHM ontogenesis?
Live-cell imaging revealed that GFP–HaRxL17 localized to the EHM at a late stage of infection between Hpa Emoy2 and Col-0. Previous searches for EHM-resident proteins had limited success. In the present study, however, live-cell imaging showed that RPW8.2 localized into vesicle-like bodies surrounding the Hpa haustorium during the early stage of infection, but was absent from the EHM. In mature haustoria, RPW8.2 localized to callose-rich encasements, as described for haustoria from powdery mildew. Ultrastructural characterization of the haustorial interface in the Golovinomyces orontii/Arabidopsis interaction confirmed the recruitment of RPW8.2 to the EHM of mature haustoria (Micali et al., 2011). As no EHM-localized proteins have been identified at the EHM of young haustoria, the mechanism by which the EHM is formed remains elusive, preventing complete understanding of compatible interactions between haustoria-forming pathogens and their hosts (Koh and Somerville, 2006). Recently, Bozkurt et al. (2011) showed that GFP:AVRblb2 fusion protein preferentially accumulated around P. infestans haustoria.
Ultrastructural studies of haustorium-invaded cells have identified that the EHM is proximal to the tonoplast, sandwiching a very thin layer of host-cell cytoplasm (approximately 0.1 μm) between the EHM and the tonoplast (Chou, 1970; Mims et al., 2004). Using live-cell imaging, we detailed the behaviour of the tonoplast over a time course of Hpa infection, and showed that the tonoplast intrinsic protein γ-TIP–GFP strongly labelled membranes within the EHM in young haustoria. The EHM and the tonoplast are too close in infected cells to distinguish between them using light microscopy, and thus higher-resolution techniques are required to determine whether or not this marker labels the EHM. The difference in localization of HaRxL17 in infected cells (EHM) and uninfected cells (tonoplast), and the changes to the vacuolar sub-structure induced by infection, indicate that vacuole-associated membranes are highly dynamic during haustoria formation. Indeed, the role that the tonoplast plays in haustoria formation and secretion of defence compounds requires further examination. The association of GFP–HaRxL17 with the EHM of mature haustoria can be exploited to further investigate the composition of this particular membrane. Biochemical analysis of plants expressing GFP-tagged HaRxL17 during infection with Hpa may allow us to identify plant-interacting proteins that localize to the EHM and to understand the plant cell vesicle trafficking that allows formation of the final membrane layer separating the host cell from its invader. In addition, localization of HaRxL17 in the native system at the cellular and subcellular level using immunocytological techniques (Kemen et al., 2011) will help us to understand the function of this effector during haustoria ontogenesis.
Characterization of nuclear and membrane-localized HaRxL functions will be critical for understanding the basis of plant pathogenesis. Further understanding of the cell biology and compartmentalization of signalling and metabolic processes during plant–pathogen interactions will highlight the mechanisms employed in the fight against infection.
Cloning of HaRxLs
To produce the effector candidate collection, primers for amplification were designed from the Hpa Emoy2 genome version 8.3 (Baxter et al., 2011). The HaRxL sequences are available in Fabro et al. (2011). Selected HaRxLs were amplified from the signal peptide cleavage site to the stop codon using genomic DNA extracted from conidiospores of Hpa Emoy2 and Accuprime Pfx proof-reading polymerase (Invitrogen, http://www.invitrogen.com/) under standard PCR conditions. The PCR fragment was inserted into the pENTR donor vector (Invitrogen) and then into plant expression vectors pK7WGF2, pK7FWG2, pH7WGR2 or pH7WGY2 using Gateway technology (Invitrogen). The constructs were sequenced by the Genome Analysis Centre in Norwich (UK).
The expression pattern of HaRxL candidates was determined using cDNA for Illumina sequencing (from tissue at 3 and 7 days post-infection) (Fabro et al., 2011). For semi-quantitative RT-PCR, 50 ng of total RNA was used and amplified with an initial denaturation step at 95°C for 3 min followed by 25 cycles of 45 sec at 94°C, 30 sec at 55°C and 72°C for 30 sec. In the last cycle, a final elongation step at 72°C for 5 min was added. PCR products were separated on 1.5% TAE agarose gels. The RPL4 gene was used to control.
Transient and stable expression in planta
For in planta expression of the fluorescent protein-tagged HaRxLs, A. tumefaciens strains GV3101 and GV3103 were used to deliver transgenes into N. benthamiana leaves (Caillaud et al., 2009). Protein stability was assessed using Western blotting as described by Schornack et al. (2010). For stable expression in planta of selected candidates, Col-0 Arabidopsis plants were transformed using the dipping method (Clough and Bent, 1998) and selected on MS medium per 0.7% agar plates containing 50 μg ml−1 kanamycin. Transformed plants were transferred to soil and seeds were collected. Ten transformed plants were analysed for each construct. T3 homozygous plants were used for in vivo confocal microscopy and pathogen assays. For fractionation analysis, two independent Arabidopsis lines expressing GFP–HaRxL17 or free GFP were used in two independent experiments. Fractionation was performed as described by Boyes et al. (1998).
For infection, 10-day-old plants were spray-inoculated to run-off using a spore suspension of 5 × 104 conidiospores per ml. Plants were kept in a growth cabinet at 16°C for 6 days with a 10 h photoperiod. To evaluate conidiospore production, ten pools of two plants for each Arabidopsis line were harvested in 1 ml of water. After vortexing, the amount of conidiospores released was determined using a haemocytometer as described by Robert-Seilaniantz et al. (2011). Statistical analyses were performed using anova for three independent experiments.
To evaluate hyphae growth and oospore production, infected cotyledons were stained with trypan blue. For Hpa Emoy2 infection in Col-0 and transgenic lines, the number of oospores per cotyledons was manually scored by observing the abaxial and adaxial surfaces of each cotyledon of 30 plants per line. For imaging of Hpa Emoy2 development in the cotyledons, the samples were observed using a Leica DM R fluorescence microscope (http://www.leica.com/). Photographs were taken using a Leica DFC 300 FX digital camera.
For determination of the subcellular localization of the HaRxLs in N. benthamiana, cut leaf samples were mounted in water and analysed on a Leica DM6000B/TCS SP5 confocal microscope using the following excitation wavelengths: GFP, 488 nm; YFP, 516 nm; RFP, 561 nm. For in vivo localization in Arabidopsis transgenic lines, 10-day-old infected seedlings were mounted in water and analysed on a Leica DM6000B/TCS SP5 confocal microscope using the following excitation wavelengths: CFP, 458 nm; GFP, 488 nm; RFP, 561 nm. The HaRxL sub-nuclear localizations were named according to the nomenclature in the Arabidopsis nucleolar protein database at http://bioinf.scri.sari.ac.uk/cgi-bin/atnopdb/home.
We thank Peter Shaw’s laboratory (John Innes Centre, Norwich, UK) for the generous gift of H2A–GFP Arabidopsis seeds. We thank Jordi Chan (John Innes Centre, Norwich, UK) and Volker Lipka (Albrecht-von-Haller, Göttingen, Germany) for access to Arabidopsis fluorescent markers, and Grant Calder (John Innes Centre, Norwich, UK) for his technical support. We thank Matthew Smoker (TSL, Norwich, UK) and Jodie Pike (TSL, Norwich, UK) for stable transformation of Arabidopsis. We thank Fabio Galeotti (UEA, Norwich, UK) for his help with statistics. We thank Freddy Boutrot (TSL, Norwich, UK) and Tolga Bozkurt (TSL, Norwich, UK) for their help with fractionation analysis. We would also like to thank Christine Faulkner (John Innes Centre, Norwich, UK) Ali Pendle (John Innes Centre, Norwich, UK), Silke Robatzek (TSL, Norwich, UK) and Valérie Nicaise (TSL, Norwich, UK) for fruitful discussions and their help with the manuscript. We thank Sophien Kamoun, Sebastian Schornack and Tolga Bozkurt for early sight of their discovery of focal recruitment of Avr-blb2 to P. infestans haustoria, and for helpful discussion to improve the manuscript. This work was supported by long-term fellowships EMBO ALTF 614 and Marie Curie FP7-PEOPLE-2009-IEF to M.C.C., Biotechnology and Biological Sciences Research Council grant G015066 to J.B. and the Gatsby Foundation. This paper is dedicated to Madeleine and Jean-Pierre Caillaud.