Pre- and Postinvasion Defenses Both Contribute to Nonhost Resistance in Arabidopsis

See allHide authors and affiliations

Science  18 Nov 2005:
Vol. 310, Issue 5751, pp. 1180-1183
DOI: 10.1126/science.1119409


Nonhost resistance describes the immunity of an entire plant species against nonadapted pathogen species. We report that Arabidopsis PEN2 restricts pathogen entry of two ascomycete powdery mildew fungi that in nature colonize grass and pea species. The PEN2 glycosyl hydrolase localizes to peroxisomes and acts as a component of an inducible preinvasion resistance mechanism. Postinvasion fungal growth is blocked by a separate resistance layer requiring the EDS1-PAD4-SAG101 signaling complex, which is known to function in basal and resistance (R) gene–triggered immunity. Concurrent impairment of pre- and postinvasion resistance renders Arabidopsis a host for both nonadapted fungi.

Host cell entry represents a critical step during pathogenesis of invasive animal and plant parasites (1, 2). This is usually not a barrier to adapted phytopathogenic fungi that are able to infect a plant species. However, in plant species beyond the host range of a pathogen, called nonhost plants, parasitic fungi typically fail to enter attacked plant cells (3, 4). In the nonhost interaction between Arabidopsis and the grass powdery mildew fungus, Blumeria graminis f. sp. hordei (Bgh), three Arabidopsis pen (penetration) mutant loci were recovered that permit, at high frequency, entry of the nonadapted parasite (5), thereby providing initial evidence for the existence of a plant-controlled process terminating fungal ingress at the cell periphery. AtPEN1 encodes a soluble NSF (N-ethylmaleimide–sensitive factor) attachment protein receptor (SNARE)–domain-containing and plasma-membrane resident syntaxin, which becomes recruited into plasma membrane microdomains beneath incipient fungal entry sites (6, 7). Because SNARE proteins play a key role in vesicle trafficking in eukaryotic cells (8), these findings have been interpreted as evidence for the existence of a vesicle-associated resistance mechanism preventing powdery mildew ingress.

Although each of the isolated pen mutants (pen1, pen2, and pen3) permits efficient Bgh entry, initiation of postinvasive fungal growth invariably ceases, and this coincides with a cell death response of epidermal cells with haustorial complexes (5). We performed a time-course experiment and compared Bgh entry rates in wild type, pen1-1, pen2-1, and pen1 pen2 double null mutants (Fig. 1). Entry rates were seven- and fivefold higher than wild type in pen1 and pen2 mutants, respectively. An 11-fold increase over wild type was seen in the pen1 pen2 genotype, suggesting PEN1 and PEN2 act in separate defense pathways (Fig. 1). Elevated fungal entry rates were associated with an increased incidence of invasion-associated cell death (Fig. 1). To assess the importance of PEN1 and PEN2 functions in a further powdery mildew nonhost interaction, we examined the mutant Arabidopsis genotypes with Erysiphe pisi, which colonizes dicotyledonous pea plants in nature. Like Bgh, E. pisi fails to reproduce on Arabidopsis but is phylogenetically more closely related to the Arabidopsis-infecting powdery mildew E. cruciferarum than to Bgh (9). Absence of PEN1 or PEN2 greatly enhanced the entry rates of E. pisi. However, E. pisi successfully invaded wild-type Arabidopsis more often than Bgh [∼25% versus ∼5% at 24 hours post-inoculation (Fig. 1)], and lack of PEN2 alone permitted an E. pisi entry frequency similar to those of the pen1 pen2 double mutants (Fig. 1). Invasiveness of Bgh on pen1 pen2 double mutants and of E. pisi on pen2 or pen1 pen2 plants is essentially indistinguishable from that of the adapted Golovinomyces orontii powdery mildew on wild-type Arabidopsis (70 to 80%) (Fig. 1). Invasiveness of G. orontii does not increase on either pen mutant genotype, indicating that PEN1 and PEN2 exert entry-limiting functions only in nonhost powdery mildew interactions.

Fig. 1.

Frequency of invasive growth and epidermal single-cell death at B. g. hordei, E. pisi, and G. orontii interaction sites on Arabidopsis wild type (WT) and pen1-1, pen2-1 and pen1-1 pen2-1 mutants. hpi, hours post–conidiospore inoculation. Error bars indicate standard deviations for triplicate measurements.

A difference between pen1 and pen2 plants was revealed after inoculation with the nonadapted hemibiotrophic oomycete Phytophthora infestans. Only pen2 mutants allowed frequent initiation of invasive growth of this potato pathogen (fig. S1A). Similar to invasion-associated cell death in interactions with the inappropriate powdery mildew species (Fig. 1 and fig. S1B), invasive growth of the oomycete in pen2 plants is linked to a localized plant cell death response (fig. S1C). Likewise, inoculation with the broad host range ascomycete Plectosphaerella cucumerina resulted in enhanced disease susceptibility only in pen2 mutants (fig. S1D). Thus, PEN2 limits growth of a wide spectrum of pathogens, whereas PEN1 function is limited to nonadapted powdery mildew species.

PEN2 was isolated by map-based cloning (fig. S2A) and encodes 1 out of 48 predicted Arabidopsis family 1 glycosyl hydrolases (F1GHs) (fig. S2B). Both chemically induced pen2 mutant alleles, pen2-1 and pen2-3, are characterized by point mutations, each generating stop codons that lead to truncated peptides (fig. S2, B and C). We also isolated a homozygous transfer DNA (T-DNA) insertion line resulting in loss of detectable PEN2 transcript and protein (designated pen2-2; fig. S2, A to C). The infection phenotype of pen2-2 plants is indistinguishable from the two chemically induced mutants. F1GHs are present in all eukaryotic and prokaryotic organisms and hydrolyze O or S glycosidic bonds between two or more carbohydrates or between a carbohydrate and a noncarbohydrate (i.e., aglycone). In plants, F1GHs have been implicated in a wide range of processes, including development, cell wall modification, and chemical defense against pathogens (10).

Cyanogenic β-glucosidase CBG1 from Trifolium repens shares 48% sequence identity (65% similarity) with PEN2 (fig. S2B). This enabled us to use the known three-dimensional structure of CBG1 as template for homology modeling of a PEN2 structure (Fig. 2A and Materials and Methods). The topview ribbon diagram of PEN2 shows the typical F1GH barrel fold structure with a central cleft gate to the active site, which harbors the catalytic machinery consisting of two conserved acid/base and nucleophilic glutamates (Glu183 and Glu398, denoted by E183 and E398, respectively), located at the bottom of the active site pocket (Fig. 2A) (11). To test whether PEN2 catalytic activity is required for pathogen entry control, we generated transgenic lines in a pen2-1 mutant background, expressing a Glu183→Asp183 (E183D) substitution variant of PEN2 (PEN2E183D) under the control of native PEN2 5′ regulatory sequences (fig. S2D). Unlike control transgenic lines expressing wild-type PEN2, none of the tested PEN2E183D lines complemented the pen2 mutant phenotype, although a PEN2-specific antiserum detected comparable PEN2E183D protein amounts in leaf extracts (Fig. 2B). This suggests that PEN2 catalytic activity is required to restrict pathogen entry.

Fig. 2.

Analysis of PEN2 function and subcellular localization. (A) Topview ribbon diagram of a PEN2 structure model. The catalytic machinery consists of two conserved acid/base and nucleophilic glutamates (E183 and E398). The unique C-terminal extension (90° rotated view shown separately on the right) includes a low complexity region (white ball and stick presentation) followed by a predicted helical region of 28 amino acids. (B) Immunodetection of PEN2 in crude leaf protein extracts using a PEN2 antibody. PEN2 accumulation in independent transgenic lines (labeled 1 and 2) expressing wild-type PEN2 or catalytically inactive (pen2E183D) or C-terminally truncated (pen2Δ28) versions driven by native 5′ regulatory sequences (PPEN2). Complementation of the pen2 phenotype after Bgh inoculation experiments is denoted by + or – below each lane. Ponceau S detection of RuBisCo was used as protein loading control. (C) Confocal laser scanning microscopy of leaf epidermal cells in transgenic pen2-1 mutants expressing a functional PEN2-GFP fusion reveals association with vesicle-like bodies. Scale bar, 10 μm. (D) Peroxisomes containing PEN2-GFP accumulate at Bgh entry sites (marked by red arrowhead). Fungal structures are stained in red color by propidium iodide. Scale bar, 10 μm. ap, appressorium.

An unusual feature of the PEN2 F1GH is the presence of a C-terminal extension containing a low-complexity region of 15 amino acids followed by a predicted helical region of 28 amino acids (Fig. 2A and fig. S2B). To test a potential functional contribution of the C-terminal extension, we generated transgenic lines in the pen2-1 background expressing a truncated PEN2 protein that lacked the C-terminal 28 amino acids (PEN2Δ28; fig. S2, B and D). Although the truncation did not affect protein stability (Fig. 2B), PEN2Δ28 failed to complement the pen2 phenotype upon Bgh or E. pisi challenge, indicating that the C-terminal extension may regulate enzyme activity and/or serve as a determinant for subcellular localization.

To examine PEN2 localization in living cells, we generated transgenic lines expressing various PEN2–green fluorescent protein (GFP) fusion constructs driven by native 5′ regulatory PEN2 sequences in the pen2-1 null mutant background. Because both N- and C-terminal fusions of PEN2 to GFP were found to produce nonfunctional proteins, we searched for candidate GFP insertion sites in PEN2 on the basis of the deduced structural model (Fig. 2A). Indeed, insertion of GFP between the predicted globular PEN2 enzyme and the C-terminal extension resulted in a fluorescent fusion protein that complemented the pen2 phenotype. Confocal imaging revealed a distinctive subcellular GFP fluorescence pattern, indicating association of PEN2 to mobile vesicle-like bodies (Fig. 2C). Coexpression of fluorophore-tagged marker proteins for the Golgi, mitochondria, or peroxisomes identified these as peroxisomes in time-lapse imaging experiments (fig. S3A), which is in agreement with the PEN2 subcellular localization predicted by PSORT (12). The two-color imaging experiments showed the expected matrix localization for red fluorescent protein (RFP) containing a type 1 peroxisomal targeting sequence, whereas PEN2-GFP appeared to be confined to the periphery of peroxisomes (fig. S3A). A striking focal accumulation of PEN2-GFP-tagged peroxisomes was seen upon inoculation with Bgh conidiospores at incipient entry sites (14 to 16 hours postinoculation) (Fig. 2D). This pathogen-inducible PEN2 accumulation at fungal entry sites coincides with its predicted biological function at the cell periphery and is distinct from the recruitment of PEN1 syntaxin in a plasma membrane microdomain beneath fungal appressoria (6, 7). Independent biochemical fractionation of crude leaf protein extracts in soluble and microbody membrane fractions corroborated an association of wild-type PEN2 with microbody membrane fractions and revealed only trace amounts in enriched plasma membrane vesicles (fig. S3B). A substantial PEN2 portion was also found in corresponding soluble fractions, possibly due to a partial dissociation from peroxisomal membranes during the fractionation procedure.

Although the identified pen single mutants (pen1, pen2, and pen3) or the double mutant pen1 pen2 compromise nonhost resistance at the cell periphery, each mutant line retains the ability to mount effective postinvasion immunity, which is linked to a cell death response (Fig. 1 and fig. S1B). Sustained post-entry growth of powdery mildews requires a haustorial complex in living epidermal cells to supply nutrients for hyphal growth on the leaf surface (epiphytic hyphae). The lipase-like EDS1 protein and its sequence-related interaction partners PAD4 and SAG101 are known to play a critical role in a subset of R protein–triggered and basal immune responses to invasive biotrophic pathogens (13). Bgh entry rates on single eds1, pad4, and sag101 mutants were not significantly different from those found in wild type. Although interaction sites containing haustorial complexes showed an increased incidence of epiphytic fungal growth on each of these mutants relative to wild type, colonization attempts usually ceased after the formation of two to three elongating hyphae. However, we noted that about 2% of such sites on pad4 and eds1 plants permitted sustained epiphytic hyphal growth (i.e., microcolony formation) (Fig. 3A). Although the incidence of epiphytic fungal growth was further enhanced upon introgression of the pen2 null mutation in the single mutants (forming pen2 sag101, pen2 pad4, and pen2 eds1 double mutants), the frequency of microcolony formation remained low (Fig. 3A). Because the function of PAD4 and SAG101 in basal and R protein–triggered immunity were recently found to be partially redundant (13), we also challenged pad4 sag101 double mutants with Bgh. Although pathogen entry rates on pad4 sag101 lines were indistinguishable from that of wild type, microcolony formation was enhanced relative to those of the single mutants, suggesting similarly redundant activities of sequence-related PAD4 and SAG101 proteins in nonhost resistance (Fig. 3A). The incidence of microcolony formation was further increased in pen2 pad4 sag101 triple mutants, revealing synergistic interactions between PEN2 and the redundant functions of PAD4 and SAG101 (Fig. 3A). We occasionally found conidiophores containing mature conidiospores on microcolonies of pen2 pad4 sag101 leaves, indicating breakdown of nonhost resistance to the grass powdery mildew fungus (fig. S3D).

Fig. 3.

Separable functions for PEN2 in preinvasion and the EDS1-PAD4-SAG101 signaling complex in postinvasion nonhost immunity. (A) Frequency of Bgh microcolonies on leaves of wild-type and mutant genotypes 7 days post–conidiospore inoculation (determined by calculating the incidence of microcolonies found per interaction site supporting epiphytic growth). The two light microscopic inserts show examples of a small (top) and a large (bottom) microcolony. The latter were only detectable on pad4-1 sag101-2 and pen2-1 pad4-1 sag101-2 lines. Error bars indicate standard deviations for triplicate measurements. (B) Macroscopic infection phenotypes of wild type and the indicated mutant genotypes 7 days post–E. pisi conidiospore inoculation. Red arrowheads denote leaves of pen2-1 pad4-1 sag101-2 plants that are densely covered by sporulating E. pisi fungal mycelium.

In interactions with E. pisi, entry rates of single eds1, pad4, and sag101 mutants were not altered relative to those of wild-type plants, whereas epiphytic hyphal growth increased substantially on the former two mutant lines (fig. S3C). A synergistic effect was seen after introgression of pen2 in eds1 or pad4 backgrounds (pen2 eds1 and pen2 pad4 plants), indicating separable functions for and synergistic activities of the respective wild-type genes in pre- and postinvasion resistance. Although sporulation was not detectable on the single mutants, both pen2 eds1 and pen2 pad4 plants supported E. pisi conidiospore formation (fig. S3C inserts). Similar to interactions with Bgh, SAG101 and PAD4 exerted redundant functions against E. pisi colonization, and impairment of these two postinvasion resistance components alone was sufficient to allow occasional E. pisi sporulation (fig. S3C). Absence of PEN2 in a pad4 sag101 background (pen2 pad4 sag101 triple mutants) greatly enhanced disease susceptibility such that the timing and extent of E. pisi colonization became macroscopically indistinguishable from wild-type Arabidopsis interactions with the virulent G. orontii species (Fig. 3B and fig. S3D). Collectively, these findings strongly suggest separate functions for PEN2 in preinvasion and the EDS1-PAD4-SAG101 signaling complex in postinvasion nonhost resistance to both nonadapted powdery mildews.

The pathogen-triggered accumulation of PEN2-containing peroxisomes reported in this study (fig. S3A) as well as the recruitment of PEN1 syntaxin in a plasma membrane microdomain at fungal entry sites (6, 7) are inconsistent with a preformed and passive barrier against fungal ingress, implying instead the existence of an inducible resistance mechanism at the cell periphery. We have shown here conserved PEN1 and PEN2 functions in preinvasion resistance to both tested nonadapted powdery mildew species that, together, explain almost all failed entry attempts (pen1 pen2 double mutants) (Fig. 1). A corollary of this finding is that host powdery mildews such as G. orontii must have evolved means to overcome or bypass PEN activities. Two lines of evidence suggest that PEN1 and PEN2 represent components of two distinct entry resistance mechanisms: the broader spectrum of biological activity of PEN2 and the accumulative effect seen in pen1 pen2 double null mutants in interactions with Bgh. Thus, we consider a simple model in which cargo of PEN2-containing peroxisomes is secreted via a putative PEN1 ternary SNARE complex unlikely. Failure of the PEN2E183D variant to complement the pen2 mutant phenotype indicates that catalytic activity is required for PEN2 function. This makes it unlikely that PEN2 acts indirectly through interactions with other proteins. However, the peroxisomal localization and focal accumulation of PEN2 at fungal entry sites suggests that subcellular localization is an additional determinant for PEN2 function. The congregation of PEN2-containing peroxisomes at fungal entry sites might provide a mechanism for the activation and release of a small molecule at a high concentration. Because PEN2 inhibits in planta the pathogenesis of diverse fungal parasites, its catalytic product might directly or indirectly exert broad-spectrum toxic activity. Irrespective of this, our findings reveal a link between immune responses and a genetically defined peroxisomal component. Conceptually, the proposed pathogen-triggered focal secretory process, including PEN1 syntaxin and concentration of PEN2 F1GH at fungal entry sites, is reminiscent of the polar secretion machinery in cytotoxic T cells that is induced upon T cell receptor recognition, leading to targeted release of lytic proteins and killing of target cells (14).

Although previous reports demonstrated the involvement of lipase-like EDS1 in nonhost resistance to the grass powdery mildew fungus (15, 16), we have shown here that the combined PAD4 and SAG101 contributions to postinvasion nonhost resistance greatly exceed those of the single components EDS1, PAD4, or SAG101. Our findings of redundant PAD4 and SAG101 activities in nonhost resistance are reminiscent of recently reported redundant PAD4 and SAG101 signaling functions in basal and R protein–triggered immunity (13), thereby pointing to the involvement of plant immune receptors in triggering nonhost resistance. One possible receptor class that could operate in nonhost resistance and act via PAD4 and SAG101 signaling would be the PAMP receptors recognizing conserved pathogen-associated molecular patterns (17). Consistent with this, flagellin perception by the FLS2 receptor-like kinase leads to rapid transcriptional activation of EDS1, PAD4, and SAG101 as well as PEN1, PEN2, and PEN3 (18). Thus, it is possible that pre- and postinvasion nonhost resistance components represent two branch pathways activated by the same immune receptors or are components of two pathways that are sequentially activated by different receptors. One potential advantage of the inferred two-layer concept for nonhost resistance to powdery mildews is robustness through functional redundancy, because pre- and postinvasive immune responses are each sufficient to terminate fungal pathogenesis.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article