This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 18/12/3721    most recent tpc.106.044180v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via ISI Web of Science (10) Citing Articles via Google Scholar Google Scholar Articles by Qutob, D. Articles by Nürnberger, T. Search for Related Content PubMed PubMed Citation Articles by Qutob, D. Articles by Nürnberger, T. Agricola Articles by Qutob, D. Articles by Nürnberger, T.

Phytotoxicity and Innate Immune Responses Induced by Nep1-Like Proteins^[W]

^a Agriculture and Agri-Food Canada, London ON N5V 4T3, Canada
^b Center for Plant Molecular Biology–Plant Biochemistry, University of Tübingen, D-72076 Tübingen, Germany
^c PLANTA Angewandte Pflanzengenetik und Biotechnologie, D-37555 Einbeck, Germany
^d Department of Genetics, Technical University Munich, D-85350 Freising, Germany
^e Department of Plant Biotechnology, University of Freiburg, D-79104 Freiburg, Germany
^f Unité Mixte de Recherches Biochimie et Physiologie Moleculaire des Plantes, AgroM, Centre National de la Recherche Scientifique, Institut National de la Recherche Agronomique, Université Montpellier II, F-34060 Montpellier, France
^g Biotechnology Center for Agriculture and the Environment, Rutgers, The State University of New Jersey, New Brunswick, New Jersey 08901-8520
^h Department of Stress and Developmental Biology, Leibniz-Institute of Plant Biochemistry, D-06120 Halle/Saale, Germany
ⁱ Department of Cellular Biochemistry, Interfaculty Institute of Biochemistry, University of Tübingen, D-72076 Tübingen, Germany
^j Department of Biology, University of North Carolina, Chapel Hill, North Carolina 27599

² To whom correspondence should be addressed. E-mail nuernberger{at}uni-tuebingen.de; fax 49-7071-295226.

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
We show that oomycete-derived Nep1 (for necrosis and ethylene-inducing peptide1)–like proteins (NLPs) trigger a comprehensive immune response in Arabidopsis thaliana, comprising posttranslational activation of mitogen-activated protein kinase activity, deposition of callose, production of nitric oxide, reactive oxygen intermediates, ethylene, and the phytoalexin camalexin, as well as cell death. Transcript profiling experiments revealed that NLPs trigger extensive reprogramming of the Arabidopsis transcriptome closely resembling that evoked by bacteria-derived flagellin. NLP-induced cell death is an active, light-dependent process requiring HSP90 but not caspase activity, salicylic acid, jasmonic acid, ethylene, or functional SGT1a/SGT1b. Studies on animal, yeast, moss, and plant cells revealed that sensitivity to NLPs is not a general characteristic of phospholipid bilayer systems but appears to be restricted to dicot plants. NLP-induced cell death does not require an intact plant cell wall, and ectopic expression of NLP in dicot plants resulted in cell death only when the protein was delivered to the apoplast. Our findings strongly suggest that NLP-induced necrosis requires interaction with a target site that is unique to the extracytoplasmic side of dicot plant plasma membranes. We propose that NLPs play dual roles in plant pathogen interactions as toxin-like virulence factors and as triggers of plant innate immune responses.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Both plants and animals possess innate defense mechanisms to resist microbial infection (Akira et al., 2006; Chisholm et al., 2006). Although innate immune systems from both lineages share conceptual and mechanistic features, they are likely the result of convergent evolution (Ausubel, 2005). Efficient plant disease resistance is based on two evolutionarily linked forms of innate immunity. The primary plant immune response is referred to as PAMP-triggered immunity (PTI) and has evolved to recognize invariant structures of microbial surfaces, termed pathogen- or microbe-associated molecular patterns (PAMPs/MAMPs) (Nürnberger et al., 2004; Ausubel, 2005; Zipfel and Felix, 2005; Chisholm et al., 2006). Subversion of PTI by microbial effectors is believed to be one of the key strategies of successful pathogens to grow and multiply on host plants (Alfano and Collmer, 2004). In the coevolution of host–microbe interactions, individual plant cultivars have acquired resistance (R) proteins that guard microbial effector-mediated perturbations of host cell functions and thereby trigger plant immune responses. This type of plant defense is referred to as effector-triggered immunity (ETI) and is synonymous to pathogen race/host plant cultivar-specific plant disease resistance (Ausubel, 2005; Chisholm et al., 2006).

Activation of either type of plant immunity requires sensitive host perception systems that recognize microbe-derived determinants of nonself (Chisholm et al., 2006). PTI is initiated upon recognition of conserved microbial structures (PAMPs) by plant surface receptors (Zipfel and Felix, 2005). Importantly, PAMP-induced immune responses have recently been demonstrated to contribute to basal resistance of host plants against virulent pathogens and have been shown to be crucial for the stability of nonhost resistance (Zipfel et al., 2004, 2006; Kim et al., 2005; He et al., 2006). Activation of plant cultivar-specific disease resistance is mediated by direct or indirect recognition of microbial effectors through R proteins. Microbial effectors that are supposed to serve as virulence factors in the absence of their cognate plant R protein are thus turned into avirulence (AVR) factors (Alfano and Collmer, 2004). This type of effector recognition has been genetically characterized as gene-for-gene resistance (Chisholm et al., 2006).

In addition to PAMP or AVR effector-mediated nonself recognition, breakdown products of the plant cell wall are known to serve as endogenous danger signals that monitor distress of host structures and elicit plant immune responses (Vorwerk et al., 2004). Such plant-derived elicitors that are probably released by glucohydrolytic activities from attacking microbes may conceptually be compared with animal stress proteins that are produced upon microbial infection and function as danger signals, alerting the immune system by induction of innate immune responses (Gallucci and Matzinger, 2001).

Microbial toxin-induced plant innate immunity constitutes a seemingly paradoxical phenomenon that is not well understood. Phytopathogenic microorganisms produce a wide range of cytolytic compounds that function as key virulence determinants (van't Slot and Knogge, 2002; Glazebrook, 2005). In particular, phytopathogenic necrotrophic fungi synthesize numerous host selective and host nonselective toxins that facilitate killing of host plant tissue (van't Slot and Knogge, 2002; Wolpert et al., 2002; Gijzen and Nürnberger, 2006). An intriguing characteristic of many of these toxins is that they trigger individual facets of the plant defensive arsenal. For example, certain Fusarium spp produce the sphinganine toxin fumonisin B1 (FB1) that elicits cytolysis of plant and animal cells most probably through competitive inhibition of ceramide synthase, a key enzyme in sphingolipid biosynthesis (Wang et al., 1996; Tolleson et al., 1999). In addition to cell death, FB1 triggers accumulation of reactive oxygen species (ROS), deposition of callose, defense-related gene expression, and production of the phytoalexin camalexin in Arabidopsis thaliana (Asai et al., 2000; Stone et al., 2000). Likewise, the cell death–inducing toxins fusicoccin from Fusicoccum amygdali or AAL toxin from Alternaria alternata trigger expression of pathogenesis-related (PR) genes in tomato (Solanum lycopersicum) or Arabidopsis, respectively (Schaller and Oecking, 1999; Gechev et al., 2004). Moreover, the host selective cell death–inducing toxin victorin from Cochliobolus victoriae was shown to elicit the production of avenanthramide phytoalexins in oat (Avena sativa) (Tada et al., 2005). In all cases, it remains unclear whether plant immune responses constitute an unavoidable consequence of toxin action or, alternatively, if activation of plant defense is essential for the virulence function of fungal toxins. In summary, cytolytic toxins appear to play dual roles in plant–pathogen interactions as virulence determinants and, like PAMPs or AVR effectors, act as nonself recognition determinants for the activation of plant innate immune responses (Gijzen and Nürnberger, 2006). Such an activity spectrum is not unique to plants as various bacteria-derived cytolytic toxins were shown to trigger both innate immune responses and cell death in mammalian cells (Huffman et al., 2004; Srivastava et al., 2005).

Programmed cell death (PCD) is a common consequence in both compatible and incompatible plant–pathogen interactions (Greenberg and Yao, 2004; Glazebrook, 2005). The hypersensitive response (HR) is a type of PCD that is frequently observed in ETI. PAMPs may also cause PCD by direct or indirect interactions with pattern recognition receptors. For example, an ethylene-inducing xylanase from Trichoderma viride causes PCD in tomato cells, apparently by binding to a cell surface receptor (Ron and Avni, 2004). Less is known about the host cell death that occurs in susceptible plants, but increasing evidence suggests that resistance and susceptibility-associated PCD share regulatory and mechanistic features (Greenberg and Yao, 2004). The timely induction of PCD may present a formidable barrier to pathogen establishment, especially to biotrophic organisms that rely on living host cells for nutrients. Defense strategies that culminate in PCD may nonetheless become a dangerous liability to the host when it is engaged with necrotrophic pathogens (Greenberg and Yao, 2004). Inappropriate PCD can accelerate disease and foster the growth of necrotrophic pathogens that live off of dead or dying cells (Gijzen and Nürnberger, 2006).

Fusarium oxysporum f. sp erythroxyli–derived Nep1 constitutes the founding member of a family of microbial proteins that are secreted by plant pathogenic oomycetes, fungi, and bacteria (Pemberton and Salmond, 2004; Gijzen and Nürnberger, 2006; Kamoun, 2006). Nep1-like proteins (NLPs) trigger plant defense responses and, subsequently, cell death. NLPs are relatively small proteins of 24 kD that exhibit a high degree of sequence conservation, including a pair of Cys residues that are predicted to form a disulfide bridge. Moreover, their necrosis and defense-inducing activity is heat-labile, suggesting that an intact three-dimensional structure and enzymatic activity are important for NLP activity. Among angiosperms, dicotyledonous plants are considered susceptible to the effects of NLPs, whereas monocots are insensitive (Bailey, 1995; Veit et al., 2001; Fellbrich et al., 2002; Keates et al., 2003; Mattinen et al., 2004; Pemberton et al., 2005). Studies in various dicot plants have shown that NLPs can activate defense-associated responses, such as the synthesis of phytoalexins and ethylene, the accumulation of defense-related transcripts, and cell death (Veit et al., 2001; Fellbrich et al., 2002; Keates et al., 2003; Mattinen et al., 2004; Pemberton et al., 2005; Bae et al., 2006). Despite the fact that NLPs rapidly activate plant defense responses, these proteins have been shown to contribute to the virulence of necrotrophic fungal and bacterial pathogens. Several arguments support the view that NLP action on plants may resemble that of host nonselective toxins. (1) NLPs exert cytolytic activity that causes cell maceration and death in dicotyledonous plants in a manner that is similar to disease symptom development during natural infections of host plants (Pemberton and Salmond, 2004; Gijzen and Nürnberger, 2006; Kamoun, 2006). (2) Loss or gain of NLP expression affects virulence and disease symptom development in dicotyledonous plants, suggesting that NLPs act as positive virulence factors during infection of plants (Amsellem et al., 2002; Mattinen et al., 2004; Pemberton et al., 2005). For example, inactivation of the NLP-encoding genes (NLP_Ec) in different Erwinia carotovora strains resulted in significantly reduced levels of soft rot disease on potato (Solanum tuberosum), indicating that NLP_Ec contributes to bacterial fitness and disease symptom development (Mattinen et al., 2004; Pemberton et al., 2005). Likewise, overexpression of Nep1 in the hypovirulent fungus Colletotrichum coccodes dramatically increased its aggressiveness toward the host plant Abutilon theophrasti and even enlarged the host range of this pathogen (Amsellem et al., 2002). (3) Finally, increased transcript accumulation of a Phytophthora sojae NLP (NLP_Ps) coincided closely with the transition from biotrophy to necrotrophy during infection of soybean (Glycine max) (Qutob et al., 2002), also suggesting that NLP_Ps may act as a virulence factor by facilitating host cell death.

Here, we show that NLPs exhibit a wide taxonomic distribution pattern that is unusual for microbial virulence factors. This, together with the wealth of NLP-sensitive plants, makes these proteins well suited to serve as nonself recognition determinants in plant–pathogen interactions. We present a comprehensive analysis of innate defense reactions that are mounted in intact Arabidopsis plants in response to various oomycete-derived NLPs. Our analyses suggest that NLPs trigger a spectrum of plant immune responses that largely resembles that of regular PAMPs, such as bacterial flagellin. In addition to alerting the plant immune system, NLPs function as toxins by causing host cell death in dicotyledonous plants. NLP-mediated necrosis is an active process with features that are both shared and distinct from PCD mediated by other known triggers of plant cell death. Moreover, NLP-induced cell death is light dependent and requires membrane side-specific interaction with a dicot plant–specific target site.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Taxonomic Diversity of the NLP Protein Family
The NPP1 domain has been recognized by the Conserved Domain Database in GenBank, by Pfam (PF05630), and by InterPro (IPR008701) as an identifiable protein motif. NPP1 refers to the original name of the Phytophthora parasitica–derived NLP_Pp (Fellbrich et al., 2002). Databases of known and predicted protein sequences in GenBank were searched using the Conserved Domain Database to find sequences containing an NPP1 domain or a fragment thereof. A total of 62 protein sequences encoding NLPs could be retrieved that upon correction for redundant sequences represent 44 different NLPs from 22 species (Figure 1 ). Thus, such proteins appear to be common molecular patterns that are associated with both prokaryotic and eukaryotic microorganisms but cannot be found in the genomes of any higher organisms, including plants (see Supplemental Figures 1 and 2 online). NLP sequences are present in gram-negative and gram-positive bacteria as well as among fungi and stramenopiles but are predominantly present in organisms that at least partially rely on heterotrophic (either hemibiotrophic, necrotrophic, or saprophytic) growth. Consequently, many plant pathogens that favor such an infection strategy were shown to harbor NLP sequences. In contrast with certain saprophytic and plant pathogenic bacterial species that possess a single NLP-encoding gene, phytopathogenic fungi or oomycetes (belonging to the eukaryotic stramenopile lineage) were shown to harbor NLP-encoding gene families (see Supplemental Figure 2 online). A total of 10 NLP-encoding sequences were reported from four plant pathogenic fungal species, while 16 sequence entries were found for four species of the oomycete genus Phytophthora, all of which are destructive plant pathogens. This apparent gene diversification suggests that NLPs are important to the hemibiotrophic or necrotrophic lifestyle of fungi and oomycetes in general and of Phytophthora species in particular.

View larger version (18K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 1. Phylogeny of NLPs. The Nep1 protein sequence and 43 related sequences are shown. The scale bar represents 20% weighted sequence divergence. GenBank identifier numbers for each protein sequence are shown along with the species of origin. Sequences with special relevance to this study are additionally labeled: Nep1, necrosis and ethylene inducing peptide 1; NLPPp, NLP from Phytophthora parasitica; NLPPs, NLP from Phytophthora sojae; NLPPya, NLP from Pythium aphanidermatum.

Posttranslational activation of mitogen-activated protein kinase (MAPK) activity is commonly associated with plant immunity (Pedley and Martin, 2005). Infiltration of recombinant NLP from P. parasitica (NLP_Pp) into Arabidopsis leaves and subsequent immunodetection of MAPK activity using an antibody specific for the enzymatically active form of MAPK was performed. As shown in Figure 2A , NLP_Pp treatment resulted in rapid but transient phosphorylation of two MAPK species of 44 and 46 kD, respectively. This pattern closely resembles that obtained upon stimulation of an Arabidopsis cell culture with the PAMP, flg22 (Nühse et al., 2000). Production of nitric oxide (NO) is another hallmark of immune responses in both animals and plants (Zeidler et al., 2004). As shown in Figure 2B, treatment with NLP_Pp of Arabidopsis resulted in a dosage-dependent increase in NO production within 30 min. Likewise, application of 1 µM flg22 triggered an NO burst similar to that produced by NLP_Pp (data not shown). While MAPK activation and NO production are likely implicated in signal transduction processes, production of the antimicrobial phytoalexin, camalexin, is part of the executing arsenal of the plant defense system. Both NLP_Ps and NLP from Pythium aphanidermatum (NLP_Pya) triggered the production of similar camalexin levels in Arabidopsis plants (Figure 2C). The maximum concentration produced was 93.9 µg/g dry mass in plants that were treated with NLP_Ps for 48 h. In addition, both NLP preparations initiated camalexin production with comparable kinetics. The earliest time point when substantial amounts of the phytoalexin were found to accumulate was 8 h after infiltration.

View larger version (31K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 2. NLP-Induced Activation of Plant Immune Responses in Arabidopsis. (A) Five-week-old Arabidopsis plants were infiltrated with 2 µM recombinant NLPPp or glutathione S-transferase (GST) as control (Fellbrich et al., 2002) for the times indicated. Proteins were extracted and subjected to protein blot analysis using a 1:1000 dilution of phospho-p44/42 MAP kinase antibody as described in Methods. (B) Arabidopsis cell culture aliquots (2.5 x 104/50 µL) were treated with the indicated recombinant NLPPp concentrations or Escherichia coli protein extracts as control. NO production is given as relative fluorescence units (rfu). (C) Camalexin accumulation in 5-week-old Arabidopsis rosette leaves after infiltration with 2 µM recombinant NLPPs (black bars), 2 µM recombinant NLPPya (gray bars), or protein renaturation buffer as control (white bars). Camalexin was extracted at the time points indicated and determined as described in Methods. All experiments shown in (A) to (C) were performed at least three times with identical results. Data in (B) and (C) show average values + SD. (D) and (E) Transcriptome analysis in 5-week-old Arabidopsis plants treated with 1 µM recombinant NLPPp (GST as control) or 1 µM synthetic flg22 (water as control). All experiments were performed in triplicate, and expression levels for each probe set were analyzed as described in Materials. (D) Behavior of 12,557 genes significantly expressed at 1 or 4 h after treatments. Scatterplot analysis of fold induction of the probe sets for NLPPp trials (4 h) versus fold induction for flg22 trials (4 h). For each treatment versus control condition, genes that changed were assigned based on a one-way analysis of variance (ANOVA) test combined with a Benjamini and Hochberg false discovery rate algorithm. The numbers given in the inset refer to those genes of which expression was statistically significantly altered more than twofold at least at one of the two time points tested. The rectangular box comprises those genes that are unaltered in expression upon stimulation. Probe sets in areas enclosed by dotted lines are coordinately upregulated (red dots) or downregulated (green dots). (E) Venn diagram showing the total number of genes that are coordinately expressed by both stimuli (overlap) or of which expression is specifically upregulated by either NLPPp or flg22 treatment.

An overall analysis of NLP_Pp and flg22-specific transcriptome profiles revealed a high degree of coexpression, strongly suggesting that both stimuli have a comparable impact on plant gene expression (Figure 2D). The number of genes of which expression was found to be induced more than twofold upon either 1 or 4 h of NLP_Pp treatment was 681 (3% of all genes arrayed), while the corresponding number of flg22-induced genes was 733 (3.2%) (Figure 2E). Individual data sets of NLP_Pp or flg22-induced genes are provided in Supplemental Tables 1 to 5 online. Importantly, expression of 377 out of 681 genes (55.4%) induced by NLP_Pp treatment was similarly induced upon recognition of flg22. Again, this finding strongly supports the view that microbe-derived toxin-like molecules, such as NLP, and genuine PAMPs trigger similar alterations in the plant transcriptome. A classification of the encoded proteins according to their proposed molecular function showed that each elicitor not only activated the expression of largely overlapping gene sets but also of genes that fall into the same functional categories (data not shown). The latter finding is important because it is also based upon the analysis of the nonoverlapping gene set.

A detailed analysis of NLP_Pp-induced genes revealed two important findings. First, numerous genes encoding receptor-like protein kinases, disease resistance-like proteins, and pathogenesis/defense-related proteins were responsive to NLP_Pp treatment (Table 1 ). Second, expression of the vast majority of these genes was responsive to NLP_Pp but was also affected by flg22 treatment. For example, seven out of 16 genes encoding Leu-rich repeat receptor–like protein kinases (LRR-RLKs) were induced by either stimulus. Expression of the FLS2 gene was induced by flg22 but was not induced above threshold levels by NLP_Pp (see Supplemental Tables 3 and 4 online). LRR-RLKs build a large monophyletic gene family in Arabidopsis, comprising 235 members (Shiu et al., 2004), two of which have been implicated in noncultivar-specific plant immunity (Gomez-Gomez and Boller, 2000; Zipfel et al., 2006). Likewise, numerous LRR-containing disease resistance proteins are known to contribute to plant cultivar-specific immunity (Nürnberger et al., 2004; Zipfel and Felix, 2005). Although it is premature to assign gene functions merely on the basis of stimulus-induced alterations in the corresponding transcript profiles, it is assumed that many LRR-RLKs contribute to nonself recognition and microbial containment during attempted infection (Zipfel and Felix, 2005; Chisholm et al., 2006).

As shown in Figure 2C, infiltration of NLP_Ps or NLP_Pya induced camalexin production. Based on our microarray experiments, we conclude that camalexin biosynthesis is preceded by production of the corresponding biosynthetic enzymes. Transcripts encoding anthranilate synthase (ASA1), Trp synthase (TSA1), and cytochrome P450 enzymes (PAD3/CYP71B15 and CYP79B2) were found to accumulate in plants infiltrated with NLP_Pp at times before camalexin accumulation.

The plant hormones ethylene, jasmonic acid (JA), and salicylic acid (SA) have been implicated in various aspects of plant disease resistance signaling (Pieterse and Van Loon, 2004). To analyze a possible involvement of hormone signaling in NLP_Pp-induced plant responses, we investigated the expression of hormone biosynthesis enzyme-encoding genes. Genes encoding ethylene biosynthesis enzymes 1-aminocyclopropane-1-carboxylate (ACC) synthase (ACS) and ACC oxidase were strongly induced upon NLP_Pp treatment. These results are in good agreement with previous studies that showed NLP_Pp-induced ethylene production (Fellbrich et al., 2002). Likewise, transcript levels of some genes encoding ethylene response proteins or ethylene-responsive element binding proteins were altered in plants treated with NLP_Pp. Strikingly, none of the genes encoding various isoforms of JA biosynthetic enzymes were altered in expression, such as phospholipase A, lipoxygenase, allene oxide synthase, allene oxide cyclase, 12-oxophytodienoate reductase, or jasmonate-O-methyl transferase (see Supplemental Tables 1 and 2 online). These results indicate that NLP_Pp treatment does not rapidly activate de novo synthesis of JA biosynthetic enzymes in plants. By contrast, accumulation of transcripts encoding SA biosynthetic enzymes was observed as transcripts for isochorismate synthase 1 (ICS1/SID2) and Phe ammonia lyase accumulated rapidly in plants treated with NLP_Pp (Table 1).

Among the 304 genes that were induced at early time points by NLP_Pp, but not by flg22 (Figure 2D; see Supplemental Tables 1 and 2 online), there were 20 sequences encoding PR proteins and disease resistance-like proteins with LRR, LRR-RLK, or TIR-NBS signatures. Moreover, four cytochrome P450-encoding genes and five NLP_Pp-responsive genes that encode lipases, a lipid-modifying enzyme, or a phospholipid transport protein were found.

Reduction of plant growth in the presence of flg22 constitutes a surprising yet inexplicable phenomenon that facilitated the identification and isolation of the flagellin receptor FLS2 by forward genetic screening (Gomez-Gomez and Boller, 2000; Zipfel et al., 2006). To test whether growth and development of Arabidopsis is affected by the presence of NLP_Ps (NLP from Phytophthora sojae), surface-sterilized seeds were germinated on solid half-strength Murashige and Skoog (MS) medium supplemented with increasing concentrations of the purified recombinant protein. As shown in Figure 3 , seedling root growth and vigor were significantly reduced at NLP_Ps concentrations of 0.1 µg/mL (4 nM). Germination and growth were completely inhibited at concentrations of 100 µg/mL or greater (Figure 3A), and root necrosis and macroscopic cell death could be observed similar to that reported recently on Nep1-treated seedlings (Bae et al., 2006). By contrast, seedling germination and development on half-strenth MS supplemented with heat-denatured NLP_Ps was normal relative to that observed on half-strenth MS alone (Figure 3C). The previous description of a naturally occurring flg22-insensitive Arabidopsis ecotype (Gomez-Gomez and Boller, 2000) prompted us to compare the responses of 17 different Arabidopsis ecotypes when germinated and grown in the presence of 2.0 µg/mL NLP_Ps. Each of the 17 ecotypes exhibited reduced growth and vigor. None of the selected subspecies appeared to be more or less sensitive toward 2.0 µg/mL NLP_Ps on a comparative basis (Figure 3B). Altogether, NLP_Ps negatively affects Arabidopsis seedling growth and development similar to flg22, but no evidence for genetic variation in this response was found among the Arabidopsis ecotypes tested.

View larger version (61K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 3. Germination of Arabidopsis Seedlings in the Presence of NLPPs. (A) Seeds of Arabidopsis ecotype Col-0 were sown in sterile media containing a range of concentrations of NLPPs, and root lengths were measured at intervals as indicated. Shown are average values and SD from measurements of 10 to 20 plants (per treatment) from a representative experiment. The experiment was performed three times with similar results. (B) Seeds of 17 Arabidopsis ecotypes were sown in sterile media, with and without added NLPPs, and root lengths were measured after 8 d. Shown are average values and SD from measurements of 10 to 20 plants (per ecotype) from a representative experiment. The experiment was performed three times with similar results. (C) Seeds of Arabidopsis ecotype Col-0 were sown in sterile half-strength MS medium alone (top panel), on half-strength MS medium supplemented with 1.0 µg/mL NLPPs (middle panel), or on half-strength MS medium containing 1.0 µg/mL heat-denatured NLPPs (bottom panel). Photographs were taken 5 d after sowing.

View larger version (118K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 4. NLP-Induced Cell Death Requires Active Plant Metabolism. Four-week-old tobacco plants were infiltrated either with the calcium channel blocker LaCl3 (1 mM), with the DNA transcription inhibitor -amanitin (100 µM), or the protein translation inhibitor cycloheximide (CHX; 100 µM) alone, in combination with buffer as control, or 1 µM NLPPp or 50 nM ß-megaspermin (ß-MG), respectively. PCD symptoms shown here were obtained after 2 d.

View larger version (70K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 5. Light Dependence of NLPPp-Induced Cell Death. Five-week-old tobacco (top panel) or Arabidopsis plants (bottom panel) were treated with 1 µM NLPPp, 1 µM heat-denatured NLPPp, 50 nM ß-megaspermin (ß-MG), or 5 x 106 colony-forming units/mL P. syringae pv tomato strain DC3000/AvrRpm1 (PstAvrRpm1) under normal light conditions or 30 min upon transfer into the dark as indicated. PCD symptoms shown here were obtained after 2 d.

View larger version (65K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 6. NLP-Induced PCD Is Independent of Caspase and Bax Inhibitor Activity. (A) Tobacco leaves were infiltrated with 4 µM NLPPp or P. syringae pv phaseolicola strains NPS4000 (PCD-noninducing strain) or NSP3121 (PCD-inducing strain) in the presence or absence of 100 µM of the caspase inhibitors Ac-YVAD-CHO, Ac-DEVD-CHO, or zVAD-FMK. Plants were kept at room temperature under continuous illumination. (B) Five-week-old Arabidopsis Col-0, atbi1-1, or atbi1-2 mutant plants were treated with 4 µM NLPPp, 4 µM heat-denatured NLPPp, or buffer as control. Photographs in (A) and (B) were taken 24 h after infiltration.

HR PCD in plants infected with avirulent pathogens requires SA. Likewise, FB1-induced cell death in Arabidopsis was shown to be dependent on SA (Asai et al., 2000). NLP_Pp-induced lesions, however, still developed in SA-deficient nahG Arabidopsis plants (Table 2 ), suggesting no SA requirement for this response. This is surprising because NLP_Pp-mediated expression of the PR-1 gene was previously reported to be SA dependent (Fellbrich et al., 2002). Mutant plants impaired in NDR1 and PAD4 activity also developed wild-type-like lesions. Moreover, unlike FB1-induced cell death (Asai et al., 2000), NLP_Pp-triggered necrosis was observed on coi1 and ein2 genotypes, suggesting that neither JA nor ethylene contributes to this phenotype. Kinetics of symptom development were indistinguishable from those observed on wild-type plants (data not shown).

Two isoforms of Arabidopsis HSP90 (HSP90.1 and HSP90.2) are reported to compromise HR PCD in plant cultivar-specific immunity (Hubert et al., 2003; Takahashi et al., 2003). We tested two independent mutant alleles of each gene for impaired responsiveness to NLP_Pp. As shown in Table 2, we observed a partial reduction in NLP_Pp-mediated lesion formation on various HSP90 mutant alleles. Thus, HSP90 chaperone activity contributes to NLP_Pp-induced cell death. Plant cytosolic HSP90 also interacts with another chaperone-like protein, RAR1, which has been shown to play a critical role for the function of the Arabidopsis resistance proteins RPM1 and RPS2 (Hubert et al., 2003; Takahashi et al., 2003). However, rar1 mutants did not exhibit altered NLP_Pp sensitivity (Table 2).

Sensitivity to NLP Is Restricted to Dicotyledonous Plant Cells
The apparent universal sensitivity of dicot plants to NLPs contrasts with the apparent insensitivity of monocot plants (Pemberton and Salmond, 2004; Gijzen and Nürnberger, 2006). Such an activity spectrum is unprecedented among known elicitors, including PAMPs, but is similar to that reported for host nonselective toxins, such as fusicoccin or FB1. It is also known that an NLP gene from Vibrio pommerensis maps to a genomic region that is indispensable for bacterial virulence and hemolysis of animal erythrocytes, perhaps pointing to an even larger range of cell types that are susceptible to NLPs (Jores et al., 2003). In contrast with PAMPs that bind to plant plasma membrane protein receptors, toxins may interact with host membranes in different ways. For example, CryA-type bacterial toxins (Bacillus thuringiensis toxin) act as membrane-disrupting cytolysins on insect or nematode cells upon docking to specific glycolipid plasma membrane constituents (Griffitts et al., 2005). To explore the interaction of NLP with phospholipid bilayers in greater detail, we tested living and synthetic membrane systems for susceptibility to this protein. In the first set of experiments, we attempted to clarify if NLP sensitivity is indeed restricted to dicot plant cells or, alternatively, whether NLP treatment would destabilize phospholipid membrane systems in general but would leave monocot membranes intact due to some unknown counteractive measure. NLP concentrations similar to or higher than those reported to cause cell death in dicot cells were used. As shown in Table 3 , addition of NLP to monolayers of human fibroblasts (line GM5756) or African green monkey kidney-derived COS-7 cells did not significantly increase cell mortality. This effect could be observed independent of the culture media that were used to grow these cell layers. A possible cell type–specific NLP sensitivity of animal cells was tested by supplementing sheep erythrocytes for up to 24 h with 1 µM NLP_Pp. However, at no time point did hemolysis caused by NLP_Pp exceed that observed in control treatments (Table 3). Moreover, a possible Ca²⁺ requirement for NLP-induced erythrocyte death similar to that described for some bacterial cytolysins could not be demonstrated. Similarly, membranes of lower eukaryotes proved insensitive to NLP, as both Pichia pastoris cells and spheroplasts derived thereof survived treatment with this protein (Table 3). Likewise, the moss Physcomitrella patens was tested for NLP sensitivity. Moss cultures were grown either on solid medium or in liquid culture supplemented with 2 µM NLP_Pp or with heat-inactivated NLP_Pp as control. Under no circumstance was viability of the culture (Table 3) affected by NLP treatment nor was spore germination rate and differentiation affected (data not shown).

View larger version (82K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 7. NLP-Induced Cell Death in Dicot Plants Does Not Require Intact Plant Cell Walls. Parsley, Arabidopsis, or maize protoplasts (5 x 105/mL) prepared from cultured cells were treated with 1 µM NLPPp for 24 h, and viability staining was performed as described in Methods.

View larger version (31K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 8. NLPs Possess Binding Affinity to Phospholipid Bilayers but Do Not Show Pore-Forming Activity on Artificial and Biological Membranes. (A) TRANSIL beads coated with POPE/POPC (20:80), POPE/POPC (5:95), or POPC alone were incubated for 1 h with 1 µM NLPPp (total protein [T]). After separation of lipid-bound (B) from free unbound material (F), proteins were analyzed by SDS-PAGE and silver staining. (B) Na+ influx into Sodium Green–filled liposomes. Sodium-mediated fluorescence was measured without protein and in the presence of 1 µM NLPPp (open symbols) or 1 µM HrpZPsph (closed symbols). Fluorescence values obtained without protein were subtracted from values obtained in the presence of either protein. Values given represent the means ± SD from assays performed in triplicate. (C) Two-electrode voltage clamp measurements on X. laevis oocytes. The current/voltage plots obtained before and after the application of 1 µM NLPPya (open squares and closed circles, respectively) or 1 µM HrpZPsph (open circles) are shown. Steady state currents were measured following 4-s pulses. The results presented are representative of those obtained in three experiments ±SD.

NLPs are secretory proteins with a likely exposure to the apoplastic side of the plant plasma membrane during infection. Thus, interaction of NLPs at the extracytoplasmic side rather than the cytoplasmic side of the host plasma membrane can be predicted to occur. To test a possible side-specific activity of NLPs at the plant plasma membrane, three independent experimental systems were used to express the protein with (+SP) or without (–SP) a signal peptide. As shown in Figure 9 , transient biolistic cotransformation of Arabidopsis leaves with the NLP_Ps(–SP) gene and the reporter gene ß-glucuronidase (GUS) resulted in detectable GUS activity. By contrast, no GUS activity could be observed in experiments when NLP_Ps(+SP) was expressed. This finding indicates that NLP_Ps(+SP)-induced cell death occurred prior to GUS expression. In a similar experimental setup, cobombardment of soybean leaves with the same cassette or of sugar beet (Beta vulgaris) leaves with a NLP_Pp(–SP)/Renilla reniformis luciferase cassette yielded significant reporter enzyme activity, while luciferase activity remained at control levels upon expression of NLP_Pp(+SP) fused to a plant signal peptide (see Supplemental Figure 3 online). In summary, delivery to the apoplastic side of the plant cell surface is a requirement for NLP-induced cell death. Altogether, our findings suggest that NLP sensitivity is not a consequence of nonspecific membrane disruption but requires a specific target site that is unique to the extracytoplasmic side of dicot plant cell membranes.

View larger version (140K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 9. NLP-Induced PCD Requires Delivery to and Recognition at the Extracytoplasmic Side of Dicotyledonous Plant Cells. Activity of NLPPs in Arabidopsis leaves as determined by a cobombardment and transient expression assay. The photographs show Arabidopsis leaves after bombardment with tungsten beads and histochemical staining for GUS activity, performed as described in Methods. The tungsten beads were treated as follows: (A), coated with pFF19G containing a GUS expression cassette; (B), uncoated tungsten beads alone; (C), coated with pFF19G (GUS) plus a pFF19 construct encoding NLPPs lacking a signal peptide [NLPPs(-)SP]; (D), coated with pFF19G (GUS) plus a pFF19 construct encoding NLPPs, including the complete open reading frame with the native signal peptide [NLPPs(+)SP].

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

The taxonomic distribution pattern of NLP raises concerns regarding the physiological role of these proteins during infection. The cytolytic activities of NLPs, their contribution to fungal and bacterial virulence, and induced NLP_Ps transcript and protein accumulation during transition from biotrophic to necrotrophic growth seem to support a role as toxin-like effectors during plant infection (Pemberton and Salmond, 2004; Gijzen and Nürnberger, 2006). Moreover, conditional expression of the NLP_Ps gene in Arabidopsis resulted in rapid wilting and cell death (E. Huitema and S. Kamoun, unpublished data), and transformation of an avirulent E. carotovora strain that lacked NLP_Ec with NLP_Pp rescued virulence on potato tubers (M. Pirhonen, unpublished data), which indicates that the protein is toxic to plant cells. Our current finding that highly conserved NLP sequences occur predominantly in organisms that live at least partially heterotrophically also supports an important physiological role of these proteins. It is thus conceivable that NLPs may contribute to a heterotrophic lifestyle by either directly killing the host and/or by facilitating access to nutrient sources through breaking down biological membranes. Moreover, several studies, including our recent work, have now shown that NLP sensitivity is restricted to dicot plant cells in a non-species-specific manner. Such a relatively wide activity spectrum is another feature that is characteristic of toxins. Likewise, the ability to trigger defense responses in a large number of plant species also distinguishes NLPs from other stimuli, such as PAMPs, AVR effectors, or endogenous elicitors. In most cases, these signals exert plant immune-stimulating activities in only a limited number of plant species or plant cultivars.

However, the broad taxonomic distribution of NLPs, in particular their occurrence in both prokaryotic and eukaryotic species, is quite unusual for known microbial phytotoxins. The production of phytotoxins is often restricted to a narrow range of microbial species (van't Slot and Knogge, 2002; Wolpert et al., 2002); therefore, NLPs seem to represent an unprecedented case. Thus, alternative molecular functions of NLPs may be considered. Several publications reported that NLP activity is heat-labile and is not restricted to a certain domain of the protein, suggesting that NLPs may possess enzymatic activity (Veit et al., 2001; Fellbrich et al., 2002; Qutob et al., 2002). Database analyses with intact or partial NLP sequences, however, did not yield any meaningful alignment with enzyme-encoding nucleotide sequences (data not shown). Moreover, heat labile and tertiary structure-dependent activities are also characteristics of proteinaceous toxins that are known to possess cytolytic activities on animal cells (Parker and Feil, 2005). Taken together, we propose that NLPs are virulence-promoting microbial effectors that exhibit toxin-like characteristics, but we admit that a classification of NLPs as genuine toxins requires a detailed understanding of their molecular mode of action and the identification of host cell targets.

NLPs Trigger a Complex Immune Response in Arabidopsis
A broad taxonomic distribution and a wide variety of sensitive plant species are appropriate characteristics of nonself recognition determinants in plant–pathogen interactions. In this report, we have comprehensively characterized the immune response of one particular plant, Arabidopsis, to NLP_Pp. Results from this study and from previous work allow us to conclude that NLPs evoke a complex immune response. This response includes MAPK activation, production of NO, ethylene, camalexin, and callose, and extensive reprogramming of the transcriptome. Cell death and tissue necrosis terminates this massive defense response. The effects of NLPs resemble those triggered by the genuine PAMP flg22, with the exception that flg22 does not induce necrosis formation. In particular, very early responses, such as the production of NO and posttranslational activation of MAPK activity, are indistinguishable upon stimulation with NLPs or flg22. This is important because these responses occur at time points (within 30 min after stimulation) that clearly precede the onset of NLP-induced necrosis, suggesting that cell death may not be the cause for the induction of plant defense responses. The most compelling evidence that immediate and early plant responses to NLP and flg22 are comparable originates from whole-genome array-based transcriptome analyses. Large qualitative and quantitative overlaps were found in gene sets whose expression was altered upon application of either elicitor (Figures 2D and 2E). Gene sets whose expression is upregulated by either stimulus could be grouped into very similar functional categories. Intriguingly, genes implicated in pathogen recognition, such as receptor-like kinases, resistance signaling (disease resistance proteins, WRKY transcription factors, and hormone biosynthesis), and plant defense execution (PR proteins) were found to be coexpressed, suggesting that both signals are perceived as equivalent determinants of microbial nonself by the plant and similarly trigger activation of the plant surveillance system.

Overall, the microarray data indicate that NLPs and flagellin have a similar potential to trigger vital plant immune responses and may thus play similar roles in plant–microbe interactions, insofar as activation of plant defense is concerned. This view is further supported by experiments that showed that phytopathogenic bacteria-derived virulence factors suppress plant defense gene expression triggered by NLP_Pp or flg22, respectively, and thereby intercept with plant immunity (He et al., 2006).

Several recent studies have addressed the impact of pathogen-derived toxins on the plant transcriptome or on plant defense gene expression. For example, a comprehensive array experiment conducted on Arabidopsis plants treated with the cell death–inducing AAL toxin reported upregulation of oxidative stress and ethylene-responsive genes (Gechev et al., 2004), some of which were found to be upregulated in our experiment as well. Moreover, WRKY18 transcript accumulation in Arabidopsis in response to foliar application of Nep1 was reported (Keates et al., 2003). This finding was confirmed in our experiments, as was the accumulation of transcripts encoding ACS in Arabidopsis plants that were treated with Verticillium dahliae NLP_Vd (Wang et al., 2004).

A prime difference between flg22 and NLP is the ability of the latter to cause cell death. Thus, some of the genes that were exclusively found to be expressed in NLP-treated plants may found the basis for this particular plant response. We compared the transcriptome response of NLP_Pp with that caused by flg22 as well as with that caused by another cell death–inducing agent, FB1. Interestingly, among the 320 genes that were specifically induced by FB1, but not by flg22, we found no genes for which expression was also triggered by NLP_Pp. Thus, different cell death–inducing agents appear to have rather different effects on the Arabidopsis transcriptome. Another promising experimental approach lies in searching for NLP-insensitive Arabidopsis mutants. This strategy may take advantage of NLP-based inhibition of se