Flavohaemoglobin HmpX from Erwinia chrysanthemi confers nitrosative stress tolerance and affects the plant hypersensitive reaction by intercepting nitric oxide produced by the host

Authors


(fax +39 045 8027929; e-mail massimo.delledonne@univr.it).

Summary

Host cells respond to infection by generating nitric oxide (NO) as a cytotoxic weapon to facilitate killing of invading microbes. Bacterial flavohaemoglobins are well-known scavengers of NO and play a crucial role in protecting animal pathogens from nitrosative stress during infection. Erwinia chrysanthemi, which causes macerating diseases in a wide variety of plants, possesses a flavohaemoglobin (HmpX) whose function in plant pathogens has remained unclear. Here we show that HmpX consumes NO and prevents inhibition by NO of cell respiration, indicating a role in protection from nitrosative stress. Furthermore, infection of Saintpaulia ionantha plants with an HmpX-deficient mutant of E. chrysanthemi revealed that the lack of NO scavenging activity causes the accumulation of unusually high levels of NO in host tissue and triggers hypersensitive cell death. Introduction of the wild-type hmpX gene in an incompatible strain of Pseudomonas syringae had a dramatic effect on the hypersensitive cell death in soya bean cell suspensions, and markedly reduced the development of macroscopic symptoms in Arabidopsis thaliana plants. These observations indicate that HmpX not only protects against nitrosative stress but also attenuates host hypersensitive reaction during infection by intercepting NO produced by the plant for the execution of the hypersensitive cell death programme.

Introduction

Flavohaemoglobins have been isolated from phylogenetically distant micro-organisms and appear to be widely distributed among bacteria and fungi. The sequence of the hmp gene (Vasudevan et al., 1991), encoding the prototype of this class of proteins, Escherichia coli Hmp, has revealed two major domains. The N-terminal domain has homology to haemoglobins while the C-terminal domain shows homology with ferredoxin-NADP+ oxidoreductase (Wu and Poole, 2003). Although globins may be broadly defined as haemoproteins that reversibly bind oxygen, it is now clear that flavohaemoglobins are involved in responding to nitric oxide (NO) and nitrosative stresses (Poole and Hughes, 2000) catalysing the production of nitrate from NO and O2 in the so-called NO denitrosylase reaction (Gardner et al., 1998; Hausladen et al., 2001).

Macrophages generate NO as a cytotoxic weapon to help destroy tumour cells or invading microbes (Stevanin et al., 2002). Flavohaemoglobins, such as those of bacterial and fungal human pathogens like Salmonella enterica and Cryptococcus neoformans, play crucial roles in the protection of these micro-organisms during infection (de Jesus-Berrios et al., 2003). Reminiscent of the mammalian immune system in which reactive oxygen species (ROS) function together with NO in the macrophage killing of bacteria (Nathan, 1995), the plant hypersensitive disease resistance response (HR) is characterized by the rapid accumulation of ROS and NO through the activation of enzyme systems similar to the neutrophil NADPH oxidase (Keller et al., 1998) and NO synthase (NOS) (Guo et al., 2003). This signal system triggers hypersensitive cell death, induces sets of defence genes and mediates a network involved in the establishment of systemic acquired resistance (Delledonne et al., 1998; Durner et al., 1998). Flavohaemoglobins-like proteins have been identified in numerous plant pathogens (Table 1), but whether or not their putative enzymatic activities correspond to a physiologically relevant process remains to be established.

Table 1.  Occurrence of putative flavohaemoglobins in plant pathogens as determined by searching NCBI-annotated databases. Where possible, the accession number indicates the NCBI reference sequence (http://www.ncbi.nlm.nih.gov/RefSeq/)
Plant pathogenAccession
Botrytis cinereaCNS01CP4
Cladosporium fulvumBE187718
Erwinia carotovoraYP_051340
Erwinia chrysanthemiX75893
Fusarium oxysporumAB016807
Gibberella zeaeXP_384634
Pseudomonas aeruginosaF83311
Xylella fastidiosaAE003859

Erwinia chrysanthemi causes macerating diseases in a wide variety of plants and plant organs (Barras et al., 1994). However, Saintpaulia ionantha plants inoculated with E. chrysanthemi deficient in the flavohaemoglobin HmpX exhibit no symptoms or necrosis compared with maceration of the whole leaf and the spread of the disease on plants infected by the wild-type strain (Favey et al., 1995). As mutation did not affect growth under either aerobic or micro-aerobic conditions, it was speculated that the loss of virulence is due to the increased sensitivity to ROS produced by the host (Favey et al., 1995). In the present study, we demonstrate that HmpX exhibits a robust, oxygen-dependent NO consumption activity that provides a potential defence mechanism to E. chrysanthemi during infection and its proliferation within the host, when NO is generated for restricting survival of the pathogen. We also provide evidence indicating that the removal of NO by HmpX impairs the capacity of the plant to trigger a normal hypersensitive cell death of infected tissue. This latter function appears crucial for successful infection.

Results

E. chrysanthemi hmpX encodes a functional NO denitrosylase

The properties of E. chrysanthemi flavohaemoglobin HmpX were examined spectrophotometrically after heterologous expression in E. coli DH5α carrying the plasmid pRK415-hmpX (Figure 1). As expected, the CO-difference spectra revealed features similar to those of purified E. coli Hmp and of an Hmp-overexpressing strain (Ioannidis et al., 1992). The 421 nm peak is due to the CO adduct of ferrous HmpX, whereas the 437 nm trough arises from loss of the ferrous form on reaction with CO. Weaker α and β signals were evident at 530–580 nm. Cells expressing HmpX possessed a 10-fold higher concentration of flavohaemoglobins (0.05 nmol mg−1) compared with the control strain constitutively expressing the endogenous Hmp.

Figure 1.

Detection of HmpX by visible absorbance spectroscopy. CO difference spectra were carried out on undiluted high-speed cell-free extracts of Escherichia coli strains DH5α (DH5α; 5.9 mg protein ml−1) and DH5α-expressing HmpX (DH5α HmpX; 7.2 mg protein ml−1).

In E. coli and S. enterica, expression of Hmp protects against NO inhibition of cell respiration (Crawford and Goldberg, 1998; Stevanin et al., 2002). To test the ability of HmpX to consume NO and protect cell respiration, O2 uptake rates were measured polarographically. At both high (approximately 65 μm) and low (approximately 25 μm) O2 concentrations, the addition of NO (36 μm) to glucose-stimulated respiring E. coli cells resulted in a rapid and immediate uptake of oxygen, which was followed by a period of inhibition of renewed respiration (Figure 2a, trace 1). Respiration in the E. coli strain expressing HmpX was inhibited only transiently (Figure 2b, trace 1) compared with control strains, as shown in Table 2. The prolonged period of respiratory inhibition correlated with an increased half-life of the added NO, as shown in trace 2 (Figure 2a,b) and Table 2. A comparison of the respiratory responses to NO added either at 65 or 25 μm (Table 2) shows that the inhibition of respiration increased at lower oxygen tensions. This is consistent with previous results obtained for wild-type E. coli expressing its native Hmp and can be attributed to the competition by NO and oxygen for the active site(s) of the oxidases.

Figure 2.

Effects of NO on oxygen uptake in Escherichia coli-expressing HmpX and evidence for NO-oxygenase activity.
(a and b) Measurements of respiration and NO concentrations in suspensions of washed whole cells were made using a closed chamber equipped with oxygen and NO electrodes. The chamber contained 2 ml 50 mm NaCl buffer. Cells (C) were added at the indicated points. Glucose (G; 5 mm final) was added to stimulate respiration; then 36 μm anoxic NO solution (40 μl) was added at the time indicated by the arrows on the oxygen tracings (1), both at high and low O2 concentrations. In (a), 0.17 mg protein E. coli DH5α cells (DH5α) were used and in (b), 0.2 mg protein E. coli DH5α-expressing HmpX (DH5α HmpX) were used. Trace 1 shows O2 concentration in the chamber and trace 2 shows NO concentration.
(c and d) Measurements of oxygen uptake and NO concentration of high-speed cell-free extracts (CFE) were made using a closed chamber equipped with oxygen and NO electrodes. The chamber contained 2 ml 50 mm NaCl buffer. NADH (500 μm) was added to stimulate oxygen uptake at the time indicated. Anoxic NO solution (36 μm, 40 μl) was added at the time indicated by the arrows on the oxygen trace (1) at both high and low oxygen concentrations. In (c) 0.52 mg E. coli DH5α (DH5α) protein extract were used and in (d), 0.53 mg of E. coli DH5α-expressing HmpX (DH5α HmpX) were used. Trace 1 shows O2 concentration in the chamber and trace 2 shows NO concentration.

Table 2.  Effects of NO on cells and cell-free extracts. O2/NO ratios, periods of inhibition of O2 uptake and half-times of NO decay were calculated after the addition of 36 μm NO to respiring whole cells or cell-free extracts of Escherichia coli strains DH5α (DH5α) and DH5α-expressing HmpX (DH5α HmpX). O2 uptake was stimulated using glucose in intact cells or NADH in cell-free extracts
 High O2 conc.Low O2 conc.
O2/NO ratioPeriod of inhib. (min) of O2 uptakeHalf-time for NO decay (min)O2/NO ratioPeriod of inhib. (min) of O2 uptakeHalf-time for NO decay (min)
  1. Values are mean (number of assays, standard deviation in brackets); inhib., period of inhibition of O2 uptake; n/a, not appropriate.

  2. aInhibition of soluble oxidase activity.

E. coli whole cells
 DH5αn/a4.25 (2, 0.21)1.04 (2, 0.2)n/a8.3 (2, 0.42)1.59 (2, 0.17)
 DH5α HmpXn/a1.9 (2, 0.5)0.53 (2, 0.29)n/a4.4 (2, 1.4)0.21 (2, 0.08)
Cell-free extracts
 DH5α0.39 (2, 0.07)0.95a (2,0.5)0.85 (2, 0.07)0.36 (2, 0.12)4.4a (2, 0.28)1.5 (2, 0.42)
 DH5α HmpX0.83 (5, 0.05)0a (3, 0)0.3 (2, 0.14)0.92 (4, 0.07)0a (3, 0)0.6 (2, 0)

In intact bacteria, the major contributor to O2 uptake rates is respiration catalysed by membrane-bound oxidases. To study the consumption of O2 and NO by the flavohaemoglobin, membrane-free extracts were used. O2 uptake was stimulated by adding NADH and the effects of added NO on O2 uptake were monitored (Figure 2c,d). Upon addition of NO, an immediate uptake of O2 (trace 1) was accompanied by the decay of the NO signal (trace 2). The NO-stimulated O2 uptake was quantified and used to determine the O2/NO ratio. Escherichia coli DH5α-expressing HmpX exhibited greater ‘NO-oxygenase’ activity, as indicated by the higher O2/NO ratio and by the faster rate of NO decay compared with the control strain (Table 2).

In the absence of NO, the E. coli flavohaemoglobin Hmp generates Oinline image and H2O2 (Membrillo-Hernández et al., 1996). Quenching of the fluorescent substrate scopoletin (Figure 3a) revealed that the cell-free extract of E. coli DH5α-expressing HmpX generated H2O2 approximately five times faster than the control strain.

Figure 3.

Evidence for H2O2 generation and HmpX localization in Escherichia coli cells expressing HmpX.
(a) Assay of H2O2 generation by quenching of scopoletin fluorescence was measured in cell-free extracts of E. coli DH5α (DH5α) and DH5α-expressing HmpX (DH5α HmpX). Additions of cell extract, horseradish peroxidase and scopoletin were made before the addition of glucose (G) as shown. Each assay contained about 1.4 mg protein.
(b) Western blot analysis of cytoplasmic (odd numbered lanes) and periplasmic fractions (even numbered lanes) from E. coli RKP4600-expressing HmpX (lanes 1 and 2) and RKP4600 (lanes 3 and 4) probed with antibody to Hmp. Approximately 25 μg protein was used for each lane.

Although haemoglobins are soluble proteins with cytoplasmic functions, Vitreoscilla haemoglobin expressed in E. coli is localized, at least partly, in the periplasm (Khosla et al., 1990). Moreover, the flavohaemoglobin Hmp is distributed between the periplasm and the cytoplasm in E. coli (Vasudevan et al., 1995). Therefore, both the cytoplasmic and periplasmic fractions of E. coli-expressing HmpX were probed by immunoblotting. In order to remove interference by constitutively expressed Hmp, the plasmid pRK415-hmpX was used to transform the E. coli Hmp-deficient mutant strain RKP4600 (Mills et al., 2001). The cytoplasmic and periplasmic fractions of E. coli RKP4600-expressing HmpX showed a weak band of approximately 44 kDa in the cytoplasm, but a much stronger band in the periplasm (Figure 3b). No bands were seen in the E. coli RKP4600 in either the cytoplasmic or periplasmic fractions. Catalase assays on all subcellular fractions showed that the total amount of catalase activity in the periplasmic fraction was <10% of that in the whole cell, indicating minimal contamination of periplasmic fractions.

As the recombinant HmpX primarily accumulated in the periplasm, we examined the periplasmic HmpX for evidence of NO consumption in the dual-electrode apparatus. NADH was added to stimulate the uptake of O2 and a number of sequential additions of 36 μm NO were made (traces not shown). The resulting rapid and immediate uptake of O2 was measured to determine the O2/NO ratio. For E. coli DH5α, this ratio was 0.48 (SD 0.076, eight assays), whilst for E. coli DH5α expressing HmpX a ratio of 0.84 (SD 0.055, eight assays) was found. This near-stoichiometric consumption of O2 and NO is consistent with the HmpX-catalysed NO denitrosylase activity in periplasmic fractions of E. coli.

E. chrysanthemi HmpX-deficient mutant induces an HR-like response in S. ionantha

The HmpX-deficient mutant of E. chrysanthemi strain 3937 was not pathogenic and produced no signs of maceration in S. ionantha plants as already reported (Favey et al., 1995). Inoculation with 108 bacterial cells led to a 20-fold increase in colony-forming units 24 h after infiltrating the wild-type strain (2 × 109 ± 0.12 CFU) whereas only a twofold increase was observed after infiltrating the E. chrysanthemi HmpX-deficient mutant (2 × 108 ± 0.21 CFU). As HmpX possesses the characteristics of a NO denitrosylase, the sensitivity of the E. chrysanthemi HmpX-deficient mutant to nitrosative stress was tested by measuring bacterial growth in the presence of increasing concentrations of the nitrosating agent sodium nitroprusside (SNP) in the liquid medium. The HmpX-deficient mutant was more sensitive to SNP at any of the concentrations tested, growing between 50 and 75% to that of the wild-type strain (Figure 4).

Figure 4.

Bacterial growth of Erwinia chrysanthemi wild type (black bars) and E. chrysanthemi HmpX-deficient mutant (white bars) in liquid medium containing indicated amounts of the NO donor SNP.

As up to 50% of leaves inoculated with the HmpX-deficient mutant showed some necrotic spots 2–3 days after infection (Figure 5a), trypan blue was used to visualize dying cells during the early phase of infection. Despite the lack of macroscopic symptoms, leaves inoculated with the HmpX-deficient mutant presented numerous patches of blue cells as soon as 7 h after infiltration, which were not observed in leaves inoculated with the wild-type strain 3937 (data not shown). To test whether the HmpX-deficient mutant triggered cell death through activation of an HR-like response, ROS and NO production were examined. 3,3′-diaminobenzidine (DAB) polymerizes on contact with H2O2 in a peroxidase-dependent reaction, enabling the visualization of H2O2in situ (Thordal-Christensen et al., 1997). A strong, reddish-brown precipitate was observed in leaves infected with either the HmpX-deficient mutant or the wild-type strain (data not shown). NO accumulation was examined with the fluorophore 4,5-diaminofluorescein diacetate (DAF-2 DA), which is converted to the fluorescent triazole derivative DAF-2T in the presence of NO. By confocal laser scanning microscopy, green fluorescence was observed 1 h post-inoculation in S. ionantha leaves infected either with the wild-type strain or the HmpX-deficient mutant. DAF-2T fluorescence was maintained for several hours in leaves inoculated with the HmpX-deficient mutant, whereas in leaves inoculated with wild-type E. chrysanthemi fluorescence was no longer visible at 3 h after infiltration (Figure 5b). Quantification of DAF-2T fluorescence revealed that the accumulation of NO was significantly higher in leaves inoculated with the HmpX-deficient mutant compared with the wild-type strain already at 1 h (Figure 5c). Although this method does not allow a precise measurement of NO, the quantification of DAF-2T fluorescence emitted by leaves 1 h after infiltration with 2 mm SNP indicates that the amount of NO accumulated during infection with E. chrysanthemi unable to express HmpX is quite high (Figure 5c). The massive and sustained accumulation of NO together with the premature death of host cells observed during infection with the HmpX-deficient mutant suggests that HmpX scavenges NO, and that the detoxification of NO produced by the host is required for successful infection.

Figure 5.

Analysis of Erwinia chrysanthemi pathogenicity in Saintpaulia ionantha.
(a) Macroscopic symptoms 3 days after inoculation with 10 mm MgCl2 (control), 108 CFU ml−1E. chrysanthemi wild type (WT) or 108 CFU ml−1E. chrysanthemi HmpX-deficient mutant strain (Hmp-deficient).
(b) Confocal laser scanning microscopy of DAF-2 DA fluorescence 1 and 3 h after infection with 108 CFU ml−1E. chrysanthemi WT or 108 CFU ml−1E. chrysanthemi HmpX-deficient mutant strain (Hmp-deficient). Four independent samples were recorded for each condition, and representative leaf areas are shown.
(c) Quantification of DAF-2T fluorescence shown as mean green channel values of at least four leaves from different plants 1 and 3 h after infection with 108 CFU ml−1E. chrysanthemi WT or 108 CFU ml−1E. chrysanthemi HmpX-deficient mutant strain (Hmp-deficient) and 1 h after infiltration of 2 mm SNP.

Expression of HmpX in the plant pathogen Pseudomonas syringae

For gain-of-function experiments, the plasmid pRK415-hmpX was mobilized into P. syringae pv. glycinea carrying the avirulence gene avrA (Psg) (Keen and Buzzell, 1991). Cells expressing HmpX exhibited a slower growth both on plates and in liquid medium compared with control cells. When grown in rich liquid media Psg-expressing HmpX exhibited a doubling time of roughly 2.4 h versus 2 h for wild-type Psg. As reported for other micro-organisms (Buisson and Labbe-Bois, 1998; Membrillo-Hernández et al., 1996), overproducing flavohaemoglobin may be detrimental to cells, a finding consistent with plasmid loss experiments showing that in Psg the plasmid pRK415 is lost at very high frequency when it carries hmpX (data not shown). This loss is most likely a consequence of oxidative stress caused by generation of Oinline image in the absence of NO (Membrillo-Hernández et al., 1996).

Expression of HmpX in avirulent P. syringae blocks hypersensitive cell death in soya bean cells

Psg carries the avirulence gene avrA and is recognized by soya bean cultivar Williams 82 that possesses the corresponding Rpg2 resistance gene, leading to rapid induction of a hypersensitive response (Keen and Buzzell, 1991). Soya bean cell cultures were challenged with either Psg or the isogenic strain expressing HmpX. Whereas Psg induced a strong HR cell death at 24 h, Psg-expressing HmpX did not (Figure 6a). This lack of execution of the cell death programme was not determined by loss of pathogen recognition, as soya bean cells responded to challenge with Psg-expressing HmpX by activating a sustained oxidative burst, as demonstrated by a scopoletin assay of H2O2 accumulation (data not shown) and by massive accumulation of transcripts encoding cellular protectants such as glutathione S-transferase (GST; Figure 6b). However, careful analysis of H2O2 accumulation revealed that while Psg induced the typical avirulence gene-dependent oxidative burst characterized by a first transient burst of ROS accumulation followed by a second burst, sustained for 4–6 h, Psg-expressing HmpX induced a massive, single phase, oxidative burst that was maintained for at least 48 h (Figure 6c).

Figure 6.

Analysis of the hypersensitive response in soya bean cell suspensions.
(a) Cell death was measured 24 h after challenge with Pseudomonas syringae pv. glycinea carrying the avirulence gene avrA (Psg) or with the isogenic strain expressing HmpX (Psg HmpX).
(b) Accumulation of transcripts encoding GST as monitored by Northern analysis of total cellular RNA isolated at indicated times after treatment of soya bean cells with P. syringae pv. glycinea carrying the avirulence gene avrA (Psg) or with the isogenic strain expressing HmpX (Psg HmpX). RNA loading was checked by gel staining with ethidium bromide (data not shown).
(c) Kinetics of ROS accumulation in response to P. syringae pv. glycinea carrying the avirulence gene avrA (•) or to the isogenic strain expressing HmpX (○).
(d) ROS accumulation 4 h after challenge with P. syringae pv. glycinea carrying the avirulence gene avrA and expressing HmpX. The contribution of HmpX to H2O2 accumulation was determined by removing soya bean cells and measuring ROS accumulation in the soya bean medium (supernatant). The assay was validated by adding to the supernatant 50 U of SOD (+SOD), 0.5 mm xanthine/0.5 U xanthine oxidase (+X/XO) or 10 μm H2O2 (+H2O2; see text for clarity).
(e) NO accumulation in response to P. syringae pv. glycinea carrying the avirulence gene avrA (Psg) or to the isogenic strain expressing HmpX (Psg HmpX) measured with 3 μm (black bars) or 10 μm (white bars) oxyhaemoglobin 4 h after bacterial infection.

To determine whether the substantial accumulation of H2O2 in soya bean cell suspensions was caused by the direct enzymatic action of the flavohaemoglobin, we investigated the contribution of Psg-expressing HmpX to the oxidative burst. The removal of soya bean cells during oxidative burst, either by centrifugation for 1 min at 800 g or by filtration, abolished the peroxidase-mediated scopoletin fluorescence quenching, indicating that H2O2 cannot be detected in the absence of host cells (Figure 6d). It should be noted, however, that this assay relies on the H2O2-dependent degradation of scopoletin by extracellular peroxidases, which are continuously produced by soya bean cell suspensions. As HmpX generates Oinline image, the effect of exogenous superoxide dismutase (SOD) was tested. Whereas the addition of SOD to the filtered soya bean medium did not alter the levels of scopoletin fluorescence, the addition of H2O2 or xanthine/xanthine oxidase (generating Oinline image) to the medium caused the degradation of scopoletin (Figure 6d), indicating that soya bean-conditioned medium contains sufficient levels of both SOD and peroxidases for the detection of Oinline image potentially generated by HmpX. Therefore, the lack of ROS accumulation when soya bean cells were first removed from the liquid medium indicates that the massive oxidative burst detected was caused by the host cells and not by Psg-expressing HmpX. This data was further confirmed by examining the effect of Ca2+ blockers and protein kinase inhibitors on the oxidative burst. In soya bean cell suspensions challenged with Psg-expressing HmpX, either the Ca2+ channel blocker lanthanum (La3+) or the protein kinase inhibitor K252-a abolished H2O2 production (data not shown).

The oxyhaemoglobin assay of NO accumulation revealed that soya bean suspension cells accumulate lower amounts of NO when challenged with Psg-expressing HmpX (Figure 6e). However, the amount of NO measured in the presence of higher concentrations of oxyhaemoglobin (10 μm) was significantly higher (Figure 6e), suggesting that the amount of NO trapped is directly related to oxyhaemoglobin concentration. This phenomenon, markedly less pronounced when soya bean suspension cells were challenged with Psg, suggests that HmpX competes for NO with oxyhaemoglobin and therefore indicates that HmpX traps NO outside from host cells.

HmpX expression in avirulent P. syringae attenuates the HR in Arabidopsis thaliana

To test whether HmpX also affects hypersensitive cell death in planta, we examined the effects of its constitutive expression in a genetically well-defined host–pathogen interaction. Infiltration of leaves of A. thaliana ecotype Col-0, which contain the RPM1 resistance gene, with P. syringae pv. tomato carrying the avirulence gene avrB (Pst) results in colourless, dry HR lesions at the infection site. The plasmid-enabled expression of HmpX in Pst markedly delayed the development of macroscopic symptoms and promoted the spreading chlorosis observed in the compatible interaction with the isogenic virulent bacteria lacking avrB (Figure 7a). Trypan blue analysis of dead cells revealed that the number of HR patches decreased significantly when leaves were infected with bacteria expressing HmpX (Figure 7b,c). Moreover, as assessed by DAB staining, H2O2 accumulation during the oxidative burst was significantly lower when Col-0 plants were challenged with Pst (HmpX) compared with Pst (Figure 7d,e).

Figure 7.

Analysis of the hypersensitive response in Arabidopsis thaliana leaves.
(a) Macroscopic HR symptoms 3 days after challenge with 5 × 106 CFU ml−1Pseudomonas syringae pv. tomato carrying the avirulence gene avrB (Pst) or with the isogenic strain expressing HmpX (Pst HmpX).
(b) Trypan blue staining (100-fold magnification) of representative A. thaliana leaves 24 h after challenge with 5 × 106 CFU ml−1P. syringae pv. tomato carrying the avirulence gene avrB (Pst) or with the isogenic strain expressing HmpX (Pst HmpX).
(c) Quantification of dead cells expressed as percentage of stained pixels inside the infiltration zone of at least five leaves from different plants.
(d) DAB staining (100-fold magnification) of representative A. thaliana leaves 4 h after challenge with 5 × 106 CFU ml−1P. syringae pv. tomato carrying the avirulence gene avrB (Pst) or with the isogenic strain expressing HmpX (Pst HmpX).
(e) Quantification of H2O2 accumulation expressed as percentage of stained pixels inside the infiltration zone. Bars represent mean ± SD of six leaves from different plants.

Discussion

Micro-organisms that evade host microbicidal activity are a prerequisite for invasive disease (Fields et al., 1986). Inflammatory cells such as macrophages and neutrophils produce large amounts of ROS and NO (Beckman et al., 1990). Therefore, during the infection process pathogens exhibit an increase in the expression of antioxidant enzymes such as SOD and catalase (De Groote et al., 1997; Nunoshiba et al., 1993) and increase their defences against nitrosative stress (Pathania et al., 2002). The concerted mode of protection against reactive oxygen and nitrogen species produced by cells of the invaded host is also revealed in plant pathogens by the essential role of both SOD and HmpX in E.chrysanthemi pathogenicity (Favey et al., 1995; Santos et al., 2001). HmpX is the first flavohaemoglobin from a plant pathogen shown to be an NO denitrosylase. CO difference spectra revealed that recombinant HmpX could be detected spectroscopically in a manner similar to E. coli flavohaemoglobin Hmp (Ioannidis et al., 1992), and was detectable as a CO adduct closely resembling carbon monoxy-Hmp. HmpX was also found to consume NO and to prevent NO-mediated inhibition of cell respiration, thus demonstrating its function in protection from nitrosative stress. Although the possibility that HmpX is cytoplasmic unless over-expressed in a heterologous host cannot be ruled out, the fact that most HmpX was found in the periplasm suggests a role in the detoxification of NO accumulating extracellularly as part of the defence mechanism from the host plant. It is clear that NO per se (but not necessarily NO+ or NO) can penetrate the outer and cytoplasmic membrane, as NO added externally is consumed (Stevanin et al., 2002) and more tellingly, NO generated in the cytoplasm is readily detected in whole cell suspensions using an NO-selective electrode (Corker and Poole, 2003). It is therefore not surprising that loss of HmpX functionality causes loss of pathogenicity in E. chrysanthemi (Favey et al., 1995) as revealed by the inability of the HmpX-deficient mutant to macerate S. ionantha leaves and by its very limited growth in planta, an effect commonly observed during incompatible interactions.

Infection by necrotrophic pathogens has been extensively shown to be associated with oxidative stress in the host to promote infection (Govrin and Levine, 2000). Recently, Botrytis elliptica, a necrotrophic pathogen with a very narrow host range, has been shown to kill lily cells through stimulation of their production of both ROS and NO, but only during compatible interactions. A partially purified B. elliptica culture filtrate was able to trigger ROS and NO accumulation as well as cell death, and to confer pathogenicity on lily to incompatible Botrytis species (Van Baarlen et al., 2004). Similarly, E. chrysanthemi causes oxidative stress at an early stage of infection before the massive production of pectinolytic enzymes (Boccara et al., 1988; Santos et al., 2001). However, E. chrysanthemi causes soft-rotting on a wide range of host plants due to the action of endopectate lyases that rapidly macerate host tissues, and the typical HR that causes tissue necrosis, phytoalexin accumulation and loss of virulence in non-host plants (Pouteau et al., 1994; Toth et al., 2003). Therefore, whereas necrotrophs typically induce plant cell death to facilitate colonization of the host tissue (Govrin and Levine, 2000), for E. chrysanthemi the HR correlates with the inability to establish successful infection (Shevchik et al., 1998).

Saintpaulia ionantha leaves infected with E. chrysanthemi accumulate massive levels of H2O2 as detected by DAB staining. However, the same analysis carried out on plants infected with the HmpX-deficient mutant produced very similar results, indicating that S. ionantha plants trigger the oxidative burst also in the absence of leaf maceration. The use of the NO-specific fluorescent dye DAF-2DA revealed that both the wild type and the HmpX-deficient mutant of E. chrysanthemi caused an early accumulation of NO. However, the burst of NO was transient when leaves were inoculated with the wild-type strain whereas it was sustained and massive when S. ionantha leaves were infected with the HmpX-deficient mutant, indicating that the lack of NO denitrosylase activity causes the maintenance of an unusually strong burst of NO in host tissue. Furthermore, the HmpX-deficient mutant strain showed a reduction in growth in liquid medium under nitrosative stress compared with the wild type that, alone, does not appear sufficient to cause the complete loss of virulence and dramatic reduction in bacterial growth when the mutant strain is inoculated to S. ionantha plants. These data do not contradict the widely accepted model of action of enzymatic defences against oxidative and nitrosative stress mounted by the host; they suggest that the loss of virulence of HmpX-deficient E. chrysanthemi involves increased sensitivity to nitrosative stress as well as host activation of an HR-like response. This hypothesis is supported by the sustained burst of NO and by the dramatic increase in the number of dying cells as early as 7 h after infection that accompanies the infection of S. ionantha plants with the HmpX-deficient mutant strain. To assess this function, HmpX was ectopically expressed in genetically well-defined host–pathogen interactions. Its constitutive expression in incompatible P. syringae dramatically suppressed hypersensitive cell death in soya bean cell suspensions and markedly delayed the development of macroscopic symptoms and promoted the spreading chlorosis observed in the compatible interaction when infiltrated in leaves of A. thaliana.

A remarkably similar defence behaviour was observed in A. thaliana expressing Hmp from E. coli (Zeier et al., 2004). These transgenic plants had a diminished nitrate reductase-derived NO accumulation, and after challenge with avirulent P. syringae they showed an attenuated NO burst. Furthermore, the oxidative burst and induction of PAL transcripts were significantly attenuated, hypersensitive cell death was reduced, and the expression of PR-1 was delayed. Thus, flavohaemoglobin expression in transgenic plants had very similar effects on plant defences compared with its expression in the bacterial pathogen (Figure 7), and essentially confirmed previous pharmacological and genetic studies indicating a pivotal role of NO in the plant signalling network during pathogen infections (Romero-Puertas et al., 2004; Zeidler et al., 2004). However, whereas flavohaemoglobin expression in the host plant and P. syringae led to a decrease in H2O2 accumulation during the HR in the A. thalianaP. syringae system, the dramatic enhancement of the oxidative burst and accumulation of GST transcripts when soya bean cell suspensions were challenged with avirulent P. syringae pv. glycinea-expressing HmpX suggests that the molecular mechanisms underlying NO-mediated regulation of the oxidative burst might differ in the two systems.

NO signalling functions depend on many complex conditions affecting its reactivity, such as the rate of production and diffusion, with ROS being the key mediator in channelling NO into the cell death pathway (Murgia et al., 2004) through a still unknown mechanism (Delledonne et al., 2001). The evidence that flavohaemoglobin expression in transgenic plants has very similar effects on hypersensitive cell death compared with its expression in the bacterial pathogen suggests a crucial function of intracellular NO in the establishment of a proper defence reaction. This hypothesis is strengthened by our observation that, in soya bean suspension cells, Psg-expressing HmpX competes with exogenous oxyhaemoglobin for NO, indicating that HmpX intercepts NO in the extracellular space. Other recent reports support the view of NO as an intracellular signal that functions in cell-to-cell spread of the HR. In fact, A. thaliana plants challenged with avirulent P. syringae were found to initially accumulate NO exclusively in the extracellular space, after which NO was seen in the cytoplasm of nearby cells: many of these cells died soon after (Zhang et al., 2003). Moreover, pharmacological studies indicated that NO does not elicit hypersensitive cell death but is required for the onset of apoptotic cell death in the adjacent cells during the defence response of oat to avirulent Puccinia coronata (Tada et al., 2004).

In conclusion, NO and ROS play important antimicrobial roles in host defence, and pathogens counteract these molecules with specific detoxifying mechanisms. However, NO and ROS are also critical signalling molecules for the establishment of disease resistance mechanisms during HR, and elaborate enzymatic defences against oxidative and nitrosative stresses may scavenge the redox signals that host cells employ in disease resistance. It is therefore intriguing to hypothesize that the antioxidant apparatus of at least some plant pathogens may also function by suppressing the activation of defence reactions through interception of the crucial signalling molecules of the host.

Experimental procedures

Bacteria

Escherichia coli cultures were grown in Luria-Bertani medium (Miller, 1972) at 37°C unless specified. Pseudomonas syringae and E. chrysanthemi cultures were grown at 30°C in King's medium B (King et al., 1954) and M9 synthetic medium (Miller, 1972) supplemented with 0.5% (w/v) glycerol. The HmpX-deficient mutant of E. chrysanthemi strain 3937 was described previously (Favey et al., 1995). For plant infection, bacteria were pelleted, washed three times with 10 mm MgCl2, resuspended, and diluted in 10 mm MgCl2 to the indicated concentration.

Expression of HmpX in E. coli and P. syringae

The pRK415-hmpX vector plasmid contains a 1200-bp fragment that was amplified from the plasmid pB413 carrying the hmpX genomic region (Favey et al., 1995). The gene was amplified using the PCR Supermix High Fidelity (Invitrogen, Carlsbad, CA, USA) with specific primers carrying restriction sites for directional cloning and proper reading frame (Forward: 5′-AAGCTTGGGATCCATGCTGGATCAACAAACCATCGC-3′; Reverse: 5′-GAGCTCCGGTTATATAACCTTGTGCGG-3′). The amplified fragment was cloned into pGEM-T vector (Promega, Madison, WI, USA), sequenced and then introduced into pRK415, a broad host range expression vector for Gram-negative bacteria (Keen et al., 1988). The resulting plasmid (pRK415-hmpX) was transferred to E. coli strains DH5α and RKP4600, the latter being a hmp-null mutant (Mills et al., 2001) and then mobilized in P. syringae pv. glycinea race 4 carrying the avrA avirulence gene (Keen and Buzzell, 1991) and P. syringae pv. tomato DC3000 carrying the avrB avirulence gene (Innes et al., 1993). Western blot analysis was performed using a polyclonal antibody to purified Hmp (Stevanin et al., 2000) and the ECL chemiluminescent system (Amersham Biosciences, Little Chalfont, UK).

Periplasmic and cytoplasmic cell fractions

The method used for preparing periplasmic fractions was based on previous reports (French et al., 1996; Poole et al., 1994; Vasudevan et al., 1995). Cells were grown to an OD420 of approximately 0.4, divided into 200 ml aliquots and centrifuged for 10 min at 5000 g after the addition of 6 ml 1 m NaCl and 6 ml 1 m Tris-HCl, pH 7.3, to each aliquot. The pellets were resuspended together in 6 ml of their own supernatant and 6 ml buffer containing 40 % sucrose, 2 mm EDTA and 33 mm Tris-HCL, pH 7.3, was added. After 20 min at room temperature, the sample was centrifuged for 5 min at 10 000 g. The pellet was then resuspended in 6 ml ice-cold water after which 70 μl 100 mm MgCl2 was added and the sample was left on ice for 10 min. After centrifugation at 10 000 g for 5 min, the supernatant (periplasmic fraction) was concentrated in a 15 ml Vivaspin concentrator (VivaScience, Hannover, Germany) retaining proteins with molecular weight above 30 000 Da. The remaining pellet was sonicated and centrifuged for 1 h at 132 000 g to yield a supernatant (cytoplasmic fraction). Periplasmic fractions were also prepared by using lysozyme at a final concentration of 50 μg ml−1 as described (Vasquez-Laslop et al., 2001).

Cell-free extracts

Cultures were grown until they reached the stationary phase, unless otherwise stated, and harvested by centrifugation for 10 min at 13 000 g and washing in PBS buffer. Cells were disrupted by sonication using an MSE/Sanyo Soniprep 150 sonicator (Sanyo Gallenkamp Plc., Loughborough, UK) delivering four 20-sec periods at full power, with 1 min cooling intervals. The volume of cells used for sonication was typically 5 ml. The material was then centrifuged at 132 000 g for 1 h at 4°C. The amount of H2O2 produced by cell-free extracts was measured using the scopoletin assay as described (Søballe and Poole, 2000).

Polarography and spectroscopy

Polarographic studies were carried out as previously detailed (Mills et al., 2001). The oxygen electrode was calibrated with air-saturated water and sodium dithionite crystals. The NO electrode was calibrated with freshly prepared solutions of anoxic NO as described (Mills et al., 2001). Experiments were carried out in 50 mm MOPS, 50 mm NaCl, pH 7.0. Electronic absorption spectra were recorded using a custom-built SDB4 dual-wavelength scanning spectrophotometer as described (Kalnenieks et al., 1998).

Determination of bacterial growth under nitrosative stress

Freshly grown bacterial cultures were diluted to 105 cells ml−1 in 2.5 ml of appropriate medium, transferred to 10 ml tubes and incubated with vigorous shaking in the presence of different concentrations of the nitrosating agent SNP. After 16 or 24 h, bacterial growth was determined spectrometrically (OD600) and expressed as a percentage of the OD reached by the same bacterial culture not exposed to SNP.

Soya bean cell suspensions

Experiments with Glycine max cv. Williams 82 cell suspensions were performed 3 days after subculture (Levine et al., 1994). Cells were incubated in 12-well tissue culture plates (1 ml/well), agitated at 60 rpm and challenged with 108 bacteria. The accumulation of H2O2 was assayed by measuring quenching of scopoletin fluorescence (Levine et al., 1994). Cell death was assayed 24 h after the indicated treatments by spectrophotometric measurement of the uptake of Evan's blue stain (Levine et al., 1994). NO accumulation was assayed by monitoring the conversion of oxyhaemoglobin to methaemoglobin as previously described (Delledonne et al., 1998). Oxyhaemoglobin was added to the indicated final concentrations. After 2 min, changes in absorbance of the medium at 421 and 401 nm were measured and NO levels were calculated using an extinction coefficient of 77 mm−1 cm−1 [A401 (metHb)–A421 (HbO2)].

Plant inoculations

Growth of S. ionantha cv. Blue Rhapsody, infiltration of leaves with E. chrysanthemi (100 μl of a bacterial suspension containing 108 CFU ml−1) and measurement of bacterial growth in planta were as described (Favey et al., 1995). Growth of A. thaliana ecotype Col-0, infiltration of leaves with P. syringae pv. tomato DC3000 (5–10 μl of a bacterial suspension containing 5 × 106 CFU ml−1) and measurement of bacterial growth in planta were as described (Cameron et al., 1994). Dead cells were visualized by trypan blue staining as described (Koch and Slusarenko, 1990). The production of H2O2 was visualized in situ with DAB staining as described (Thordal-Christensen et al., 1997). The number of stained pixels inside the infected leaf area was quantified by reduction of microscopic photographs to greyscale mode followed by quantification of pixels with a grey tone value <125 as described (Zeier et al., 2004). The production of NO was visualized using dDAF-2 DA as described (Foissner et al., 2000). Fluorescence for DAF-2DA-infiltrated leaves was quantified by determining the mean green channel values for captured images with the histogram function of Adobe Photoshop 7.0 (Adobe Systems, Mountain View, CA, USA). To account for background fluorescence, the corresponding mean value for non-infected leaves infiltrated with the fluorescent dye (8.2 ± 1.4) was subtracted.

RNA blot hybridization

Total RNA was isolated using Trizol reagent (Invitrogen). RNA blot hybridization was performed with the Gmhsp-26 gst cDNA probe (Levine et al., 1994). Equal loading was verified by gel staining with ethidium bromide and by hybridization with an rRNA gene sequence (Pepper et al., 1994).

Acknowledgements

We thank Jeff Dangl for helpful discussion and critical reading of the manuscript. We thank Federica Zaninotto and Davide Ederle for technical support. This research was supported by the UK Biotechnology and Biological Sciences Research Council (C.L.; R.K.P.), the US Department of Agriculture (C.L.) and the EMBO Young Investigators Program (M.D.).

Ancillary