Plants respond to pathogen attack via a rapid burst of reactive oxygen species (ROS). However, ROS are also produced by fungal metabolism and are required for the development of infection structures in Magnaporthe oryzae.
To obtain a better understanding of redox regulation in M. oryzae, we measured the amount and redox potential of glutathione (EGSH), as the major cytoplasmic anti-oxidant, the rates of ROS production, and mitochondrial activity using multi-channel four-dimensional (x,y,z,t) confocal imaging of Grx1-roGFP2 and fluorescent reporters during spore germination, appressorium formation and infection.
High levels of mitochondrial activity and ROS were localized to the growing germ tube and appressorium, but EGSH was highly reduced and tightly regulated during development. Furthermore, germlings were extremely resistant to external H2O2 exposure ex planta. EGSH remained highly reduced during successful infection of the susceptible rice cultivar CO39. By contrast, there was a dramatic reduction in the infection of resistant (IR68) rice, but the sparse hyphae that did form also maintained a similar reduced EGSH.
We conclude that M. oryzae has a robust anti-oxidant defence system and maintains tight control of EGSH despite substantial oxidative challenge. Furthermore, the magnitude of the host oxidative burst alone does not stress the pathogen sufficiently to prevent infection in this pathosystem.
Rice blast disease caused by Magnaporthe oryzae is a major global emerging infectious disease (EID) causing rice losses sufficient to feed 3–10.6% of the world's population 2000 calories a day for 1 yr (Fisher et al., 2012). Infection starts with the germination of a three-celled spore on the host surface, producing a short germ tube that develops a domed infection structure, termed the appressorium (Wilson & Talbot, 2009). Development requires progression through a single cell cycle in the apical cell, followed by autophagy (Veneault-Fourrey et al., 2006) and mobilization of storage reserves to fuel appressorium development (Thines et al., 2000; Wang et al., 2007; Patkar et al., 2012a). The appressorium becomes melanized and builds up sufficient turgor pressure to drive an infection peg into the host epidermal cell (Howard et al., 1991; Money & Howard, 1996; Wilson & Talbot, 2009). In a susceptible host, the infection hyphae ramify through the infected cell and then colonize adjacent cells through plasmodesmata (Kankanala et al., 2007). In resistant varieties, infection is halted at the penetration stage or during early colonization, and is associated with a hypersensitive response (HR) (Torres, 2010; Heller & Tudzynski, 2011).
Plants respond to infection by the rapid production of reactive oxygen species (ROS) using membrane-bound NADPH oxidases (Marino et al., 2012), or secreted peroxidases and amine oxidases (Bolwell et al., 2002), as part of the general pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) or more specific effector-triggered immunity (ETI) responses (Yoshioka et al., 2009; Torres, 2010; Heller & Tudzynski, 2011; Thomma et al., 2011; Tudzynski et al., 2012). ROS produce cross-linked plant wall polymers to form a barrier to penetration, attack the pathogen directly or act as diffusible signals in the plant to up-regulate pathogenesis-related proteins (Lamb & Dixon, 1997; Shetty et al., 2008; Heller & Tudzynski, 2011). In a resistant host challenged by an avirulent pathogen, the initial ROS burst is followed by a longer lasting second phase, culminating in an HR (Mur et al., 2008) and programmed cell death (Levine et al., 1994). In biotrophic and hemibiotrophic fungi, such as M. oryzae, which require living hosts, the first phase of ROS production still occurs, but the second phase is suppressed in susceptible hosts, probably through the secretion of effectors through the biotrophic interfacial complex (Valent & Khang, 2010), which re-programme the metabolic pathways involved in host ROS production (Parker et al., 2009).
Fungi require effective anti-oxidant defence systems to operate in such an environment in which oxidative stress is endemic. An abundant array of anti-oxidant genes exists in the M. grisea genome (Dean et al., 2005; Egan & Talbot, 2008; Morel et al., 2008), although less is known about the low-molecular-weight anti-oxidants in fungi more generally (Georgiou & Petropoulou, 2001; Patsoukis & Georgiou, 2004). Nevertheless, the need to detoxify host ROS can be inferred from fungal mutants that lack critical anti-oxidant enzymes, or from treatments that manipulate ROS levels during infection. Thus, Magnaporthe mutants in glutathione peroxidase, Hyr1, are less tolerant to ROS and produce smaller lesions on susceptible plants (Huang et al., 2011). Likewise, several redox-sensitive transcription factors, such as MoAP1 (Guo et al., 2011) and MoSwi6 (Qi et al., 2012), and the defence suppressor Des1 (Chi et al., 2009) increase resistance to external H2O2, with mutants showing reduced pathogenicity and pleiotropic changes in gene expression, including decreases in extracellular peroxidases. Conversely, the reduction of external ROS levels during infection through the addition of exogenous catalase (Tanabe et al., 2009) or the NADPH oxidase inhibitor diphenylene iodonium (DPI) (Chi et al., 2009) promotes increased infection of compatible and incompatible strains, or particular H2O2-sensitive mutants.
Nevertheless, the significance of other putative anti-oxidant defences is less clear. Thus, the deletion of the major secreted M. oryzae catalase-peroxidase, CPXB, increases sensitivity to exogenous H2O2, but does not affect overall pathogenicity (Tanabe et al., 2011). Likewise, mutants lacking the large subunit catalase, catB, are less pathogenic, but through changes in normal fungal wall strengthening rather than by detoxification of host-derived H2O2 (Skamnioti et al., 2007). Furthermore, expression data from transcriptome profiling show that the genes most highly up-regulated in M. oryzae during infection are related to nutrient limitation rather than oxidative stress (Mathioni et al., 2011). This echoes results for the necrotrophic fungus Botrytis cinerea, in which the redox-sensitive AP1 transcription factor homologue, Bap1, is critical for ROS resistance in vitro, but deletion mutants do not show reduced virulence, and the suite of downstream target genes regulated by Bap1 is not highly expressed in planta (Temme & Tudzynski, 2009). The authors concluded that B. cinerea does not suffer H2O2 stress in planta, in contrast with conventional expectations of the role of the oxidative burst in restricting infection (Temme & Tudzynski, 2009), but actually exploits the host HR as part of its necrotrophic habit (Govrin & Levine, 2000).
The interplay between ROS and anti-oxidant defences is further complicated, as ROS are produced by normal cell metabolism and act as signalling intermediates associated with key transitions in microbial development, including differentiation, sexual reproduction, conidiation, spore germination, secondary metabolism and apoptosis (Hansberg & Aguirre, 1990; Aguirre et al., 2005; D'Autreaux & Toledano, 2007; Gessler et al., 2007; Takemoto et al., 2007; Egan & Talbot, 2008; Scott & Eaton, 2008; Shetty et al., 2008; Aguirre & Lambeth, 2010; Heller & Tudzynski, 2011; Tudzynski et al., 2012). Indeed, Hansberg and Aguirre originally proposed that microbial cell differentiation might be triggered by transient oxidation that initiates a shift between developmental states (Hansberg & Aguirre, 1990; Aguirre et al., 2005). ROS production is well documented during pre-penetration in M. oryzae, whereas scavenging external ROS reduces infection rates, all consistent with a role for ROS in the developmental programme (Egan et al., 2007; Ryder et al., 2013).
The major cytoplasmic anti-oxidant that mitigates oxidative stress in eukaryotes is glutathione (Belozerskaya & Gessler, 2007; Gessler et al., 2007; Meyer, 2008). However, little is known about the glutathione concentrations and dynamics in filamentous fungi. Tools are now available to quantify both the amount of glutathione in vivo (Fricker et al., 2000; Meyer & Fricker, 2000, 2008; Fricker & Meyer, 2001; Meyer et al., 2001) and the electrochemical potential of the reduced glutathione:oxidized glutathione (GSH:GSSG) redox couple (EGSH) using transgenic redox green fluorescent protein (GFP)-based reporters (Dooley et al., 2004; Schwarzlander et al., 2008; Meyer & Dick, 2010). In particular, Grx1-roGFP2 includes a glutaredoxin (Grx) subunit to improve the response kinetics (Gutscher et al., 2008) and is known to function correctly in fungi (Heller et al., 2012). Notably, the difference in the mid-point potential between the Grx1-roGFP redox couple and the GSH:GSSG redox couple makes Grx1-roGFP exquisitely sensitive to small changes in the degree of glutathione oxidation from the highly reduced level typically found in vivo (Meyer & Dick, 2010). Thus, in this article, we use multi-parameter live-cell confocal imaging and a range of fluorescent reporters to determine, first, whether there is physiological evidence for the redox control of early development in M. oryzae mediated by changes in EGSH; second, what is the relative level of endogenous ROS production during development; third, what is the capacity of the glutathione anti-oxidant system to deal with an imposed oxidative burst, as might be encountered during host infection; and fourth, what impact is exerted by the actual host oxidative burst on EGSHin vivo during susceptible and resistant interactions.
Materials and Methods
Fungal strains and growth conditions
Wild-type Magnaporthe oryzae strain Guy11 and transgenic Grx1-roGFP2 strain were cultured at 24°C under a 14 h : 10 h light : dark cycle according to Talbot et al. (1993). Spores were scraped from 10-d-old cultures, filtered through two layers of miracloth (Calbiochem, San Diego, CA, USA), washed with sterile deminineralized water (dH2O), pelleted by centrifugation (4000 g, 5 min) and resuspended in dH2O at a spore concentration of c. 5 × 105 ml−1.
Cloning and generation of pRP27::Grx1:roGFP2 strains
Standard molecular techniques and cloning (Ausubel et al., 1999) were used to prepare the pRP27::Grx1:roGFP2 construct. A set of transformation vectors was created based on pUCAP (van Engelen et al., 1995), which carries pUC19 MCS and unique AscI and PacI sites.
The bialophos resistance marker (GenBank AF013602) was removed as a SalI fragment from pGEMTEasy-BAR vector (Samalova et al., 2013), blunt-ended and cloned into the pUCAP SacI site, and digested with Ecl136II to re-create the SalI sites and generate pUCAP-BAR. The polyadenylation signal pATrpC was amplified from pMJK142.2 (kindly provided by Professor N. Talbot, Exeter, UK) using P1 and P2 primers (Table 1), digested with BamHI and KpnI, and cloned into the same restriction sites of the pUCAP-BAR vector, creating pUCAP-pATrpC/BAR, and confirmed by sequencing. The M. oryzae promoter RP27 (Czymmek et al., 2002) was amplified from Guy11 genomic DNA using P3 and P4 primers (Table 1). Grx1-roGFP2 was amplified from pBinCM-GRX1-roGFP2 (Gutscher et al., 2008) using P5 and P6 primers (Table 1). The two fragments were fused by overlapping PCR using primers P3 and P6. The PCR product was digested with AscI and BamHI, and cloned into the same restriction sites of pUCAP-pATrpC/BAR. The final pUCAP-pRP27::Grx1:roGFP2 vectors were confirmed by sequencing, and digested with PacI, before transformation into M. oryzae.
DNA-mediated protoplast transformation (Talbot et al., 1993) was used to generate putative transformants that were selected on defined complex medium supplemented with 60 μg ml−1 bialophos (Goldbio, St Louis, MO, USA). They were subjected to PCR using P6 and P7 primers (Table 1) to confirm the presence of Grx1-roGFP2. The most fluorescent line was chosen for this study.
Leaf sheath assays
Rice cultivar CO39 (susceptible) and IR68 (resistant; Gilbert et al., 2006; Hubbart et al., 2007) were grown at 24°C, 80% humidity in a 14 h light cycle. Leaf sheaths of 2–3-wk-old plants were inoculated with 20–40 μl of conidial suspension (5 × 105 spores ml−1) and placed onto water agar (2% w/v) in the growth chamber. Leaf sheaths were mounted using double-sided tape, sliced open with a razor blade and imaged by confocal microscopy.
Time series were collected at 30–120-s intervals for 15–45 min, as z-stacks of 5–10 sections taken at 0.67–3-μm intervals apart, using a Zeiss 40× 1.2 NA PlanApo water-immersion lens. Pixel sizes were 0.22–0.44 μm in x and y. Pinhole settings were adjusted individually for each channel to give an estimated optical section thickness of 2–3 μm (c. 2–3 Airy units) for each of the wavelength combinations. This provided some degree of optical sectioning, but with sufficient signal-to-noise to allow long-term physiological measurements with low laser intensities. Laser power was measured from the defocused beam using a Newport 1815-C power meter (Newport Corp., Irvine, CA, USA). Values ranged from 1.3 to 7.3 μW. Non-confocal bright-field images were collected simultaneously with a transmission detector.
Coverslips were placed in a perfusion chamber mounted on a Zeiss LSM510META confocal microscope and Grx1-roGFP2 (emission, 505–530 nm) was imaged with 405- and 488-nm excitation in multi-track mode with line switching. Autofluorescence was measured in parallel (excitation, 405 nm; emission, 435–485 nm).
Dual-excitation confocal ratiometric analysis
Image time series were imported into a custom MatLab (The MathWorks, Natick, MA, USA) program (available on request from M.D.F.). z-stacks were averaged in x, y and z using a 3 × 3 × 3 or 5 × 5 × 3 kernel. The z-position of the brightest pixel in a maximum z-projection of the main channel of interest was used to extract the corresponding z-pixel for each of the other wavelengths to give the average intensity from all channels in a volume around the same, bright pixel in x, y and z. The resultant image is termed an optimum plane projection. The average background intensity was measured adjacent to the spores on coverslips, or in the vacuole for in planta experiments, and subtracted. The autofluorescence bleed-through contribution into the Grx1-roGFP2405 channel was measured from Guy11 spores, and the corresponding scaling factor was used to subtract autofluorescence from the Grx1-roGFP2405 image. Ratio images were calculated on a pixel-by-pixel basis as I405/I488. Pixels with intensity values < 2 standard deviation units above background, within 10% of saturation, or where the local coefficient of variance (CV) in a 3 × 3 neighbourhood was above 20%, were masked.
For pseudo-colour display, the masked ratio was coded by hue on a spectral colour scale ranging from blue (most reduced) to red (most oxidized), with the limits set by the in situ calibration. Individual germlings were segmented using Otsu's method (Otsu, 1979) and the extracted images were rotated to give a montage of selected time points for display.
Quantitative measurements were calculated as the ratio of the mean intensity from each channel from the regions of interest. In situ calibration was performed using 10 mM dithiothreitol (DTT) for 5 min, followed by 100 mM H2O2 for 5 min, to drive Grx1-roGFP2 to a highly reduced and highly oxidized form, respectively. The degree of oxidation (OxDGrx1-roGFP2) and pH-corrected GRx1-roGFP redox potential (EGrx1-roGFP2) were calculated according to Schwarzlander et al. (2008), assuming a mid-point potential (E°Grx1-roGFP2) of −280 mV (Dooley et al., 2004; Hanson et al., 2004; Gutscher et al., 2008; Morgan et al., 2011, 2013) and a cytoplasmic pH of pH 7.6 (Parton et al., 1997; Hesse et al., 2002).
For population studies, a Gaussian mixture model (GMM) with two to four components was fitted to the pixel intensities at each wavelength, with the contribution of each pixel weighted by the average intensity. Contour maps from the GMM were overlaid on two-dimensional histograms generated from the weighted pixel intensities. The intensity value of the Gaussian peak at each wavelength was used to calculate the ratio, and hence degree of Grx1-roGFP2 oxidation, EGrx1-roGFP2 and EGSH, respectively, for each component.
Estimation of cytosolic glutathione concentrations in spores of M. oryzae
The available glutathione pool was labelled in situ following conjugation to 100 μM monochlorobimane (MCB) to give a fluorescent glutathione–bimane (GSB) conjugate (excitation, 405 nm; emission, 435–485 nm) according to Fricker et al. (2000), Fricker & Meyer (2001), Meyer et al. (2001) and Meyer & Fricker (2002). Calibration solutions were made by reacting 10 mM monobromobimane with an excess of GSH. As GSB was transported into the vacuole, estimation of the initial cytosolic GSH concentration required correction for the cytoplasmic : vacuolar volume ratio. GSB signals from overlapping vacuoles were separated using a three-dimensional watershed and the regional maximum intensity for each vacuole was determined. Each vacuolar volume was estimated as the number of pixels above the 50% threshold between this local maximum and the background (White et al., 1996; Errington et al., 1997).
The cytoplasmic volume was the difference in volume between the vacuole volume and the total volume of the spore compartment, estimated after labelling the cell wall with 15 μM propidium iodide (PI; excitation, 543 nm; emission, 585–6150 nm). Three-dimensional z-stacks were rotated in x, y and tilted in z to align the long axis of the spore with the x-axis. The median wall outline was segmented from the maximum z-projection, filled and the volume was measured assuming radial symmetry from the cylinder of rotation about the x-axis.
Measurement of ROS levels and mitochondrial activity
Spores were labelled with 0.2–2.5 μM 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA) (excitation, 488 nm; emission, 500–530 nm) or CellROX Deep Red (CRDR, 2.5 μM) simultaneously with tetramethyl rhodamine methyl ester (TMRM, 100 nM) for mitochondria, and Grx1-roGFP2 for GSH redox potential, using quadruple excitation with paired line switching at 405 and 633 nm, and 488 and 543 nm. Emission wavelengths were 435–485, 500–530 and 657–721 nm for wall autofluorescence, oxidized Grx1-roGFP2 and CRDR, respectively, and 500–530 nm and 561–603 nm for reduced Grx1-roGFP2 and TMRM, respectively. A non-confocal, bright-field transmission image was collected in parallel.
Results and Discussion
Grx1-roGFP2 reports EGSH in M. oryzae
To measure the electrochemical potential of the GSH:GSSG redox couple (EGSH) in vivo, Grx1-roGFP2 (Gutscher et al., 2008; Heller et al., 2012) was expressed from the ribosomal protein RP27 promoter in M. oryzae and visualized using confocal ratio imaging (Meyer & Fricker, 2008; Schwarzlander et al., 2008; Morgan et al., 2011). Spores expressing Grx1-roGFP2 showed normal rates of germ tube extension, swelling and appressorium development (e.g. Fig. 1e), and were fully pathogenic on barley and rice (data not shown). Fluorescence was observed in the cytoplasm and, to a lesser extent, vacuoles (Fig. 1a–d, Supporting Information Video S1). Excitation at 405 nm increased with probe oxidation, but also caused autofluorescence bleed-through into the Grx1-roGFP2405 channel, typically around the site of germ tube emergence and the appressorium wall. To correct the Grx1-roGFP2405 images, autofluorescence images were collected at 435–485 nm (Fig. 1f), scaled by an empirically determined bleed-through correction factor and subtracted. Localization of Grx1-roGFP2 in vacuoles was not expected, but parallels other cytoplasmically expressed fluorescent proteins in M. oryzae germlings, such as mCherry (data not shown, see also Czymmek et al., 2002), and may reflect high levels of autophagy. Nevertheless, the Grx1-roGFP2 signal in both cytoplasmic and vacuolar compartments responded to in situ calibration (Fig. 1g,h). Changes in the Grx1-roGFP2 spectrum were visualized as pseudo-colour-coded ratio images, scaled to the minimum and maximum of these calibration values (Fig. 1i).
Grx1-roGFP2 responses were rapid, with oxidation by H2O2 within 1–2 min, and reversible, with spontaneous recovery to a reduced state after washout (Fig. 1j–l), and reduction with subsequent exposure to DTT (Fig. 1j–l). Although H2O2 caused bleaching of the probe (Fig. 1j,k), the ratio corrected for the changes in intensity (Fig. 1l), giving a dynamic range around four, similar to previous reports with the wavelength combinations used here (Gutscher et al., 2008). It is notable that M. oryzae was very resistant to imposed oxidative loads, with 100 mM H2O2 needed to drive Grx1-roGFP2 to an oxidized state in vivo. Using this calibration, the degree of oxidation of cytoplasmic Grx1-roGFP2 (OxDGrx1-roGFP2) was 10–12% in all three germling cells (Fig. 1m). Interestingly, vacuolar Grx1-roGFP2 still responded to calibration, with the same dynamic range, giving a value equivalent to 50–60% oxidation (Fig. 1d,k,l). OxDGrx1-roGFP2 was converted to the redox potential (EGrx1-roGFP) (Gutscher et al., 2008; Schwarzlander et al., 2008; Meyer & Dick, 2010) using a consensus mid-point potential (E°′Grx1-roGFP2) of −280 mV (Dooley et al., 2004) and a cytoplasmic pH of pH 7.6, as measured in filamentous fungi (Parton et al., 1997; Hesse et al., 2002). With these assumptions, 10% oxidation of Grx1-roGFP2 equates to EGrx1-roGFP2 of −316 mV. Assuming that EGrx1-roGFP2 equilibrates with EGSH (Morgan et al., 2011), this requires tight regulation of GSSG at submicromolar concentrations (Meyer & Dick, 2010; Morgan et al., 2013). Thus, we infer that Grx1-roGFP2 provides a rapid and highly sensitive readout of EGSH in M. oryzae, similar to previous reports for the necrotrophic fungus B. cinerea (Heller et al., 2012). Nevertheless, high-resolution imaging is also required to separate signals from cytoplasm and vacuole.
The glutathione pool shows only slight shifts in the degree of oxidation during development
To determine whether there were shifts in glutathione oxidation associated with developmental transitions (Hansberg & Aguirre, 1990), we quantified OxDGrx1-roGFP2 at different developmental stages, classified as: spores (0–0.5 h post-inoculation, hpi), germination (0.5–1 hpi), germ tube growth (1–2 hpi), germ tube swelling (2–3 hpi), appressorium formation (3–5 hpi) and appressorial melanization (5–8 hpi).
A highly reduced cytoplasmic Grx1-roGFP2 signal (< 5% oxidized) was measured during germination in all three spore cells (Fig. 2i), becoming slightly more oxidized (8–12%) as the germ tube developed at 1–2 hpi (Fig. 2a,b,i, see also Fig. 1). Ratios from the mid and basal cells were progressively dominated by the oxidized Grx1-roGFP2 vacuolar signal once appressoria started to form at 5 hpi (Fig. 2c,d,i), making it difficult to extract a clean cytoplasmic value. Absolute fluorescence intensities were also reduced from the basal and mid cells c. 8 hpi (Fig. 2e,f), and disappeared following isolation and autophagic cell death at 24 hpi (Fig. 2g,h). At this stage, the apical cell was more vacuolate, but the cytoplasmic probe was still responsive to calibration. The appressorium itself showed a highly reduced cytoplasmic Grx1-roGFP2 signal throughout development, with a characteristic ring of more oxidized vacuoles around the periphery (Fig. 2f,h,i). However, Grx1-roGFP2 in the melanizing appressorium (Fig. 2f) was protected from H2O2 calibration, with slower responses at 8 hpi and complete insensitivity to 100 mM H2O2 at 24 hpi (Fig. 2 h).
We infer that M. oryzae maintains a highly reduced cytoplasmic EGSH during early development, and there was only very limited evidence that EGSH increases during appressorium formation, indicative of a glutathione-mediated redox-associated developmental switch (Hansberg & Aguirre, 1990).
There are high levels of endogenous ROS production during development
In the absence of a host, EGSH reflects the rate of detoxification of endogenous metabolic ROS. Previous measurements of ROS levels in M. oryzae have used the relatively non-specific probe H2DCFDA (Egan et al., 2007; Kim et al., 2009; Guo et al., 2011; Huang et al., 2011). H2DCFDA is membrane permeant, but trapped intracellularly following the cleavage of the di-acetate groups by intracellular esterases, where oxidation gives fluorescent 2′,7′-dichlorofluorescein (DCF). Nevertheless, it is not clear to which ROS the probe responds, as intrinsic reaction rates with H2O2 and superoxide are low, and the probe is likely to be oxidized by more reactive species or through the action of peroxidases or haem proteins (Halliwell & Whiteman, 2004; Rhee et al., 2010).
DCF fluorescence accumulated mainly in the cytoplasm in all three cells at all stages of germling development (Fig. 3b,d, see also Video S2). The greatest increase in fluorescence was observed in the mid cell (Fig. 3e,f), at about twice the rate in the other cells. Significant DCF also accumulated in the appressorium wall (Fig. 3d,f, see also Video S2). These data are similar to previous reports of DCF staining in conidia and in septal cell walls before germination, followed by labelling in germ tubes and a burst during appressorium formation, which might reflect the activity of the Nox1 NADPH oxidase (Egan et al., 2007; Ryder et al., 2013). However, this would also imply that there was sufficient de-esterified H2DCF in the apoplast to react with externally produced ROS.
Washout of the labelling medium led to rapid loss of the wall-associated label, consistent with free dye in the apoplast (Fig. 3d,f). However, there was also substantial loss of cytoplasmic fluorescence in all three cell types, together with some recovery in the appressorial wall (Fig. 3d,f). We infer that DCF is either very membrane permeant in M. oryzae, or there is an additional plasma membrane xenobiotic detoxification system, which would also explain the inverse kinetics of cytoplasmic and wall signals during washout (Fig. 3d,f). Previously, we have reported a similar phenomenon for the NO reporter diaminorhodamine-4M (DAR-4M) in M. oryzae (Samalova et al., 2013). Indeed, Magnaporthe is well endowed with potential transporters, with 50 members of the ATP-binding cassette (ABC) transporter family and 251 members of the major facilitator superfamily (MFS) (Coleman & Mylonakis, 2009). Furthermore, several of these transporters are up-regulated during appressorium formation (Oh et al., 2008), under stress treatments or in planta (Mathioni et al., 2011), and are required for pathogenicity (Urban et al., 1999; Sun et al., 2006; Patkar et al., 2012b), possibly to protect against the buildup of peroxides and oxidative damage (Sun et al., 2006).
Given our concerns over H2DCFDA localization and specificity, we also examined the more sensitive, specific and photostable long-wavelength ROS probe CRDR. CRDR showed a time-dependent increase in fluorescence in M. oryzae spores, but, by contrast with H2DCFDA, CRDR labelled discrete punctate structures, predominantly in the apical cell (Fig 3h,j). Despite this apparent asymmetric ROS production, simultaneous imaging with Grx1-roGFP2 showed a consistent low ratio in all three cell types (Fig. 3k). We speculate that CRDR might highlight mitochondrial ROS production, and probably reacts with superoxide rather than H2O2, given the lack of response to H2O2 addition (Fig. 3k).
Mitochondria show different activity in each germling cell type during development
To investigate whether the increase in CRDR fluorescence principally reflects the high level of ROS produced by mitochondrial activity (Schwarzlander & Finkemeier, 2013), we tested whether the punctate CRDR structures co-localized with active mitochondria, identified by potential-dependent accumulation of TMRM, and whether the mitochondria were differentially active in the different spore cells. In addition, we examined whether there were any changes in mitochondrial ROS production associated with developmental switches, particularly as mitochondria are involved in autophagic cell death (Scherz-Shouval & Elazar, 2011), which forms part of the developmental programme in M. oryzae (Veneault-Fourrey et al., 2006; Talbot & Kershaw, 2009).
TMRM labelling revealed highly motile tubular mitochondria, particularly in the apical cell during germ tube growth (Fig. 4b) and appressorium formation (Fig. 4j). In parallel, CRDR showed a time-dependent increase in punctate fluorescence in swelling germ tubes (Fig. 4c, see also Video S3) and developing appressoria (Fig. 4k, see also Video S4), which overlapped the TMRM signal (Fig. 4d,l). Lower levels of TMRM and CRDR accumulation were observed in the other cells throughout development (e.g. Fig. 4b–d,j–l, see also Fig. 3 h), even after extended periods of incubation. We infer that the mitochondria in these cells are in a different metabolic state, possibly geared towards the synthesis of metabolic intermediates (e.g. Patkar et al., 2012a), rather than ATP production.
The Grx1-roGFP2 signal measured in parallel from the swelling germ tube apical tip (Fig. 4e,f,h, see also Video S3) and developing appressorium (Fig. 4m,n,p, see also Video S4) showed the expected reduced profile in the cytoplasm and a more oxidized signal in punctate vacuoles. However, there was no evidence for any substantial changes in EGrx1-roGFP2 that might reflect a burst of ROS production (Fig. 4 h,p) during these and more than 30 other time series spanning 2–5 hpi when appressoria develop. Interestingly, the internal CRDR signal was also lost rapidly during the calibration sequence in these cells. We infer, as with DAR-4M and DCF, that CRDR can be rapidly expelled by M. oryzae.
The apparent compartmentalization of mitochondrial activity and ROS production was unexpected if the three cells of the spore have direct cytoplasmic continuity, as has been argued previously from transmission electron microscopy (TEM) images for M. oryzae conidia, which show open septal pores (Money & Howard, 1996; Soundararajan et al., 2004). Equally, differential turgor pressures in adjacent cells have been interpreted as evidence for the regulation of pore opening and closure to permit the transfer of materials between cells (Money & Howard, 1996). We found evidence using fluorescence recovery after photobleaching (FRAP) of Grx1-roGFP2 that spore cells were functionally isolated during germination and hyphal growth, but allowed communication during appressorium formation (Fig. S1). Furthermore, TEM images (Methods S1) showed occlusion of some septal pores by Woronin bodies (Fig. S2). These data are consistent with regulated septal conductivity during early development, with an increased probability of opening when transfer of reserves to the developing appressorium is required.
It is clear that there are differential rates of endogenous ROS production in each cell of germlings of M. oryzae during early development. However, at this stage, we do not have strong evidence for oxidative shifts during developmental transitions from either ROS measurements or changes in EGrx1-roGFP2. Equally, quantitative interpretation of ROS levels and localization are challenging, as the signal represents a dynamic balance between dye uptake, intracellular hydrolysis, ROS production, reaction rates and subsequent probe sequestration (Fricker et al., 2001; Halliwell & Whiteman, 2004; Meyer & Fricker, 2008). It is also possible that the glutathione pool may not be the most sensitive marker for redox events involved in signalling (Patsoukis & Georgiou, 2004; Belozerskaya & Gessler, 2007; Winterbourn, 2008; Winterbourn & Hampton, 2008), with more reactive targets, such as thioredoxin and peroxiredoxins, being responsible for the initiation of developmental responses (D'Autreaux & Toledano, 2007; Meyer, 2008; Heller & Tudzynski, 2011). Indeed, it has been proposed in yeast that the primary function of cytosolic GSH is to drive extra-mitochondrial iron–sulfur cluster maturation, and GSH only acts as a back up to thioredoxin in redox homeostasis (Kumar et al., 2011; Toledano et al., 2013). The development of probes for these other redox couples is underway (Meyer & Dick, 2010) and would greatly facilitate physiological measurements in parallel with Grx1-roGFP2 to test such hypotheses. Likewise, the solution to understanding the spatial and temporal dynamics of ROS production would benefit from the further development of specific, transgenic ratio probes for the different ROS species, such as HyPer (Belousov et al., 2006) or roGFP2-Orp1 (Gutscher et al., 2009; Meyer & Dick, 2010).
M. oryzae can tolerate extreme oxidative stress
Whilst the maintenance of a highly reduced cytoplasm is required for normal cell physiology, one of the major defences in plants and animals to pathogen attack is through pronounced localized production of ROS to high levels, often within a few hours of inoculation (Xu et al., 2009). We therefore tested how well M. oryzae might handle transient exposure to increasing concentrations of H2O2 to mimic attack by a host-derived oxidative burst.
In five extended time-series experiments, with a total of 89 germlings classified into different developmental stages, cells initially showed a rapid transient oxidation of Grx1-roGFP2 in response to short (5 min) exposure to increasing concentrations of H2O2, which returned to a more reduced state, even in the continued presence of the oxidant. A typical experiment is shown in Fig. 5 for a field of developing germlings (Fig. 5a). The average ratio for each apical cell is shown in Fig. 5(i). Ratio images are shown at low resolution for all cells at rest (Fig. 5b), following DTT (Fig. 5c), at the peak of each of the H2O2 exposures (Fig. 5d–f,h) and for the penultimate recovery (Fig. 5g). More detailed time-series images are shown for specific germlings to highlight the variability in response (Fig. 5j). Each increasing concentration of H2O2 gave a marked transient oxidation of Grx1-roGFP2 in these germlings, which rapidly returned towards resting levels, even in the continued presence of H2O2. The peak responses in other experiments were not as strong, but were captured in the average peak response for multiple cells and replicates (Fig. 5k). Washout of the H2O2 allowed OxDGrx1-roGFP2 to return to near resting levels, even with treatments up to 40 mM H2O2 (Fig. 5i–k). Germ tube growth continued with 10 and 20 mM H2O2 shocks, but showed a reduction or ceased completely with 40–80 mM H2O2.
Within this overall pattern, there was variability in response at different developmental stages, most notably the rate of spontaneous recovery (Fig. 5i,j). During germination (0–1 hpi) and germ tube growth (2–3 hpi), the apical spore cell showed the greatest excursions in response to H2O2, but also the most rapid recovery (Fig. 5k). Developing appressoria (3–5 hpi) did not respond as strongly to oxidative stress, but also showed a lower subsequent rate of reduction of Grx1-roGFP2 (Fig. 5k). As noted previously, after c. 8 hpi, melanized appressoria were essentially insensitive to shocks of even 100 mM H2O2 (Fig. 2).
We infer that germlings of M. oryzae are able to withstand an aggressive imposed oxidative burst of H2O2, with transient excursions rapidly brought down to a new steady state. Germ tube growth continued in short-term exposures to 10–20 mM H2O2, whilst the melanizing appressorium was impervious to much higher H2O2 concentrations. The difference in and makes Grx1-roGFP2 particularly sensitive to subtle shifts in EGSH, with submicromolar cytoplasmic concentrations of GSSG, even with millimolar concentrations of exogenous H2O2. In addition to the rapid reduction of GSSG by glutathione reductase, it is possible that GSSG is sequestered in the vacuole to maintain tight control of cytoplasmic EGSH, as has recently been documented for yeast (Morgan et al., 2013).
GSH levels are at millimolar concentrations in M. oryzae
Given the variable levels of endogenous ROS observed, the highly reduced EGSH and the ability to tolerate substantial external bursts of H2O2, we sought to measure the absolute cytoplasmic concentration of glutathione to determine whether M. oryzae has a particularly high anti-oxidant capacity compared with non-pathogenic fungi and other organisms. Total levels of GSH were measured following in situ reaction with MCB to give a fluorescent GSB conjugate (Fricker et al., 2000; Meyer & Fricker, 2000, 2008; Fricker & Meyer, 2001; Meyer et al., 2001). MCB requires a glutathione transferase to react at an appreciable rate in vivo, and thus preferentially labels the existing cytoplasmic GSH pool, together with any GSSG that can be re-reduced by glutathione reductase (Fricker et al., 2000; Meyer & Fricker, 2000; Fricker & Meyer, 2001; Meyer et al., 2001). GSB was formed in the cytoplasm initially, and subsequently sequestered in the vacuoles over c. 30 min (Fig. 6a,b, see also Video S5), with similar kinetics in all three cells of the spore (Fig. 6e). This matches the xenobiotic detoxification pathways in plant and animal systems, where GSB is exported from the cytoplasm by glutathione-S-conjugate pumps (Coleman et al., 1997; Fricker & Meyer, 2001). To measure the original cytoplasmic GSH concentration, the total GSB was corrected for the relative volume of the cytoplasmic and vacuolar compartments. The volume of the numerous vacuoles was determined from segmentation of the bimane images (Fig. 6c,d, see also Video S6), whereas the total cell volume was estimated from a maximum projection of the PI-stained cell wall, assuming that the spore was rotationally symmetric about the long axis (Fig. 6f). The cytoplasmic volume was calculated as the difference between the total cell and vacuolar volumes (Fig. 6g). At this stage of development, all three cells in the spore were densely cytoplasmic, at c. 70% of the total cell volume. Using the estimated cytoplasmic and vacuolar volumes, the total GSH concentration ranged from 1.29 ± 0.08 mM in the basal cell to 0.89 ± 0.05 mM in the mid cell, n = 81 (Fig. 6 h).
These GSH concentrations are comparable with those of aerial plant tissues, but at the low end for root tissues, yeast and mammalian cells (Meister & Anderson, 1983; Meyer, 2008; Noctor et al., 2012). Thus, it appears that anti-oxidant defences in M. oryzae are not based on a massive constitutive GSH pool; the highly reduced EGSH must be based on either rapid re-reduction of any GSSG produced or rapid removal of GSSG, or both (Morgan et al., 2013).
M. oryzae does not appear to experience oxidative stress in vivo
Having determined the amount of GSH, and the capacity of M. oryzae to withstand imposed oxidative stress, we asked what happens to EGSH during infection. On susceptible hosts, M. oryzae formed appressoria and penetration pegs within 16 hpi (Fig. 7g, see also Video S7), which developed to form swollen, branched infection hyphae by 24 hpi (Fig. 7m, see also Video S7). Over the next 24 h, M. oryzae colonized and spread through neighbouring cells to form small, microscopic lesions by 48 hpi (Fig. 7s, see also Video S8).
The quantification of EGSH in hyphae ramifying in three dimensions was challenging, as single-point measurements did not capture the potential variation throughout the colony. Estimates were further complicated by vacuolar Grx1-roGFP2 signals, and high levels of host autofluorescence. Thus, we developed a pixel population approach to quantify OxDGrx1-roGFP2 within the entire network structure. The fluorescence signal at both excitation wavelengths of Grx1-roGFP2 was fitted with a Gaussian mixture model (GMM) comprising two to four components (Fig. 7f,l,r,u), reflecting different cytoplasmic, vacuolar or autofluorescence signals in different regions of the colony. Ratios were compared with in situ calibration of germinating spores imaged under the same conditions. It was not possible to perform a comparable in planta calibration, because of the strong reaction of the plant tissue to the high H2O2 concentrations needed to calibrate Grx1-roGFP2 within intracellular fungal hyphae (data not shown). Occasionally, however, cells containing dying hyphae were observed, which provided an internal control for oxidized Grx1-roGFP2 (Fig. 7a–f), and confirmed the applicability of the ex planta calibration.
Cytoplasmic EGSH was highly reduced throughout the infection process when quantified by ratio imaging (Fig. 7k,q,w, see also Videos S7, S8) or Gaussian modelling (Fig. 7 l,r,u). Both approaches highlighted the more oxidized signal from the numerous vacuoles present, particularly at later stages of infection (Fig. 7w). There were increasing levels of host wall fluorescence visible as a red/purple colour, resulting from the overlap between the autofluorescence channel, coded in blue, and bleed-through into the GRx1-roGFP2405 channel, coded in red (Fig. 7h,n,v), indicative of a host ROS response. Around 48 hpi, the centre of the lesion showed some cell necrosis (Fig. 7v,w), a slightly higher average cytoplasmic redox potential compared with hyphae from the expanding colony margin, and an increased number of vacuoles with oxidized Grx1-roGFP2 (Fig. 7x). In general, Grx1-roGFP2 reported a highly reduced EGSH throughout, with no evidence for shifts during early penetration events at c. 16–24 hpi, or during subsequent internal colonization.
In a resistant IR68 cultivar, cell penetration and intracellular development of M. oryzae were dramatically reduced with few successful penetration events and much higher levels of host wall autofluorescence or cell death, consistent with a strong oxidative host defence response (Fig. 8b,h,p, see also Videos S9, S10). Nevertheless, spores, germ tubes and the numerous appressoria unable to form infection pegs still maintained a reduced cytoplasmic GSH pool (see upper appressorium in Fig. 8e), with no evidence for extensive oxidation of Grx1-roGFP2 in the ratio images (Fig. 8e, see also Videos S9, S10) or GMM (Fig. 8f). When sparse intracellular hyphae did form, they were thinner than in the susceptible interaction, more vacuolated, and appeared to migrate between cells with little or no intracellular branching (Fig. 8 h,p). Nevertheless, the cytoplasmic EGSH of these hyphae remained as highly reduced as in the susceptible plants (Fig. 8e,k,i,q, see also Videos S9, S10).
This represents something of a conundrum – at a physiological level, it is not obvious that ROS toxicity plays a dominant role in host resistance to rice blast as, under normal circumstances, M. oryzae has sufficient anti-oxidant defences to deal with both endogenous and exogenous ROS production and tightly regulates EGSH with submicromolar levels of oxidized GSSG present. Equally, fungal mutants that have compromised anti-oxidant pathways suffer decreased pathogenicity. This parallels the results reported for the necrotroph B. cinerea (Temme & Tudzynski, 2009), in which the key redox-sensitive regulatory transcription factor, Bap1, was required for tolerance to exogenous H2O2ex planta, but the corresponding Bap1 target genes were not induced during infection, with the clear implication that the fungus was not experiencing oxidative stress in planta. Furthermore, Bap1 was not essential for pathogenesis (Temme & Tudzynski, 2009), similar to results for the deletion of AP1 homologues in the human pathogen Aspergillus fumigatus (Lessing et al., 2007) and the plant pathogen Cochliobolus heterotrophus (Lev et al., 2005). The situation in M. oryzae is not identical, as deletion of the homologous gene, MoAP1, causes a reduction in pathogenicity, but during later intracellular infection and hyphal growth, rather than during appressorium formation and host penetration. MoAP1 has a wide range of targets, but reduced secretion of peroxidases and laccases might provide the strongest mechanistic link to ROS tolerance (Guo et al., 2010, 2011). Similarly, deletion of AP1 homologues in Ustilago maydis (Molina & Kahmann, 2007) and Alternaria alternata (Lin et al., 2009, 2011; Yang et al., 2009) also affects both H2O2 tolerance and pathogenicity.
The overall picture emerging is that M. oryzae tightly regulates EGSH and has sufficient anti-oxidant capacity to handle even a strong host response, with the consequence that ROS toxicity alone is not sufficient to kill the pathogen. We infer that a distinction needs to be drawn between the presence of a high oxidative load during the host response, and the extent that a fungus actually experiences oxidative stress in planta, which cannot necessarily be inferred from ex planta responses to oxidative stress, but requires in vivo measurements.
Funding was provided by the Biotechnology and Biological Sciences Research Council (BBSRC) (BB/G00207X/1) to S.J.G. and M.D.F. We would like to thank Sarah Rodgers for the TEM images and Ian Moore for reading the manuscript.