Freshly collected spores of Bga race 5, and Bgh CC1 were used. Bga and Bgh isolates were maintained on plants of susceptible oat cv. Selma and barley cv. Pallas, respectively in spore proof growth chambers. Plants were shaken to remove aging conidia 1 day before the inoculation.

Three oat cvs (Cory, Charming and Orblanche), previously identified as resistant to Bga race 5 and one susceptible (Selma) (Sanchez-Martin et al., 2011), were used for study of the oat-powdery mildew interaction. In addition the barley genotype P01 (Kølster et al., 1986) previously used in the characterisation of the stomatal dysfunctions in barley was added for comparison.

Plants were grown individually in 30 × 110 mm plastic centrifuge tubes (with two 5 mm drainage holes) filled with peat: sand (3:1). Tubes were stood in trays filled to a ∼50 mm depth in compost which was watered freely throughout. Plants were grown in a room with 20°C, 65% relative humidity and under 12 h dark/12 h light. Plants grew usually under a light intensity regime of 250 μmol m^-2 s^-1 photon flux density (hereafter referred to as low light, LL) but for the experiments of increased light intensity, after inoculation plants were subjected to a moderate light intensity regime of 450 μmol m^-2 s^-1 photon flux density supplied by high-output white fluorescent tubes (hereafter referred to as high light, HL).

For powdery mildew inoculation, the first fully expanded leaf (12 days seedlings), of the oat or barley plants was inoculated with Bga or Bgh, respectively, using a settling tower (Sanchez-Martin et al., 2011) to give about 30 conidia mm^-2 (checked by counts made from glass slides laid adjacent to leaves).

For histological studies, plants were maintained in the above mentioned growth chamber. Then at 36 h after inoculation (h.a.i) the central 30 mm leaf segment was excised and fixed on pads moistened with 3:1 ethanol:glacial acetic acid (v/v), and cleared with lactoglycerol (equal parts lactic acid, glycerol and water), as described before (Prats et al., 2007b) to avoid displacement of ungerminated conidia and loosely attached germlings. Four plants per genotype were analyzed at each fixation time under white and ultraviolet light incident fluorescent microscopy (330 nm excitation/380 nm emission) using a Leica DM LS phase contrast microscope (Leica Microsystems, Wetzlar, Germany; 40x objective).

Percentages of germlings hampered in the infection process before or at time of cell penetration (penetration resistance), percentage of germlings inducing early or late cell death and percentage of established colonies were determined from 100 germinated urediniospores per leaf segment. Death of attacked epidermal cells was recognized by whole-cell autofluorescence (Koga et al., 1990; Zeyen et al., 1995; Sanchez-Martin et al., 2011, 2012).

Leaf water conductance (g_l) was measured in ten plants per genotype with an AP4 cycling porometer that allows a non-destructive and rapid method for stomatal conductance measurement (Delta-T Devices Ltd, Cambridge, UK) as described in Prats et al. (2006). g_l is the sum of cuticular and stomatal conductance, but as cuticular conductance of oat is low (Bengtson et al., 1978), changes in g_l largely reflect changes in stomatal aperture. Stomatal conductance was measured on the center of the adaxial surface of leaf laminae (covering an area of 17.5 mm × 2.5 mm), of fully expanded second leaves twice a day, 3 h after the onset of the light period and 2 h before the end of the dark period. In light, a single measurement took <20 s, and in darkness slightly longer as the g₁ was lower. Consequently, 10 plants were measured under 5 min. In each experiment, sets of 10 healthy and 10 inoculated plants of the chosen genotypes were measured, each set being held in adjacent trays on the growth room bench. The porometer was wiped clean after measuring inoculated leaves to avoid transferring the pathogen. Measurements in the powdery mildew highly susceptible cv. Selma were stopped earlier than in the other cvs in order to avoid experimental interference with sporulation.

To better visualize differences between treatments and genotypes the area under the conductance progress curve (AUCPC) of treated plants with respect to the control curve of non-stressed plants was calculated using the following formula

where S_i is the conductance at assessment date i, t_i is the number of days after the first observation on assessment date i and k is the number of successive observations.

Fluorescence quenching analyses were measured using modulated fluorescence on second leaves of dark-adapted plants with a PAM 2100 Fluorometer (PAM-2000; Walz, Effeltrich, Germany). Measurements were made in four different replications according to (Sánchez-Martín et al., 2015).

The Fo was determined after dark adaptation (at least 30 min) with a pulsed low red measuring light (ML) (0.1 μmol photons m^-2 s^-1). Then, a 1-s saturating light pulse (∼6.000 μmol photons m^-2 s^-1] of white light was applied to measure the maximal fluorescence (Fm) value. When fluorescence returned again to the F_o level, plants were illuminated by a non-saturating continuous red “AL” (655nm) for 5 min to drive photosynthesis and gives F’ values. During the AL induced fluorescence kinetics, saturating pulses were applied with 20-s intervals in order to keep track of the fluorescence parameters Fm’ (the maximum fluorescence level in the light-adapted state). After several minutes under AL, and when the F value remained low and constant, AL was turned off and a far-red radiation (735 nm) that excited preferentially PS I was switched on during 4 s to determine the value of Fo′. During the induction kinetic induced by AL the ML is automatically switched to a higher frequency of 100 kHz in order to achieve a better signal to noise ratio and time resolution. In ML, while measuring F₀, the frequency of 100 k Hz is replaced by 1.6 k Hz to avoid any chlorophyll fluorescence induction kinetics. Based on these basic parameters obtained, other parameters with significant physiological relevance such as i) Maximum Quantum Yield of PSII photochemistry (F_v/F_m) and ii) PSII operating efficiency (F′_q/F′_m also termed ΔF/F_m‘) were derived.

H₂O₂ content was measured in five plants per genotype according to Wei et al. (2015) with some modifications. Approximately 100 mg of fresh weight was homogenized in 1 mL 0.1% trichloroacetic acid (TCA) in an ice bath. The homogenate was centrifuged at 12000 g for 20 min at 4°C. Then, the reaction mixture consisting of 75 μL of supernatant, 75 μL of 10 mM potassium phosphate buffer (pH 7.0) and 150 μL of KI was added to a microtiter well. The absorbance was measured at 390 nm 10 min later and was stable at least 30 min afterward. A calibration curve was performed with H₂O₂ standards at different concentrations in a similar way.

Cell membrane stability (CMS) was measured in five plants per accession according to Sánchez-Martín et al. (2015). Samples collected were washed three times in deionized water to remove electrolytes adhered on the surface. The samples were then inserted into a capped vial (20 mL) containing 10 mL of deionized water and incubated in the dark for 24 h at room temperature. The conductivity was measured with a conductivity meter (CMD 510, WPA, UK). After the first measurement, the vials were autoclaved for 15 min to kill the leaf tissue and release the remaining electrolytes. After cooling, the second conductivity reading was taken. These two measurements were carried out individually for all the samples from both the control and stress treatments. The control gave a measure of leakage solely due to the cutting and incubation of leaf disks. The conductance of the stressed sample was a measure of electrolyte leakage due to water stress, in addition to damage caused by cutting and incubation, and was assumed to be proportional to the degree of injury to the membranes. CMS was calculated as the reciprocal of cell-membrane injury after Blum and Ebercon, 1981:

where T and C refer to the treated and control samples, respectively; the subscripts 1 and 2 refer to the initial and final conductance readings, respectively.

All experiments were performed according to completely randomized designs. For ease of understanding, means of raw percentage data are presented in tables and figures. However, for statistical analysis, data recorded as percentages were transformed to arcsine square roots (transformed value = 180/II × arcsine [√(%/100)]) to normalize data and stabilize variances throughout the data range. Data were subjected to analysis of variance (ANOVA) using SPSS software for comparison of treatments and analysis of interactions between factors, after which residual plots were inspected to confirm data conformed to normality. Significance of differences between means was determined by contrast analysis (Scheffe’s). Pearson correlations were calculated to detect statistical correlations between traits measurements. In addition, least significant difference (LSD) values were added to tables for comparison.

To further test the correlation between the execution of different resistance responses and stomatal lock-up (Prats et al., 2007a, 2010), we evaluated the resistance responses and also the timing of HR development following Bga inoculation in several oat cultivars with different genetic backgrounds (Table 1).

As expected, Selma was highly susceptible to Bga with more than 65% of attack sites forming colonies with no sign of visible HR. The remaining sites exhibited penetration resistance due to papilla formation. All other tested oat cvs were significantly more resistant than the susceptible Selma (Table 1). Cory, Charming and Orblanche showed similar levels of HR to Bga, being observed at nearly 50% of attack sites, with most of the remaining exhibiting penetration resistance (∼45%). These proportions were similar to that elicited by Bgh in the barley cv. P01 which was included to allow comparison with our previously published results (Prats et al., 2006, 2007a). However, whereas in P01 most of the HR developed very rapidly; before any haustoria could be seen in the epidermal cells, Charming, Cory and Orblanche, exhibited a higher and similar proportion of late cell death.

It is widely known that a transient oxidative burst is generated during the execution of the penetration resistance and HR (i.e., Lamb and Dixon, 1997, Piffanelli et al., 2004). In addition H₂O₂ is required for the complex signaling pathway that orchestrates stomatal movements (Pei et al., 2000). We therefore explored whether different H₂O₂ content in the resistant genotypes following inoculation could be related to the different stomatal and photosynthetic dysfunctions observed. Interestingly, most cultivars exhibited reduced H₂O₂ content 2 days after pathogen attack, albeit significant interactions between cultivars and inoculation treatment were observed. Thus, resistant cv. Charming showed no significant differences in H₂O₂ content whereas Cory, Orblanche and the resistant barley P01 showed decreases of approximately 25–30% in H₂O₂ content (Figure 3A). In some cvs this could reflect a diversion of reductants (e.g., NADPH) from a generalized whole tissue pattern of H₂O₂ to a more focused infection site specific oxidative burst. In this context, infection linked H₂O₂ content at 5 d.a.i. was higher than at 2 d.a.i. with no significant interaction between time and cultivars.

To investigate the link between general and localized oxidant/antioxidant metabolism on stomatal locking we explored increasing overall oxidative stress through slightly increase of the light intensity up to 450 μmol m^-2 s^-1. Fungal development and leaf epidermal cell responses under this light regime were assessed to check for any possible change of the resistance responses (Table 2). Overall, mildew development and plant cell responses to the fungi under the moderate high light intensity did not differ from that observed under the normal regime (Prats et al., 2007a, 2010). However, there were slight changes in penetration resistance.

The increase of light intensity increased also significantly (P < 0.001) the overall H₂O₂ content but this was again reduced (P < 0.001) in response to pathogen attack in resistant cultivars, with the exception of Charming (Figure 3B). The susceptible cv. Selma did not exhibit the clear response observed in the resistant cultivars. There were interactions between all factors indicating that the different genotypes responded in a different manner to the light and inoculation treatment. However, overall, under the moderate high light regime we observed a slightly higher reduction of the H₂O₂ content in inoculated respect to control plants (Figure 3B). Assessment of CMS showed no cell damage consequence of oxidative stress under any of the light regimes used and after pathogen attack (Figure 4).

Next, in order to determine whether the changes in the oxidant/antioxidant metabolism could influence further the stomatal and the photosynthetic dysfunctions previously observed in the resistant genotypes, we assessed the effect of the light-induced changes on stomatal conductance and chlorophyll fluorescence parameters.

Following light intensity increase at time of inoculation, we observed a strong interaction regarding the responses of the different genotypes to the increased light and pathogen attack (P < 0.001). Thus, whereas Charming and Selma still did not show any differences in daytime conductance following inoculation, Cory, Orblanche and P01, showed higher decreases in diurnal g₁ after pathogen attack than those observed under the normal light regime (Figure 5A). A similar effect was observed regarding night-time g₁. Thus, there was a significantly (P < 0.001) higher stomatal lock-up in all inoculated resistant cvs under the increased light intensity compared with the normal intensity (Figure 5B). Interestingly there was a strong and significant interaction (P < 0.001) between cultivars, light intensity and inoculation indicating that not all genotypes responded in the same way. Thus, whereas under the normal light intensity the increases of night-time g₁ were approximately of 131.7 and 82.7% for Cory and Orblanche, under the increased light intensity the night-time g₁ increased, respectively, up to 203.1 and 110.8% with respect to non-inoculated plants. In addition Charming, that previously did not increase night-time g₁, showed a slight stomatal lock-up when superimposing pathogen inoculation and higher light intensity. Overall, P01 barley did not respond to the increased light intensity with further nocturnal g₁ increases, although it indeed increased g₁ at the later time points of the time course (Supplementary Figure S1).

Regarding the effect of the moderate light increase in chlorophyll fluorescence parameters, there was an overall slight but significant (P < 0.001) decrease in the Fv/Fm values of the area under the progress curve from 5.73 to 5.58 (Figure 6A) but no significant effect on photoinhibition (Figure 6B). Confirming the previous results, under the moderate high light, pathogen challenge reduced significantly (P < 0.001) Fv/Fm ratios in all cultivars tested. However, there was no significant interaction between inoculation and light intensity, indicating that the overlapping of the increased light with the pathogen challenge did not change further the proportion of the Fv/Fm reduction previously observed. However, there were interactions between cultivars and inoculation and between cultivars and light intensity treatment. Thus, for instance, the resistant cv. Charming, that under the normal light assayed did not show photoinhibition following pathogen attack, showed it under the higher light intensity (Figures 2B and 6B).

Figure 7 shows a scheme of statistically significant correlations found between H₂O₂ content, stomatal responses and chlorophyll fluorescence parameters assessed. It showed a significant correlation between H₂O₂ content and g₁, this correlation being negative with night-time conductance but positive with daytime conductance. This data suggest that lower H₂O₂ content would favor high night-time and low daytime conductance as observed. There was no direct correlation between H₂O₂ content and chlorophyll fluorescence parameters. However, there was a strong correlation between stomatal conductance and photosynthetic electron flow estimates. Thus, night-time g₁ was negatively correlated with Fv/Fm values, with a r² coefficient of c.a 0.8, and positively correlated with q_i. This suggested that high night-time g₁ strongly influenced a reduction in Fv/Fm ratio and increased photoinhibition. On the other hand, daytime g₁ followed an inverse trend suggesting that lower daytime g₁ also affected negatively Fv/Fm values and increased photoinhibition. As expected Fv/Fm values were strongly and negatively correlated with q_i values. Interestingly, when taking into consideration only controls plants growing under the different light intensities a slight but significant positive correlation was observed between daytime and night-time g_s (r² = 0.3, P < 0.01). However, when overlapping the inoculation treatment correlation turned to be negative (r² = -0.55, P < 0.001).