Using a modified seedbox screening technique [13] ten-day-old seedlings were uniformly infested with 3–4 second-instar BPH nymphs with free choice to settle on their preferred host. Hopperburn symptoms were observed at different intervals (Table 1). Following infestation, continuous feeding by growing second generation BPH nymphs caused wilting of the seedlings, leading to hopperburn (browning of stem and leaves) symptoms first on D1131, followed by IR64 and finally on D518 (Figure 1). Early on infestation (T1 and T2), damage symptoms were not detected on infested plants. This is likely due to a low number of nymphs that were initially released on plants, which did not cause enough damage and plants were able to overcome low level of insect stress. The difference in phenotype among the mutants and IR64 was more obvious at T3 and T4 (28 DAI and 34 DAI, respectively). The average leaf damage rate was recorded on a modified 1–9 scale (1 = resistant, 9 = highly susceptible) [23]. Leaf damage at T3 was lowest for D518 (3.5), intermediate for IR64 (5.2), and highest for D1131 (6.8).

The proteome response of wild type IR64 during BPH infestation over 5-week period after infestation was first studied. This is a condition that simulates natural infestation on rice under field conditions. Among 1500 protein spots visualized on silver stained 2-D polyacrylamide gel (3–10 pH), 65 protein spots were found altered (p < 0.001) with BPH infestation (Figure 2) at pI 4–7, whereas the remaining spots were detected with pI > 7.0 (figure not shown). Mixed models ANOVA using BPH induced proteins in the control and BPH infested IR64 treatments shows that a larger cohort of these proteins was changed only during T3 and T4 stage, indicating higher stress response at the later stage (Figure 3). Since the effect of BPH stress was more evident at T3 (28 DAI), we compared the protein abundance at T3 in isolation using control and BPH infested plants. Comparison of protein abundance (spot volume of infested/control at T3 showed that a total of 36 proteins increased >1.5 fold while 29 proteins showed <0.5 fold decrease with BPH infestation. The protein abundance showed a reduction through time as the plants entered senescence at T4 (34 DAI).

Based on matrix assisted laser desorption ionization time-of-flight (MALDI-TOF) and quadrupole time-of-flight (Q-TOF) mass spectrometry, the identity of 52 proteins was generated; 27 proteins with increased abundance and 25 proteins with decreased abundance (Table 2). Peptide mass of the remaining 13 of total 65 protein spots did not match with any known proteins in the NCBI protein database. These BPH responsive proteins were classified into 11 functional categories [24] of which 39% belonged to energy category, whereas 16% were stress and plant defense related. The identity and function of 20% of BPH responsive protein spots in IR64 are not known. In general, the dominating category of BPH affected the functional group involved photosynthesis and metabolism related proteins. BPH induced proteins related to photosynthetic processes were identified as Rubisco activase (Ract), various rubisco large subunits, ferredoxin [(flavodoxin-NADP(H)] reductase (FNR) and oxygen evolving enhancer protein 3 (OEE3) in IR64. This indicates that photosynthesis was one of the common responses to BPH infestation. Likewise, oxidative stress response proteins such as ascorbate peroxidase (APX), GSH dependent dehydro-ascorbate reductase, and CuZn superoxide dismutase (SOD) were identified as BPH stress response proteins in IR64. Abundance of multiple spots of ribulose bisphosphate carboxylase large (rubisco, rbcl) subunits (4 spots), ascorbate peroxidase (APX) (5 spots), unnamed protein (2 spots), oxygen evolving enhancer protein 3 (2 spots), and enolase (2 spots) may represent post translational modifications during BPH stress or presence of multiple gene copies of these proteins in rice.

Abundance of several oxidative stress-response proteins, drought (#LD7) and two salt stress (#23 and #27) response proteins was altered with BPH stress as observed at T3 (Figure 4). Repeated measures analysis with individual spot abundance in control and BPH infested plants indicated that the spots #13, #14 and #28 were consistently increased (p < 0.05) with BPH stress over time whereas spots #12, #21, #23, #49 and #LD7 showed significant decrease as compared to the control (p < 0.05) over time (Figure 4). Although the protein “#LD7” (S-like RNase) was less changed with BPH infestation as compared to the control plants, protein levels increased through time during infestation (p < 0.05). The abundance of protein spots #23 and #27, which showed similarity to salt stress root protein “RS1” (Gi34904362) [25], also changed differentially with BPH infestation at different times, particularly at T3 and T4 (p < 0.05). At T3, the abundance of protein #23 decreased > 2 times (p < 0.05) than in control plants, while the protein spot #27 which remained suppressed in control plants, was however more abundant with infestation through all four time points (Figure 4).

The abundance of 16 protein spots (spot #20, #32, #38, #39, #−39a, #40, #43, #45, #47, #50, #53, #57, #59–61 and #64) was observed (Figure 5) at different time points only in BPH infested plants. Interestingly, a change in the protein levels of the spot #20 (proteophosphoglycan, PPG), spot #50 and spot #64 (EFTu1) was also observed at T1 and or T2 indicating that these proteins accumulate in IR64 during early BPH-induced stress (Figure 5). Induction of proteophosphoglycan (#20), putative 1,4-benzoquinone reductase (#47), Putative defective chloroplasts and leaves (DCL) protein (#53), Putative FH protein NFH2 (#59), hypothetical protein P0677B10.12 (#60), putative oxygen evolving enhancer protein 3-1, chloroplast precursor (#61) and chloroplast translation elongation factor Tu1 (#64) have not been reported earlier in BPH-rice interactions and may have role in rice resistance to BPH infestation. The highest levels of these proteins was observed with spot #64 (spot density = 12.58 ± 1.52) at T3 as compared to the abundance of other proteins whereas the spot #39 (0.20 ± 0.06) was least induced with BPH infestation. The abundance of all these BPH induced proteins, except spots #32 (enolase), #43 (unknown), and #47 (putative 1,4-benzoquinone reductase) showed declining trend at T4 as the plants started to senesce.

A few proteins identified in this study were also non-rice proteins (#38, #39, #39a and #68). Spot #38 was identified as “ATP-dependent DNA helicase UvrD (Shewanella denitrificans OS217)”. Also #39a with molecular weight of 97.5 kDa showed similarities to leech derived protease inhibitor protein (LDPI) and #39 showed a similarity with “Vitellogenin” from BPH. Spot #68 matched to “Succinyl-CoA ligase [ADP-forming] subunit beta OS = Mesorhizobium sp. (strain BNC1)”. These proteins could be either BPH associated proteins injected into rice sheath during feeding or environmental contaminants that colonized BPH wounded rice plants.

To understand the defense response of rice against BPH infestation, the protein levels in control and BPH infested IR64 were compared with gain (D518) and loss of resistance (D1131) mutants of IR64 at T3. These mutants were previously identified during a screening of chemically generated IR64 mutants against BPH using a modified seedbox screening technique [23]. Field performance of these mutants did not show compromise in agronomical traits due to mutations.

By comparing the protein abundance (protein volume in BPH infested/control) between IR64 and the mutants, 22 proteins were identified that showed differential abundance (Table 3). Ten proteins were altered in a unique manner in the susceptible mutant (D1131) when compared to IR64 and resistant mutant (D518). Among these proteins, eight proteins (spot #7, #43, #45, #47, #53, #57, #59, #B) were significantly increased (p < 0.05) whereas two proteins (#21 and #32) were highly decreased in D1131 (p < 0.05) than in D518 and IR64. The protein #27 generally increased with BPH stress, however showed little change in D1131 whereas comparatively, the abundance of this protein was in greater quantities in IR64 and D518 following BPH infestation (p = 0.018). In contrast, twelve proteins were linked to a D518 related response to BPH (Table 3). Three proteins (#35; #38 and #40) were significantly (p < 0.05) reduced in D518 during BPH stress; another five proteins (#9, #21; #29, #30 and #31) were least affected in D518 whereas the same proteins were decreased in IR64 and D1131 (p < 0.05). Similarly, two proteins (#27 and #LD7) showed higher levels (p < 0.05) in D518 as compared to IR64 and D1131 The abundance of protein #64 was higher than D1131 but this difference was not significant than IR64. Two proteins (#8 and #41) though increased in abundance, but to a lesser extent (p < 0.05) in D518 compared to IR64 and D1131. The abundance of spot “LD7” exceptionally increased in D518 but reduced in IR64 and D1131 with BPH stress. When compared over time after BPH infestation, the protein spot #LD7 remained unchanged at T1 and T2, increased to greater quantities at T3 and decreased thereafter at T4 in D518.

From the biplot analysis, it is clear that the variation in the levels of specific proteins was associated with specific factors. For example, the variation in the abundance of proteins #12, #29, #23, #35, #48 and #49 were associated with the control. Furthermore, all proteins whose eigenvectors are travelling in the same direction as the thick eigenvectors, are associated with that factor. Likewise, #LD7 was associated with D518 and to a lesser extent IR64. Several proteins including #64, #28, #13, #32 were associated with the “BPH infested” treatment. Variability in the protein 11 (unknown protein) was the only one clearly associated with D1130 (Figure 6a). Similarly, broader random experimental factors can be included to evaluate responses to covariates (Figure 6b). Variation in the abundance of proteins located on the right side of the biplot (Figure 6b) indicates that change in protein levels was associated with the progression of “time” and the presence of “BPH infestation”, whereas proteins on the left side of the biplot were associated with the lack of treatment or control as well as the earlier time points. Interestingly, levels of protein #a (unknown protein) was strongly associated with D518 and conversely, variation in the induction of protein #4 (Putative ribosomal protein s12) was associated to a lesser degree with D1131 and to a greater degree BPH infestation. IR64 did not explain a significant proportion of variation in the protein abundance data.

Brown planthopper (BPH), Nilaparvata lugens (Stal) populations were continuously maintained on the susceptible variety “Taichung Native 1” (TN1) at the International Rice Research Institute (IRRI), Los Baños, Philippines. The parent BPH population was collected from rice fields around IRRI, Laguna. Gravid females were used to get a synchronized hopper stage for infestation.

The Indica rice cultivar IR64 along with its two mutants, i.e., D518 (gain-of resistance) and D1131 (loss-of-resistance) generated through the chemical and radioactive mutagenesis of IR64 [23] were used for this study. The mutant D518 shows enhanced resistance during BPH infestation whereas D1131 is susceptible. The mutants were used following six generations of selfing and after confirmed field evaluation showing absence of any deleterious effect of mutations. The field trials of these mutants revealed no agronomical differences from IR64 [23] whereas analysis using IR64 specific molecular markers suggested that the mutants are essentially near-isogenic (unpublished data). The experimental plants were maintained under greenhouse conditions at 28 ± 2 °C with a photoperiod of 16 h day/8 h night cycle.

Phenotypic response of IR64 to BPH infestation was determined using a modified seedbox screening technique under greenhouse conditions [13]. This technique provides free choice to BPH nymphs to colonize the plants in the seedbox. Briefly, pre-germinated seeds were sown in seedboxes (45 cm × 35 cm ×15 cm) containing heat-sterilized soil in six equally spaced rows (two rows for each entry) and 15 seedlings per row. Each row (mutant or wild type) was randomized within a seedbox and replicated in three independent seedboxes. Ten-day-old seedlings were uniformly infested with 3–4 second-instar BPH nymphs per plant and allowed to settle on plants of their choice. Hopperburn symptoms were observed 34 days after infestation (DAI).

Since phenotypic response of IR64 differed with two mutants, a no-choice setup was planned to allow equal number of BPH stress to feed on these genotypes. Fifteen seeds of mutants or wild type plants were sown in individual nine inch circular pots using three technical and three biological replicates. The seedlings were maintained in the greenhouse and before infestation with three nymphs per plant 10 days after sowing, pots were randomized between entries and covered with mylar cage and infested. Control plants were not infested but were covered with mylar cage and arranged randomly. For protein extraction, the plants from three experimental and biological replicates were sampled at four time points after infestation. For the first sample (T1), plant tissue was harvested 2 DAI when the infested nymphs were still in 3^rd–4^th instar stage; the second sampling (T2) was done at 13 DAI when the majority of nymphs were at the adult stage; the third sampling (T3) was performed 28 DAI following the emergence of second generation nymphs; the fourth sampling (T4) was done when the susceptible mutant (D1131) started wilting (34 DAI). For protein analysis, a 10 cm sample above ground portion of leaf sheath was harvested and stored immediately in liquid nitrogen. For control, plants were harvested at same time points using non-infested plants.

Protein extraction. Total leaf sheath proteins were extracted in a precipitation solution (10% Trichloroacetic acid, 89.93% Acetone, 0.07% Dithiothreitol) using a modified method of Damerval et al.[54]. The protein concentration was determined using a Protein-Assay-Kit (Bio-Rad) following the manufacturer’s instructions.

Protein separation and image analysis of 2D Gels. Gel electrophoresis was performed using non-linear (NL) 18-cm IPG strips with pH 4–7 and 3–10 (Amersham Pharmacia Biotech, Uppsala, Sweden). The IPG strips were rehydrated overnight in 350 μL of rehydration buffer and 100 μg of sample protein. The isoelectric focusing (IEF) of proteins was performed on a Multiphor II Electrophoresis unit (Amersham Biosciences) at 20 °C with constant 200 V for the first hour, 500 V for next 2 h and finally 16 h at 2950 V. Proteins from DTT/IAA equilibrated IEF strips were separated on 15% sodium dodecyl sulfate (SDS) polyacrylamide gels using a Protean-II Multi cell (Bio Rad: Hercules, CA, USA) at 4 °C.

The gels were stained with silver nitrate (Sigma Aldrich) for scanning or spot quantification analysis whereas coomassie blue stain (Sigma Aldrich) was used for protein identification with mass spectrometry using standard staining protocols. The gels were scanned with a GS-800 Calibrated Densitometer (Bio-Rad) at a resolution of 600. For spot detection, protein quantification and spot analysis, Melanie-3 image analysis software (GeneBio, Geneva, Switzerland) was used. Spot detection parameters were as follows: number of smooths, 5; Laplacian threshold, 5; partial threshold, 1; saturation, 90; peakness increase, 100; minimum perimeter, 10. The Melanie software automatically normalized the spot intensity (the relative volume) i.e., the volume divided by the total volume over the whole image (Melanie 3 user manual). The percent spot volume detected by software was used to match spots for intensity differences and predict BPH induced proteins.. The protein spots were categorized as BPH altered (increased or decreased in abundance) if protein abundance in a rice line increased or decreased with BPH infestation compared to mean control value. Abundance ratio (protein volume in infested plants/control plants) was compared with control at a time point to determine fold change in proteins. An arbitrary cutoff was used to express highly altered proteins [>1.5 (increased), <0.5 (decreased) or >0.5 and <1.5 (least altered)].

The proteins spots from Coomassie Brilliant Blue (G-250) stained gels were manually excised using a sterilized scalpel and submitted to the Australian Proteome Analysis Facility (APAF) Macquarie University, Sydney, Australia [55] for characterization. Protein samples were analyzed with matrix assisted laser desorption ionization time of flight (MALDI-TOF) mass spectrometry using a Micromass Tofspec time-of-flight mass spectrometer (Micromass, Manchester, UK) at APAF following standard procedures. If proteins could not be identified with MALDI-TOF, a further analysis was performed on Q-TOF LCMS. For protein identification, peak lists were used and peptide masses were searched against SWISS-PROT and NCBInr databases using the Mascot search engine [56] supported by Matrix Science Ltd., London. In MS/MS Ion Search, following parameters were used for database queries on monoisotopic peptide masses using the Viridiplantae and Oryza sativa as taxonomic categories; peptide mass tolerance of 150 ppm; fragment mass tolerance: ±0.6 Da; variable modifications: Oxidation (M), Propionamide (C); and the maximum number of missed tryptic cleavages, 1. Peptide masses that yielded a significant ion score (p < 0.05) were considered positively identified.

Data analysis was performed with Statistical Analysis Software (SAS) (Version 9.1) and JMP-IN (Version 5.1) (SAS Institute, Cary, NC, USA) using protein abundance values in control and BPH infested plants of three genotypes (wild type IR64 and two mutants) and compared at each respective time point. Protein abundance ratio in relation to each control group (IR64 or mutants) was calculated by dividing the spot abundance in the BPH infested plants by the mean spot abundance of the control plants and expressed as fold change with statistical significance at p-value lower than 0.05. A 2-way ANOVA was used to compare the protein abundance between IR64 and the mutants and the means were separated with the Tukey’s HSD multiple means comparison test (p < 0.05). Ordination statistics were performed on protein abundance and genotypes to measure interactions between the BPH and rice proteins (Canoco V.4.5) [57]. Initially, detrended correspondence analysis (DCA) was performed to measure eigenvector length of expressed proteins variables (control, infested) [26]. Redundancy analyses (RDA) were performed and the significance of the first two axes, as well all four axes, were tested using a Monte Carlo test with 1000 permutations in reduced space. The reason RDA was chosen in this particular instance rather than another multivariate method, is that the variable data showed linear responses as opposed to unimodal responses. Multivariate biplots allow one to explore trends through numerical data analysis above and beyond simple hypothesis testing. Where relationships and covariation between variables is not evident with simple univariate statistics, multivariate methods clearly show the abundance of specific proteins as variables in relation to experimental factors. In this case it is clear that specific proteins covary with specific treatments, and the treatments themselves also show covariation.