Genes and Pathways Induced in Early Response to Defoliation in Rice Seedlings

How plant gene expression respond to grazing defoliation is critical for plant re-growth, survival, and composition in the natural and dairy farming grassland environments. Rice, with genome sequence available, was used as a model plant to study grazing-induced pathway selections. When seedlings were 18 cm in height, the top 12 cm was removed by simulated grazing. The gene expression activities within 2 to 24 hours in the remaining aboveground tissues were profiled using the Affymetrix Rice GeneChips and RT-qPCR. The seedlings responded to grazing by immediately adjusting their global gene expression, e.g., enhancing anaerobic respiration, starch-to-sugar conversion, sucrose synthesis, and sugar transport. The results suggest that (1) remaining aboveground tissues used anaerobic respiration as an emergency measure for energy/substrates supply; (2) Sink tissues reduced its demand after 2 h; (3) Sucrose synthesis enhancement around the 24th hour is likely driven by shoot re-growth. In total, the expression activity of 466 genes, involved in signal transduction, miRNA regulation, cell wall modification, metabolism, hormone synthesis, and molecule transporters, had been significantly changed at least twofold. These genes and their biochemical pathways identified provide insights into how plants respond to grazing at the molecular physiology level.


Introduction
Grassland grass defoliation can be caused mainly by grazing, although by icy-storms and insects at certain degrees.Considerable variation in plant composition and sustainability of natural or dairy-farming grasslands can be due to grazing defoliation by wild or dairy animals such as cows (Brueland et al., 2003), sheep (Jauregui et al., 2007), deer (Vourch' et al., 2002), rabbits (Crawley, 1990), and kangaroos (Meers and Adams, 2003).Without grazing, grasslands would not be the same as we see now.Grazing influences the abundance, distribution, and persistence of plant species in environment with recurrent grazing defoliation (Donaghy and Fulkerson, 2002;Pecetti and Piano, 2002;Del-Val and Crawley, 2005).For example, rabbit grazing increased the abundance of Anthoxanthum odoratum and the Rumex acetosella grasses and decreased the abundance of Festuca rubra and Agrostis capillaris grasses (Crawley, 1990).Grazing is a multiple-component process that includes wounding, saliva depositing, and defoliation, etc.The sustainability of various plant species to grazing has likely resulted from the adaptation in various physiological and stress-response pathways.Although grass photosynthesis, secondary metabolism, carbohydrates relocation, and re-growth after grazing have been documented (Detling et al., 1979;Nowak and Caldwell, 1984;Gold and Caldwell, 1990;Orodho and Trlica, 1990;Belesky and Hill, 1997;McPherson and Williams, 1998), it is unclear how rapidly and through which biochemical pathways the seedlings achieve the observed growth-level response.
Among the biochemical pathways usually involved in the response to stresses in plants, anaerobic respiration under anoxia is known to play an important role in the generation of both cellular energy and biosynthesis-required substrates (Givan, 1999).Flooding and gas limitation are among the most known anoxic stresses (Kato-Noguchi et al., 2003;Agarwal and Grover, 2006).Anaerobic respiration is required during anoxia but not yet found for other environmental stresses, although anaerobic respiration is suspected to play an important role in non-anoxic tissues (Kursteiner et al., 2003).In the present study, we hypothesized that anaerobic respirations may be activated as an emergencyrescue approach even in leaf tissues in the air when grasses are grazed.
Recent breakthroughs in molecular biology have provided scientists with exciting new tools to pursue eco-genomics research.Microarray technique has been successfully applied to ecology studies (Gibson, 2002) and has identified genes that may be involved in plant-insect interactions (Reymond et al., 2000;Moran et al., 2002;Reymond et al., 2004;Voelckel et al., 2004).After Arabidopsis plants were infested with two types of caterpillars (Pieris rapae and P. brassicae), the plants showed highly similar gene expression profiles (Reymond et al., 2000;Reymond et al., 2004).In Arabidopsis plants, gene expression profile changes induced by caterpillar and thrips showed highest degrees of overlap, followed by thrips and aphid induced changes.However, caterpillar and aphid induced changes hardly overlapped at all (De Vos et al., 2005).Although gene expression profiling has proved to be very informative in studying Arabidopsis-herbivore interaction, to the best of our knowledge, there is no report that evaluates the gene expression profile of the early response in grasses to grazing.Because of the critical importance of gene activities for grass re-growth and survival after grazing, this study aimed to identify the candidate genes and their biochemical 82 Chen et al. pathways using rice, the model monocot plant, mainly because its whole genome has been sequenced (Goff et al., 2002).

Plant materials
Rice (Oryza sativa sp.japonica) cultivar Zhonghua No. 10 was grown in a greenhouse with a 12h light cycle, at 29ºC during the day and 21ºC during the night.The relative humidity was approximately from 50% to 70%.Germinated rice seeds were grown in a plastic container (LxWxH: 9×9×10 cm) filled with rice nursery culture soil (containing 0.15 g each of nitrogen, phosphate, and potassium).Twelve rice seedlings were planted in each plastic container.When the seedlings were at approximately 18 cm in height (about 3-4 leaves per seedling), the top 12 cm part including all the full-size leaf blades was removed by simulated grazing with cow saliva daubing using cotton stick at cut surface immediately after clipping.The shoot meristem was not damaged.The inside main meristem remained unhurt.The remaining aboveground parts (nearly no leaf blades) were collected 2, 6, and 24 h after grazing (Figure 1).For the control seedlings (unclipped), the corresponding parts in ungrazed seedlings were collected.Microarray (2 h) hybridizations and/or RT-qPCR (2, 6 and 24 h) experiments were separately for each of the three independent biological replicates.Each sample was created from pooling at least 20 rice seedlings from two plastic containers.The samples were immediately frozen in liquid nitrogen and stored at -80° for RNA analysis.

RNA isolation, cRNA synthesis, and hybridization to Affymetrix Rice GeneChips
Total RNA was extracted with Trizol reagent (Invitrogen, Carlsbad, CA, USA) and NucleoSpin® RNA Clean-up (CapitalBio company, China).RNA quality was assessed by agarose gel electrophoresis and spectrophotometry.RNA was processed for use on the 51,279-gene Affymetrix Rice Genome GeneChip arrays (Affymetrix, Inc, Santa Clara, CA, USA), according to the manufacturer's protocol.In brief, 10 mg of total RNA was used in the reverse transcription reaction (Ambion MessageAmp kit) to generate the firststrand cDNA.After the second-strand synthesis, the doublestrand cDNA was transcribed in vitro to generate biotinylated cRNA.After purification and fragmentation, the biotinylated cRNA was used for hybridization.The hybridization, washing, staining, and scanning procedures were performed by CapitalBio company (Beijing, China), as described in the Affymetrix technical manual.The biotin-labelled targets were hybridized to the GeneChip Rice Genome Array for 16 h at 45°C with rotation at 60 r.p.m. in an Affymetrix GeneChip Hybridization Oven 640.Washing and staining were carried out in an Affymetrix Fluidics Station 450, following the protocol for the standard format of antibody amplification for eukaryotic targets (EuKGEWS2v4).The processed arrays were scanned with an Agilent GeneArray Scanner (Agilent, Palo Alto, CA, USA).

Data analysis
The hybridization data were analyzed using GeneChip Operating Software (GCOS 1.4) and dChip software (Li andWong, 2001, 2001) (http://www.dchip.org).The scanned images were examined for any visible defects.Satisfactory images were analyzed to generate raw data saved as CEL files using the default settings of GCOS 1.4 from Affymetrix.
A global scaling factor of 500, a normalization value of 1, and default parameter settings for the rice genome array were used.The detection calls (present, absent, or marginal) for the probe sets were made by GCOS.Further analysis was done using dChip.The normalization of all arrays was performed using an invariant set approach.For calculating the expression index of probe sets, we used the perfect match/mismatch model and setup "absent" by truncating the low-expression values to the tenth percentile of the probe set signals.The expression values were log 2 transformed Figure 1.Rice seedlings at 2, 6, and 24 h after simulated grazing.From left to right: the control plants and the grazing plants at 2, 6, and 24 h, respectively.after calculating the expression index.The signal log ratio is the change in the expression level of a transcript between the control and experimental samples, and log 2 (ratio) ≥1 (2-fold change) or log 2 (ratio) ≤-1 (0.5-fold change) was used as cutoff.The Significance Analysis of Microarrays software package (Tusher et al., 2001) was utilized to analyze the three repeated samples between grazing and control using q value ≤ 0.05 and fold-change ≥2 as cutoff.
We used GeneBins to define a generic hierarchical classification.GeneBins can be used to identify the functional categories associated with a set of sequences (e.g.differentially expressed) and thus find the metabolic pathways or other cellular functions up-or down-regulated in microarray experiments (Goffard and Weiller, 2007).The sum of the percentages can be more than 100% as a gene can be assigned to several BINs.

RT-qPCR
The total RNA was treated with DNase I to avoid DNA contamination.One microgram of RNA was reverse transcribed using the Superscript II reverse transcriptase (Invitrogen) with an oligo d(T18) primer according to manufacturer's instructions.PCR experiments were conducted using RealMasterMix (SYBR Green) (Tiangen Biotech, Beijing, China).Reactions were carried out on a Rotor-Gene 3000 multicolor real-time detection system.The following standard thermal profile was used for all PCR experiments: 95 °C for 10 min; 45 cycles of 95 °C for 15 s, 55 °C for 20 s and 68 °C for 20 s; then 72 °C for 10 min.Fluorescence signals were captured at the end of each cycle, and the melting curve analysis was performed from 72 °C to 95 °C.Gene-specific primers were designed using the Primer3 software (Table S1).Data analysis was performed with the Rotor-Gene 6 software.The relative quantification method was used to evaluate quantitative variation between replicates examined.The amplification of actin gene sequence (Os03g0718100) was used as the internal control to normalize all the data.

Gene expression profile in grazing-treated rice seedlings
Within 24 hours after clipping/grazing, the young leaves at the centre of the seedling shoot re-grew to approximately 3 cm longer (Figure 1).When RNA from the aboveground parts harvested 2 h after clipping was used in hybridization analysis against the 51,279 genes on the GeneChip arrays, the expression activity of 466 probe sets changed at least 2 fold (q≤ 0.05) in the grazed seedlings, compared to the corresponding ungrazed seedlings (Table S2).
In the generic hierarchical classification of the 466 genes (Figure 2), the main classes were carbohydrate metabolism (15%), biosynthesis of secondary metabolism (11.8%),
The activity changes for four functional gene groups including signal transduction and transcription factors, hormonal signaling pathways, secondary metabolism and defense, and carbohydrate metabolism genes are demonstrated in Figure 3.
The validation of microarray results was conducted using RT-qPCR analysis of 30 differentially expressed genes covering all the four functional groups and both upregulated and down-regulated genes (Table 1).These genes were chosen, based on biochemistry knowledge, from those involved in flavonoid biosynthetic pathways, carbohydrate metabolism, hormone signaling, signal transduction and transcription factors.The expression activities determined by RT-qPCR data correlate strongly with the changes identified by the microarray experiments (R 2 = 0.93, Figure S1).

Signal transduction and transcription factors
Several receptor-like kinase genes in the signal transduction pathways were activated (Figure 3A), including genes encoding receptor-like serine-threonine protein kinase, and receptor-like protein kinase homolog RK20-1.A number of non-receptor protein kinase genes were also induced, including serine/threonine-protein kinases, GHMP kinases putative ATP-binding protein, calmodulin-binding protein, leucine rich repeat family protein, and cyclin delta-2.Expression changes were also observed in several transcription factors such as AP2, WRKY, MYB, Helixloop-helix, NAC, EF hand, F-box motif, bHLH, and bZIP families.

Hormonal signaling pathways
A cytochrome P450 74A1 gene, a lipoxygenase 2.1 gene, a 12-oxophytodienoate reductase gene, and two SAM dependent carboxyl methyltransferase family protein genes were up-regulated by grazing (Figure 3B).Each of these genes are involved in jasmonic acid (JA) synthesis.The genes up-regulated by grazing include also an ethylene synthesis gene encoding 1-aminocyclopropane-1carboxylate oxidase 1 (ACO 1), two ethylene-responsive protein genes, two AUX/IAA family protein genes, and a gibberellin-20 oxidase-2 gene (Figure 3B).Cell wall synthesis, secondary metabolism, and defense-related genes Among the genes induced by grazing (Figure 3C), there are also those involved in cell wall construction, modification, cell wall expansion, maintenance of cell wall architecture, and cellulose degradation, such as alpha-expansin, endoglucanases, and class III peroxidases.Gene expression of several fungal cell wall polysaccharides degradation enzymes such as glucan endo-1,3-beta-glucosidase genes were considerably induced.Genes encoding several enzymes involved in the flavonoid biosynthetic pathways were significantly induced after grazing (Figure 3C and Figure 5), including two dihydroflavonol-4-reductase genes, one flavonol synthase genes, four naringenin 3-dioxygenase genes, two flavonoid 3-monooxygenase genes, two flavonol 3-O-methyltransferase genes, six flavonol 3-O-glucosyltransferase genes, and two chalcone synthase genes.
An esterase precursor, a protein involved in alkaloid biosynthesis, was also significantly up-regulated (Figure 3C).Several cytochrome family genes which are involved in secondary metabolite biosynthesis were also elevated (Figure 3C).
Grazing treatment enhanced the expressions of several protease inhibitor genes including Bowman-Birk type inhibitors (Figure 3C).Two late embryogenesis abundant genes (Figure 3C), also known as dehydrins were also strongly induced by grazing.Several defense -related enzymes, including regulatory protein NPR1, pathogenesisrelated protein 1, and nematode-resistance protein were also induced by grazing (Figure 3C).

Carbohydrate metabolism
Within two hours after grazing, genes involved in starch synthesis were down-regulated (Figure 3D), including genes encoding glucose-1-phosphate adenylyltransferase, granule-bound starch synthase 1b, 1,4-alpha-glucan branching enzyme, and glucose-6-phosphate/phosphate translocator.The expression of alpha-amylase, which is involved in starch degradation, was strongly activated (Figure 3D).For sucrose synthesis and transportation, genes encoding sucrose phosphate synthase and sugar transporter were up-regulated (Figure 3D).Among the activated genes, there were also pyruvate anaerobic metabolism pathway genes (Figure 3D), including those encoding enzymes of pyruvate dehydrogenase, aldehyde dehydrogenase, and L-lactate dehydrogenase.The metabolism pathways of these differential expression genes are shown in Figure 6.

Interaction between grazing and wounding or insect feeding
Grazing-induced genes in this study were compared to the genes induced by wounding (Lan et al., 2005) and insect feeding (Yuan et al., 2008) in rice.The genes responsive to grazing, wounding, or insect feeding were clustered and shown in Figure 7.A number of genes for metabolic enzymes, signal transduction and transcription factors, and defense-related genes were also induced by insect feeding.Some genes were induced by wounding but not by insect feeding such as peroxidase genes, cysteine synthase gene, proteasome, bZIP transcription factor, Myb-like DNA-binding domain, DnaJ domain containing protein, etc.Some genes encoding WRKY transcription factor, Leucine Rich Repeat family protein, transporters, proteinase inhibitor, hydrolase, oxidoreductase, lipoxygenase, etc. were both induced by wounding and insect feeding.

Discussion
In grasslands, since the wounding, saliva depositing, and defoliation occur during grazing as a joint process, this first study focused on the integrated effects of the simulated grazing.Effects of each separated factor of the simulated grazing will be studied in the follow-up experiments and will be reported elsewhere.
The strong stimulation of grazing immediately activates genes involved in signal transduction, cell wall metabolism, secondary metabolism, biostress defense and carbohydrate metabolism, either related to the preparation of resistance to wounding and potential pathogen attacks or the seedling regrowth.Genes responsive to grazing identified from rice may serve as candidate genes for molecular physiological studies and genetic improvement of other grass species since rice shows great synteny and high sequence similarity with other cereal genomes (Goff et al., 2002).

Hormonal pathways related to regrowth or cellular functions
Two genes encoding AUX/IAA family proteins, OsIAA2 and OsIAA6, were activated by grazing (Figure 3B).It was reported that the expression of Aux/IAA proteins represses the auxin signaling through heterodimerization with auxin response factors (ARFs) (Liscum and Reed, 2002).The binding of these transcription factors to auxin-responsive elements (AuxREs) in promoters of primary auxin-response genes may activate or repress transcription (Hagen and Guilfoyle, 2002).Auxin signaling in plants is involved in resistance to bacterial pathogens (Navarro et al., 2006).
As part of a plant-induced immune response, bacterial pathogen-associated molecular pattern (PAMP) recognition down-regulates auxin signaling in Arabidopsis by targeting auxin receptor transcripts.These results indicate that decreasing plant auxin signaling can increase the resistance to bacterial pathogens (Navarro et al., 2006).These previous reports suggested that the up-regulated expressions of two Aux/IAA proteins (OsIAA2 and OsIAA6) might repress auxin signaling and increase the resistance to potential pathogens induced by wound under stimulated grazing.Wounding can repress auxin actions (Cheong et al., 2002) and cause IAA levels to decline (Thornburg and Li, 1991).Ethylene participates in a variety of defense and abiotic stress responses (Ecker, 1995).In this study, the induction of genes that are involved in the ethylene biosynthetic pathway, such as 1-aminocyclopropane-1-carboxylate oxidase 1 may increase the ethylene level and subsequently activate ethylene response genes such as genes involved in senescence.
The gibberellin-20 oxidase gene up-regulated by grazing (Figure 3B) in rice seedling encodes a gibberellinsynthesis-required key enzyme that catalyzes the conversion of GA53 to GA20 via a three-step oxidation at C-20 of the GA skeleton.The upregulation of this gene suggests that grazing may increase gibberellin biosynthesis within two hours after grazing.

Stress-and defense-related gene expression
Many studies have also indicated that JA and its derivatives play an important role in the defense mechanism against different types of herbivores in Arabidopsis (McConn et al., 1997;Stintzi et al., 2001;Stotz et al., 2002;Reymond et al., 2004;Van Poecke and Dicke, 2004).Grazing-induced insect resistance cross-talks with the salicylic acid and jasmonic acid signal transduction pathways and the pathogeninduced systemic, acquired resistance (Felton et al., 1999).A regulatory protein NPR1 gene and a pathogenesis-related protein PRB1-3 gene were up-regulated in this study (Figure 3C).The expression of these genes have been previously reported to be upregulated after wounding in Arabidopsis (Zhang et al., 1999).NPR1 is a critical component of the salicylic acid (SA)-mediated signal transduction pathway leading to the induction of defense genes, such as the pathogenesis-related (PR-1) gene, and the enhancement of disease resistance (Zhou et al., 2000).Pathogenesisrelated genes were also induced after wounding (Kim et al., 2005) and insects feeding (Figure 7) (Yuan et al., 2008) in rice and in Arabidopsis suspension cell cultures (Guan and Nothnagel, 2004).
Protease inhibitors are often induced by both herbivory and wounding.They are defense-related proteins that occur naturally in a wide range of plants and are considered to be an essential part in the plant direct defense against herbivores.A Bowman-Birk type protease inhibitor and a seed storage protease inhibitor responsive to grazing were also wounding-induced proteins (Figure 7).Most of protease inhibitors identified in this study were also induced by insect feeding (Figure 7).Genes encoding subtilisinchymotrypsin inhibitor and metallothionein-like protein were found to exhibit rapidly and highly induced expression after insect feeding ( Yuan et al., 2005;Hua et al., 2007;Yuan et al., 2008), but down-regulated under grazing at 2h. Genes either shared by multiple stresses or specific to the grazing stress are subject to further investigation.Plant cell walls play critical roles during the life cycle of a plant, including response to environmental stresses.In this study, a number of cell wall-related genes such as expansins and peroxidases were induced by grazing (Figure 3C).These genes are involved in cell wall construction, modification, and maintenance of cell wall architecture.Expansins are involved in cell wall expansion (Cosgrove et al., 2002).Peroxidases play a wide range of functions in removal of peroxides, oxidation of toxic reductants, biosynthesis and degradation of lignin, suberization, and auxin catabolism.The peroxidase gene expression activity was up-regulated in response to environmental stresses such as wounding (Figure 7) and pathogen attacks (Tognolli et al., 2002).
Cytochrome P450s participate in many biochemical pathways, such as those for the synthesis/metabolism of phenylpropanoids, alkaloids, terpenoids, lipids and involve in the protection of plant against environmental stresses.P450 genes identified in this study can also activated by wounding (Yang et al., 2005) and insects feeding (Yang et al., 2006).In the current study, several secondary metabolic pathways, such as flavonoid and terpene pathways, were significantly induced after grazing.Flavonoids, synthesized via the phenylpropanoid pathway, are a diverse group of secondary metabolites with a wide array of biological functions, including the defense against various stresses, such as pathogens, wounding, and UV light damage (Winkel-Shirley, 2002).Terpene synthase gene, indole-3glycerol phosphate lyase gene, lipoxygenase gene, and phenylalanine ammonia-lyase gene induced by this study were involved in biosynthesis of plant volatiles of terpenoids, indole, fatty acid derivatives, and phenylpropanoids/ benzenoids.These genes were also responsive to insect feeding and/or wounding (Figure 7).Volatile terpenes are the most abundant armyworm induced volatile in rice (Yuan et al., 2008).Volatiles induced by cow saliva in plants need future investigation.
Signal transduction pathways target transcription factors and chromatin modifying factors leading to their phosphorylation by protein kinases or dephosphorylation in response to alterations in the environment.These modifications influence transcription factor activities and regulate gene expression networks, which results in appropriate changes in cell behavior (Hunter and Karin, 1992;Hill and Treisman, 1995).Grazing-induced activity changes of several signal transduction pathway genes such as protein kinases and many transcription factors such as AP2 and WRKY were observed in rice seedlings (Figure 3A).Most of these protein kinases and transcription factors were also reported to have response to insects feeding in rice (Figure 7).The C 2 H 2 -type zinc finger proteins represent a large and transcription factor family.These proteins play various roles in developmental processes.They also respond to multiple stresses including wounding, pathogen infection, and abiotic stresses and may be regulators in the ROS scavenging mechanism (Takatsuji, 1999;Davletova et Figure 7. Comparative cluster analysis of the genes induced by biostresses.Data sources: grazing (this study), wounding (Lan et al., 2005), and insect feeding (Yuan et al., 2008(Yuan et al., ). al., 2005;;Vogel et al., 2005).AP2 gene families and WRKY proteins have been reported to be induced by wounding, SA, JA, ethylene, and pathogen attack (Rushton et al., 1996;Eulgem et al., 2000;Maleck et al., 2000;Schenk et al., 2000).
Among grazing induced genes, thirteen genes were found to be the targets of known miRNA (Table 2).These genes involved in signal transduction, transcription factor, and stress defense.It has been reported that miRNAs can be induced by environmental stress (Jones-Rhoades and Bartel, 2004;Sunkar and Zhu, 2004), suggesting that miRNAs may play an important role in regulating plant response to grazing.

Carbohydrate metabolism
The stimulation of genes involved in starch degradation detected in this study can be interpreted by the following hypothesis: the energy and sugar demand in sink tissues, such as the roots, took some time before it slowed down to a balanced level.The tissues met this demand by stimulating starch degradation within the first 2 hours after grazing.Twenty-four hours after grazing, the sucrose transporter was no longer upregulated, which may suggest that the sink demand slowed down and sucrose transporter proteins had reached their balanced requirement.
Starch degradation can be either through starch phosphorylases or amylases, depending on the plants and conditions.Starch phosphorylases are responsible for starch degradation in stored potato tubers (Sowokinos, 2001).However, enhanced alpha-amylase gene expression has been reported in rice cell culture (Ho et al., 2001), water-stressed barley leaves (Jacobsen et al., 1986), virus-infected tobacco leaves (Heitz et al., 1991), and wounded mung bean cotyledons (Koizuka et al., 1995).The microarray-based findings in the rice seedling in this study also indicated that starch degradation in the rice leaves after grazing was through the alpha amylase pathway, not the starch phosphorylase pathways (Figure 6).
The immediate reduction in expression of genes involved in the starch synthesis pathways as well as glucosephosphate transporters suggested that the supply of starch synthesis-required energy and substrates was suddenly reduced by grazing off the mature leaf blades (Figure 6).This result might suggest that the immature leaves at the 3-to 4-leaf stage in non-clipped rice seedlings not only use translocated sucrose for growth but also use some of the over-flowed sucrose from the top leaves to conduct starch synthesis.To the best of our knowledge, this finding has not been previously reported in plants.
Grazing suddenly reduce the stomata uptake of oxygen at the plant level because of the sudden removal of most leaf blades that were rich in stomata on the surfaces.This reduction of oxygen uptake might limit the power of mitochondria to emergently produce sufficient energy for the plants.Therefore, the up-regulation of the anaerobic respiration pathway observed in this study is likely an emergency approach for the seedlings to meet the demand for providing essential energy after the sudden loss of photosynthetic leaves.This result is intriguing and may suggest a positive role of anaerobic respiration for aboveground non-water-stressed leafy tissues.
The lactate dehydrogenase was up-regulated at 2h, then declined at 6h and 24h, while the aldehyde dehydrogenase remained up-regulated during the 24-hour period (Table 1, Figure 4).These results suggest that the aldehyde dehydrogenase pathway is the main anaerobic respiration pathway in rice seedlings under grazing-induced stress, but the relatively more toxic aldehyde dehydrogenase pathway can be also needed when there is a sudden loss of energy supply.This study focuses on identifying genes and pathways in early response to grazing.Further studies may be needed to investigate other factors such as grazing intervals, effects of different animals, saliva types, teeth sharpness, grazing severity, clipping methods, seedling stages, grass species, and grass growing conditions.

Conclusions
This study illustrated that grazing caused the rice seedlings to immediately adjust their gene expression profile within two hours.In the aboveground tissues, gene expression data suggest that starch synthesis in the seedlings slowed down, while starch degradation increased.Sucrose synthesis and transportation immediately enhanced likely due to the demand for sugars in the sink from the remained tissues, and then likely due to the re-growth of the shoot.Anaerobic respiration pathways were enhanced in leafy tissues in the air without obvious anoxic stress, likely in Os.19003 Genes and Pathways Induced by Simulated Grazing Defoliation 91 response to emergency demand for the supply of energy/ substrates.Gene expression data favors the possibility that there is starch synthesis in the immature leaves or the lower leaf portion in non-clipped rice seedlings using some of the over-flowed sucrose from the full functional leaf blades.
The activity of signal transduction and hormone pathways suggest that both wound-related recovery and many abiotic stress pathway-shared genes are stimulated in grazed plants.Many genes induced by simulated grazing are shared by wound response and insect feeding response.This provides us a clue that animal feeding and insects feeding may have some common response mechanism in plants.The genes and pathways with significantly modified expression activities detected in this study but not shared with wounding or insect-feeding may represent grazingspecific genes and may provide insights into the plant specific response to grazing.

Supplemental Material
Figure S1.Correlation between microarray data and RT-qPCR data for the gene expression activities 2h after grazing.Logtransformed fold change (base 2) for 30 differentially expressed genes.The gene IDs, products, and involved pathways and data regarding activity changes are presented in Table 1.
94 Chen et al.Table S1.Genes and primers used in RT-qPCR analysis.

Figure 2 .Figure 3 .
Figure 2. Functional distribution in the GeneBins ontology analysis of differential expression probe sets at 2 h.The percentage represents the proportion of submitted probe sets that have been assigned in the corresponding categories (BIN).

Figure 4 .
Figure 4. Clustering analysis of the 30 selected grazing responsive genes based on the gene activities detected by microarray hybridization (2h) and RT-qPCR (2h, 6h and 24h).Green to bright red: down regulation to up-regulation.

4-Coumaroyl-CoA Naringenin chalcone Chalcone synthase Os.18431.1.S1_at (2.31) Os.11952.1.S1_at (1.49)
PDF !"# "pdfFactory Pro" $#%&'( www.fineprint.cnFigure5.A simplified scheme of the flavonoid biosynthesis pathway.Enzymes encoded by genes that are induced by grazing at 2 h are in red.The expression difference (log2) between the control plants and the grazed plants is listed in brackets.Note: Genes that are significantly induced in response to grazing are shown in red letters.
Figure 6.The genes involved in carbohydrate metabolism induced by grazing in rice seedlings.Red letters highlight the metabolic steps that are strongly up-regulated, whereas blue letters indicate down-regulation.Brackets indicate the expression difference (log2) between the control plants and the grazed plants at 2 h.