Rice plants fed on by fall armyworm (Spodoptera frugiperda, FAW) caterpillars emit a blend of volatiles dominated by terpenoids. These volatiles were highly attractive to females of the parasitoid Cotesia marginiventris. Microarray analysis identified 196 rice genes whose expression was significantly upregulated by FAW feeding, 18 of which encode metabolic enzymes potentially involved in volatile biosynthesis. Significant induction of expression of seven of the 11 terpene synthase (TPS) genes identified through the microarray experiments was confirmd using real-time RT-PCR. Enzymes encoded by three TPS genes, Os02g02930, Os08g07100 and Os08g04500, were biochemically characterized. Os02g02930 was found to encode a monoterpene synthase producing the single product S-linalool, which is the most abundant volatile emitted from FAW-damaged rice plants. Both Os08g07100 and Os08g04500 were found to encode sesquiterpene synthases, each producing multiple products. These three enzymes are responsible for production of the majority of the terpenes released from FAW-damaged rice plants. In addition to TPS genes, several key genes in the upstream terpenoid pathways were also found to be upregulated by FAW feeding. This paper provides a comprehensive analysis of FAW-induced volatiles and the corresponding volatile biosynthetic genes potentially involved in indirect defense in rice. Evolution of the genetic basis governing volatile terpenoid biosynthesis for indirect defense is discussed.
Many plants release elevated levels of volatile organic compounds upon insect herbivory. Some herbivore-induced plant volatiles function in indirect plant defense by attracting predators and/or parasitoids that are natural enemies of the feeding herbivores (reviewed by Takabayashi and Dicke, 1996; Pichersky and Gershenzon, 2002; Dudareva et al., 2006). Volatile-mediated indirect defense has been reported in more than 20 plant species (Turlings and Wäckers, 2004). In contrast to the relatively well-studied ecological aspects of indirect defense, the molecular and genomic basis of indirect defense, i.e. how many genes are directly involved in the production of herbivory-induced volatiles and how the biosynthesis is regulated, is poorly characterized. Our inadequate understanding is partly due to the complex nature of herbivore-induced volatiles, which are often a mixture of metabolites derived from multiple biosynthetic pathways (Dudareva et al., 2006). The lack of knowledge regarding specific genes for volatile biogenesis is partly responsible for the paucity of mutant plants that differ from wild-type plants only with respect to the production of specific herbivore-induced volatiles, which in turn makes it difficult to assess the relevance of individual volatiles in attracting natural enemies of herbivores (D’Alessandro and Turlings, 2006; Kappers et al., 2005). Therefore, the molecular and genomic basis of herbivore-induced volatile production has been recognized as an important area of indirect defense research that requires much more emphasis (D’Alessandro and Turlings, 2006; Kessler and Baldwin, 2002).
Rice (Oryza sativa spp.) is an enormously useful model for studying the molecular and genomic basis of indirect plant defense because of the availability of a full genome sequence and extensive genetic and genomic resources (International Rice Genome Sequencing Project, 2005). Rice is arguably the most important food crop, as more than half of the world’s population relies on rice as the main food staple (Hirochika et al., 2004). Insects are important pests of rice production. A number of insect species, such as brown planthopper (Nilaparvata lugens, BPH), water weevil (Lissorhoptrus oryzophilus) and fall armyworm (Spodoptera frugiperda, FAW), can cause severe yield reduction (Karban and Chen, 2007). Understanding the natural defense mechanisms in rice should provide better strategies for insect control. For indirect defense, rice is well studied with respect to its interaction with BPH. BPH-infested rice plants were shown to attract Anagrus nilaparvatae, an egg parasitoid of BPH (Lou et al., 2005). The volatiles from rice plants potentially responsible for this attraction were determined to be a complex mixture of compounds that includes terpenoids, indole and methyl salicylate (Lou et al., 2005, 2006). Very little is known about how these volatiles are synthesized and regulated in rice at the molecular level.
Terpenoids, including monoterpenes and sesquiterpenes, are often the most common and diverse group among insect-induced plant volatiles (Pare and Tumlinson, 1999). The ecological relevance of certain terpenoids in indirect defense has been demonstrated by the in vitro administration of pure, synthetic compounds (Kessler and Baldwin, 2001), as well as through genetic manipulation of terpene synthase (TPS) genes (Kappers et al., 2005; Schnee et al., 2006). All terpenoids are synthesized by the action of TPSs, which convert geranyl diphosphate (GPP), farnesyl diphosphate (FPP) and geranylgeranyl diphosphate (GGPP) to monoterpenes (C10), sesquiterpenes (C15) and diterpenes (C20), respectively. Terpene synthases are encoded by large gene families (Bohlmann et al., 1998). Thirty-two TPS genes have been identified in Arabidopsis (Aubourg et al., 2002). The rice genome also contains a large number of TPS genes. Some of the rice TPS genes have been found to encode diterpene synthases. Some of these diterpene synthases are involved in gibberellic acid biosynthesis (Prisic et al., 2004). Others are involved in the production of diterpenes that function as phytoalexins and allelochemicals (Prisic et al., 2004; Wilderman et al., 2004; Xu et al., 2004, 2007). In addition, two jasmonic acid-induced sesquiterpene synthase genes were recently characterized (Cheng et al., 2007). Overexpression of one of these genes in rice enhances the effectiveness of transgenic rice plants in attracting the natural enemies of BPH (Cheng et al., 2007). The molecular and biochemical function for the majority of rice TPS genes, however, remains unknown.
To elucidate the molecular and genomic basis underlying volatile biosynthesis for indirect defense in rice, we investigated the rice plant–FAW–Cotesia marginiventris system. The rice plant–BPH–A. nilaparvatae system has been relatively well studied (Lou et al., 2005, 2006). BPH is a sap-sucking insect. To better understand the effect of a chewing insect, we chose FAW as a model. In addition, the Spodoptera–Cotesia system has been analyzed in numerous plant species in the context of indirect defense. Our study in rice will provide an improved foundation for cross-species comparison of indirect defense.
Production of insect-induced plant volatiles involves multiple biochemical pathways. Genes in these biochemical pathways often belong to gene families with multiple members (Pichersky et al., 2006). Individual members of such families are often highly homologous to each other at the protein sequence level but may have significant substrate and product variations (Pichersky and Gang, 2000). We therefore chose to use an integrated genomics approach that combines targeted metabolic profiling, expression profiling and biochemical analysis to systematically identify candidate genes for production of insect-induced volatiles in rice. In this paper, the biochemical characterization focuses on terpene metabolism.
Rice plants emit a blend of volatiles when damaged by FAW larvae
Headspace collection coupled with GC-MS analysis showed that intact control rice plants emit only trace amounts of volatiles, which included limonene and nonanal (Figure 1). The volatile profile of FAW-damaged rice plants was significantly different from that of control plants. While some volatiles such as limonene and nonanal showed elevated emission levels, a total of 28 volatiles were only detected from FAW-damaged rice plants (Figure 1 and Table S1). Similar to volatiles emitted from other plant species upon herbivory, insect-induced volatiles from rice can be categorized into three major groups: terpenes, shikimic acid-derived metabolites and fatty acid-derived metabolites. The terpene group contained two monoterpenes, S-linalool and limonene, and 19 sesquiterpenes. S-linalool was the most abundant compound in FAW-induced volatiles, with an emission rate of 165.2 ng per plant per hour. Zingiberene was the most abundant sesquiterpene. Two compounds are derived from the shikimate pathway: methyl salicylate and indole. Decanal is a representative of fatty acid-derived volatiles.
The response of Cotesia marginiventris, a generalist parasitic wasp and natural enemy of FAW (Loke and Ashley, 1984), to FAW-induced volatiles emitted from rice plants was tested using a Y-tube two-choice olfactometer bioassay. Naive female parasitic wasps were given a choice between the odor of rice plants damaged by FAW and the odor of untreated control rice plants. Of all wasps tested, 89.7% walked to the arm of the Y-tube that carried the odor of FAW-damaged rice plants. The remaining 10.3% either did not make a choice or chose the arm that carried the odor of untreated rice plants (Figure 2). These results indicate that C. marginiventris could clearly discriminate between the odor from FAW-treated rice plants and that from control rice plants.
Rice genes induced by FAW infestation revealed by microarray analysis
To identify candidate genes for the production of FAW-induced volatiles, we performed microarray experiments to analyze gene expression changes in FAW-damaged rice plants versus control plants. Three biological samples and two technical samples were analyzed. Using a two-fold change as the cut-off value, 196 rice genes (P < 0.01) were found to be significantly upregulated by FAW feeding (Table S2). The gene identities given are based on the TIGR rice genome pseudomolecules (http://www.tigr.org/tdb/e2k1/osa1/), except that the ‘LOC_’ prefix was omitted. The abundance of genes in various functional groups is shown in Figure 3.
Genes encoding enzymes composed the largest group of upregulated genes with known or putative functions (Figure 3). More than half of the genes in the metabolism category appear to be involved in secondary metabolism. These include TPS genes, P450 genes, lipoxygenase genes, methyltransferase genes and BAHD acyltransferase genes (Table S2). Many secondary metabolites produced by the action of the above-mentioned enzymes have roles in plant defense against insects and pathogens (D’Auria, 2006; Qi et al., 2006; Reymond et al., 2004; Schnee et al., 2006; Vellosillo et al., 2007). Transcription factors are the second largest group of induced genes with known function that are annotated (Figure 3). Upregulation of transcription factors represents a general shift in transcriptional regulation in response to insect damage (Reymond et al., 2004). Notably, three WRKY transcription factor genes were upregulated. Some WRKY genes have been shown to be involved in plant defense (Wang et al., 2006). Protein phosphorylation plays an important role in many plant signal transduction pathways involved in plant defense responses (Mishra et al., 2006; Zheng et al., 2006). Eleven genes that contain a kinase domain or LRR repeat were induced by FAW herbivory. In addition, two phosphatase genes were found to be upregulated by FAW herbivory. Membrane transport is important for many physiological processes (Higgins, 1995). Several genes encoding sugar transporters and amino acid transporters were induced by herbivory. Additionally, three ABC transporter genes were induced. The active involvement of ABC transporters in plant defense has been described previously (Campbell et al., 2003). A number of genes encoding proteins with functions in insect and pathogen resistance were also induced by FAW. Four Bowman–Birk serine protease inhibitor genes belong to this group. Twenty-six genes with putative functions that do not belong to any of the above described functional categories were grouped together (Figure 3). Many of these genes, such as those encoding later embryogenesis-abundant proteins and senescence-associated proteins (Table S2), have roles in the general stress response.
Candidate genes involved in the production of FAW-induced rice volatiles
Volatile terpenes are the most abundant FAW-induced volatiles in rice (Figure 2). Ten rice TPS genes were found to be significantly upregulated by FAW feeding (Table 1). Os08g04500 is an additional rice TPS gene that showed 1.5-fold induction by FAW herbivory in the microarray analysis. Because this gene is closely related to Os08g07100 and Os08g07080, we included it in our later analysis.
Table 1. Candidate rice genes for production of FAW-induced volatiles
*Ratio average values of FAW-damaged plants over average values of control plants.
Terpene synthase genes
Putative terpene synthase
Putative terpene synthase
Putative terpene synthase
Putative terpene synthase
Putative terpene synthase
Terpene synthase family
Putative terpene synthase
Putative terpene synthase
Putative terpene synthase
Similar to sesquiterpene synthase 1
Terpene synthase family
SABATH methyltransferase genes
SAM-dependent carboxyl methyltransferase
SAM-dependent carboxyl methyltransferase
Indole pathway genes
Putative indole-3-glycerol phosphate lyase
Lipoxygenase pathway genes
Similar to lipoxygenase
Methyl salicylate and indole are synthesized from the shikimate pathway. Methyl salicylate is a methyl ester of salicylic acid synthesized in plants from salicylic acid by the action of salicylic acid methyltranferase (SAMT; Zubieta et al., 2003). SAMTs isolated from a number of plant species (Chen et al., 2003a; Negre et al., 2002; Pott et al., 2002; Ross et al., 1999) all belong to the protein family SABATH (Chen et al., 2003a). Two rice SABATH genes, Os02g48770 and Os05g01140, were found to be significantly upregulated by FAW herbivory (Table 1). Indole is a product of the tryptophan branch of the shikimate pathway (D’Alessandro et al., 2006; Hansen and Halkier, 2005). Indole-3-glycerol phosphate lyase and anthranilate phosphoribosyl transferase are two key enzymes involved in indole biosynthesis (Hansen and Halkier, 2005). Genes encoding these two enzymes, Os03g58300 and Os03g03450 respectively, were found to be significantly upregulated by FAW feeding (Table 1).
Fatty-acid derived products are generally synthesized from fatty acids such as α-linolenic acid and linoleic acid via their respective hydroperoxides (Noordermeer et al., 2001). The enzymes catalyzing fatty acid oxidation are lipoxygenases (Feussner and Wasternack, 2002; Kessler et al., 2004), and three rice lipoxygenase genes, Os12g37320, Os12g37260 and Os03g52860, were found to be significantly upregulated by FAW herbivory (Table 1).
Confirmation of expression of FAW-induced rice TPS genes
The relevance of some individual terpenes (e.g. linalool) or a group of terpenes (e.g. sesquiterpene products of maize TPS10) in indirect defense has been demonstrated previously (Kessler and Baldwin, 2001; Schnee et al., 2006). Because terpenes are the most abundant among FAW-induced volatiles in rice (Figure 2), we attempted to identify all rice TPS genes involved in producing FAW-induced volatile terpenes. Microarray experiments showed that ten TPS genes were upregulated by FAW feeding (Table 1). False-positive results could be produced in microarray experiments from cross-hybridization (Xu et al., 2001). To confirm the induction of the ten rice TPS genes plus Os08g04500, we performed real-time RT-PCR experiments using gene-specific primers. Upregulation for seven of the 11 TPS genes, including Os02g02930, Os03g22634, Os04g27190, Os04g27670, Os08g07110, Os08g04500 and Os08g07080, was confirmed (Figure 4).
Biochemical characterization of three rice TPS genes
Monoterpenes and dipterpenes are synthesized in plastids and sesquiterpenes are synthesized in the cytosol (Tholl, 2006). Thus, monoterpene synthases and diterpene synthases contain a transit peptide, while sesquiterpene synthases do not (Aubourg et al., 2002; Bohlmann et al., 1998; Tholl, 2006). Protein sequence analysis using TargetP revealed that the proteins encoded by three of the seven TPS genes, Os08g07100, Os08g04500 and Os08g07080, do not contain a transit peptide (Figure 5), which suggests that these three genes encode sesquiterpene synthase. Os08g07080 was determined to be a pseudogene (Figure S1). The remaining four TPS proteins contain a transit peptide and are therefore either monoterpene synthases or diterpene synthases. From phylogenetic analysis, Os0g804500 and Os08g07100 were found to be most closely related to the maize sesquiterpene synthase TPS10 (Figure 6). Os02g02930 clustered with a group of monoterpene synthases that include linalool synthase from Arabidopsis (Figure 6). In this paper, we describe the biochemical activity of Os02g02930, Os08g07100 and Os08g04500.
Full-length cDNAs of Os02g02930, Os08g07100 and Os08g04500 were cloned from FAW-damaged rice leaves using RT-PCR. Escherichia coli-expressed recombinant Os02g02930, Os08g07100 and Os08g04500 proteins were assayed for TPS activity. Many monoterpene synthases have been shown to be more active when expressed as pseudomature proteins (i.e. without the transit peptide) (Chen et al., 2003b; Williams et al., 1998). We produced two constructs with and without the N-terminal sequence containing the transit peptide of Os02g02930 (Figure 5). Both the full-length and truncated form of Os02g02930 catalyzed formation of the single product S-linalool using GPP as substrate (Figure 7), with the truncated form displaying higher activity. Os02g02930 can also use FPP as a substrate to produce nerolidol, but had no activity with GGPP (data not shown).
Escherichia coli-expressed recombinant Os08g07100 catalyzed the formation of 14 sesquiterpenes using FPP as the substrate, with zingiberene and β-sesquiphellandrene as the major products (Figure 8a). Recombinant Os08g04500 catalyzed the formation of five sesquiterpenes using FPP as the substrate, with (E)-β-caryophyllene as the major product (Figure 8b). Os08g07100 can also use GPP as substrate to produce multiple monoterpenes, with β-myrcene as the major product. In contrast, Os08g04500 had no activity with GPP. In addition, neither Os08g07100 nor Os08g04500 showed any activity with GGPP (data not shown).
Expression analysis of terpenoid pathway genes
In plants, the precursors for terpenes, GPP, FPP and GGPP, are synthesized through two separate biochemical pathways (Tholl, 2006): a cytosol-localized mevalonate pathway leading to the formation of FPP, and a plastid-localized non-mevalonate pathway leading to the formation of GPP and GGPP. In the mevalonate pathway, 3-hydroxy-3-methylglutaryl (HMG) CoA reductase (HMGR) catalyzes the first committed step by converting HMG CoA to mevalonic acid (Learned and Fink, 1989). In the non-mevalonate pathway, 1-deoxy-d-xylulose-5-phosphate synthase (DXPS) catalyzes the first committed step by converting d-glyceraldehdye-3-phosphate and pyruvate to 1-deoxy-d-xylulose-5-phosphate (Eisenreich et al., 2001). Microarray analysis showed that one putative HMGR gene, Os05g02990, was upregulated 1.5-fold by FAW herbivory, and one putative DXPS gene, Os07g09190, was upregulated 5.3-fold. In addition to the above two genes, other terpenoid pathway genes that were upregulated by FAW herbivory included a putative 1-deoxy-d-xylulose-5-phosphate reductoisomerase gene Os01g01710 (DXR, 2.1-fold) of the non-mevolonate pathway, and a putative isopentenyl diphosphate isomerase gene Os02g55030 (IPPS, 1.4-fold) and a putative FPP synthase gene Os01g50760 (FPPS, 1.8-fold) of the mevalonate pathway. To confirm expression changes of these pathway genes, we performed real-time RT-PCR analysis. Induction by FAW feeding was verified for all selected pathway genes (Figure 9).
The specificity of FAW-induced rice volatiles for attracting C. marginiventris
In this paper, we demonstrate that FAW-damaged rice plants can attract parasitoid C. marginiventris (Figure 2). Previous studies showed that rice plants infested by BPH can attract the egg parasitoid A. nilaparvatae (Lou et al., 2005, 2006). Both FAW-damaged rice plants and BPH-infested rice plants emit a complex mixture of volatiles including monoterpenes, sesquiterprenes, methyl salicylate, indole and fatty-acid derived metabolites (Figure 1) (Lou et al., 2005, 2006). Despite the similarity in overall quality, some volatiles showed differences in quantity. For example, limonene is one of the major compounds emitted from BPH-infested rice plants (Lou et al., 2006). However, it is a minor constituent in the volatile blend emitted from FAW-damaged rice plants (Figure 1). Such differences in quantity may contribute to the specificity of volatiles for attracting different species of carnivorous insects. In addition, some rice volatiles may be herbivore-specific. The identity of a number of BPH-induced rice volatiles has not been determined (Lou et al., 2006). Further chemical analysis will verify whether there is a difference in the composition of rice volatiles induced by different herbivorous insects and whether such differences in quality play a critical role in determining the specificity of herbivore-induced volatiles in indirect defense.
In addition to their role in attracting natural enemies, herbivory-induced volatiles from rice may have other ecological functions. Recent studies showed that herbivore-induced plant volatiles can act as intra-plant signaling molecules to activate defense responses in distal parts of the same plant being infested by herbivores (Heil and Silva Bueno, 2007). These volatiles could also act as inter-plant signaling molecules to prime the neighboring healthy plants for defense responses (Karban et al., 2000; Ton et al., 2007). In addition, emission of herbivore-induced volatiles can have an impact on plant–pathogen interactions (Kishimoto et al., 2005; Shiojiri et al., 2006). Whether FAW-induced rice volatiles have these aforementioned functions, and, if so, which volatiles are the key signals, remains to be determined.
A genomic view of the molecular basis of FAW-induced volatile production in rice
Global gene expression changes in rice plants in response to FAW feeding were determined using microarray technology. Some patterns of gene expression changes in rice, such as the significant upregulation of genes in defense pathways and transcription factors, are similar to those observed in other herbivore-challenged plant species, such as Arabidopsis (Reymond et al., 2004), poplar (Ralph et al., 2006) and Sorghum (Zhu-Salzman et al., 2004). In this paper, special attention is paid to the identification and characterization of genes involved in the production of herbivore-induced volatiles at the whole-genome level. Based on the chemical identities of FAW-induced rice volatiles and existing knowledge on pathways and genes for the production of same or similar compounds in other plants, 18 genes were identified as candidates involved in production of FAW-induced volatiles (Table 1). These genes belong to three biochemical pathways: the terpenoid pathway, the shikimic acid pathway and the lipoxygenase pathway.
Among the 18 candidate genes, some encode enzymes catalyzing the very last step for production of specific FAW-induced volatiles, and others are involved in upstream pathways to produce intermediates or an immediate substrate for a specific volatile. Our detailed expression analysis of genes in the terpenoid pathway provides evidence that regulation of insect-induced terpenoid production in rice occurs at the pathway level. Key genes in the mevalonate and non-mevalonate pathways were found to be upregulated in rice plants damaged by FAW (Figure 9). This observation is in agreement with a study on floral terpenoid biosynthesis in Antirrhinum majus (snapdragon), in which expression of both TPS and the key pathway genes including DXPS and DXR was found to be upregulated during floral scent biogenesis (Dudareva et al., 2005). Our results, however, are dissimilar to a recent finding obtained from a study with hybrid poplar. When fed on by forest tent caterpillars, hybrid poplar trees emit sesquiterpenes that are implicated in indirect defense (Arimura et al., 2004a). The expression of one HMGR gene and one DXR gene from poplar was not significantly affected by forest tent caterpillar feeding (Arimura et al., 2004a). Whether terpenoid production for indirect defense involves species-specific mechanisms remains to be determined.
Three TPS genes are responsible for the production of the majority of FAW-induced terpenes
In most plant systems where there is volatile-mediated indirect defense, terpenoids are often the most diverse group. Although a number of TPS genes have been isolated from a variety of plant species where they are involved in producing herbivore-induced volatile terpenoids (Arimura et al., 2004a,b; Mercke et al., 2004; Miller et al., 2005; Schnee et al., 2002, 2006), the genomic basis of volatile terpenoid production for indirect defense is not fully understood. Our volatile profiling showed that FAW-damaged rice plants emit more than 20 volatile terpenoids. Surprisingly, three rice TPS genes, Os08g07100, Os08g04500 and Os02g02930, are responsible for production of the majority of these terpenoids. Os08g07100 produces 14 sesquiterpenes and Os08g04500 produces five (Figure 8). While the products of Os08g07100 and Os08g04500 do not overlap, all were present in FAW-induced rice volatiles (Figure 1). Biochemical characterization of Os08g04500 and Os08g07100 has recently been reported, in which Os08g04500 was shown to be a sesquiterpene synthase producing 14 products using FPP as a substrate, three of which were detected in sesquiterpene profiles of rice plants. In the same paper, Os08g07100 was reported to have no activity with FPP (Cheng et al., 2007). These discrepancies may be due to the different rice cultivars used in the two studies. In our study, Os02g02930 was determined to be an S-linalool synthase (Figure 7). Although genetic evidence is still needed, our results strongly suggest that Os08g07100, Os08g04500 and Os02g02930 are the major, if not the only, TPSs that are responsible for FAW-induced terpenoids. This result is similar to observations on biogenesis of Arabidopsis floral terpenoids. Arabidopsis flowers emit more than 20 sesquiterpene terpenoids (Chen et al., 2003b). Biochemical and genetic studies have revealed that the mixture of sesquiterpenes emitted from Arabidopsis flowers is produced by two TPS genes (Chen et al., 2003b; Tholl et al., 2005). Although the volatile terpenoids emitted from insect-damaged plants are often complex, their genetic basis can be rather simple.
Evolution of volatile terpenoid biosynthesis for indirect defense
Volatile terpenoids appear to play a critical role in determining the specificity of plant–carnivore interactions (Kappers et al., 2005; Tholl, 2006). The variation in the composition of terpenoids in herbivory-induced volatiles from different plant species is enormous. For example, Arabidopsis, when damaged by Pieris rapae caterpillars, emit a blend of volatiles that attract Cotesia rubecula, a specialist parasitic wasp (Van Poecke et al., 2001). The blend contains two monoterpenes but no sesquiterpenes (Van Poecke et al., 2001), suggesting that sequiterpenes are not major signals for indirect defense in Arabidopsis, at least in the P. rapae–C. rubecula system. The lack of herbivory-induced sesquiterpenes may be explained by the genetic makeup of Arabidopsis, which contains four sesquiterpene synthase genes, two of which are flower-specific (Chen et al., 2003b; Tholl et al., 2005) and two are expressed only in roots (Ro et al., 2006). In contrast, many plant species, including maize (Schnee et al., 2002, 2006), poplar (Arimura et al., 2004a), cucumber (Mercke et al., 2004) and rice (this study), contain herbivory-inducible sesquiterpene synthase genes. None of the sesquiterpene synthase genes that have been characterized in these species have identical biochemical activities, although some of their products may overlap. Although the vast chemical diversity of terpenoids involved in plant indirect defense is intriguing, it is difficult to determine the evolutionary relationships of the TPS genes that are responsible for producing these terpenoids because little knowledge is available on the TPS family and the function of individual members in these plant species. A phylogenetic analysis of all TPSs from Arabidopsis and the TPSs identified from the rice genome in our preliminary genomic analysis showed that sesquiterpene synthases from Arabidopsis and known sesquiterpene synthases from rice do not form a monophyletic group (F. Chen, unpublished results). This suggests independent evolution of sesquiterpene synthase genes within a specific lineage, which probably has been under selection partly due to the specific interactions of the plant with specific herbivores and carnivores.
In contrast to the above notion, TPS10 from maize and Os08g07100 from rice may have evolved by divergent evolution. Like Os08g07100, TPS10 is an important gene for producing insect-induced sesquiterpenes in maize that can attract parasitic wasps (Schnee et al., 2006). TPS10 catalyzes the formation of a group of six sesquiterpenes (Schnee et al., 2006), all of which are also the products of Os08g07100. However, the proportions of individual sesquiterpenes in product profiles of Os08g07100 and TPS10 are different. For example, the two major products of TPS10, (E)-α-bergamotene and (E)-β-farnesene, are not major products of Os08g07100. Maize TPS10 is most related to Os08g07100 (Figure 6), suggesting that they are probably orthologous genes. If true, this would suggest that lineage-specific evolution of a sesquiterpene synthase gene for indirect defense had occurred since the divergence of rice and maize lineages.
The function of TPS genes in other biological processes should also be taken into consideration. In addition to herbivore-induced emission, volatile terpenoids are often produced by plants in responses to pathogen infection (Keeling and Bohlmann, 2006) or abiotic stresses (Gouinguene and Turlings, 2002). Therefore, one specific TPS gene might have numerous roles and a complex fitness landscape depending on interacting environmental factors. Novel insight into the evolution of terpenoid metabolism, especially in the context of indirect defense, will be revealed as the molecular and genomic basis governing the production of herbivore-induced terpenoids in rice, maize and other plant species is fully elucidated.
Plants, insects and plant treatments
Rice (Oryza sativa ssp. japonica ‘Nipponbare’) seeds were de-hulled and germinated at 30°C in the dark for 5 days. Seedlings were planted at eight plants per 200 ml glass jar, and grown at 26°C with 14 h of light per day for 2 weeks. FAW was used as the model herbivore, and individuals were obtained from a colony maintained in the integrated pest management/biological control laboratory of the Department of Entomology and Plant Pathology at the University of Tennessee. Eggs and newly emerged larvae of FAW were transferred from 3.78 l glass jars (in which adults were contained to mate and lay eggs) to 37.5 ml cups containing approximately 15–20 ml of a pinto bean-based artificial diet as a food source. These diet cups containing larvae were maintained in an incubator (24°C and 16 h light per 8 h dark). For plant treatment, two second-instar FAW larvae were removed from the diet cups and placed on the leaves of a single 2-week-old rice seedling at 16:00 h. After 18 h (when approximately 20% of the leaf area had been consumed), insects were removed and the rice plants were subjected to tissue collection for RNA extraction, volatile collection or the Y-tube olfactometer bioassay.
The solitary endoparasitoid Cotesia marginiventris, a natural enemy of FAW, was used as the model carnivore. Cocoons of C. marginiventris were obtained from the Department of Entomology at the University of Georgia (Tifton, GA, USA). Wasp cocoons were placed in Plexiglas® cages (31.5 cm × 31.5 cm × 41 cm) and kept in the laboratory under ambient light and temperature conditions. Newly emerged adults were provided a food source of 10% honey in water. To rear parasitoids, 5–10 mated female wasps were placed in a Petri dish (100 × 15 mm) with 100–150 young FAW larvae (2–4 days old) for 24 h. Exposed FAW larvae were then transferred to 37.5 ml diet cups and maintained in an incubator (24°C and 16 h light per 8 h dark) until parasitic wasp larvae emerged and formed cocoons. Wasp cocoons were removed and placed in a Petri dish until adult emergence. Newly emerged adults were sexed, and males and females were kept in separate cages, with honey drops and moistened cotton balls. Four-day-old naïve female wasps without prior exposure to plant volatiles, were used for Y-tube olfactometer bioassays.
Two-choice Y-tube olfactometer bioassay
To determine whether FAW-damaged rice plants preferentially attract parasitic wasps, a Y-tube olfactometer bioassay was performed. The Y-tube olfactometer was purchased from Analytic Research Systems (http://www.ars-fla.com). The system consists of a Y-shaped glass body, a pair of odor source adapters with two glass chambers, an insect inlet adapter, and a regulated air delivery system. The dimensions of the olfactometer were 2.8 cm diameter, 15.25 cm main body length, and 8.89 cm branch length. The air flow was maintained at 0.8 l min−1. Inexperienced female wasps (4 days old) were released individually at the base of the olfactometer and observed for 5 min. If a wasp did not make a choice during this period, it was removed from the olfactometer and recorded as a ‘no choice’. Wasps that flew or walked to the end of one of the arms and stayed there for at least 10 sec were recorded as having made a choice of the odor offered through that arm. After ten individuals had been tested, treatment and control arms were swapped to avoid directional bias. The apparatus was washed with acetone and air-dried after each trial. The bioassays were performed between 12:00 and 15:00 h.
Plant volatile collection
Volatiles emitted from FAW-damaged rice plants and control rice plants were collected in an open headspace sampling system (Analytical Research Systems). Eight plants grown in a single glass jar wrapped with aluminum foil were placed in a glass chamber 10 cm in diameter and 30 cm high, with a removable O-ring snap lid with an air outlet port. Charcoal-purified air entered the chamber at a flow rate of 0.8 l min−1 from the top through a Teflon® hose. Volatiles were collected for 4 h by pumping air from the chamber through a SuperQ volatile collection trap (Analytical Research Systems). Volatiles were eluted with 40 μl of CH2Cl2, and 1-octanol was added as an internal standard for quantification. The results of the three replicates were similar, and the results for one replicate are shown in Figure 1.
Plant volatiles and volatile terpenoids from terpene synthase enzyme assays (see below) were analyzed on a Shimadzu 17A gas chromatograph coupled to a Shimadzu (http://www.shimadzu.com) QP5050A quadrupole mass selective detector. Separation was performed on a Restek SHR5XLB column (30 m × 0.25 mm internal diameter × 0.25 μm thickness, Shimadzu). Helium was used as the carrier gas (flow rate of 5 ml min−1), a splitless injection (injection injector temperature 250°C) was used, and a temperature gradient of 5°C min−1 from 40°C (3 min hold) to 240°C was applied. Volatile terpenoids from enzyme assays were also analyzed using a Hewlett-Packard (http://www.hp.com) model 6890 gas chromatograph with the carrier gas helium at 1 ml min−1, splitless injection (injector temperature 220°C), a Chrompack CP-SIL-5 CB-MS column [5% (phenyl)-methylpolysiloxane, 25 m × 0.25 mm internal diameter, 0.25 μm film thickness] (Varian (http://www.varian.com)) and a temperature program from 40°C (3 min hold) at 5°C min−1 to 240°C (3 min hold). The coupled mass spectrometer was a Hewlett-Packard model 5973 with a quadrupole mass selective detector. Because germacrene A is converted to β-elemene at high temperatures in the GC injector, separation of Os08g04500 products was performed with a decreased injector temperature of 150°C. The enantiomers of linalool were separated and identified by GC-MS using a Hydrodex®-β-3P column (2,6-di-O-methyl-3-O-pentyl, 25 m × 0.25 mm internal diamater, Macherey-Nagel, http://www.macherey-nagel.com) and a temperature program from 45°C (1 min hold) at 100°C min−1 to 65°C, and then further at 1°C min−1 to 120°C. Products were identified by comparison of retention times and mass spectra with those of authentic reference compounds obtained from Fluka, Sigma (http://www.sigmaaldrich.com/) and W. König at the Hamburg University.
Total RNA was isolated from appropriate rice tissues using Plant RNA Isolation Reagent (Invitrogen, http://www.invitrogen.com/) according to the manufacturer’s protocol. DNA contamination was removed using an on-column DNase (Qiagen, http://www.qiagen.com/) treatment. Isolated total RNA was used for real-time RT-PCR analysis, gene cloning and microarray experiments.
The rice half genome oligonucleotide array (version 2.0) provided by the Microarray Core Facility at the University of California at Davis was used for global gene expression profiling. mRNA was isolated from total RNA using an Oligotex mRNA kit (Qiagen). One microgram of mRNA was labeled using a Superscript III direct labeling kit (Invitrogen) according to the manufacturer’s instructions. The purified probes were mixed and hybridized with the oligonucleotide arrays using a microarray hybridization kit (Corning, http://www.corning.com) according to the manufacturer’s instructions. Reverse labeling experiments were included to eliminate dye-specific bias. For each sample set of FAW-treated rice versus control, the treated mRNA was first labeled with Cy5 and the control with Cy3. In the reverse experiment, the labeling dyes were swapped. The labeling reactions and dye-swapped microarray hybridizations were performed in parallel. For the reverse labeling experiments, a total of three biological replicates and two technical replicates were included.
After hybridization, the microarray slides were washed and scanned using a GenePix 4000 scanner (Axon Instruments, http://www.axon.com), and the images were processed using GenePix Pro software. The microarray gpr files obtained were analyzed using the R-based open source software Bioconductor (http://www.bioconductor.org), with local background subtraction and Lowess normalization performed for each microarray slide. Linear models and empirical Bayes methods from the limma package of Bioconductor were applied to derive a P-value and mean of log2-based ratio across six slides. The quality of the microarray data was evaluated by examining the MA plot, in which log-transformed ratios of fluorescence intensities [M = log2(R/G)] were plotted against log-transformed multiples of intensities [A = log2(R*G)/2] (Yang and Speed, 2002). The plot suggested that dye-dependent effects were effectively removed after normalization. Changes in gene expression pattern were considered statistically significant at P < 0.01. A ratio cut-off of two and degrees of freedom greater or equal to three were included as quality controls.
Full-length cDNAs of rice TPS genes were cloned from FAW-damaged rice leaves using RT-PCR as previously described (Chen et al., 2003b). The primers used were 5′-ATGGTTTGCCACGTCTTCTCG-3′ (forward) and 5′-CGCCATTATGCATGGACGA-3′ (reverse) for Os02g02930, 5′-ATGTCATCGACACCTGCAGCTAA-3′ (forward) and 5′-TTAAATGCTATATGGCTCAACGTAAA-3′ (reverse) for Os08g07100, 5′-ATGTCGTCGCCACCTGCAGC-3′ (forward) and 5′-TCTTGCCACGATTTTTGGT-3′ (reverse) for Os08g07080, and 5′-ATGGCAACCTCTGTTCCGAGTGTACT-3′ (forward) and 5′-TTAAACAGAGAGGATGTAGATGGAGTGT-3′ (reverse) for Os08g04500. In addition, a forward primer 5′-ATGGCCACCGTCGACCACCT-3′ and the same reverse primer were used to amplify the truncated form of Os02g02930.
Protein expression in E. coli and terpene synthase assay
An E. coli BL21 Codon Plus strain (Invitrogen, http://www.invitrogen.com), transformed with the appropriate expression construct, was used for protein expression. Liquid cultures of the bacteria harboring the expression constructs were grown at 37°C to an attenuance at 600 nm of 0.6. Expression was induced by addition of isopropyl-1-thio-d-galactopyranoside to a final concentration of 1 mm. After 20 h incubation at 18°C, the cells were collected by centrifugation at 10 000 g for 15 min, and disrupted by a 4 × 30 sec treatment with a sonicator in chilled extraction buffer (50 mm Mopso, pH 7.0, with 5 mm MgCl2, 5 mm sodium ascorbate, 0.5 mm PMSF, 5 mm dithiothreitol and 10% v/v glycerol). The cell fragments were removed by centrifugation at 14 000 g, and the supernatant was desalted into assay buffer (10 mm Mopso, pH 7.0, 1 mm dithiothreitol, 10% v/v glycerol) by passage through a Econopac 10DG column (Bio-Rad, http://www.bio-rad.com/). To determine the catalytic activity of the recombinant proteins, enzyme assays were performed in a Teflon®-sealed, screw-capped 1 ml GC glass vial containing 20 μl of the bacterial extract and 80 μl assay buffer with 10 μm substrate (GPP and FPP, respectively), 10 mm MgCl2, 0.2 mm NaWO4 and 0.1 mm NaF. An SPME (solid phase micro-extraction) fiber consisting of 100 μm polydimethylsiloxane (Supelco, http://www.sigmaaldrich.com) was placed into the headspace of the vial for 60 min incubation at 30°C. For analysis of the adsorbed reaction products, the SPME fiber was directly inserted into the injector of the gas chromatograph. To determine the potential catalytic activity of the recombinant proteins with GGPP, enzyme assays were performed in a Teflon®-sealed, screw-capped 1 ml GC glass vial containing 50 μl of the bacterial extract, 10 μl GGPP (440 ng μl−1), 1 μl MgCl2 (1 M), 1 μl NaF (10 mm), 2 μl NaWO4 (10 mm) and 36 μl assay buffer. The assays were overlaid with 100 μl pentane containing 10 ng μl−1 nonyl acetate as an internal standard, and incubated for 3 h at 30°C. To extract the enzyme products, the assays were mixed for 60 sec. The organic phase was then removed and analyzed by GC-MS using a temperature program from 60°C (2 min hold) at 8°C min−1 to 280°C.
We thank Dr John Adamczyk at the USDA-ARS for providing us with rice fall armyworms for preliminary study, Dr Kimberley Gwinn and David Trently for use of GC-MS at the initial stage of this project, and Dr John Ruberson at the University of Georgia for providing starter colonies for C. marginiventris and FAW. We also thank Dr Neal Stewart and Dr Beth Mullin for critical reading of the manuscript. The critical comments and suggestions made by Dr Eran Pichersky and two anonymous reviewers are also highly appreciated. This project was supported by research start-up funds from the University of Tennessee (to FC).