Insect Biochemistry and Molecular Biology 50 (2014) 58e67
Contents lists available at ScienceDirect
Insect Biochemistry and Molecular Biology journal homepage: www.elsevier.com/locate/ibmb
Host plant-specific remodeling of midgut physiology in the generalist insect herbivore Trichoplusia ni Marco Herde a, *, Gregg A. Howe a, b a b
Department of Energy-Plant Research Laboratory, East Lansing, MI 48824, USA Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, MI 48824, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 December 2013 Received in revised form 19 March 2014 Accepted 31 March 2014
Species diversity in terrestrial ecosystems is influenced by plant defense compounds that alter the behavior, physiology, and host preference of insect herbivores. Although it is established that insects evolved the ability to detoxify specific allelochemicals, the mechanisms by which polyphagous insects cope with toxic compounds in diverse host plants are not well understood. Here, we used defended and non-defended plant genotypes to study how variation in chemical defense affects midgut responses of the lepidopteran herbivore Trichoplusia ni, which is a pest of a wide variety of native and cultivated plants. The genome-wide midgut transcriptional response of T. ni larvae to glucosinolate-based defenses in the crucifer Arabidopsis thaliana was characterized by strong induction of genes encoding Phase I and II detoxification enzymes. In contrast, the response of T. ni to proteinase inhibitors and other jasmonateregulated defenses in tomato (Solanum lycopersicum) was dominated by changes in the expression of digestive enzymes and, strikingly, concomitant repression of transcripts encoding detoxification enzymes. Unbiased proteomic analyses of T. ni feces demonstrated that tomato defenses remodel the complement of T.ni digestive enzymes, which was associated with increased amounts of serine proteases and decreased lipase protein abundance upon encountering tomato defense chemistry. These collective results indicate that T. ni adjusts its gut physiology to the presence of host plant-specific chemical defenses, and further suggest that plants may exploit this digestive flexibility as a defensive strategy to suppress the production of enzymes that detoxify allelochemicals. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: Herbivore Gut physiology Solanum lycopersicum Trichoplusia ni Arabidopsis thaliana Transcriptome Jasmonate
1. Introduction
Abbreviations: aos, allene oxide synthase; BLAST, Basic Local Alignment Search Tool; COI1, Coronatine Insensitive 1; JA, jasmonate; jai1, jasmonate insensitive 1; LAP-A, leucine amino peptidase A; LC/MS, liquid chromatography/mass spectrometry; NCBI, National Center for Biotechnology Information; nr, non redundant; PI, proteinase inhibitor; qPCR, quantitative polymerase chain reaction; SRA, short read archive; RbcL, ribulose-1,5 bisphosphate carboxylase-oxygenase large subunit; Rbcs, ribulose-1,5 bisphosphate carboxylase-oxygenase small subunit; RNA-seq, Ribonucleic acid sequencing; RuBisCO, ribulose-1,5 bisphosphate carboxylaseoxygenase; TD2, threonine deaminase 2; TGG1, beta-thioglucoside glucohydrolase 1; TGG2, beta-thioglucoside glucohydrolase 2; TnCXE, Trichoplusia ni carboxylesterase; TnCYP, Trichoplusia ni cytochrome P450; TnGST, Trichoplusia ni glutathione S-transferase; TnLP, Trichoplusia ni lipase; TnSP, Trichoplusia ni serine protease; TnUGT, Trichoplusia ni UDP-glycosyltransferase; WT, wildtype. * Corresponding author. Department of Plant Biochemistry, Dahlem Center of Plant Sciences, Freie Universität Berlin, 14195 Berlin, Germany. Tel.: þ49 (0) 30 838 56262. E-mail addresses: [email protected] (M. Herde), [email protected] (G. A. Howe). http://dx.doi.org/10.1016/j.ibmb.2014.03.013 0965-1748/Ó 2014 Elsevier Ltd. All rights reserved.
Insect herbivores and their host plants are engaged in a coevolutionary battle to eat or not be eaten. The extraordinary diversity of plant-insect herbivore associations is shaped by the evolution of novel phytochemical traits that deter herbivory and, in the case of the herbivore, reciprocal coevolution of biochemical and behavioral mechanisms to detoxify or otherwise avoid plant defensive compounds (Berenbaum et al., 1986; Ehrlich and Raven, 1964). These iterative cycles of plant adaptation and herbivore counteradaptation provide a rich opportunity to understand the molecular underpinnings of plant-insect coevolution. In addition to genetic adaptations played out over evolutionary time scales, herbivores adjust their feeding habits and digestive physiology to the nutritional landscape of available food. Many arthropod herbivores, for example, acclimate to plant secondary metabolites by inducing the expression of detoxification enzymes in gut tissues (Brattsten et al., 1977; Dermauw et al., 2013; Li et al., 2007). Likewise, plant proteinase inhibitors (PIs) that inactivate insect digestive proteases also elicit major changes in the
M. Herde, G.A. Howe / Insect Biochemistry and Molecular Biology 50 (2014) 58e67
complement of digestive enzymes secreted into the midgut. The effectiveness of PIs as a mechanism to deplete essential amino acids may also be neutralized by increased tissue consumption by the insect, or by the expression of protease isoforms that are insensitive to dietary PIs (Bolter and Jongsma, 1997; Bown et al., 2004; Jongsma et al., 1995; Zhu-Salzman et al., 2003). Interestingly, however, the capacity of insects to adapt to individual defense compounds (e.g., PIs) may be attenuated by mixtures of allelochemicals (Berenbaum, 1985; Berenbaum and Neal, 1985). Nicotine production in Nicotiana attenuata, for example, prevents compensatory feeding of Spodoptera exigua larvae in response to PIs in leaf tissue (Steppuhn and Baldwin, 2007). Although it is likely that synergy between chemical defenses is a common strategy for plant anti-insect resistance (Berenbaum, 1985; Dyer et al., 2003; Steppuhn and Baldwin, 2007), little is known about how multiple defenses interact to increase plant resistance to herbivory. The ability to perceive and respond to complex mixtures of plant defense compounds is particularly important for polyphagous (socalled generalist) herbivores that encounter a wide variety of plant diets. The lepidopteran herbivore Trichoplusia ni (cabbage looper) provides a suitable model system to investigate this question. T. ni exhibits a feeding preference for hosts within the Brassicaceae but is also well endowed in its ability to physiologically acclimate to a broad range of plants, including many crop species of economic importance (Hill, 1987; Broadway, 1997). In the present study, we describe a genomics-guided approach to address the question of how T. ni midgut physiology is affected by chemical defense systems in Solanum lycopersicum (tomato) and Arabidopsis thaliana. Although anti-insect defenses in both species are induced at the transcriptional level by the plant defense hormone jasmonate (JA) (Bodenhausen and Reymond, 2007; Howe and Jander, 2008), the chemical nature of these defense compounds in tomato and Arabidopsis is highly distinct. The major defense in Arabidopsis is the glucosinolate-myrosinase system in which two myrosinases (TGG1/TGG2) convert glucosinolates to toxic aglycones upon tissue damage (Halkier and Gershenzon, 2006). In tomato, induced defenses are dominated by PIs and various hyperstable enzymes that degrade essential amino acids in the insect gut (Chen et al., 2005; Gonzales-Vigil et al., 2011; Green and Ryan, 1972). Glandular trichomes rich in terpenoids and flavonoids, together with other JAregulated secondary metabolites, provide additional layers of chemical defense in tomato (Chen et al., 2006; Duffey and Felton, 1989; Kang et al., 2010; Kennedy, 2003; Peiffer et al., 2009; Kang et al., 2014). Here, we show that the multi-tiered defense system of tomato is significantly more effective in impeding T. ni growth than glucosinolate-based defenses in Arabidopsis. We use RNA-seq analysis to demonstrate that the midgut transcriptional response of T. ni is highly specific for the defense system of each host plant. Strikingly, changes in the composition of T. ni digestive enzymes in response to JA-regulated defenses in tomato was associated with strong repression of genes encoding Phase I/II detoxification enzymes, which were a major component of T. ni’s response to glucosinolate production in Arabidopsis. This pattern of midgut gene expression suggests that the poor performance of T. ni on tomato may reflect a plant defense strategy in which diet-induced reprogramming of digestive enzymes suppresses the insect’s ability to detoxify secondary metabolites. 2. Material and methods 2.1. Biological material and T. ni feeding assays A. thaliana (accession Col-0) and S. lycopersicum (cv. Castlemart) were used as wildtype (WT) lines for all experiments. Arabidopsis
59
plants were grown in a chamber maintained at 20 C under 10 h light (100 mE m2 s1) and 14 h dark. Tomato was grown under 17 h of light (300 mE m2 s1) at 28 C and 7 h of darkness at 18 C. The tomato jasmonate-insensitive1-1 (jai1-1) mutation corresponds to a deletion in the COI1 gene, which encodes a component of the JA receptor (Katsir et al., 2008). This mutation was initially isolated in the Microtom cultivar and subsequently backcrossed into cv Casltemart (Li et al., 2004; Chen et al., 2005). The jai1-1 line used for T. ni feeding experiments exhibits strong JA-related phenotypes, including loss of production of JA-regulated defense proteins and secondary metabolites (Chen et al., 2005, 2006, 2007), defective development of glandular trichomes and associated trichomederived terpenoids, as well as defects in male and female fertility (Li et al., 2004). The Arabidopsis aos mutant (SALK 017796) is blocked in the biosynthesis of JA and also exhibits strong JA-related phenotypes, including male sterility, increased growth, and loss of wound- and JA-induced defense responses (Park et al., 2002; Schilmiller et al., 2007; Yan et al., 2007; Mafli et al., 2012). The myrosinase-deficient line (tgg1 tgg2) of Arabidopsis that is defective in glucosinolate breakdown was previously described (Barth and Jander, 2006). An aos tgg1 tgg2 triple mutant of Arabidopsis was identified as a male-sterile plant in an F2 population derived from a cross between the tgg1 tgg2 and aos homozygous parents. PCR was used to confirm the presence of homozygous tgg1 and tgg2 T-DNA alleles. Seed for the aos tgg1 tgg2 mutant was propagated by treatment of flowers with methyl-JA (Schilmiller et al., 2007). 2.2. Feeding assays and midgut isolation T. ni eggs were obtained from Benzon Research. Neonate larvae were caged on individual tomato or Arabidopsis plants to prevent movement of insects between plants. Arabidopsis plants (4.5weeks old) were challenged with a single neonate larva caged on individual plants as described (Herde et al., 2013). Four-week-old tomato plants were challenged with three neonate larvae per plant, and each plant was enclosed within a Clear Makrolon Tuffak Lexan Polycarbonate (0.01 inch thick) sheet (Ridout Plastics Co. Inc., San Diego). Larvae were allowed to feed until they obtained a weight of w150 mg, at which time actively feeding individual larva were dissected for harvesting of midgut tissue. Because larvae developed more slowly on defended (i.e., WT) genotypes compared to the defenseless genotypes, larvae were harvested on different days (at the same time of day) after the beginning of the feeding trial in order to obtain the threshold weight of w150 mg. Thus, the weight of larvae reared on the different host genotypes at the time of gut dissection was not significantly different. At the time of dissection, all larvae were in the same instar and had neither emerged from the previous instar nor transitioned into pupation, as indicated by the dark-green appearance of plant material within the midgut. Dissected midguts were washed three times in a modified insect Ringer’s solution (13 mM NaCl, 4.7 mM KCl, 1.9 mM CaCl2) to remove the food bolus. Washed tissue was stored in a solution of RNAlater (Qiagen) at 4 C until further use for RNA isolation. 2.3. Midgut RNA isolation and sequencing Three RNA-seq experiments were performed with midgut RNA isolated from larvae reared on matched WT and mutant plants (Col0/aos; Col-0/tgg; tomato WT/jai1). RNA samples comprising three biological replicates per host genotype were sequenced for each experiment, for a total of 18 sequencing runs (Table S1). For each biological replicate, midgut tissue from 12 individual larvae (reared on a given host genotype) was pooled. Total RNA was isolated using a Qiagen RNAeasy kit and samples were assessed for quality on a
60
M. Herde, G.A. Howe / Insect Biochemistry and Molecular Biology 50 (2014) 58e67
Bioanalyzer. Libraries were prepared using the mRNA-Seq8 Sample Prep Kit (Illumina), according to the manufacturer’s instructions. All libraries were sequenced with a 75 bp paired-end sequencing run performed on the Illumina Genome Analyzer II. Each of the three sequencing experiments was performed in an independent flow cell. The number of 75-mer pair-end reads obtained for each sample ranged between 14.6 and 33.4 106 (Table S1). Raw sequence files are available at the NCBI Short Read Archive (SRA: SRP025999). 2.4. Transcriptome assembly Paired-end reads were filtered for quality with a minimum Sanger score of 27. The FastX toolkit was used to remove bases with a Sanger score lower than 27, adapter sequences, and reads with 99% sequence identity were placed in the same cluster and designated functionally equivalent. A single representative contig from each cluster was chosen on the basis of the best quality score as described below. In an attempt to identify contigs that are exclusively expressed in T. ni grown on susceptible genotypes (aos, tgg1 tgg2, jai1), we created three different assemblies corresponding to each susceptible genotype. Only the longer contig was retained if one or more contigs with 99% sequence identity was found within these three assemblies. A contig was removed from the final assembled list genes if it its sequence was >99% identical with an existing contig in the master assembly (Dataset S1). Contigs whose expression was downregulated by a particular defense and their expression pattern are reported in Dataset S3. Contigs
Contents lists available at ScienceDirect
Insect Biochemistry and Molecular Biology journal homepage: www.elsevier.com/locate/ibmb
Host plant-specific remodeling of midgut physiology in the generalist insect herbivore Trichoplusia ni Marco Herde a, *, Gregg A. Howe a, b a b
Department of Energy-Plant Research Laboratory, East Lansing, MI 48824, USA Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, MI 48824, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 December 2013 Received in revised form 19 March 2014 Accepted 31 March 2014
Species diversity in terrestrial ecosystems is influenced by plant defense compounds that alter the behavior, physiology, and host preference of insect herbivores. Although it is established that insects evolved the ability to detoxify specific allelochemicals, the mechanisms by which polyphagous insects cope with toxic compounds in diverse host plants are not well understood. Here, we used defended and non-defended plant genotypes to study how variation in chemical defense affects midgut responses of the lepidopteran herbivore Trichoplusia ni, which is a pest of a wide variety of native and cultivated plants. The genome-wide midgut transcriptional response of T. ni larvae to glucosinolate-based defenses in the crucifer Arabidopsis thaliana was characterized by strong induction of genes encoding Phase I and II detoxification enzymes. In contrast, the response of T. ni to proteinase inhibitors and other jasmonateregulated defenses in tomato (Solanum lycopersicum) was dominated by changes in the expression of digestive enzymes and, strikingly, concomitant repression of transcripts encoding detoxification enzymes. Unbiased proteomic analyses of T. ni feces demonstrated that tomato defenses remodel the complement of T.ni digestive enzymes, which was associated with increased amounts of serine proteases and decreased lipase protein abundance upon encountering tomato defense chemistry. These collective results indicate that T. ni adjusts its gut physiology to the presence of host plant-specific chemical defenses, and further suggest that plants may exploit this digestive flexibility as a defensive strategy to suppress the production of enzymes that detoxify allelochemicals. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: Herbivore Gut physiology Solanum lycopersicum Trichoplusia ni Arabidopsis thaliana Transcriptome Jasmonate
1. Introduction
Abbreviations: aos, allene oxide synthase; BLAST, Basic Local Alignment Search Tool; COI1, Coronatine Insensitive 1; JA, jasmonate; jai1, jasmonate insensitive 1; LAP-A, leucine amino peptidase A; LC/MS, liquid chromatography/mass spectrometry; NCBI, National Center for Biotechnology Information; nr, non redundant; PI, proteinase inhibitor; qPCR, quantitative polymerase chain reaction; SRA, short read archive; RbcL, ribulose-1,5 bisphosphate carboxylase-oxygenase large subunit; Rbcs, ribulose-1,5 bisphosphate carboxylase-oxygenase small subunit; RNA-seq, Ribonucleic acid sequencing; RuBisCO, ribulose-1,5 bisphosphate carboxylaseoxygenase; TD2, threonine deaminase 2; TGG1, beta-thioglucoside glucohydrolase 1; TGG2, beta-thioglucoside glucohydrolase 2; TnCXE, Trichoplusia ni carboxylesterase; TnCYP, Trichoplusia ni cytochrome P450; TnGST, Trichoplusia ni glutathione S-transferase; TnLP, Trichoplusia ni lipase; TnSP, Trichoplusia ni serine protease; TnUGT, Trichoplusia ni UDP-glycosyltransferase; WT, wildtype. * Corresponding author. Department of Plant Biochemistry, Dahlem Center of Plant Sciences, Freie Universität Berlin, 14195 Berlin, Germany. Tel.: þ49 (0) 30 838 56262. E-mail addresses: [email protected] (M. Herde), [email protected] (G. A. Howe). http://dx.doi.org/10.1016/j.ibmb.2014.03.013 0965-1748/Ó 2014 Elsevier Ltd. All rights reserved.
Insect herbivores and their host plants are engaged in a coevolutionary battle to eat or not be eaten. The extraordinary diversity of plant-insect herbivore associations is shaped by the evolution of novel phytochemical traits that deter herbivory and, in the case of the herbivore, reciprocal coevolution of biochemical and behavioral mechanisms to detoxify or otherwise avoid plant defensive compounds (Berenbaum et al., 1986; Ehrlich and Raven, 1964). These iterative cycles of plant adaptation and herbivore counteradaptation provide a rich opportunity to understand the molecular underpinnings of plant-insect coevolution. In addition to genetic adaptations played out over evolutionary time scales, herbivores adjust their feeding habits and digestive physiology to the nutritional landscape of available food. Many arthropod herbivores, for example, acclimate to plant secondary metabolites by inducing the expression of detoxification enzymes in gut tissues (Brattsten et al., 1977; Dermauw et al., 2013; Li et al., 2007). Likewise, plant proteinase inhibitors (PIs) that inactivate insect digestive proteases also elicit major changes in the
M. Herde, G.A. Howe / Insect Biochemistry and Molecular Biology 50 (2014) 58e67
complement of digestive enzymes secreted into the midgut. The effectiveness of PIs as a mechanism to deplete essential amino acids may also be neutralized by increased tissue consumption by the insect, or by the expression of protease isoforms that are insensitive to dietary PIs (Bolter and Jongsma, 1997; Bown et al., 2004; Jongsma et al., 1995; Zhu-Salzman et al., 2003). Interestingly, however, the capacity of insects to adapt to individual defense compounds (e.g., PIs) may be attenuated by mixtures of allelochemicals (Berenbaum, 1985; Berenbaum and Neal, 1985). Nicotine production in Nicotiana attenuata, for example, prevents compensatory feeding of Spodoptera exigua larvae in response to PIs in leaf tissue (Steppuhn and Baldwin, 2007). Although it is likely that synergy between chemical defenses is a common strategy for plant anti-insect resistance (Berenbaum, 1985; Dyer et al., 2003; Steppuhn and Baldwin, 2007), little is known about how multiple defenses interact to increase plant resistance to herbivory. The ability to perceive and respond to complex mixtures of plant defense compounds is particularly important for polyphagous (socalled generalist) herbivores that encounter a wide variety of plant diets. The lepidopteran herbivore Trichoplusia ni (cabbage looper) provides a suitable model system to investigate this question. T. ni exhibits a feeding preference for hosts within the Brassicaceae but is also well endowed in its ability to physiologically acclimate to a broad range of plants, including many crop species of economic importance (Hill, 1987; Broadway, 1997). In the present study, we describe a genomics-guided approach to address the question of how T. ni midgut physiology is affected by chemical defense systems in Solanum lycopersicum (tomato) and Arabidopsis thaliana. Although anti-insect defenses in both species are induced at the transcriptional level by the plant defense hormone jasmonate (JA) (Bodenhausen and Reymond, 2007; Howe and Jander, 2008), the chemical nature of these defense compounds in tomato and Arabidopsis is highly distinct. The major defense in Arabidopsis is the glucosinolate-myrosinase system in which two myrosinases (TGG1/TGG2) convert glucosinolates to toxic aglycones upon tissue damage (Halkier and Gershenzon, 2006). In tomato, induced defenses are dominated by PIs and various hyperstable enzymes that degrade essential amino acids in the insect gut (Chen et al., 2005; Gonzales-Vigil et al., 2011; Green and Ryan, 1972). Glandular trichomes rich in terpenoids and flavonoids, together with other JAregulated secondary metabolites, provide additional layers of chemical defense in tomato (Chen et al., 2006; Duffey and Felton, 1989; Kang et al., 2010; Kennedy, 2003; Peiffer et al., 2009; Kang et al., 2014). Here, we show that the multi-tiered defense system of tomato is significantly more effective in impeding T. ni growth than glucosinolate-based defenses in Arabidopsis. We use RNA-seq analysis to demonstrate that the midgut transcriptional response of T. ni is highly specific for the defense system of each host plant. Strikingly, changes in the composition of T. ni digestive enzymes in response to JA-regulated defenses in tomato was associated with strong repression of genes encoding Phase I/II detoxification enzymes, which were a major component of T. ni’s response to glucosinolate production in Arabidopsis. This pattern of midgut gene expression suggests that the poor performance of T. ni on tomato may reflect a plant defense strategy in which diet-induced reprogramming of digestive enzymes suppresses the insect’s ability to detoxify secondary metabolites. 2. Material and methods 2.1. Biological material and T. ni feeding assays A. thaliana (accession Col-0) and S. lycopersicum (cv. Castlemart) were used as wildtype (WT) lines for all experiments. Arabidopsis
59
plants were grown in a chamber maintained at 20 C under 10 h light (100 mE m2 s1) and 14 h dark. Tomato was grown under 17 h of light (300 mE m2 s1) at 28 C and 7 h of darkness at 18 C. The tomato jasmonate-insensitive1-1 (jai1-1) mutation corresponds to a deletion in the COI1 gene, which encodes a component of the JA receptor (Katsir et al., 2008). This mutation was initially isolated in the Microtom cultivar and subsequently backcrossed into cv Casltemart (Li et al., 2004; Chen et al., 2005). The jai1-1 line used for T. ni feeding experiments exhibits strong JA-related phenotypes, including loss of production of JA-regulated defense proteins and secondary metabolites (Chen et al., 2005, 2006, 2007), defective development of glandular trichomes and associated trichomederived terpenoids, as well as defects in male and female fertility (Li et al., 2004). The Arabidopsis aos mutant (SALK 017796) is blocked in the biosynthesis of JA and also exhibits strong JA-related phenotypes, including male sterility, increased growth, and loss of wound- and JA-induced defense responses (Park et al., 2002; Schilmiller et al., 2007; Yan et al., 2007; Mafli et al., 2012). The myrosinase-deficient line (tgg1 tgg2) of Arabidopsis that is defective in glucosinolate breakdown was previously described (Barth and Jander, 2006). An aos tgg1 tgg2 triple mutant of Arabidopsis was identified as a male-sterile plant in an F2 population derived from a cross between the tgg1 tgg2 and aos homozygous parents. PCR was used to confirm the presence of homozygous tgg1 and tgg2 T-DNA alleles. Seed for the aos tgg1 tgg2 mutant was propagated by treatment of flowers with methyl-JA (Schilmiller et al., 2007). 2.2. Feeding assays and midgut isolation T. ni eggs were obtained from Benzon Research. Neonate larvae were caged on individual tomato or Arabidopsis plants to prevent movement of insects between plants. Arabidopsis plants (4.5weeks old) were challenged with a single neonate larva caged on individual plants as described (Herde et al., 2013). Four-week-old tomato plants were challenged with three neonate larvae per plant, and each plant was enclosed within a Clear Makrolon Tuffak Lexan Polycarbonate (0.01 inch thick) sheet (Ridout Plastics Co. Inc., San Diego). Larvae were allowed to feed until they obtained a weight of w150 mg, at which time actively feeding individual larva were dissected for harvesting of midgut tissue. Because larvae developed more slowly on defended (i.e., WT) genotypes compared to the defenseless genotypes, larvae were harvested on different days (at the same time of day) after the beginning of the feeding trial in order to obtain the threshold weight of w150 mg. Thus, the weight of larvae reared on the different host genotypes at the time of gut dissection was not significantly different. At the time of dissection, all larvae were in the same instar and had neither emerged from the previous instar nor transitioned into pupation, as indicated by the dark-green appearance of plant material within the midgut. Dissected midguts were washed three times in a modified insect Ringer’s solution (13 mM NaCl, 4.7 mM KCl, 1.9 mM CaCl2) to remove the food bolus. Washed tissue was stored in a solution of RNAlater (Qiagen) at 4 C until further use for RNA isolation. 2.3. Midgut RNA isolation and sequencing Three RNA-seq experiments were performed with midgut RNA isolated from larvae reared on matched WT and mutant plants (Col0/aos; Col-0/tgg; tomato WT/jai1). RNA samples comprising three biological replicates per host genotype were sequenced for each experiment, for a total of 18 sequencing runs (Table S1). For each biological replicate, midgut tissue from 12 individual larvae (reared on a given host genotype) was pooled. Total RNA was isolated using a Qiagen RNAeasy kit and samples were assessed for quality on a
60
M. Herde, G.A. Howe / Insect Biochemistry and Molecular Biology 50 (2014) 58e67
Bioanalyzer. Libraries were prepared using the mRNA-Seq8 Sample Prep Kit (Illumina), according to the manufacturer’s instructions. All libraries were sequenced with a 75 bp paired-end sequencing run performed on the Illumina Genome Analyzer II. Each of the three sequencing experiments was performed in an independent flow cell. The number of 75-mer pair-end reads obtained for each sample ranged between 14.6 and 33.4 106 (Table S1). Raw sequence files are available at the NCBI Short Read Archive (SRA: SRP025999). 2.4. Transcriptome assembly Paired-end reads were filtered for quality with a minimum Sanger score of 27. The FastX toolkit was used to remove bases with a Sanger score lower than 27, adapter sequences, and reads with 99% sequence identity were placed in the same cluster and designated functionally equivalent. A single representative contig from each cluster was chosen on the basis of the best quality score as described below. In an attempt to identify contigs that are exclusively expressed in T. ni grown on susceptible genotypes (aos, tgg1 tgg2, jai1), we created three different assemblies corresponding to each susceptible genotype. Only the longer contig was retained if one or more contigs with 99% sequence identity was found within these three assemblies. A contig was removed from the final assembled list genes if it its sequence was >99% identical with an existing contig in the master assembly (Dataset S1). Contigs whose expression was downregulated by a particular defense and their expression pattern are reported in Dataset S3. Contigs
