Microarray Analysis of Gene Regulations and Potential Association with Acephate-Resistance and Fitness Cost in Lygus lineolaris

The tarnished plant bug has become increasingly resistant to organophosphates in recent years. To better understand acephate resistance mechanisms, biological, biochemical, and molecular experiments were systematically conducted with susceptible (LLS) and acephate-selected (LLR) strains. Selection of a field population with acephate significantly increased resistance ratio to 5.9-fold, coupled with a significant increase of esterase activities by 2-fold. Microarray analysis of 6,688 genes revealed 329 up- and 333 down-regulated (≥2-fold) genes in LLR. Six esterase, three P450, and one glutathione S-transferase genes were significantly up-regulated, and no such genes were down-regulated in LLR. All vitellogenin and eggshell protein genes were significantly down-regulated in LLR. Thirteen protease genes were significantly down-regulated and only 3 were up-regulated in LLR. More than twice the number of catalysis genes and more than 3.6-fold of metabolic genes were up-regulated, respectively, as compared to those down-regulated with the same molecular and biological functions. The large portion of metabolic or catalysis genes with significant up-regulations indicated a substantial increase of metabolic detoxification in LLR. Significant increase of acephate resistance, increases of esterase activities and gene expressions, and variable esterase sequences between LLS and LLR consistently demonstrated a major esterase-mediated resistance in LLR, which was functionally provable by abolishing the resistance with esterase inhibitors. In addition, significant elevation of P450 gene expression and reduced susceptibility to imidacloprid in LLR indicated a concurrent resistance risk that may impact other classes of insecticides. This study demonstrated the first association of down-regulation of reproductive- and digestive-related genes with resistance to conventional insecticides, suggesting potential fitness costs associated with resistance development. This study shed new light on the understanding of the molecular basis of insecticide resistance, and the information is highly valuable for development of chemical control guidelines and tactics to minimize resistance and cross-resistance risks.


Introduction
During the last decade, widespread adoptions of transgenic Bt cotton and altered chemical control schemes have allowed sucking insect populations to increase. Of these pests, the tarnished plant bug (TPB), Lygus lineolaris, emerged as the most economically significant [1]. Management of tarnished plant bug relies almost exclusively on chemical control. Commonly used insecticides include pyrethroids, organophosphates, carbamates, and neonicotinoids. Acephate, an organophosphorus insecticide, is among the most widely used insecticides for TPB control. In order to suppress feeding damages from tarnished plant bug and bollworm/tobacco budworm, cotton is more frequently sprayed than other major crops in the South. Over the years, tarnished plant bug has become increasingly resistant to several chemical insecticides [2,3,4], including acephate [5].
Acephate (Orthene) has been widely used to control tarnished plant bug on cotton in the Delta region of Mississippi, Arkansas, and Louisiana. TPB populations have developed 3-to 5-fold resistance, and control of these field populations may become difficult once resistance ratios reached 3-fold or higher [5]. Our field surveys indicate that susceptibility to acephate decreases as the growing season progresses. During late season (Aug. to Oct.), TPBs are substantially less susceptible to insecticides than those collected in May to July. Susceptibility reaches its lowest level in October. Our field surveys also indicate that TPB populations around cotton fields are less susceptible than populations around corn or soybean fields.
Another important phenomenon in studying TPB resistance to insecticides is the difficulty of maintaining resistant colonies in the laboratory. Less susceptible strains tend to have high mortality and low egg production. Although insecticide selections may increase resistance level to 10-fold or higher, selected colonies often die or fail to reproduce enough progeny for further research. Without further selection with acephate, a resistant colony can be sustained for a few generations. Resistance gradually decreased and sustained at 2-to 3-fold higher LC 50 than susceptible strain. All of these factors limit further investigation of resistance mechanisms in this important insect pest.
With multiple generations per year, high mobility, and existence of differential activities of major detoxification enzymes [6], field populations of TPB across the Delta region have potential to evolve high levels of resistance to multiple insecticides, especially when they are under high selection pressure. Examinations of acephate resistance mechanisms have not been conducted previously, except for a few limited sequence and gene expression level comparison of one esterase and one glutathione S-transferase related to malathion-resistant TPB [7,8]. Increased economic importance and reports of pyrethroid and organophosphate resistance [4,9] prompted our research to better understand the mechanisms of acephate resistance in this cotton pest. By integrating biological and biochemical assays with novel microarray and other molecular analytic tools, this study compared more than seven thousand genes simultaneously in acephatesusceptible (LLS) and -selected (LLR) tarnished plant bugs. Results are expected to generate a general picture of acephate resistance in TPB and help explain how TPB evolves resistance to the chemical. In addition, this study provides insight into potential fitness costs associated with acephate resistance and possible development of multiple and cross resistance to other classes of insecticides.

Insect laboratory colony and field collection
A laboratory colony was provided by Kathy Knighten and Fred Musser at Mississippi State University. The colony has been maintained on artificial diet for seven years without exposure to insecticide. Wild adults are introduced into this colony in the spring of each year. This colony was used as a standard susceptible strain (LLS). Resistance ratios (LC 50 of field-collected population/ LC 50 of laboratory strain) were calculated for field and laboratory strains relative to LLS. A field population (Lula) was collected in October 2010 from pigweed around a cotton field west of Lula, Mississippi. Acephate-selected bugs (Lula600) were from an acephate (OrtheneH 90WP, Valent, Walnut Creek, CA) treatment of the Lula field population. Approximately 45,000 bugs were collected and selected with Orthene 90WP at 600 mg/L (or 540 mg a.i./L acephate). Adults were held in a cage (W6D6H: 60660660 cm) covered with 16616 mesh metal screen on all sides except the bottom (metal sheet). Approximately 20 mL of acephate solution was used to treat the cage thoroughly from the top and all 4 sides using a hand sprayer. After treatment with Orthene (600 mg/L 90WP) for approximately 12 h, survivors were transferred into a 3.8 L container covered with a fine net cloth (10 grids/cm) and fed fresh green beans. After the first acephate treatment for 6 d, survivors were subjected to a doseresponse bioassay. Six acephate concentrations were prepared at 300, 500, 750, 1,000, 1,500, and 2,000 mg/L in d-H 2 O. A water only control was also included. Three replicates were used for each concentration, and 17 adults were used for each replicate. The adults were placed in 3.8 L container and sprayed with acephate solutions. After exposure for 10 m, the adults, along with 4 fresh green bean pods, were transferred into a clean plastic container (diameter[D]6height[H]: 10.567 cm). Bugs were maintained at 21uC and 14:10 (L:D) h. Mortality was recorded after 48 h and LC 50 s were calculated using SAS Probit analysis [10]. Survivors from Orthene 90WP treatment at 2,000 mg/L (LLR with resistance ratio .25-fold) were used for comparison of gene regulation using microarray analysis.

Enzyme activity assay
Esterase and glutathione S-transferase activities were comparatively examined using the protocols described by Zhu et al. [6]. In brief, nine individual tarnished plant bugs from each sample were homogenized in sodium phosphate buffer. The homogenate was centrifuged at 10,0006g for 5 minutes at 4uC. Protein concentrations were determined using the Pierce protein assay kit which utilizes the Bradford method [11]. To determine esterase activity, micro-titer plate assays were conducted using a-NA, b-NA, and PNPA as substrates. A Bio-Tek ELx808iu plate reader (Winooski, VT) was used to monitor a-NA and b-NA reactions at 450 nm for 10 minutes with measurements taken every 15 seconds [12]. For assays with PNPA, reactions were assayed at 405 nm for 10 minutes with readings taken every 15 seconds [13]. To determine glutathione S-transferase activities, micro-titer plate assays were conducted using CDNB as substrate. The reactions were monitored at 340 nm for 10 minutes with readings taken every 15 seconds [12].

RNA preparation, cDNA library construction and sequencing
Three adults (per tube) of L. lineolaris were homogenized in 1000 mL TriZol reagent (Invitrogen, Carlsbad, CA). Three to five replicates were included for each sample. mRNA was purified from total RNA using NucleoTrap mRNA purification kit (BD Bioscience Clontech, Palo Alto, CA). The Creator Smart cDNA Library Construction Kit (BD Bioscience Clontech) was used for cDNA library construction, by following manufacturer's instructions and modified protocols described by Zhu et al. [14]. Approximately 1 mg mRNA was used for reverse transcription and cDNA library construction. cDNA was ligated into pDNR-LIB vector (Clontech). The ligation was used to transform TOP10 competent cells (Invitrogen), which then were plated on chloramphenicol-agar plates. Approximately 30,000 clones were obtained and sequenced with a M13 forward primer on an ABI 3730XL sequencer (Applied Biosystems Inc., Foster City, CA) located at the Genomics and Bioinformatics Research Unit, USDA-ARS, in Stoneville, MS.

Sequence data processing and expression gene chips
After vector trimming and assembling using DNAStar (Ver. 8, Madison, WI), sequences were subjected to a similarity search for putative identity against protein and nucleotide databases of the GenBank in the National Center for Biotechnology Information (http://blast.ncbi.nlm.nih.gov/Blast.cgi) using Blastx NR, Blastn, and tBlastX protocols of Blast2GO software (http://www. blast2go.org/) [15,16] with 10 23 for cutoff E-value.

Acquiring microarray data
Roche NimbleGen 72 K gene expression chips in 4-plex format (Roche NimbleGen, Inc., Madison, WI) were used to compare global gene expression between the acephate-selected (LLR) and non-selected (LLS) strains of TPB. A 60-bp specific oligonucleotide was designed and synthesized as a probe. Approximately 35,000 probes (average of 5 probes per cDNA) were synthesized and printed on each gene expression chip. Microarray analysis was processed using standard NimbleGen array protocols. Total RNA was extracted from adults using TriZol reagent (Invitrogen). Double strand cDNAs were synthesized by using the SuperScript Double-Stranded cDNA Synthesis Kit (Invitrogen) according to the manufacturer's protocols. Double strand cDNA samples were labeled with One-color DNA Labeling Kit and hybridized to the microarray chips. Microarray data were acquired according to NimbleScan v.25 User's Guide through Florida State University Microarray processing facility. Four arrays (4 replicates) of 72 K NimbleGen expression chips were processed for each sample.

Analysis of microarray data
After gene expression data were obtained from 4672 K array processing, ArrayStarH software (DNAStar, Inc., Madison, WI) was used to analyze and compare microarray data between LLS and LLR. Expression data were log2-transformed and normalized through quantile normalization [17], and gene calls were generated using a Robust Multichip Average (RMA) algorithm [18]. Normalized data were analyzed using classical parametric statistics. P-values were calculated using Modified t-test. Clustering techniques, including the most popular ''hierarchical'' clustering and ''k-means'' clustering, were applied in the microarray data analysis. Clustering aims at dividing data points (gene or samples) into groups (clusters) using measures of similarity, such as correlation or Eucliden distance [19]. Hierarchical clustering creates a hierarchical, tree-like structure of data, and may be displayed using a ''heat map''. By using hierarchical clustering in this study, the expression levels of each gene could be quantitatively compared side by side between LLS and LLR strains. Scatter plotting was also applied to generate a distribution of more than 6,688 genes tested in an effort to locate differentially expressed genes between LLS and LLR. A fold-change cutoff of 2 and p-value threshold of 0.05 were used to determine differential gene expression (www.illumina.com/documents/products/ whitepapers/whitepaper_RNASeq_to_arrays_comparison.pdf).
Cloning full-length cDNAs coding for esterases cDNA library sequencing yielded several clones which matched esterase cDNAs in GenBank. Cloning of full-length cDNAs was achieved using procedures similar to those of Zhu et al. [7] with some modifications as described by Yang et al. [20]. Briefly, to obtain the full length cDNAs, total RNA was extracted from adults. The SuperScript First Strand cDNA Synthesis kit (Invitrogen) was used in a reverse transcriptase polymerase chain reaction (RT-PCR) with 5 mg of total RNA and an oligo-dT primer for cDNA synthesis. Forward primers were designed and used with oligo-dT primer in RT-PCR reaction to amplify 39-end of the esterase cDNA. The 59-end of cDNA for each of the esterases was obtained by using the 59 rapid amplification of cDNA end (59 RACE) system (Invitrogen). Two to three specific reverse primers for each of the three esterases were designed and used in semi-nested amplifications with a forward abridged anchor primer from 59-RACE kit. The 59-end of the cDNA was isolated and C-tailed, and then cloned into a pGEM-T vector (Promega). Plasmid DNAs were prepared and sequenced using an ABI 3730XL DNA analyzer to confirm full coding sequences of the esterases. To obtain error-proof full-length cDNAs, total RNAs from LLS and LLR were used for synthesizing RT-cDNA. RT-PCR amplification was repeated with a Platinum High Fidelity Taq DNA polymerase (Invitrogen). The PCR products were purified using Qiaquick PCR purification kit (Qiagen) and sequenced from both directions as described above.
Verification of esterase gene expressions using real-time PCR (qRT-PCR) Differential expression of 4 up-regulated esterase genes (LLE1-4), detected by the microarray analysis, were validated by qRT-PCR according to methods described by Yang et al. [20]. TPB adults were collected in October 2011 from Lula, Mississippi, the same location of the original LLR colony. Collected bugs were divided into two groups. One group of bugs were treated with 1,000 mg/L acephate and maintained on treated green bean (21uC, L:D = 14:10) for 7 days (Lula1000). Another group of the bugs were untreated and maintained under the same conditions for 7 days as an aging control (Lula CK). LLS (untreated) was also included as a control.
The qRT-PCR assays were performed in a 25 ml reaction volume using iScript TM One-step RT-PCR Kit with SYBR Green (Bio-Rad, Hercules, CA) in a thermal cycler PTC-200 with Chrome4 detector attached (Bio-Rad). The qRT-PCR thermal cycling profile was programmed to run at 50uC for 10 min of cDNA synthesis, 95uC for 5 min of iScript reverse transcriptase deactivation, followed by 40 cycles of 95uC for 10 s and 55uC for 30 s. A melting curve thermal cycling from 55uC to 95uC with an increment of 1uC for 10 s was added to check amplification specificity. Opticon Monitor 3 (Bio-Rad, Hercules, CA) was used to control all PCR reactions and data output. To obtain absolute quantities of each gene, two steps of qRT-PCR [20] were performed for each esterase gene. First qRT-PCR was conducted to amplify ribosomal 18S gene to estimate RNA concentration for each sample with 18S housekeeper gene as an internal standard. RNA samples were adjusted to the same concentration based on the first qRT-PCR with 18S as the internal standard. Second qRT-PCR was conducted to achieve an absolute estimation using standardized RNA samples and target gene (LLE1-4) cDNA as an internal standard. Three replications (samples) were used for each treatment, and 5 TPB adults were included in each RNA preparation. Gene expression quantity was automatically calcu-

Examination of dose responses of field populations to acephate and imidacloprid
To detect potential resistance to multiple insecticides in TPB, dose responses to Orthene 90WP (80 mg/L or 72 mg a.i./L acephate) and imidacloprid 2 L (85 mg/L or 18.19 mg a.i./L) were evaluated in different field populations. The concentration for each insecticide was similar to the LC 50 value (Note: this value was lower than the LC 50 in Table 1, because both bugs and food were sprayed in this experiment) against the laboratory susceptible strain (LLS). TPBs were collected in September 2011 from northwest Mississippi. Approximately 25 bugs were used for each replicate and 3 replicates were used for each population. Bugs along with 4 green bean pods were placed in a plastic container (D6H: 10.567 cm). A 9-cm hole was cut on lid and covered with a fine net cloth (10 grids/cm). Five hundred ml of insecticide solution was delivered into the container using a modified Potter Spray Tower. The sprayer was set at 7.5 psi with spray distance of 30.5 cm to ensure a uniform deposition of insecticide mist on inner surface of the container, green beans, and bugs. Mortality was recorded 48 h after treatment.

Examination of synergistic effect of esterase inhibitors on acephate toxicity against TPB
A field population was collected in April 2012 from Leland, MS, and treated with 72 mg a.i./L acephate. A laboratory colony (LLX) was collected in 2010 from many locations in Delta regions of Mississippi and Arkansas, and was selected with 225 mg/L acephate. In this experiment, LLX colony was treated with 144 mg/L acephate. DEF or TPP solution at 1% was prepared by dissolving inhibitor in ethanol, thereafter, an equal volume of dd-H 2 O was added to the solution. The bugs were sprayed with inhibitor solution one hour before they treated with acephate. The bugs in control were treated with 50% ethanol in water. Spray tower settings and spray volume were the same as described above. Mortality was recorded 48 h after treatment.

Comparison of acephate-susceptibility and enzyme activities
The acephate LC 50 for the susceptible strain (LLS) was 148.3 mg a.i./L ( Table 1). The LC 50 for the Lula field population was 2.2-fold higher than LLS. After selection with 600 mg/L Orthene 90WP, the observed LC 50 reached 5.9-fold higher than LLS and 2.7-fold higher than the unselected Lula field population. Esterase activity determined with a-NA was 2.5-fold higher in Lula field population and 4.5-fold higher in the acephate selected bugs (Lula600) than LLS. The b-NA and PNPA activities also significantly increased by1.3-1.4-fold in Lula, and by 2.8-to 2.9fold in Lula600, respectively (Table 1). Unlike the esterases, glutathione S-transferase activity in both Lula and Lula600 decreased by 1.4-and 1.1-fold (Table 1).

Microarray analysis of gene regulation in LLR
Identification of differentially expressed genes in acephate-resistant bug (LLR). A total of 7,446 unique contigs and singletons were obtained from cDNA library sequencing, and 6,688 genes had valid expression values from microarray analysis. Hybridization signals were analyzed and gene expression changes in LLR were presented as mean of fold changes when compared to the LLS. Of 6,688 genes examined in the LLS and LLR strains, approximately 9.9% genes were up-or down-regulated in the LLR. Expressions of the remaining 90% of genes were not significantly different between LLS and LLR. Based on P values (P ,0.05) and fold change ($2), significant differences in mRNA levels were detected in 662 genes between the LLS and LLR, which included 329 up-regulated and 333 down-regulated genes in LLR. Among the 662 differentially expressed genes, only 225 genes were putatively identified using Blast2go search of GenBank, including 107 up-regulated (Table 2) and 118 down-regulated (Table 3) genes in the LLR. Identities of the remaining 437 genes have not been determined in similarity search of GenBank using Blast2go.
Comparison of gene expression levels by hierarchical clustering. The 329 up-regulated genes and 333 down-regulated genes (identified with microarray and Blast2go similarity search) in the LLR strain, selected based on the P value (P,0.05) and fold change ($2), were analyzed by using hierarchical clustering analysis and plotted as a heat map with the ArrayStar software (Fig. 1A). Each column represented a sample (LLS or LLR) and each row represented a gene. The gray scale depicted relative levels of gene expressions from low to high with corresponding grayscale from light (for lower expression) to dark (for higher levels). The gene clustering at the left of the heatmap indicated the presence of three groups of genes with distinct expression profiles of 662 genes ($2-fold change) across samples. Group 1 (G1) (413 genes) was separated into subgroups of G1a   and G1b. All genes in G1a (23 genes) were down-regulated in LLR strain. G1b was separated into G1b1 (202 genes) and G1b2 (188 genes). The genes in G1b1 appeared to be up-regulated, while the genes in G1b2 were down-regulated in LLR (Fig. 1A). Group 2 had subgroups G2a (122 genes) and G2b (125 genes). Genes in G2a appeared to be down-regulated. Genes in G2b were upregulated in LLR. Group 3 (G3) was a small group (2 genes), and the genes in G3 were highly up-regulated in the LLR strain.
Scatter-plot comparison of gene expression levels between LLS and LLR. A total of 6,688 valid gene expression data (log2) points (dots) of LLR were pairwise plotted against the same genes of LLS (Fig. 1B). The two sets of gene expression data showed linear correlation with R 2 0.89. When the Student's t-test was selected, there were 1,638 and 2,534 genes that showed significantly different between LLS and LLR at 99% and 95% confidence interval levels, respectively. Dots of the scatter plot in the upper left corner represented up-regulated genes, and dots in low right corner represented down-regulated genes (Fig. 1B). Distance between a dot and the regression line (R) indicated a gene expression level with longer distance for higher up-or downregulation. A total of 662 genes (dots above line 2a and below line 2b) showed 2-fold changes, 191 genes (dots above line 4a and below line 4b) showed 4-fold changes, and 113 genes (dots above line 8a and below line 8b) exhibited 8-fold changes (Fig. 1B) between LLS and LLR.
Annotation and functional analysis of up-and downregulated genes. Three hundred and twenty-nine up-regulated and 333 down-regulated ($2-fold) gene cDNAs were subjected separately to blast2go mapping and annotation to determine each putative protein's role in biological process and molecular function. Annotation with blast2go showed that 66 genes of the 662 differentially expressed genes were involved in biological processes at GO level 2, including 41 up-regulated and 25 downregulated genes in LLR. Among the 41 up-regulated genes, 18 (27.3% of the 66 biological process-related genes) genes were associated with metabolic processes ( Fig. 2A), while only 5 of the 25 (7.6% of the 66 genes) down-regulated genes were associated with the same function of metabolic process ( Fig. 2A). Among the 662 differentially expressed genes, 76 genes were involved in molecular functions, including 49 up-regulated and 27 downregulated genes at GO level 2 in LLR. Among the 49 up-regulated genes, 37 (48.7% of 76 molecular function-related genes) genes were associated with catalytic activities (Fig. 2B), while only 17 of the 27 (22.4% of the 76 genes) down-regulated genes were associated with catalytic activities. Results indicated that more than 3.6-fold of the metabolic-related genes were up-regulated compared to those down-regulated genes with the same biological function. The results also indicated a substantial increase of metabolic activities in LLR. Similarly, analysis of molecular function revealed a large portion of catalytic-related genes with significant up-regulations, suggesting an increase of catalytic activities in LLR compared to LLS.
Potential association of up-regulated genes with altered acephate-resistance in L. lineolaris Up-regulation of esterase genes and metabolic resistance development. Increase of esterase activities in LLR prompted further examination and comparison of esterase gene expressions and cDNA sequences between LLS and LLR. Six esterase genes in LLR showed significantly higher gene expression levels (2.14-7.57-fold increase) than those in LLS (Table 2). No esterase genes were down-regulated in LLR (Table 3). By using cDNA library
Validation of up-regulated esterase genes using real-time PCR (qRT-PCR). Real-time PCR was conducted to verify reproducibility of microarray data. Four representative genes (LLE1-4) were selected and subjected to real-time qRT-PCR validation in acephate-selected TPB adults (Lula1000) collected from the same location in 2011 as was done for LLR in 2010. Expression levels of LLE1 in Lula-CK and Lula1000 were significantly up-regulated by 24-and 57.9-fold compared to the gene expression in LLS (Fig. 4A). LLE2 gene expressions in Lula-CK and Lula1000 were also up-regulated by 3.4-and 4-fold over that of LLS (Fig. 4B). Lula-CK and Lula1000 showed 6.3-and 15fold higher LLE3 gene expressions than LLS (Fig. 4C). LLE4 gene expressions in Lula-CK and Lula1000 were also up-regulated by 2.9-and 6.3-fold over that of LLS (Fig. 4D). Lula1000 and Lula-CK were collected from the same location and were the same age. After acephate (1,000 mg/L) selection and removal of susceptible bugs, all four esterase gene (LLE1-4) expressions increased 2.4-, 1.2-, 2.4-, and 2.2-fold, respectively. These results suggest upregulated esterase genes are closely associated with reduced susceptibility in acephate-selected TPB (Lula600, Table 1).
Synergistic effect of esterase inhibitors on acephate toxicity against two TPB colonies. By applying esterase inhibitor, the toxicity of acephate was increased against both field and acephate-selected colonies of TPB (Fig. 5). Approximately 35.90% of Leland TPBs were killed by 72 mg/L acephate-only treatment. DEF-only killed 7.97% of the bugs. Acephate+DEF killed 40.31% of the bugs, which was numerically higher but not significantly different from acephate-only treatment. Acepha-te+TPP killed 49.04% of Leland field bugs, which was significantly higher than that of acephate-only treatment. After corrections with each inhibitor's mortality, synergistic ratios of DEF and TPP were 0.98 and 1.34, respectively. Similarly, approximately 63.330% of acephate-selected TPBs (LLX) were killed by 144 mg/L acephateonly treatment. DEF-only killed 13.33% of the bugs. Acephate+-DEF killed 88.33% of the bugs, which was significantly different from acephate-only treatment. Acephate+TPP killed 85.24% of Leland field bugs, which was also significantly higher than that of acephate-only treatment. After corrections with each inhibitor's mortality, synergistic ratios of DEF and TPP against LLX colony were 1.37 and 1.34, respectively.
Up-regulation of a glutathione S-transferase (GST) gene in LLR. By cDNA library sequencing, 19 GST cDNAs were obtained. These GST cDNAs were re-assembled into 11 highly different contigs or singletons. Multiple sequence alignment showed that average sequence identities were 30% (10-86%) among 11 predicted GST protein sequences. These GSTs had 18.6% (17-45.5%) average sequence identities with previously reported GST from L. lineolaris (ABC46449; [8]). Based on their relatively low sequence identities, we suggest that all 11 GSTs from this study are newly cloned GSTs. Microarray data indicated that only one GST gene (L6147) was up-regulated by 2.14-fold in LLR (Table 2). However, no GST genes were down-regulated in LLR.
The 691-bp full-length cDNA encoded a 203-residue GST protein. Pair-wise alignment of protein sequences showed only 19.2% amino acid identity between this up-regulated GST in LLR and the only GST reported previously [8], suggesting a new GST might be related to reduced acephate susceptibility in L. lineolaris.
Up-regulation of Cytochrome P450 monooxygenase (P450) genes in LLR. A total of 13 partial cDNAs of P450s were obtained from cDNA library sequencing. Microarray data showed that three P450 genes were up-regulated by 2.7-to 3-fold in LLR (Table 2). No P450 genes were down-regulated in LLR (Table 3). L3359 belongs to CYP 6 family, which is commonly associated with pesticide resistance in insects. LL_39 and L4510 genes are also similar to P450s in the CYP 6 family. Comparison of 74-residue peptide close to C-termini revealed 31.9-52.8% amino acid identities among the three putative P450 proteins. Showing low sequence identities (38.9-41.7%) to 3 highly similar P450s (CYP6X1v1-3; [21]) of L. lineolaris in GenBank, L3359, LL_39, and L4510 may be newly sequenced P450s with potential association with insecticide resistance.
To test whether over-expression of P450 genes confer multiple resistances to different insecticide classes, dose responses to acephate and imidacloprid were evaluated in 9 field populations collected from northwest Mississippi, including Lula. Results showed variable survival rates in different populations (Fig. 6). Responses to both insecticides appeared to be correlated. The population showing low susceptibility to acephate also had low susceptibility to imidacloprid. The correlation was significant with R 2 value 0.81 (Fig. 6).
Potential association of down-regulated genes with reduced acephate susceptibility in L. lineolaris Down-regulation of eggshell and vitellogenin genes associated with egg production. Microarray data showed that three eggshell protein genes were down-regulated by 7.56-to 147.45-fold in LLR. In addition, three vitellogenin genes were also down-regulated by 3.34-to 6.26-fold in LLR (Table 3). Consistently, no eggshell or vitellogenin genes expressed significant increases in their gene transcript abundances in LLR (Table 2). Vitellogenin coding region of L4564 and LL_168 was overlapped with that of L140, but their cDNA sequence had only 33% and 45% identity, respectively, with L140. Therefore, L4564 and LL_168 are different vitellogenin cDNAs from L140. Because L4564 and LL_168 were not overlapping, whether they are from the same vitellogenin gene needs to be determined in a future study.
Down-regulation of protease and cathepsin genes associated with protein hydrolyzation. In LLR, only two protease and one salivary trypsin were up-regulated by 30.46-36.71-and 3.52-fold (Table 2), respectively. Thirteen serine protease genes showed 2.09-to 146.15-fold decrease in expression in LLR (Table 3). Another 11 cathepsin genes were also downregulated by 2.15-to 15.51-fold in LLR (Table 3).

Up-regulation of esterase genes associated with increased detoxification
Possessing ester bonds is a common feature in organophosphate insecticides, including acephate. Esterases detoxify organophosphate insecticides directly by splitting esters into an acid and an alcohol [22]. In addition, esterases might be involved in pyrethroid resistance development [23]. In this study, we consistently demonstrated that esterases are major detoxification enzymes and are responsible for the metabolic resistance closely associated with increased acephate resistance in the tarnished plant bug. By using acephate selection, we first demonstrated that elevated resistance level was closely correlated with increased esterase activities (Table 1). A previous study showed that fluctuation of esterase activities was well synchronized with seasonal increases of resistance levels in TPB [7]. Increased esterase activity in the late season likely resulted from pesticide applications on cotton. Without further exposure to insecticides in winter, the populations become less resistant. Therefore, insecticide resistance levels are closely associated with the intensity of chemical sprays. Seasonal variability in insecticide resistance are well documented [24,25,26]. Insecticides kill susceptible individuals and resistance levels evolve as resistant individuals increase over the growing season [27,28].
Higher esterase activities might result from altered gene sequence and hence the protein sequence difference between LLS and LLR. Careful examination of amino acid substitutions between LLSE1 and LLRE1 revealed 5 amino acid substitutions, but these substituted amino acids were strongly similar. Besides 2 strongly similar residue substitutions, LLRE4 had one weakly similar amino acid substitution (G to A) and a different amino acid substitution (S to R), suggesting a quality change of the LLRE4 esterase. Esterases are often found in multigene families [29,30] and it is very likely that a complex of esterases is present in the tarnished plant bug. cDNA library sequencing and microarray  analysis in this study revealed multiple esterase genes and differential expressed esterase genes in LLR. Overproduction of gene transcripts might have resulted from the transcription of a single copy of the esterase gene [31] or multiple copies of the esterase genes [32,33]. It is possible that resistant TPBs may have additional copies of esterase genes. A total of 14 esterase cDNA fragments were obtained in this study. Further assembling of these esterase cDNAs obtained at least 7 highly different esterase cDNAs or genes in the tarnished plant bug. Of the esterase genes examined, six of them were significantly up-regulated and none were down-regulated in LLR.
Synergists are often used as a diagnostic tool to characterize resistance mechanisms in pest populations. S,S,S-Tributyl phosphorotrithioate (DEF) inhibits hydrolysis of insecticides with ester linkages [34,35]. Triphenyl phosphate (TPP) was recognized as a specific inhibitor of carboxylesterase, which synergized the effect of malathion by blocking the production of malathion mono-acid in a resistant strain of Tribolium castaneum [36]. Based on our 2010 data, Leland population had relatively lower resistance ratio (1.6) to acephate. In this study, DEF was unable to synergize acephate toxicity, indicating that the population had no elevated DEFsuppressible esterases. The acephate-selected LLX colony is a mixture of several populations collected from Delta regions of Mississippi and Arkansas. It maintains 2-3-fold resistance ratio relatively to the LLS colony. Significant DEF synergism to acephate in acephate-selected LLX colony indicated that elevated esterase gene expressions exist in LLX for detoxifying acephate. Unlike DEF, TPP synergized acephate toxicity in both Leland field population and acephate-selected LLX colony, suggesting that both colonies had elevated carboxylesterases. These synergistic data functionally proved that esterase-mediated metabolic detoxification is potential resistance mechanism in acephateresistant TPB. To knock down detoxification genes, RNAi will be conducted in future study to identify specific esterase genes that are the most responsible for acephate resistance in TPB.
Up-regulation of a glutathione S-transferase gene and potential association with metabolic detoxification in LLR Glutathione S-transferases (GSTs) catalyze transformations of a wide range of endogenous and xenobiotic compounds, including carcinogens, therapeutic drugs, products of oxidative stress, herbicides, and insecticides [37,38,39]. They have the capacity  to conjugate reduced glutathione on the thiol of cysteine to various electrophiles and to bind with high affinity to various hydrophobic compounds [40]. Elevated GST activity has been associated with resistance to all the major classes of insecticides, commonly through increases in transcriptional rate, rather than qualitative changes in individual enzymes [41,42]. Of the 19 GSTs analyzed using microarray in this study, only one GST gene showed significantly higher transcripts in LLR. However, no GST was significantly down-regulated in LLR. GST enzyme activity data were also consistent with GST gene expression data. Significantly lower GST activity was found in the Lula natural population, suggesting a lower GST activity baseline than LLS (Table 1). Selection with Orthene 600 mg/L removed approximately 80% susceptible bugs from the sample causing GST activity to increase to a level similar to that of LLS. Considering GST enzyme activity, expression fold changes, and the number of genes involved, we postulate that GSTs may play a less significant role than esterases in reduced susceptibility of L. lineolaris to acephate.
P450 oxidation and potential multiple resistance and cross resistance to different classes of insecticides in LLR Cytochrome P450 enzymes (mixed function oxidases, cytochrome P450 monooxygenases, CYP), are a large and diverse class of enzymes found in virtually all insect tissues. The function of most P450 enzymes includes catalyzing the oxidation of organic substances to fulfill many important tasks, from the synthesis, degradation, and metabolic intermediations of lipids, ecdysteroids and juvenile hormones to the metabolism of xenobiotic substances of natural or synthetic origin [43]. P450-mediated resistance is probably the most frequent type of metabolism based insecticide resistance [44,45]. This mechanism may potentially affect several classes of insecticides and thereby confer cross-resistance to unrelated compounds due to their broad substrate spectra [46]. Most cases of P450-mediated resistance result from an increase in detoxification. Resistance can occur by increased transcription of a P450 leading to both increased expression of the protein and increased detoxification of the insecticide [47]. In addition to overexpressions of several esterase genes, acephate-selected bugs (LLR) showed increased gene expressions of at least 3 different P450 genes. In the Mississippi Delta area, a variety of insecticides have been used for cotton insect control. Tarnished plant bug has a history of exposure to organophosphates and pyrethroids, and resistances to these pesticides have been reported [9,5]. In recent years, neonicotinoids have become a popular alternative for plant bug control. P450s are associated with resistance to both pyrethroids [44] and neonicotinoids [48]. Elevated expressions of at least three P450 genes may confer multiple and/or cross resistance to the three commonly used insecticide classes. Our dose-response assays on field populations of TPB (Fig. 6) highly supported this statement of multiple/cross resistance in LLR. Dose-response assays on multiple populations (including Lula) indicated a close correlation between survival rates of TPB treated separately with LC 50 s of acephate and imidacloprid. We are currently trying to develop a P450 enzyme activity assay method to link enzyme activity and elevated P450 gene expression with biological data.
Down-regulation of eggshell, vitellogenin, protease genes associated with fitness cost Insect eggshell is composed of a set of proteins (rich in proline and alanine) synthesized by the follicular epithelium during the oogenesis and organized into an inner zone (vitelline membrane) and an outer zone (chorion) [49]. Vitellogenin is a unique group of proteins that are synthesized extraovarially and become the major egg yolk protein, vitellin [50]. Vitellogenin production by females is a prerequisite for successful egg production, and is directly linked to survival and reproductive success, thereby affecting both individual and colony-level fitness [51]. Serine proteases are enzymes that cleave peptide bonds in proteins. They are responsible for coordinating various physiological functions, including digestion, immune response, blood coagulation and reproduction [52]. Very little research has been conducted to link eggshell, vitellogenin, and protease production with fitness cost and resistance to conventional insecticides. We observed a substantial decrease of egg production in survivals after treatment with Orthene 90WP at 240 mg/L. In this study, we report an association of organophosphate resistance with down-regulations of eggshell, vitellogenin, and protease genes in L. lineolaris.
Microarray data revealed significant down-regulations of eggshell, vitellogenin, and protease genes in LLR. The finding could be valuable in addressing resistance issues in L. lineolaris. When selection pressure increases, either through laboratory selections or field sprays, resistant bugs with elevated esterase and P450 levels could survive and relatively increase resistance gene frequency after removal of susceptible bugs. But, reproductive incompetency substantially limits resistant population growth. When selection pressure is low and suitable host are abundantly available, susceptible bugs, with higher quantities of proteases and eggshell/yolk proteins, take reproductive advantage and increase population density quickly. Considering seasonal fluctuation and exposure-driven natures, we suggest that the acephate resistance in TPB is controlled by multiple genes. We also suggest that the resistance is associated with certain fitness cost.
In summary, microarray gene expression, Blast2go annotation, and other molecular comparisons, in concert with bioassays and enzyme activity data, revealed a significant increase of metabolic processes in the acephate-selected strain, suggesting metabolic detoxification as a major resistance mechanism. The resistance is controlled by many genes. Esterases are critical in detoxification of acephate, and suppression of esterases with inhibitors could significantly abolish acephate resistance. P450s also played an important role, along with less significant influence of GSTs. Down-regulations of many reproductive-and digestive-related genes indicated a potential fitness cost, which might dynamically keep the resistance from becoming fixed in the population. However, up-regulation of P450s and GSTs may increase the risk of multiple and/or cross resistance to other insecticide classes. Precautions must be taken to reduce selection pressure on target insect. Genetic composition of field populations should be constantly monitored to prevent potential genetic shift.