PGE2 upregulates gene expression of dual oxidase in a lepidopteran insect midgut via cAMP signalling pathway

In insect midgut, prostaglandins (PGs) play a crucial role in defending bacterial and malarial pathogens. However, little is known about the PG signalling pathway in the midgut. A dual oxidase (Se-Duox) with presumed function of catalysing reactive oxygen species (ROS) production in the midgut was identified in beet armyworm, Spodoptera exigua. Se-Duox was expressed in all developmental stages, exhibiting relatively high expression levels in the midgut of late larval instars. Se-Duox expression was upregulated upon bacterial challenge. RNA interference (RNAi) of Se-Duox expression significantly suppressed ROS levels in the midgut lumen. The suppression of ROS levels increased insecticidal activity of Serratia marcescens after oral infection. Interestingly, treatment with a PLA2 inhibitor prevented the induction of Se-Duox expression in response to bacterial challenge. On the other hand, addition of its catalytic product rescued the induction of Se-Duox expression. Especially, PG synthesis inhibitor significantly suppressed Se-Duox expression, while the addition of PGE2 or PGD2 rescued the inhibition. Subsequent PG signals involved cAMP and downstream components because specific inhibitors of cAMP signal components such as adenylate cyclase (AC) and protein kinase A (PKA) significantly inhibited Se-Duox expression. Indeed, addition of a cAMP analogue stimulated Se-Duox expression in the midgut. Furthermore, individual RNAi specific to PGE2 receptor (a trimeric G-protein subunit), AC, PKA or cAMP-responsive element-binding protein resulted in suppression of Se-Duox expression. These results suggest that PGs can activate midgut immunity via cAMP signalling pathway by inducing Se-Duox expression along with increased ROS levels.


Introduction
The insect gut is usually exposed to various microbes. The insect foregut and hindgut are originated from embryonic ectoderm. They can protect insects from pathogen infection through their cuticle linings. The midgut lacking a cuticle barrier is known to be relatively susceptible to various microbial pathogens such as bacteria, viruses, nematodes and protozoa [1]. Instead of a cuticle layer, the midgut in many insects possesses a mucus layer and a peritrophic matrix that can act as physical barriers against microbial infections [2]. In addition, insect midguts possess commensal or mutualistic microbes required for assisting digestion, supplementing essential nutrients, detoxifying xenochemicals or defending against other pathogens [3].
In addition to these protective defence lines, insect midguts exhibit direct antimicrobial activities by producing reactive oxygen species (ROS) and antimicrobial peptides (AMPs) [4]. However, these chemical defences should be tightly regulated so that beneficial microbes could be protected in insect midguts, while pathogens could be selectively removed [5]. For example, peptidoglycans derived from invading bacterial pathogens can trigger AMP expression [6]. However, Caudal, a homeobox protein, can selectively repress the expression of IMD/NF-kB-dependent AMP genes to protect beneficial symbiotic bacteria in the midgut [7]. In addition, ROS are specifically produced by pathogens [8][9][10].
ROS are produced in insect midguts by two different kinds of enzymes: NADPH-dependent oxidase (Nox) and dual oxidase (Duox) [11]. Inducing activities of these enzymes is crucial for the regulation of toxic ROS production. Nox expression can be induced by lactic acid produced by anaerobic metabolism of pathogens [12], while Duox expression is regulated by uracil released from pathogens in Drosophila [10]. However, it remains unclear how pathogen-derived factors can activate oxidases to upregulate ROS levels.
Eicosanoids are a group of oxygenated C20 polyunsaturated fatty acids that can mediate various physiological processes including immune responses in metazoan animals [13]. In Spodoptera exigua, a lepidopteran insect, Nox gene expression and subsequent ROS production in haemocytes are regulated by eicosanoids [14]. In the midgut of Plutella xylostella, another lepidopteran insect, Duox gene expression and subsequent ROS production that are essential for defending against bacterial pathogens are inhibited by treatment with eicosanoid biosynthesis inhibitors [11]. These results suggest that eicosanoids can mediate ROS production by activating Nox or Duox in these lepidopteran insects. Among eicosanoids, PGs play crucial role in mediating various immune responses in midguts of lepidopteran and dipteran insects [11,15]. In particular, several PGs along with PGE 2 receptor (PGR) and its downstream signal components have been found in S. exigua [16]. Thus, the hypothesis of this study was that PGE 2 could mediate ROS production by inducing Duox expression in S. exigua through cAMP signalling pathway.

Identification of Se-Duox and domain analysis
Transcriptome of S. exigua was interrogated with Duox gene of Drosophila melanogaster. A highly matched sequence (Se-Duox) encoding 1594 amino acid residues (figure 1) was found. It was predicted to be a transmembrane protein because of the presence of a signal peptide and transmembrane domains (figure 1a). It contains two oxidase domains ( peroxidase at the N terminus and FAD-linked oxidase at C-terminus). When Se-Duox was phylogenetically analysed with several insect Duox genes, it formed a monophyletic cluster with other lepidopteran orthologues (figure 1b).  Se-Duox was expressed in all developmental stages of S. exigua (figure 2a). It was highly expressed in late larval instars. In the last larval instar (L5), Se-Duox was highly expressed in the midgut (figure 2b). Bacterial challenge significantly ( p < 0.05) increased its gene expression levels in all larval instars (figure 2c).

RNA interference of Se-Duox suppresses ROS level and enhances bacterial pathogenicity
Injection of dsRNA specific to Se-Duox to L5 larvae significantly ( p < 0.05) suppressed its expression. Such RNA interference (RNAi) efficiency was maintained for at least 72 h PI (figure 3a). Along with decrease in Se-Duox expression, ROS levels in the gut lumen were also significantly ( p < 0.05) reduced (figure 3b). Such a decrease in ROS levels increased insecticidal activity of S. marcescens after oral administration (figure 3c).

PGs activate Se-Duox expression
Next, the effect of eicosanoids on induction of Se-Duox expression upon bacterial challenge was analysed. PLA 2 activity was significantly ( p < 0.05) inhibited by its specific inhibitor dexamethasone (Dex). Bacterial induction of Se-Duox expression was reduced by such inhibitor treatment (figure 4a). However, an addition of arachidonic acid ('AA', a catalytic product of PLA 2 ) to Dex-treated larvae significantly ( p < 0.05) rescued such inhibition of Se-Duox expression. ROS levels in gut lumen were highly dependent on Se-Duox expression levels. Dex treatment suppressed ROS levels, whereas the addition of AA rescued ROS levels suppressed by Dex treatment (figure 4b).
Modulation of Se-Duox expression by Dex or AA treatment suggested that Se-Duox expression could be controlled by eicosanoids. To clarify the types of eicosanoids that could modulate Se-Duox expression, esculetin (Esc) known to inhibit LOX or naproxen (Nap) known to inhibit COX was subjected to assessment for their abilities to modulate Se-Duox expression (figure 4c). Both inhibitors significantly ( p < 0.05) suppressed Se-Duox expression, with Nap being much more potent than Esc. Inhibited expression of Se-Duox after Nap treatment was rescued by the addition of PGE 2 or PGD 2 . ROS level in the gut lumen was downregulated by Nap treatment (figure 4d ). However, the suppressed ROS level caused by Nap treatment was significantly ( p < 0.05) rescued by the addition of PGE 2 or PGD 2 .
pupa adult relative mRAN levels of Se-Duox relative mRAN levels of Se-Duox  Midguts were then dissected to study expression patterns of Se-Duox. A ribosomal gene, RL32, was used as a reference gene. Each treatment was replicated three times with independent tissue preparations. Different letters above standard deviation bars indicate significant differences among means at Type I error = 0.05 (LSD test).
royalsocietypublishing.org/journal/rsob Open Biol. 10: 200197 2.5. cAMP signal mediates Se-Duox expression PGR of S. exigua can activate cAMP secondary messenger [16]. To predict the involvement of cAMP signal pathway in modulating Se-Duox expression, the upstream region (708 bp) of Se-Duox open reading frame (ORF) was analysed to determine the presence of any cAMP-related promoter element (figure 5a). In addition to the TATA box, the putative promoter region contained cAMP-responsive elements (CREs) at 0 to −100 bp and −500 to −600 bp. The presence of CRE in the promoter region suggested a functional interaction with CRE-binding protein (CREB). CREB of S. exigua (Se-CREB) was predicted from the transcriptome of S. exigua (GenBank accession number: WNNL01000001.1) after interrogating with CREB gene of S. litura (NCBI GenBank accession number: XM_022964340.1). Its sequence (390 amino acids) was highly homologous to other CREB genes (figure 5b). This predicted Se-CREB was found to be expressed in different larval tissues (figure 5c).
Specific inhibitors against adenylate cyclase (AC) producing cAMP and protein kinase A (PKA) phosphorylating proteins in response to cAMP were then used to treat larvae of S. exigua (figure 6a). Larval midguts from treated larvae expressed Se-Duox at significantly lower levels compared with those from control larvae without such treatment. By contrast, the administration of cAMP analogue (8-(4-chlorophenylthio)adenosine 3 0 ,5 0 -cyclic monophosphate sodium salt) increased Se-Duox expression in a dose-dependent manner (figure 6b).
The role of cAMP in modulating Se-Duox expression suggested a signalling pathway from PGE 2 to Se-Duox expression via PGR, Gαs, AC, PKA and CREB. All these five genes were downregulated after injecting their gene-specific dsRNAs (electronic supplementary material, figure S1). When Se-Duox expression was assessed at 72 h after each dsRNA treatment, its expression levels were significantly reduced after RNAi treatment (all five gene-specific dsRNAs) compared with those in controls (figure 6c).

Discussion
This study focused on gut immunity associated with ROS production. To this end, we identified a Se-Duox gene RNAi-treated larvae were exposed to different concentrations of S. marcescens. Mortality was recorded at 48 h post feeding. Each treatment was replicated three times. Each replication used 10 larvae. Different letters above standard deviation bars indicate significant difference among means in each treatment and control at Type I error = 0.05 (LSD test). Asterisk indicates the statistical difference between control and treatment at 10 7 CFU ml −1 dose.  [17]. Our current study showed that RNAi specific to Se-Duox suppressed ROS level. Suppressed ROS level after RNAi treatment caused larvae to become susceptible to bacterial infection. This suggests that Se-Duox activity is required for immune defence but should be tightly controlled because excessive ROS production by uncontrolled Duox activity would be detrimental to insect midgut tissues.
In Drosophila, Duox activity is activated through increased gene expression or post-translational modification [9]. Uracil derived from pathogenic bacteria can bind to G-proteincoupled receptor and trigger Gαq-PLCβ-Ca 2+ pathway to activate MEKK1-MKK3-p38 MAPK with subsequent upregulation of Duox gene expression or increased enzymatic activity through Ca 2+ -binding domain of Duox. By contrast, Duox in another lepidopteran host, P. xylostella, is controlled by eicosanoid biosynthesis inhibitor [11], suggesting that there might be a cross-talk between Gαq-PLCβ-Ca 2+ pathway and eicosanoid signalling. Se-Duox expression was dependent on PLA 2 activity. The inhibition of PLA 2 activity suppressed Se-Duox expression. Such suppression in the expression of Se-Duox was then rescued by the addition of AA, suggesting that AA and/or its oxygenated eicosanoids could mediate gene expression of Se-Duox. To clarify the types of eicosanoids involved in the regulation, esculetin (leukotriene biosynthesis inhibitor) or naproxen ( prostaglandin biosynthesis inhibitor) was used for treatment. The higher inhibitory activity of naproxen against Se-Duox expression suggested the role of PGs in mediating the gene expression. This was further supported by the  Changed PLA 2 activities by its specific inhibitor, dexamethasone (DEX), modulated ROS production, suggesting that PGE 2 could activate Se-Duox to produce ROS in the midgut. To explain this pathway, we speculate that MAPK p38 could mediate ROS production via activation of Duox gene expression in Drosophila's midgut [18]. In mammalian intestine, MAPK p38 can also activate COX-2 to elevate PGE 2 levels [19], suggesting that MAPK p38 plays a role in mediating the cross-talk between the Gαq-PLCβ-Ca 2+ pathway and eicosanoid signal to produce ROS. Thus, we wondered how the upregulated level of PGE 2 could activate Se-Duox expression.
Promoter analysis of Se-Duox gene indicated the presence of a CRE at the upstream region of Se-Duox ORF. PGR of S. exigua can transmit its signal using cAMP [16]. To determine the significance of CRE, this study identified a CRE-binding protein (Se-CREB) of S. exigua. Its sequence exhibited high homologies with other known CREB genes. The presence of a highly conserved serine residue was predicted to be activated after phosphorylation by PKA [20]. The leucine zipper motif in the α-helix was predicted to form a characteristic leucine zipper motif, which binds to CRE for transactivation [21]. Se-CREB was expressed in all developmental stages and different tissues of S. exigua. These results suggest a royalsocietypublishing.org/journal/rsob Open Biol. 10: 200197 cAMP signal pathway of S. exigua from PGR to CREB via Gαs, AC and PKA signalling components. Specific inhibitors against AC or PKA prevented the induction of Se-Duox expression. By contrast, the addition of cAMP analogue significantly induced the expression of Se-Duox in a dose-dependent manner. Thus, cAMP signalling pathway is likely to mediate Se-Duox expression in response to PGE 2 in S. exigua. Indeed, RNAi of PGR (= PGE 2 receptor) gene expression prevented the induction of Se-Duox gene. Furthermore, the cAMP signalling pathway to induce Se-Duox expression was supported by suppressed Se-Duox expression after individual RNAi treatments of its signalling components. Alternatively, calcium signalling pathway may play a crucial role in activating Duox enzyme activity [4]. Especially, Se-Duox contains EF domain suggesting its catalytic activity depending on Ca 2+ . A specific PG receptor in oenocytoi haemocytes of S. exigua activates calcium signalling pathway [13]. PG-Ca 2+ -Duox activity needs to be clarified in the subsequent study.
This study showed that Se-Duox expression in the gut was mediated by PGE 2 via cAMP secondary messenger (figure 7). MAPK p38 induced by pathogenic bacterial infection might induce PG production via a COX-like enzyme in S. exigua. A specific peroxidase called peroxynectin (Pxt) is known to mediate PG production in S. exigua [22]. PGE 2 , a main PG member in S. exigua [16], can activate the cAMP signal via its specific receptor, leading to the expression of Se-Duox to produce ROS in the midgut to defend microbial pathogens. PGE 2 is produced in midguts of different insects including Manduca sexta and Helicoverpa zea (two lepidopteran insects), Periplanata americana (cockroach) and Anopheles gambiae (mosquito) [15,23]. In addition to ROS production caused by induced Duox expression, PGE 2 can also mediate other gut immune responses by recruiting haemocytes to infection foci [15] and activating gene expression of antimicrobial peptides [24] in mosquitoes. All these immune mediation by PGE 2 should be preceded by pathogen recognition in the midgut. This study suggests that uracil-based Gαq-PLCβ-Ca 2+ recognition signal can activate eicosanoid signal pathway via p38 MAPK in S. exigua. Further study is needed to identify a specific p38 kinase that can activate Pxt to produce PGE 2 .

Insect rearing and bacterial culture
All experiments were conducted with a lepidopteran model, S. exigua larvae. Rearing of S. exigua larvae followed the method described previously [25]. Under rearing conditions, S. exigua underwent five larval instars (L1-L5) before pupation. Escherichia coli Top10A, a Gram-negative bacterium, was obtained from Invitrogen (Carlsbad, CA, USA) and cultured overnight in Luria-Bertani (LB) medium at 37°C. Serratia marcescens, a Gram-negative bacterium, was cultured in nutrient broth medium for 20 h at 28°C in a shaking (200 rpm) incubator. After centrifuging the cultured broth at 8000g for 20 min, bacterial cells were collected with 100 mM phosphate-buffered saline (PBS, pH 7.4) and used for immune challenge. Escherichia coli (2 × 10 4 cells larva −1 ) and S. marcescens (10 5 , 10 6 and 10 7 colony-forming unit (CFU) ml −1 ) were used for immune challenge through feeding assays after dipping food/diet in bacterial suspensions. For each bacterial concentration, 10 test larvae were used. Each treatment was replicated three times.

Midgut tissue preparation
Midguts were isolated from larval instars (L1-L5), pupae (2-3 days old) and adults (2-3 days old). For L1-L2 stages, 50 individuals were used to collect a midgut sample. For L3-L4 stages, 10 individuals were used. By contrast, only one individual was used to collect a midgut sample from L5 larva, pupa or adult.

RNA extraction and cDNA construction
Total RNAs were extracted from all developmental stages of S. exigua (egg stage, more than 1000 eggs were used for each extracton; L1, 30 larvae; L2, 20 larvae; L3, 10 larvae; L4, three larvae; L5, one larva; one pupa; one adult) using Trizol reagent (Invitrogen) according to the manufacturer's instruction. Different tissue samples were isolated from L5 larvae, in which fat body, midgut and epidermis were collected from five larvae as an experimental unit. Haemocytes were collected from 30 larvae per replication. Extracted RNAs were dissolved in 50 µl of diethyl pyrocarbonate-treated deionized and distilled water. First-strand cDNAs were synthesized from 1 µg of RNAs using Maxime RT PreMix (Intron Biotechnology, Seoul, Korea) containing oligo dT primers according to the manufacturer's instruction.

dsRNA preparation and RNAi
Double-stranded RNAs (dsRNAs) specific to Se-Duox and other genes were prepared as described previously [11]. Briefly, a partial Se-Duox sequence was produced by PCR using genespecific primers containing a T7 promoter sequence at the 5 0 end (electronic supplementary material, table S1). The PCR product was used as a template to generate dsRNA using a MEGAscript RNAi kit (Ambion, Austin, TX, USA) according to the manufacturer's instruction. Sense and antisense RNA strands were synthesized using T7 RNA polymerase at 37°C for 4 h. A control dsRNA (dsCON) was also prepared by synthesizing 520 bp fragment dsRNA of CpBV302, a viral gene. The resulting dsRNA was purified and mixed with transfection reagent Metafectene PRO (Biontex, Plannegg, Germany) at a ratio of 1 : 1 (v/v) followed by incubation at 25°C for 30 min to form liposomes. One microgram of dsRNA was injected to larval haemocoel (3-day-old L4 larvae) using a microsyringe (Hamilton, Reno, NV, USA) equipped with a 26 gauge needle. At 24, 48 and 72 h post-injection (PI), RNAi efficacies were determined using RT-qPCR as described above. At 48 h PI, treated larvae were used for immune challenge and measurement of Se-Duox expression after an oral infection. Each treatment was replicated three times.

ROS measurement
ROS quantification was conducted using an OxiSelect Intracellular ROS Assay Kit (cat. no. STA-342, Cell Biolabs Inc., San Diego, CA, USA). After squeezing out contents of the dissected midguts, remaining gut tissues were then washed twice with 1 ml of PBS by repetitive resuspension followed by centrifugation at 800g for 5 min. Washed tissues were then suspended in TC-100 cell culture medium containing 0.1× DCFH-DA (dichlorofluorescein diacetate, 20× DCFH-DA stock, Part no. 234201) and incubated at 37°C for 30 min with gentle mix (inverting). After washing tissues three times with PBS, tissues were dissolved in 500 µl of Cell Lysis Buffer (Part no. 234203) diluted with TC100 cell culture medium and incubated at room temperature for 5 min. Cell lysate (150 µl) was then transferred to a 96-well plate and fluorescence was read at emission wavelength of 530 nm after excitation at 480 nm. For extracellular ROS measurement, DCFH-DA was mixed with midgut content and incubated at 37°C for 30 min. A calibration curve was drawn using serial dilutions of DCF (dichlorofluorescein) standard ( part no. 234202) in TC-100 cell culture medium.

Applications of different inhibitors
All inhibitors used in this study were dissolved in DMSO at 10 µM and injected into L5 larvae (1 µl larva −1 ). At 8 h PI, treated larvae were used to estimate Se-Duox expression levels. For oral administration, gene expression analysis was performed at 12 h after feeding treatment along with the bacterial challenge.

Data analysis
All data are presented as mean ± s.e. and plotted with Sigma plot. To determine statistical differences, means were compared with Fisher's least squared difference (LSD) test and discriminated at Type I error = 0.05. Significant differences in RNAi efficiency were tested using the t-test of Sigma plot software v. 12.5. A p-value of less than 0.05 was considered statistically significant. All experiments were performed using three independent biological replicates.