Glutamate–GABA imbalance mediated by miR-8-5p and its STTM regulates phase-related behavior of locusts

Significance

Outbreaks of locust plagues are primarily due to the transition of locusts from low-density solitary to high-density gregarious populations. There is an urgent need to determine how to block the phase transition via environmentally friendly governance by targeting new pathways. Here, we found that glutamate/γ–aminobutyric acid (GABA) homeostasis regulated by miR-8-5p controlled the behavioral shift between solitary and gregarious locusts. Furthermore, we generated transgenic rice expressing STTM–miR-8-5p, which strongly hampered locust aggregation when locusts fed on these plants. Thus, the results reveal an alternative regulatory pathway modulating locust phase change other than dopamine pathway, elucidating a positive feedback regulation mechanism. Importantly, STTM transgenic plants could be employed to interfere with insect behavior, providing a safe economic approach for pest control.

Abstract

The aggregation of locusts from solitary to gregarious phases is crucial for the formation of devastating locust plagues. Locust management requires research on the prevention of aggregation or alternative and greener solutions to replace insecticide use, and insect-derived microRNAs (miRNAs) show the potential for application in pest control. Here, we performed a genome-wide screen of the differential expression of miRNAs between solitary and gregarious locusts and showed that miR-8-5p controls the γ-aminobutyric acid (GABA)/glutamate functional balance by directly targeting glutamate decarboxylase (Gad). Blocking glutamate–GABA neurotransmission by miR-8-5p overexpression or Gad RNAi in solitary locusts decreased GABA production, resulting in locust aggregation behavior. Conversely, activating this pathway by miR-8-5p knockdown in gregarious locusts induced GABA production to eliminate aggregation behavior. Further results demonstrated that ionotropic glutamate/GABA receptors tuned glutamate/GABA to trigger/hamper the aggregation behavior of locusts. Finally, we successfully established a transgenic rice line expressing the miR-8-5p inhibitor by short tandem target mimic (STTM). When locusts fed on transgenic rice plants, Gad transcript levels in the brain increased greatly, and aggregation behavior was lost. This study provided insights into different regulatory pathways in the phase change of locusts and a potential control approach through behavioral regulation in insect pests.

The migratory locust, Locusta migratoria, is a notorious pest worldwide that threatens agricultural and food safety (1, 2). Outbreaks of locust plagues are due to the transition from solitary to gregarious phases, which leads to large-scale aggregation and swarm migration (1, 3). The extensive application of chemical insecticides to suppress locust plagues has resulted in environmental and health issues, including environmental pollution, declines in nontarget animals, and the development of resistance to chemicals (4, 5). Therefore, locust management requires that research aim at preventing aggregation or developing alternative novel control approaches that could replace chemical insecticides. The change between solitary and gregarious phases in locusts is associated with remarkable phenotypic variation in response to population density but not with any genotypic changes (6–7). Therefore, an in-depth understanding of the molecular regulation of phase changes in locusts is crucial for developing technology for controlling locust plagues.

In the past two decades, many research efforts have focused on gene expression and regulation to reveal the molecular mechanisms of locust phase changes related to behavior, body color, reproduction, flight, and so on (3, 6, 8, 9). In particular, dopamine, sulfation modification of dopamine as a neurotransmitter, and the genes in the dopamine pathway can positively regulate the phase change of locust from solitary to gregarious states in behavior (3, 10, 11). Even more, a Gypsy element for long noncoding RNA PAHAL-mediated Henna gene transcription activation in dopamine pathway modulate aggregative behavior of locusts (12, 13). Some of the studies involving the posttranscriptional regulation of genes have first identified target genes and then explored their regulatory functions mediated by microRNAs (miRNAs) in the phase change of locusts (14, 7, 15). The critical target genes Pale and Henna, which are involved in dopamine biosynthesis, are negatively regulated by miR-133 and are related to behavioral phase changes in locusts (3, 7). Based on the miRBase database, miR-133 is found in more than 400 animal species. Because of the ubiquitous presence of the dopamine pathway in organisms, looking for more specific microRNAs to regulate the aggregation behavior of locusts will benefit insect control and crop protection.

miRNAs participate in the regulation of almost all biological processes and generally function by inhibiting gene expression through posttranscriptional mechanisms (16). Many studies have confirmed that animal behavior is also regulated by posttranscriptional regulation and the interaction of miRNAs and target genes involved in aggregation, locomotion, pervasive behavior, feeding, climbing, and circadian rhythms (7, 17–18). In fact, behavioral change is the most remarkable trait in the phase change of locusts (3, 6). Therefore, the genome-based screening of miRNAs could elucidate the different microRNAs and regulatory pathways of phase change in locusts. However, the complete profiles of system-level miRNA network-mediated phase change in locusts have not been sufficiently explored, with the exception of a study on different profiles of miRNAs between solitary and gregarious locusts (19).

Considering that locusts are typical herbivores, transgenic plants could be applied as an ideal delivery method for special miRNAs for locust control. A plant-derived miRNA has been reported to be involved in the cross-kingdom regulation of the growth and development of insect pests (20). Moreover, transgenic plants expressing artificial microRNAs show efficient, stable viral resistance by silencing the vital genes of viruses (21). Insect-derived miRNAs were also shown to significantly increase mortality and developmental defects after lepidopterans were fed transgenic plants expressing artificial microRNAs (22, 23), similar to the results obtained in insects feeding on transgenic double-stranded RNAs (dsRNAs) (24). However, the oral delivery of highly concentrated dsRNA did not silence target genes in locusts (25). Therefore, we reasoned that locusts may be sensitive to the ingestion of endogenous miRNAs, which avoid rapid degradation by nuclease enzymes in the locust gut, unlike exogenous dsRNAs. However, the expression of locust-derived miRNAs in plants to silence vital genes for control purposes has not been investigated.

In this study, we systematically investigated the difference in the expression profiles of miRNAs in solitary and gregarious locusts and identified an important miRNA, miR-8-5p, that could regulate the GABA/glutamate functional balance by targeting glutamate decarboxylase (Gad). The downregulation of miR-8-5p suppressed gregarious behavior. The expression of short tandem target mimic (STTM)–miRNA (miR-8-5p inhibitor) in transgenic plants could block aggregation behavior without any secondary effects on plant characteristics. Therefore, the study strategy of using miRNA networks but not differentially expressed genes (DEGs) revealed a crucial regulatory mechanism of the phase change of locusts and provided an alternative approach in insect pest control.

Results

miR-8-5p Regulates Behavioral Changes in Locusts.

To conduct a genome-wide screen of miRNAs associated with different locust phases, we analyzed the miRNA expression profiles of gregarious and solitary locusts by small RNA transcriptome sequencing. Locust miRNA homologs were identified using the arthropod miRNAs released in miRbase (https://mirbase.org/) via homology searches. A total of 20 differentially expressed miRNAs were identified between gregarious and solitary locusts. Among these miRNAs, 13 and seven were highly expressed in solitary and gregarious locusts, respectively (Fig. 1A). After selecting the top eight miRNAs on the basis of the difference in the fold change of their expression in both types of locusts (>3), we investigated five and three highly expressed miRNAs in solitary and gregarious locusts, respectively, as candidates involved in the phase change of locusts.

Fig. 1.

miR-8-5p regulates the behavioral transition between gregarious and solitary locusts. (A) miRNA expression profiles of gregarious and solitary locusts. (B) Expression of miRNAs in locusts after IG and CS based on qPCR. All data are presented as the mean ± SEM (n = 6). Significant differences are denoted by letters (one-way ANOVA). (C) P_greg, a probabilistic metric of gregariousness (n ≥ 25 locusts, Mann–Whitney U test), in gregarious or solitary locusts 48 h after the knockdown or overexpression of miR-8-5p, miR-252, and miR-10a-5p, respectively. At, antagomir; Ag, agomir. (D) P_greg in solitary and gregarious locusts 48 h after the knockdown or upregulation of miR-9c-5p, miR-276-3p, and miR-21. Green arrows indicate median P_greg values. P_greg = 0 represents a fully S-phase behavioral state; P_greg = 1 represents a fully G-phase behavioral state. G, gregarious locusts; S, solitary locusts.

Given that behavioral change is the most remarkable feature in locust phase transition, we examined the dynamic expression of the eight miRNAs during the isolation of gregarious locusts (IG) and crowding of solitary locusts (CS) over time by qPCR. The results showed that only miR-8-5p, miR-252, and miR-10a-5p were up-regulated during CS and down-regulated during IG. Moreover, miR-276-3p, miR-21, and miR-9c-5p were up-regulated by IG and down-regulated by CS (Fig. 1B). In contrast, miR-6497-5p remained relatively stable (one-way ANOVA, F = 0.422, P = 0.831) during isolation and showed only a significant decrease after 64 h of crowding (one-way ANOVA, F = 6.910, P = 0.001). However, the expression of miR-8-3p significantly decreased in both IG and CS (one-way ANOVA, F = 0.993, P = 0.447), as shown in Fig. 1B. Thus, miR-6497-5p and miR-8-3p are not involved in the phase change of locusts.

To explore whether the other six miRNAs regulate behavioral plasticity, we administered an miRNA agomir (overexpression) or an miRNA antagomir (knockdown) to gregarious and solitary locusts. Phase-related behavior was quantified using the index P_greg, a summary statistic provided by this behavioral assay and a reliable indicator of the locust aggregation propensity (6). Among the six miRNAs, only miR-8-5p significantly affected behavioral change in locusts. In particular, when miR-8-5p was knocked down by injecting antagomir-8-5p into gregarious locusts, the median P_greg value changed from 0.9 to 0.28, indicating a significant behavioral change from a gregarious to a solitary state (Mann–Whitney U test, P = 0.003, Fig. 1C). In parallel, solitary locusts injected with agomir-8-5p exhibited a significant but incomplete shift to gregarious behavior (P_greg = 0.49), with 38.4% of the locusts shifting into the P_greg interval of 0.8 to 1.0 (Fig. 1C). The gregarious locusts treated with antagmir-252 showed behavioral changes toward the solitary phase (P_greg = 0.61, P = 0.003). In contrast, the solitary locusts did not display any significant aggregation behavior after being injected with agomir-252 (P_greg = 0.06) compared with the agomir controls (P_greg = 0.02, P = 0.064. Fig. 1C). When the remaining four miRNAs (miR-10a-5p, miR-9c-5p, miR-267-3p, and miR-21) were activated or inhibited, we did not observe obvious behavioral changes in the solitary and gregarious locusts (P_greg < 0.2 for solitary locusts following the injection of the miR-9c-5p/miR-276-3p/miR-21 antagomir or the miR-10a-5p agomir; P_greg > 0.8 for gregarious locusts following the injection of the miR-9c-5p/miR-276-3p/miR-21 agomir or the miR-10a-5p antagomir; Fig. 1 C and D). Therefore, miR-8-5p is probably an important miRNA involved in phase-related behavioral changes.

miR-8-5p Directly Targets Gad in Glutamate–GABA Homeostasis.

To identify the target genes of miR-8-5p, we employed the miRanda algorithm (http://www.microrna.org/microrna/home.do) and the RNAhybrid program to predict the potential target genes of miR-8-5p in the locust genome database (26). Among the candidates, 1,632 genes showed the possibility of binding with miR-8-5p (alignment score > 115, P value < 0.05, Dataset S1). Additional analyses were performed to refine the range of candidate target genes that can regulate behavioral changes.

To focus on special miR-8-5p target genes associated with locust phase change, we considered only the binding and expression relationship between miR-8-5p and DEGs identified from the transcriptomic analysis of gregarious and solitary locusts. A total of 339 candidate target genes were found among the DEGs between gregarious and solitary locusts (Dataset S2). Furtherly, Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis showed that such candidate target genes were enriched mostly in pathways involved in energy metabolism and neurotransmitter pathways (Fig. 2A and Dataset S3). Of the most enriched top 20 pathways, 38 candidate target genes were potentially bound to miR-8-5p (Dataset S4).

Fig. 2.

Identification of Gad as a miR-8-5p target in locusts. (A) Top 20 pathways from the KEGG enrichment analysis of predicted candidate miR-8-5p targeting DEGs identified through genome-wide RNA-Seq between gregarious and solitary locusts. (B and C) Verification of target genes by RIP in locusts injected with agomir-8-5p or agomir-CK. Normal mouse IgG was used as a negative control. qPCR (B) or RT PCR (C) analysis was performed to amplify target genes. The red arrow represents the positive target gene by RIP. (D) Luciferase reporter assays in S2 cells cotransfected with miR-8-5p overexpression vectors and psiCHECK2 vectors containing WT or mutant (mt, the 8 nt of the region that corresponds to the mutated miRNA seed) sequences of Gad (n = 6, Student’s t test). The results of the luciferase activity and qPCR analyses are presented as mean ± SEM (n = 6). ***P < 0.001. (E) Double-FISH for miR-8-5p and Gad in locusts. The squares and arrows specifically indicate the areas where miR-8-5p and Gad were localized. Images were visualized using an LSM 710 confocal fluorescence microscope (Zeiss) at magnifications of ×10 (the Upper row) and ×40 (the Bottom row). Control, scrambled miRNA and target gene sense probes. (Scale bars, 50 μm.)

To determine whether miR-8-5p interacts with 38 predicted targets in locusts, we performed an RNA immunoprecipitation (RIP) assay by using a monoclonal antibody against the Ago1 protein in the locust brain (Fig. 2B). Only Gad was significantly enriched in the Ago1-immunoprecipitated RNAs from brains treated with agomir-8-5p compared with those from brains treated with agomir-NC (t = 4.99, P = 0.002 for Gad; Fig. 2 B and C).

To further confirm the interaction between miR-8-5p and Gad in vitro, we performed reporter assays by using luciferase constructs fused to the binding region of Gad. The constructs with Gad binding sites produced lower luciferase activity than the reporter constructs without miR-8-5p binding sites (controls) when cotransfected with the miR-8-5p overexpression vectors in S2 cells (t = 5.5, P < 0.001, Fig. 2D). When regions homologous to the ‘‘seed’’ sequence of miR-8-5p were mutated in the Gad reporter constructs, luciferase activity returned to the original levels produced by mock transfection with the empty reporter plasmid (t = 0.749, P = 0.471, Fig. 2D). Thus, the predicted sites in Gad could be targeted by miR-8-5p.

The colocalization of miR-8-5p and Gad was detected by double labeling via fluorescence in situ hybridization (FISH) to confirm the interaction between miR-8-5p and Gad in the locust brain. Both miR-8-5p and Gad were found to colocalize in Kenyon cells of the mushroom body and in the neuronal cell body of the olfactory lobe of the locust brain (Fig. 2E), which are two important regions in determining behavioral responses mediated by output and input neurons (27, 28). The negative control showed no signal in the same part of the locust brain (Fig. 2E). Thus, the cellular colocalization of miR-8-5p and Gad in the mushroom body and olfactory lobe supports direct interactions between them.

miR-8-5p Negatively Regulates Gad to Mediate Locust Aggregation Behavior.

Given that miR-8-5p could target Gad, we quantified Gad expression in the brains of gregarious and solitary locusts. The Gad expression level in gregarious locusts was significantly lower than that in solitary locusts (t = 5.51, P < 0.001, Fig. 3A). We also determined the mRNA and protein expression levels of Gad over the time course of IG and CS processes (Fig. 3 B and C). The protein and mRNA expression levels of Gad were down-regulated by CS and up-regulated by IG. Thus, Gad expression was negatively correlated with miR-8-5p expression during the phase change of locusts (Figs. 1B and 3 B and C).

Fig. 3.

Gad hampers locust aggregation behavior via glutamate-GABA neurotransmission. (A) Expression levels of Gad in the brains of gregarious (G) and solitary (S) locusts based on qPCR. (B and C) Expression levels of Gad in the brains of locusts after IG and CS based on qPCR (B, n = 8) and western blot analyses (C, n = 3). The qPCR data are presented as mean ± SEM. Western blot bands were quantified using densitometry and are presented as mean ± SEM. The same letter indicates data that are not significantly different. (D and E) Contents of glutamate (D) or GABA (E) in the brains of gregarious and solitary locusts according to HPLC MS (n = 6). (F) RNAi-directed knockdown of Gad resulted in behavioral transition from S to G phase (n ≥ 25 locusts, Mann–Whitney U test). The median P_greg values are indicated by green arrows. (G and H) Content of glutamate (G) or GABA (H) in the brains of solitary locusts after Gad RNAi knockdown, as measured using HPLC MS (n = 10). The qPCR and HPLC MS data are shown as mean ± SEM. *P < 0.05, ***P < 0.001.

As Gad is a rate-limiting enzyme that catalyzes the synthesis of GABA from glutamate (29), we measured glutamate and GABA contents in the brains of gregarious and solitary locusts using reverse-phase high-performance liquid chromatography (HPLC) with electrochemical detection (ECD). The glutamate content in gregarious locusts was significantly higher than that in solitary locusts (t = 3.659, P = 0.004, Fig. 3D), whereas the GABA content in gregarious locusts was significantly lower than that in solitary locusts (t = 2.276, P = 0.046, Fig. 3E).

To verify the function of Gad in the behavioral transition, we injected dsRNA of Gad into the brains of solitary locusts. At 48 h after injection, Gad knockdown (SI Appendix, Fig. S1 A and B) in the brains of solitary locusts significantly increased gregarious behavior (P_greg = 0.55, P = 0.009). In particular, 44.4% of solitary locusts displayed gregarious behavior compared with 4.4% of the locusts in the dsGFP-injected group (Fig. 3F). In addition, we detected glutamate accumulation and GABA production after Gad knockdown. Gad knockdown significantly reduced GABA production (t = 2.429, P = 0.026, approximately 34.4%; Fig. 3H) and increased the glutamate content (t = 2.396, P = 0.028, approximately 24.9%; Fig. 3G) of solitary locusts. These results indicated that Gad probably plays a role in the gregarious behavior of locusts mediated by the direct effector of glutamate–GABA homeostasis.

miR-8-5p Regulates the Gregarious Behavior of Locusts by Switching Glutamate–GABA Neurotransmission.

To determine the effects of miR-8-5p on its target gene to modulate the behavioral change, we quantified the mRNA and protein levels of Gad after the overexpression or knockdown of miR-8-5p in the locust brain. After agomir-8-5p was injected into solitary locusts, the mRNA and protein levels of Gad significantly decreased compared to those in control locusts (t = 13.453, P < 0.001 for mRNA; t = 5.126, P = 0.007 for protein). Moreover, the injection of antagomir-8-5p increased the mRNA and protein levels of Gad in gregarious locusts (t = 3.679, P = 0.004 for mRNA; t = 3.321, P = 0.029 for protein; Fig. 4A and SI Appendix, Fig. S2).

Fig. 4.

miR-8-5p regulates the gregarious behavior of locusts by switching glutamate and GABA homeostasis. (A) Relative protein levels (n = 3) of Gad in the brains of gregarious locusts injected with antagomir-8-5p and solitary locusts injected with agomir-8-5p. The western blot data are presented as mean ± SEM; *P < 0.05; **P < 0.01. (B–E) Glutamate accumulation (B and D, n=10) and GABA production (C and E, n = 10) in solitary locusts injected with agomir-8-5p and in gregarious locusts injected with antagomir-8-5p according to HPLC MS. Data are shown as mean ± SEM. *P < 0.05. (F) Behavioral rescue experiments were performed by injecting dsRNA against Gad into gregarious locusts pretreated with antagomir-8-5p. (G) Behavioral change in gregarious nymphs microinjected with GABA and in solitary nymphs microinjected with glutamate. (H) Behavioral change in gregarious locusts microinjected with an IGR antagonist, kynurenic acid, or an MGR antagonist, (S)-4-carboxyphenylglycine; behavioral change in solitary locusts microinjected with a GABAAR antagonist, picrotoxin, or a GABABR antagonist, saclofen; behavioral state of gregarious locusts in the rescue experiments involving injection with an IGR antagonist and agonist; behavioral state of solitary locusts in the rescue experiments involving injection with a GABAAR antagonist and agonist. All statistical analyses of the behavioral experiments were conducted with the Mann–Whitney U test relative to control groups (n ≥ 25 locusts). The vertical lines indicate the median values of P_greg; At, antagomir; Ag, agomir.

To further confirm the effects of Gad expression mediated by miR-8-5p on glutamate accumulation and GABA synthesis, we measured glutamate and GABA contents in the brains of gregarious and solitary locusts after miR-8-5p administration. The overexpression of miR-8-5p increased glutamate accumulation (t = 1.189, P = 0.250, Fig. 4B) and reduced GABA production in the brains of solitary locusts (t = 2.336, P = 0.031, Fig. 4C). In contrast, miR-8-5p knockdown significantly reduced glutamate accumulation (t = 2.142, P = 0.046, Fig. 4D) and increased GABA content (t = 2.169, P = 0.044, Fig. 4E) in the brains of gregarious locusts. Therefore, miR-8-5p controls glutamate accumulation and GABA production by negatively regulating the expression of Gad.

To determine whether miR-8-5p knockdown induces an increase in Gad expression linked with behavioral changes, we performed rescue experiments by injecting dsRNA targeting Gad into gregarious locusts subjected to miR-8-5p knockdown. As expected, the miR-8-5p knockdown-induced behavioral phenotype was fully rescued by treatment with dsGad (P_greg = 0.58, P = 0.039) compared with the dsGFP-injected controls (P_greg = 0.22, P = 0.394, Fig. 4F). Thus, Gad is the key target gene involved in the posttranscriptional regulation of miR-8-5p in the phase transition of locusts.

Ionotropic Glutamate/GABA Receptors Tune Glutamate/GABA during Behavioral Change.

Given the linkage between behavior and glutamate/GABA (30), the requirements for glutamate/GABA biosynthesis, synaptic release, and signaling in the locust phase transition suggest that glutamate and GABA are potentially key neurotransmitters in the induction of gregarious behavior. This hypothesis was confirmed by the direct injection of glutamate into the brains of solitary locusts, leading to a significant shift toward gregarious behavior (P_greg = 0.68, P < 0.001, Fig. 4G). Additionally, the gregarious locusts injected with GABA showed typical features of solitary behavior (P_greg = 0.08, P < 0.001). Saline-injected gregarious and solitary locusts used as controls maintained their individual behavioral characteristics (Fig. 4G). Thus, glutamate and GABA induced behavioral changes in locusts in different directions.

To further verify the function of glutamate and GABA in mediating behavioral changes, we interfered with their receptors. When we injected kynurenic acid, a general blocker of ionotropic glutamate receptors (IGR), into the brains of gregarious locusts, their behavior significantly shifted toward solitary traits (P_greg = 0.08, P < 0.001, Fig. 4H). However, the injection of (S)-4-carboxyphenylglycine (S-4-CPG), an antagonist of metabotropic glutamate receptors (MGR), did not induce a more pronounced behavioral shift toward solitary traits (P_greg = 0.82, P = 0.312, Fig. 4H). We also examined the behavior of gregarious locusts by injection with (S)-5-nitrowillardiine, an IGR agonist, at different doses in locusts treated with an IGR antagonist 24 h before. The injection of this agonist robustly abolished solitary behavioral traits in the gregarious phase in locusts induced by kynurenic acid at both intermediate (i.e., 5 mM; P_greg = 0.70) and high doses (i.e., 25 mM; P_greg = 0.98) but did not change the behavioral state when applied at a low dose (i.e., 1 mM; P_greg = 0.02) (Fig. 4H and SI Appendix, Fig. S3A). Therefore, IGRs bound to glutamate signaling mediate the behavioral transition between gregarious and solitary locusts.

The two receptors of GABA in migratory locusts are the ionotropic GABA A receptor (GABAAR) and metabotropic GABA B receptor (GABABR) (31). Accordingly, the injection of solitary locusts with picrotoxin, a specific antagonist of GABAAR, resulted in an approximately 44% shift toward gregarious behavior (P_greg = 0.65, P = 0.001, Fig. 4H). In contrast, injection with saclofen, a GABABR antagonist, did not induce gregarious behavioral traits in solitary locusts (P_greg = 0.29, P = 0.138). To further determine whether the GABAAR was involved in the behavioral transition of locusts, we rescued the behavioral phenotypes by injecting the GABAAR agonist muscimol into locusts subjected to picrotoxin pretreatment. As expected, the GABAAR inhibition-induced gregarious-like behavior was fully rescued by treatment with a medium dose (i.e., 5 mM; P_greg = 0.19) or high dose (i.e., 25 mM; P_greg = 0.03) of the GABAAR agonist but not a low dose (i.e., 1 mM; P_greg = 0.6) of the GABAAR agonist after 24 h (Fig. 4H and SI Appendix, Fig. S3B). Thus, ionotropic GABAARs mediate the behavioral transition in the central nervous system in response to binding of the inhibitory neurotransmitter GABA.

Transgenic Rice Expressing the miR-8-5p Inhibitor Hampers Gregarious Behavior.

As STTMs have provided effective tools for blocking endogenous mature miRNA activity (32), we attempted to use STTMs to suppress insect miRNAs by feeding locusts STTM-expressing plants. An STTM transgenic rice containing two noncleavable miRNA-binding sites linked by a spacer (48 to 88 nt, Fig. 5A) was created to target or deactivate the endogenous miR-8-5p of locusts. The expression of STTM miR-8-5p was driven by a CaMV 35S promoter. The STTM vector was introduced into the embryonic calli of the rice variety Zhonghua (ZH11) via Agrobacterium-mediated transformation, yielding 94 rice transformants after kanamycin screening. The seedlings were then transplanted into pots of soil and grown in the greenhouse (SI Appendix, Fig. S4 A–E).

Fig. 5.

STTM–miR-8-5p transgenic rice plants hamper the gregarious aggregation behavior of locusts. (A) Vector construction and transformation of rice. STTM vectors containing the expression cassette (35S-STTM-terminator) were constructed as described in a schematic diagram of two small RNA-binding sites complementary to miR-8-5p and a spacer/linker. miR-8-5p bs, small RNA-binding sites complementary to miR-8-5p. HygR, hygromycin; NOS, Nos terminator; RB, right border; LB, left border. (B) The expression of STTM-8-5p in the brains of locusts after 3 d of ingestion of STTM or WT plants by qRT PCR (n = 6). (C and D) Relative mRNA (C, n = 6) and protein levels (D and E, n = 3) of Gad in the brains of locusts fed with STTM plants compared to those fed with WT plants. (F–H) Morphology, body length, and body weight of gregarious nymphs fed with STTM plants compared with those fed with WT plants. (I) Behavioral changes in gregarious locusts fed with STTM or WT plants (n ≥ 25 locusts, Mann Whitney U test). (J–R) STTM transgenic rice lines showed no difference in agronomic traits, including plant height at maturity (J and K, n = 8), grain number per panicle (N, n = 4), panicle length (L and N, n = 8), 1,000-grain weight (O, n = 4), and seed size (P–R, n = 9) compared with WT plants. The qPCR, western blotting, and agronomic trait data are presented as mean ± SEM; **P < 0.01; ***P < 0.001.

To confirm the insertion of the STTM miR-8-5p expression cassette in the genome of genetically modified (GM) rice, the marker gene hygromycin (Hyg) was amplified in positive transformants (SI Appendix, Fig. S5). The expression of STTM-8-5p was measured by qRT PCR in T0 GM rice (SI Appendix, Fig. S6). High expression levels were observed in the transformants of three repetitions (the first group for C#1–C#31, the second group for C#32–C#63, and the third group for C#64–C#94).

When the locusts fed on transgenic rice plants for 3 d, we detected high expression of STTM-8-5p in the locust brain, but no expression of STTM-8-5p was found in the brains of locust that fed on wild-type (WT) plants (Fig. 5B). The locusts fed with STTM plants exhibited no apparent developmental defects in their average body weight or body size compared with those fed with WT plants (Fig. 5 F–H). The locusts fed with STTM-8-5p plants for 3 d presented higher Gad expression levels in the brains than those fed with WT plants (Fig. 5 C–E).

As the gregarious behavior of locusts is correlated with Gad expression, it can be inferred that STTM plants have a negative effect on locust behavior. For verification, we conducted behavioral assays on locusts after feeding with transgenic and WT rice plants for 3 d. As expected, 58% of the gregarious locusts fed with STTM-8-5p evidently shifted their behavior toward solitary traits, with a P_greg of 0.24 after 3 d (P = 0.001, Fig. 5I).

The experiments were repeated with the T1 generation of STTM transgenic rice plants. We chose C#5 and C#28, with high expression of STTM-8-5p, to cultivate the T1 generation. Compared with the WT plants, the locusts that fed on the STTM plants showed activation of Gad expression in the brain (SI Appendix, Fig. S7A). Furthermore, the locusts that fed on C#5 presented higher Gad expression than those that fed on C#28 (SI Appendix, Fig. S7A). The plants of the C#5 lines presented higher levels of STTM-8-5p than those of the C#28 lines (SI Appendix, Fig. S7B), suggesting that the gene activation efficiency was related to STTM-8-5p expression levels in transgenic plants.

To assess the efficiency of the effects of transgenic plants, we analyzed behavioral changes over a time course when locusts were reared on C#5 STTM-8-5p or WT plants for 1, 2, and 3 d. The Gad transcript level was almost unchanged on day 1, increased on day 2, and was much higher on day 3 than that of the corresponding controls (t = 4.431, P < 0.01, SI Appendix, Fig. S8A). Accordingly, the gregarious locusts on C#5 plants did not display obvious behavioral changes after 2 d compared with the control (SI Appendix, Fig. S8B). However, 59% of the gregarious locusts fed on STTM-8-5p showed an evident shift toward solitary traits, with a P_greg of 0.18 after 3 d (P = 0.001, SI Appendix, Fig. S8B). Therefore, behavioral transition became evident at day 3 when the locusts were fed C#5 STTM-8-5p, consistent with the T0 generation-mediated behavior phenotype (Fig. 5I and SI Appendix, Fig. S8B). Therefore, the STTM insect-derived miRNA rice showed transgenerational stability, resulting in the same behavioral phenotypes of reduced aggregation behavior both in the T1 generation and the T0 generation when locusts fed on the plants.

We further investigated the developmental effects of rice plants GM by STTM insect-derived miRNAs. The characterization of the STTM transgenic rice lines showed no difference in agronomic traits. In particular, the STTM-8-5p and WT plants showed similar plant heights at maturity (Fig. 5 J and K). The STTM-8-5p lines also exhibited no change in seed size, grain length, grain number per panicle, or 1,000-grain weight compared with the WT plants (Fig. 5 L–R). Altogether, STTM miR-8-5p greatly hampered the gregarious behavior of the locust, without any side effects on plant growth and development.

Discussion

In this study, we found that the microRNA miR-8-5p can control the GABA/glutamate functional balance by directly targeting Gad, which mediates the behavioral shift between solitary and gregarious locusts by balancing glutamate accumulation and GABA production. Furthermore, the ingestion of STTM locust-derived miRNA in transgenic rice triggered miRNA-8-5p silencing and the loss of gregarious behavior in locusts (Fig. 6). Thus, we identified an important regulatory pathway in the phase change of locusts and developed STTM transgenic plants to interfere with gregarious behavior as an environmentally friendly control technology.

Fig. 6.

A diagram of the miR-8-5p-mediated control of glutamate/GABA functional balance by directly targeting Gad, resulting in the modulation of locust behavior. miR-8-5p downregulation hampers gregarious behavior in locusts following the ingestion of transgenic STTM or the injection of an miRNA antagomir. IGRs and GABA receptors tune glutamate/GABA in association with gregarious behavior. Transgenic plants expressing STTM–miR-8-5p may be an important control approach in locusts.

In the present results, the genome-based screening of miRNAs revealed glutamate–GABA neurotransmission to be an alternative pathway involved in the modulation of phase changes in locusts. The pathway is independent of the dopamine metabolic pathway, which was determined to regulate the phase change in locusts reported in our previous studies (7, 13, 33). In fact, many complex behaviors of animals are jointly controlled by the interactions among various central neuronal circuitries (34). In particular, glutamate–dopamine–GABA, as neurotransmitters, show a close interaction in motor behavior through the complicated and reciprocal regulation of the release of glutamate, dopamine, and GABA in animals (35). The present study found that miR-8-5p mediated the glutamate–GABA homeostasis in regulating the phase-related behavior of locusts. Activation of glutamate–GABA neurotransmission by a miR-8-5p inhibitor controlled the gregarious behavior of locusts. Glutamate and GABA, which act as primary excitatory and inhibitory neurotransmitters in the central nervous system, have also been reported to be important for behavior control in vertebrate and invertebrate systems, respectively (36, 37). Glutamate and GABA release underlie many behavioral reinforcements, including mood, defensiveness, feeding, and acute locomotion (38–39). Therefore, the glutamate–GABA homeostasis balanced by miRNA is another important pathway for controlling phase-related behavioral changes in locusts.

By systematically screening the network of miRNAs in locusts, miR-8-5p was confirmed to be highly expressed in gregarious locusts, leading to glutamate accumulation and GABA reduction. Based on the miRBase database, miR-8-5p is found in 17 animal species, and much more specific than miR-133. On the other hand, miR-133 is highly expressed in solitary locusts, resulting in the inhibition of locust aggregation by targeting Henna and Pale, the two key genes in the dopamine synthesis pathway (7). Thus, miRNAs probably act as key junctions of the potential crosstalk between different neurotransmitters at a precise level in which they can exhibit proper function. When we knocked down Pale expression in the dopamine pathway, the activity of miR-8-5p was not affected (t = 1.107, P = 0.294, SI Appendix, Fig. S9A). Conversely, the inhibition of Gad expression up-regulated the miR-133 level (t = 4.083, P < 0.01, SI Appendix, Fig. S9B). Thus, excitation in behavioral regulation by dopamine (7) may present some link with glutamate–GABA neurotransmission through the miRNA junction in locusts, indicating that a feedback regulatory loop may be formed to join the two neural pathways. Different miRNAs are tuned to a precise level to initiate different types of neural pathways and execute their proper functions, emphasizing the important roles of miRNA-mediated regulation in the behavior of locusts. Further research is needed to establish how miRNAs function between dopamine and glutamate–GABA homeostasis to execute the phase change of locust in behavior.

This study extended the understanding of receptor types involved in regulating phase change in locusts. Functionally, miR-8-5p regulates the ionotropic glutamate–GABA pathway, whereas miR-133 inhibits the metabolic receptors of the dopamine pathway (7). A previous study clarified that two metabotropic receptors of dopamine, Dop1 and Dop2, modulate locust phase change in two different directions: Dop1 signaling induces gregariousness, and Dop2 signaling mediates solitariness (10). In the present study, ionotropic glutamate and GABA receptors bound to the corresponding signaling molecules were found to mediate the behavioral transition between gregarious and solitary locusts. However, the respective metabotropic receptors did not play a significant role in the phase-related behavioral regulation of locusts. In fact, different neurotransmitters evoke postsynaptic electrical responses and promote behavioral change by combining with a diverse group of postsynaptic receptors (40). Thus, our results provide a more comprehensive understanding of the molecular mechanism underlying phase changes in locusts through glutamate–GABA neurotransmission and the dopamine pathway.

Various phase-related behavioral traits may be regulated by different pathways. We also analyzed each behavioral parameter of the index P_greg, which mainly retains three variables: attraction index, total distance moved, and total duration of movement (6). Our results demonstrated that locust locomotor activity, including the total duration of movement and total distance moved, was strongly suppressed by antagomir-8-5p injection in gregarious locusts but enhanced by agomir-8-5p injection in solitary locusts (SI Appendix, Fig. S10 A and B). However, the attraction index was not significantly altered by these treatments (SI Appendix, Fig. S10C). Thus, miR-8-5p plays important roles in the locust behavioral phase transition by modulating locomotor activity. A previous study showed that 4VA acts as an aggregation pheromone in the locust (2). The changes in glutamate–GABA homeostasis mediated by miR-8-5p may not affect the release and sensitivity of 4VA, which is an aggregation pheromone of locusts. Consequently, the sensitivity of peripheral olfactory attraction between gregarious and solitary locusts may be modulated by other central nervous pathways that specifically drive attraction behavior.

In this study, transgenic rice plants engineered to express STTM insect-derived miRNA hampered the aggregation behavior of locusts. This is an attempt to employ transgenic plants to interfere with insect behavior rather than killing pests. Thus, insect-derived miRNAs could be exploited as prospective targets to control insect pests via the in-plant expression of STTM–miRNA. Although miRNAs are well known to play important roles in the regulation of insect biological processes (41), only a limited number of studies have reported the exploitation of miRNAs for insect pest control (42). To date, no efficient approach is available for deploying miRNA inhibitors via feeding delivery into crops in control pests. In the present study, we employed STTM locust-derived miRNAs in rice to inhibit miR-8-5p expression in locusts. Due to huge economic losses and insecticidal environmental concerns caused by locusts (4, 5), developing alternative approaches to control locust plagues is especially noteworthy. Moreover, locusts are very sensitive to oral miRNA silencing but not to oral dsRNA delivery because of the rapid degradation of dsRNA in the locust midgut (25). Accordingly, STTM locust-derived miRNAs designed to specifically control locusts have great potential to increase the efficacy of crop protection. In practice, we could establish an isolation belt of STTM transgenic plants surrounding locust breeding areas to disturb aggregation and block the marching band behavior of locusts. At the same time, fungal pesticides will achieve much better efficiency when gregarious locusts feed on transgenic plants to break up locust swarms because solitary locusts are more sensitive to fungal pesticides (43, 44).

Due to the conservation of miR-8-5p in insects, its extended application could give rise to some concerns. However, the delivery of STTM–miRNAs (miRNA inhibitors) in transgenic plants is indeed feasible for insect-specific miRNAs. Notably, when transgenic plants were developed to produce STTM-locust miRNA, no adverse effects on rice growth were observed. Therefore, the insect miRNA silencing effect triggered by the ingestion of transgenic STTM–miRNA–producing plants is insect-specific and could be used to protect crop plants from insect damage. Further studies are needed to verify this possibility of STTM–miRNA for safe, specific, and efficacious crop protection.

Materials and Methods

Insects.

Experiments were performed using fourth-instar locusts from solitary and gregarious colonies. Gregarious (400 insects per case) and solitary (individual) locusts were reared under a 14-h light/10-h dark cycle at 30 °C ± 2 °C and were fed fresh wheat seedlings (45).

Deep Sequencing of Small RNAs.

The total RNAs in the brains of gregarious and solitary locusts were extracted using TRIzol (Invitrogen) and treated with DNase I. Each sample contained 12 brains (six males and six females). Three independent replicates were performed for each group. The RNA concentration and purity were assessed to verify RNA integrity by an Agilent 2100 Bioanalyzer (Agilent). Small RNA libraries were constructed according to the method described by Guo et al. (15). Small RNAs (18 to 35 nt) were sequenced at BGI–Shenzhen as described previously (19). The small RNA libraries were deposited in the Sequence Read Archive database (accession number: PRJNA867122).

Quantitative PCR Assays of Genes and miRNAs.

Genes and miRNAs were assayed by qPCR according to the method described by Guo et al. (15). The qPCR primers are listed in SI Appendix, Table S1. All qRT-PCR assays were performed with six biological replicates.

Western Blot Analysis.

Western blot analysis was performed according to the method described by Yang et al. (7). Membranes were incubated with the primary antibodies (Beijing Protein Innovation Co., Ltd., BPI, rabbit anti-Gad serum, 1:500) overnight at 4 °C. The specificity of the antibody against Gad was evaluated (SI Appendix, Fig. S1B).

In Vitro Luciferase Reporter Gene Assays.

The ~300-bp sequence surrounding the predicted miR-8-5p target sites in Gad was cloned into the psiCHECK-2 vector (Promega). To generate the mutated versions of the sequence, 8 nt of the binding sites, including the region complementary to the miR-8-5p seed, was mutated (GGTCAGAT for Gad). The cotransfection of S2 cells was assayed according to the method described by Yang et al. (46).

RNA-seq and Data Processing.

The brains of fourth-instar gregarious and solitary locusts were collected, and each sample contained 12 brains (six males and six females). Three independent replicates were performed for each group. Total RNA was isolated, and RNA quality was confirmed by agarose gel electrophoresis. cDNA libraries were prepared in accordance with Illumina protocols. Index codes were added to attribute sequences to each sample. Raw data were filtered, and the cleaned data were mapped to the locust genome sequence with HISAT2 software. StringTie software was used to calculate the fragments per kilobase million values of genes. DEGs were analyzed using EdgeR software. The DEGs with a fold change of P < 0.01 were selected. Gene enrichment analysis was performed by KEGG signaling pathway analysis. Additionally, RNA-seq data were deposited in the NCBI Sequence Read Archive Database (accession number: PRJNA867152).

miRNA Agomir and Antagomir Treatment In Vivo.

An miRNA agomir or antagomir (RiboBio) was used to validate the function of the miRNA in vivo. The brains of 2-d-old fourth-instar gregarious or solitary nymphs were microinjected with agomir-8-5p or antagomir-8-5p, respectively. The sequence of a Caenorhabditis elegans miR-67-3p (5′-UCA CAA CCU CCU AGA AAG AGU AGA-3′) was used as a negative control. Agomir-8-5p or antagomir-8-5p (42 pmol, 200 mM; RiboBio) was injected into locust brains. Agomir-negative and antagomir-negative controls (200 mM) were also injected into the brains of gregarious or solitary locusts (RiboBio). All injections were administered in accordance with a previous study (7).

RNA Interference.

To knock down Gad, fourth-instar nymphs were each injected with 3 μg dsRNAs on day 2. Control nymphs were injected with 3 μg dsGFP alone. A total of 30 nymphs were injected with dsRNA for each gene. The effect of RNAi on the relative mRNA and protein expression levels was investigated by qPCR and western blotting 48 h after injection (SI Appendix, Fig. S1 A and B). The primers employed for RNAi are provided in SI Appendix, Table S1. The nucleotide sequences used for RNAi are listed in SI Appendix, Table S2.

Colocalization of miRNA and Its Targets by FISH.

A combined two-color fluorescence in situ analysis of miRNA-8-5p and its target, Gad, was performed on the brains of fourth-instar nymphs by colabeling the miRNA and its target in accordance with the method described by Yang et al. (7). An antisense locked nucleic acid detection probe for miR-8-5p was subjected to double digoxigenin labeling (Exiqon), and biotin-labeled probes of Gad were generated using the T7/SP6 RNAtranscription system (Roche, Basel, Switzerland). The C. elegans miR-67-3p sequence (5′-UCACAACCUCCUAGAAAGAGUAGA-3′) and the sense sequence of the Gad probe were used as the negative controls. The primers used for Gad probe synthesis are listed in SI Appendix, Table S1.

RIP Assays.

The RIP assays were performed using a Magna RIP Quad kit (Millipore) in accordance with the method described by Guo et al. (15). Approximately 40 nymphs were microinjected with agomir-8-5p. A scrambled miRNA agomir was used as a negative control.

Behavioral Assays.

Behavioral assays of fourth-instar nymphs were conducted in a rectangular arena (40 cm × 30 cm × 10 cm) with opaque walls. The behavioral parameters were recorded and expressed as a mixture of behavioral or categorical markers (6). The measurement and quantification of the behavioral phenotypes of the injected nymphs were performed with reference to a previous logistic model (6): P_greg = e^ƞ / (1 + e^ƞ), where ƞ = β₀ + β₁X₁ + β₂X₂ + ··· + β_kX_k, and P_greg indicates the probability of a locust being considered gregarious. P_greg = 1 indicates fully gregarious behavior, whereas P_greg = 0 indicates fully solitary behavior. DsGad/miRNA agomir/antagomir-injected locusts or STTM-transgenic plant-fed locusts were used to analyze behavioral states based on the regression model.

Behavioral Rescue Experiments In Vivo.

To determine the role of Gad as the key target gene involved in the posttranscriptional regulation of miR-8-5p in the phase transition of locusts, we microinjected 42 pmol of antagomir-8-5p (200 mM) into the brains of gregarious locusts. After 24 h, 3 μg of dsGad or the dsGFP control was injected into the thoracic hemocoel of gregarious locusts. The injected locusts were subjected to behavioral analysis 24 h after injection of dsGad.

Measurement of Glutamate and GABA Levels in Locust Brains.

The glutamate and GABA contents of the locust brain were quantified through reverse-phase HPLC and ECD, as previously described (7). Ten samples of locust brains (eight individuals/sample) were collected and homogenized. Glutamate and GABA levels were quantified by referring to external standards. A standard curve was produced through the serial dilution of a standard solution containing glutamate and GABA (Sigma–Aldrich).

Behavioral Pharmacology.

To determine the role of glutamate and GABA in the behavioral shift between gregarious and solitary locusts, we microinjected glutamate (5 mM × 69 nL, Sigma-Aldrich) into the brains of solitary locusts and then assayed their behavior after 4 h. We microinjected GABA (5 mM × 69 nL, Sigma-Aldrich) into the brains of gregarious locusts and then assayed their behavior after 4 h. The control group received the same volume of saline.

For the determination of the role of IGR or MGR in the behavior of gregarious locusts, kynurenic acid (5 mM × 69 nL; Sigma-Aldrich; a general antagonist of IGRs) or S-4-CPG ( 5 mM × 69 nL; Sigma-Aldrich; an antagonist of MGRs) was microinjected into the brains of gregarious locusts. Then, their aggregation behavior was determined 4 h after microinjection. The control group received the same volume of saline.

For the determination of the role of ionotropic or metabotropic GABA receptors in the behavior of solitary locusts, picrotoxin (an antagonist of GABAAR, 5 mM × 69 nL; Sigma-Aldrich) or saclofen (an antagonist of GABABR, 5 mM × 69 nL; Sigma-Aldrich) was microinjected into the brains of fourth-stadium solitary locusts. All solitary locusts were subjected to behavioral assays 4 h after microinjection. All of the control groups received the same volume of saline before the behavioral assay. All injection procedures were performed as previously described (47).

Behavioral Rescue Experiments for Pharmacological Analysis.

In gregarious locusts, we microinjected kynurenic acid (5 mM × 69 nL; antagonist of IGRs, Sigma-Aldrich) into the brain. After 4 h, (S)-5-nitrowillardiine (5 mM × 69 nL; IGR agonist; Sigma-Aldrich) was microinjected into the brains of these locusts. The injected locusts were subjected to behavioral analysis 4 h after the injection of (S)-5-nitrowillardiine.

In solitary locusts, we microinjected picrotoxin (5 mM × 69 nL; antagonist of GABAAR; Sigma-Aldrich) into the brain. After 4 h, muscimol (5 mM × 69 nL; agonist of GABAAR; Sigma-Aldrich) was microinjected into the brains of these locusts. The injected locusts were subjected to behavioral analysis 4 h after muscimol injection.

Plant Materials and Trait Measurements.

STTM-8-5p and WT plants were grown in a greenhouse with 30 °C/24 °C ± 1 °C d/night temperatures, 50 to 70% relative humidity, and a light/dark period of 14 h/10 h. Plant height, grain length, grain width, panicle length, grain number per panicle, and 1,000-grain weight were measured at full maturity. Plant height was measured in the greenhouse. Grain length and width were measured using the ImageJ analysis system. The 1,000-grain weight was weighed after the fully filled grains were dried at 42 °C in an oven for 2 wk.

Rice Transformation and Screening for the Expression of STTM–miR-8-5p.

The STTM–miR-8-5p rice was genetically transformed by BioRun Biotechnology (Wuhan, China). STTM vectors were constructed as described by Tang et al. (48). The expression vectors were introduced into the Japonica rice variety ZH11 following the protocol of Lin et al. (22). After transformation, the calli were washed and transferred for differentiation and rooting. The regenerated plantlets were cultivated in a greenhouse. Total RNA was isolated from T0 GM and WT ZH11 plants to identify positive STTM–miRNA T0 GM rice plants. The expression of STTM–miRNA was quantified via qRT PCR. cDNA from WT plants and sterile water were used as negative controls. The sequences of all primers used are listed in SI Appendix, Table S1.

Locust Feeding Assay.

After the determination of STTM–miRNA expression levels in rice plants using qRT PCR, the fresh stems of rice lines with high expression levels of STTM–miRNA were fed to 30 to 35 locusts. Old rice plants were replaced with fresh rice plants every day. The T0 and T1 STTM-transgenic plants were fed to locusts for 1 to 3 d, and the locusts were then subjected to behavioral analysis and qPCR assays. The control insects were fed WT plants. The locusts were inspected each day during their feeding on rice plants, and the developmental phenotypes of the nymphs were recorded.

Statistical Analysis.

SPSS 20.0 software (SPSS Inc.) was used for statistical analysis. The Mann–Whitney U test was used to analyze the behavioral data due to nonnormal distribution characteristics. The differences between treatments were compared using either Student’s t test or one-way ANOVA followed by Tukey’s test for multiple comparisons. The differences were considered statistically significant at P < 0.05. All results are expressed as means ± SEM.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Data, Materials, and Software Availability

The small RNA libraries were deposited in the Sequence Read Archive database (accession number: PRJNA867122). RNA-seq data were deposited in the NCBI Sequence Read Archive Database (accession number: PRJNA867152). All study data are included in the article and/or SI Appendix.