Proteomic analysis reveals that COP9 signalosome complex subunit 7A (CSN7A) is essential for the phase transition of migratory locust

The migratory locust displays a reversible, density-dependent transition between the two phases of gregaria and solitaria. This phenomenon is a typical kind of behavior plasticity. Here, we report that COP9 signalosome complex subunit 7A (CSN7A) is involved in the regulation of locust phase transition. Firstly, 90 proteins were identified to express differentially between the two phases by quantitative proteomic analysis. Gregaria revealed higher levels in proteins related to structure formation, melanism and energy metabolism, whereas solitaria had more abundant proteins related to digestion, absorption and chemical sensing. Subsequently, ten proteins including CSN7A were found to reveal differential mRNA expression profiles between the two phases. The CSN7A had higher mRNA level in the gregaria as compared with the solitaria, and the mRNA amount in the gregaria decreased remarkably during the 32 h-isolation. However, the mRNA level in the solitaria kept constant during the crowding rearing. Finally and importantly, RNA interference of CSN7A in gregaria resulted in obvious phase transition towards solitaria within 24 h. It suggests that CSN7A plays an essential role in the transition of gregaria towards solitaria in the migratory locust. To our knowledge, it’s the first time to report the role of CSN in behavior plasticity of animals.

Scientific RepoRts | 5:12542 | DOi: 10.1038/srep12542 In recent years, many fruitful studies have been carried out to elucidate the intrinsic molecular mechanisms of phase transition in locusts from various aspects such as genomics, transcriptomics and metabolomics. A large scale of transcriptomic sequencing was carried out in the migratory locust using an expressed sequence tag (EST) technique in 2004 11 , and 532 differentially expressed unigenes were identified between the two phases. The transcriptome dynamics in the same species were further analyzed in 2010 based on a newly emerged next-generation sequencing technology 12 . A lot of genes related to neural pathway, such as dopamine receptor, adipokinetic hormone, neurotransmitter synthetase were found to be up-regulated in the gregarious locusts. Another transcriptomic analysis was performed in the desert locust 13 . The solitary locusts up-regulate genes related to antioxidant systems, detoxification and anabolic renewal, whereas gregarious locusts have a greater abundance of transcripts for genes involved in sensory processing and nervous system development and plasticity. After monitoring and comparing transcript profiles between the two phases at various developmental stages, Chen et al. found that a sharp rise in phase differences appeared during the 4 th instar and the high level difference was maintained in all the following stages. Therefore, the 4 th instar stage seems to be a turning point in the process of forming the phase differences in the migratory locust 12 . Some neuronal signaling and sensory activity related genes, such as dopamine receptor 5 , chemosensory protein (CSP) and takeout 3 were proved to play roles during the phase transition. The successful assembly of the migratory locust genome is a milestone in the study of phase transition of locust 14 . The genome is 6.5 giga base pairs (Gb), the largest animal genome sequenced so far. Significant expansion of gene families associated with energy consumption and detoxification were found in the locust genome 14 . Besides, small RNA 15 and metabolomics 4 analysis also disclosed a lot of regulators contributing for the phase transition.
Proteomic researches have also been carried out, but few significant progresses have been made till now. In 1999, polypeptide maps were generated from hemolymph of the desert locust and twenty differential spots were identified between the two phases. However, detailed information about these peptides was not available 15 . Two proteins, a 6-kDa peptide and a serine protease inhibitor were identified to have different expression patterns between the two phases in the desert locust using a combined approach of high-performance liquid chromatography (HPLC) with matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry (MALDI-TOF MS) 16 . These proteomic studies moved very slowly because of the lack of locust genome information at that time.
The behavior plasticity makes locusts to be a good model in the epigenetic researches 17,18 . Two DNA methyltransferase genes were shown to be phase-specific in certain tissues of the desert locust 19 . Further analysis revealed that the methylome of the gregarious desert locust was characterized by CpG-and exon-specific methylation, and the overall methylation levels were substantially higher than other invertebrates 20 . These findings suggest that DNA methylation may be involved in the regulation of locust phase transition. Besides, a cAMP-dependent protein kinase (PKA) was reported to play a role in the transition from solitary to gregarious behavior in the desert locust 21 . Except for these reports, few studies have been further performed in recent years.
In general, large progresses have been made in exploring the mechanisms of phase transition in the migratory and desert locusts. A lot of differentially expressed genes and pathways have been identified based on DNA sequencing techniques. However, the researches in protein areas, such as protein identification and protein modification, have been largely lagged. One of the most key reasons is the lack of genome sequence information. Fortunately, the genome assembly of the migratory locust was just finished 14 , which provides much convenience for protein identification and will do great help for exploring the complex mechanism of phase transition in another viewpoint.
In the present study, we identified 90 differentially expressed proteins between the two phases in the migratory locust by a quantitative proteomic technique. Among them, CSN7A was found to play an essential role in the transition of gregaria towards solitaria.

Results
Proteins identified in the locust head. A total of 4, 895 peptides were identified by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) from the locust head, and they were finally assembled into 1, 387 proteins. After COG classification, 1, 104 proteins were assigned to 25 COG categories, and the "R" cluster (General function predication) and "O" cluster (Posttranslational modification, protein turnover, chaperones) represent the largest two groups, and their amounts are 18% and 15% of the total identified proteins, respectively (Fig. 1). The "O" cluster proteins are mainly heat shock protein chaperones, ubiquitin-dependent proteins, proteasome-related proteins, peptidase activity-related proteins, glutathione S-transferase, protein disulfide-isomerase, and COP9 signalosome complex subunits (Supplementary Table S1). The top 70 most abundant proteins are listed in Table 1  and Supplementary Table S2 Differentially expressed proteins between the two phases. Among the 1, 387 identified proteins, 90 proteins were shown to have different expression levels between the two phases. Sixty-four were up-regulated in the gregaria (as compared with the solitaria), and twenty-six were down-regulated ( Table 2, Supplementary Table S3). Most of the up-regulated proteins are involved in the processes of structure formation (such as cuticle protein, beta-1 tubulin, profiling, and troponin), energy metabolism (electron transfer flavoprotein subunit alpha [ETFA], dehydrogenase/reductase SDR family member 11-like, 3-ketoacyl-CoA thiolase, V-type proton ATPase subunit B and isocitrate dehydrogenase Based on sequence homology, 1, 104 proteins were classified into 25 COG (Clusters of orthologous groups) categories (A). The "R" and "O" clusters represent the largest two groups. Proteins classified in the "O" cluster (Posttranslational modification, protein turnover, chaperones) were further assigned to different categories according to their molecular functions (B). Detailed information about these "O" cluster proteins can refer to Supplementary Table S1. Continued NAD subunit beta), and environmental stress response (heat shock protein 60 [Hsp60] and heat shock protein 20.6 [Hsp20.6]). Besides, four hexamerin-like proteins are also abundant in the gregaria. The down-regulated proteins are mainly related to the processes of digestion and absorption (carboxypeptidase A-like, serine protease-like protein, and 1,4-alpha-glucan-branching enzyme-like) and chemical sensing (takeout-like). In addition, the differentially expressed proteins are also enriched in the class of "Regulation of gene expression" in both the gregaria and solitaria ( Table 2). For example, wingless protein, 3′ -phosphoadenosine 5′ -phosphosulfate synthase (PAPSS), COP9 signalosome complex subunit 7A (CSN7A), and juvenile hormone binding protein (JHBP) were highly expressed in the gregaria, and splicing factor 3B subunit, ubiquitin-conjugating enzyme E2 variant 2-like isoform 1, proteasome subunit alpha type-4, and arginine/serine-rich-splicing factor RSP31 (RSP31) showed higher levels in the solitaria.
Differential expression at mRNA levels. To validate the differential expression, fourteen representative proteins were selected according to their function categories in Table 2. Their mRNA expression profiles were examined in the whole head of the two-phase locusts. Nine protein genes, including CSN7A, JHBP, PAPSS, choline transporter-like protein 4 (CTL-4), two hexamerin-like protein 2 (Hexa2 and Hexa2*), cytoplasmic actin A3a (actinA3a), ETFA and Arylphorin revealed higher mRNA level in gregaria ( Fig. 2). It was in consistent with the protein profiles in Table 2. The brain tissues, the most important part of head, were also studied. Four genes, such as CSN7A, JHBP, PAPSS, CTL-4 and takeout-like, showed similar expression patterns between the mRNA and protein levels. There were still four genes, including V-ATPase subunit B (V-ATPase), ATPsyn-d, RSP31, and NADPH--cytochrome P450 (P450) revealed constant mRNA levels between the two phases ( Supplementary Fig. S1).

Time-dependent mRNA expression during phase transition.
In order to further narrow target proteins that may play a role in the regulation of locust phase transition, CSN7A was chosen and time-dependent mRNA expression dynamics were examined in brain during the phase transition process. The CSN7A had higher mRNA level in the gregaria (Figs 2, 3), the level decreased significantly at 4, 16 and 32 h-isolation and was as low as that in the solitaria at 32 h (Fig. 3). However, the mRNA level did not change during the crowding of solitary locusts (Fig. 3).

RNA interference (RNAi) and behavioral assay.
To validate the function of CSN7A in locust phase transition, RNAi and behavioral assay were carried out. The mRNA level was suppressed by injection of CSN7A dsRNA in the gregaria (Fig. 4A), and the behavioral state shifted from gregaria (dsGFP population) to solitaria (dsCSN7A population) (Fig. 4B). The phase difference between two populations was highly significant (P Mann-Whitney U test = 1.61 × 10 −12 ). For example, 60% and 0% individuals fall into the P greg interval of 0.8-1.0 in the dsGFP and dsCSN7A population, respectively. In addition, significant difference existed in the three key behavioral parameters (attraction index, total distance moved, and total duration of movement) between the two populations (Fig. 4C). These results revealed that phase transition did happen by RNAi of CSN7A in the gregarious locust.

Discussion
The "O" cluster proteins are extremely abundant in the locust head. This phenomenon was also found in the antennae of Batocera horsfieldi based on cDNA library analysis 22 . However, similar phenomenon did not exist in the whole insect bodies [23][24][25][26] . It seems that "O" cluster proteins are mainly abundant in the head as compared with the other parts of insects. It suggests that the proteins related to post-translational modification, protein turnover and chaperone folding are highly involved in the regulation of head function in insects. Locust phase polyphenism is a typical phenomenon of epigenetics 17,19,26 . The existence of high abundant "O" cluster proteins suggests that post-translational modification may play important roles in the locust phase transition. The two locust phases differ in many aspects, especially in the body color and behavioral activity. The gregaria is darker and more active, while the solitaria is shallower and quieter. The proteomic analysis revealed that proteins related to structure formation, melanism and energy metabolism have significantly higher expression level in the gregaria. This is consistent with the facts that gregarious locusts have stronger muscles, darker color and more frequent activity. As compared with the gregaria, the solitaria owns more abundant proteins related to digestion, absorption and chemical sensing. It's apparently that the former two characteristics provide the solitary locusts with higher abilities in digestion and absorption, and the latter one gives them stronger olfactory sensation. This makes them have an advantage over the gregarious locusts in feeding and mating, and then results in higher reproductive capacity 14 .
In the present study, hexamerins and JHBP are abundant in the head of gregarious locust. Similar results have been revealed by EST library analysis in the same species 11 . Both hexamerin and JHBP have been suggested to play a role as juvenile hormone (JH) transporters, and even as regulators of JH levels and action [27][28][29][30] . This explains the involvement of hexamerins in JH-dependent differentiation of caste phenotype in some social insects, including termite Reticulitermes flavipes [31][32][33][34] , honey bee 35 and wasp Polistes metricus 36 . Besides caste-related polyphenism in social insects, JH was also reported to mediate plasticity of aggregation behavior in adult desert locusts 37 . Surgical removal of the corpora allata to terminate JH secretion increased aggregation index and behavioral activity of adult locust. This effect was caused by repressing the responsiveness of olfactory interneurons in the antennal lobe to aggregation pheromone. Thus, hexamerins and JHBP can be involved in the phase plasticity of locust by mediating JH action.
Heat shock proteins (Hsps) are a kind of stress-induced proteins that can be synthesized rapidly in response to various environmental stress signals. Hsps usually function as molecular chaperones and participate in numerous cellular functions such as folding, assembly, intracellular localization, secretion, regulation and degradation of proteins 38,39 . Gregarious locusts live at high population density. Population density can alter the expression of Hsps. For example, the mRNA levels of five Hsps (Hsp20.5, Hsp20.6, Hsp20.7 and Hsp90) are significantly higher in the gregarious locust head as compared with those in the solitaria. The mRNA levels were up-regulated by crowding of the solitary locusts (for 32 h), and down-regulated by isolation of the gregarious locusts 40 . In the present study, Hsp60 and Hsp20.6 were identified to have higher protein levels in the gregarious locust head. The over-expression of Hsps in gregaria seems to be a direct response to high-population gather of locust. It's hard to distinguish whether Hsps play a role to control the phase transition.
In the desert locusts, two phase populations display different sensitivity to aggregation pheromone 10,41,42 . Chemosensory protein (CSP) and takeout are important proteins for olfactory sensing [43][44][45][46][47] . RNA interference combined with olfactory behavioral experiments confirmed that six CSP genes (CSP-1 to 6) and one takeout gene, LmigTO1, are responsible for the formation of gregarious and solitary behaviors, respectively 3 . In our study, another CSP (CSP-7) and three new takeout proteins (TO 4 to 6) were identified from the head of Locust migratoria (Supplementary Fig. S2), and the CSPs revealed higher protein level in the gregaria, while the TOs showed higher protein levels in the solitaria. These protein expression patterns are consistent with the early report at mRNA levels 3 , and further confirm that both CSP and takeout are involved in the phase plasticity of locust.   The mRNA expression profiles were examined by qRT-PCR in both the head and brain tissues. The mRNA levels were quantified by standard curves generated with serial (10×) dilutions of plasmid DNAs. The relative expression level of each target gene was normalized against a house-keeping gene (RP49). Differences between treatments were compared by Student's t-test, and two levels (P < 0.05 or 0.01) were adopted to judge the significance of difference. Abbreviations: "G", gregaria; "S", solitaria. The abbreviation for gene names can refer to Table 3.
Scientific RepoRts | 5:12542 | DOi: 10.1038/srep12542 The CSN, an eight protein complex (CSN1-8) 48 was originally discovered as an essential regulator in light-induced development in Arabidopsis thaliana 49 . In Drosophila melanogaster, it also plays an essential role for development. Disruption of one of the subunits caused lethality at the late larval or pupal stages 50 . This role of CSN is partly due to its regulation on Hedgehog signaling by mediating proteolysis of some transcription factors 51 . In the same species, CSN was also reported to be involved in circadian rhythms by controlling the degradation of two clock proteins 52 . Interestingly, our study showed that CSN7A played a role in the phase transition from gregaria to solitaria in the migratory locusts. RNAi of CSN7A triggered the phase shift from gregaria to solitaria within 24 h (Fig. 4). Isolation (gregaria to solitaria) and crowding (solitaria to gregaia) may have different regulation mechanism. The former takes place within 4 h in the migratory locusts, whereas the latter cannot finish until 32 h 3 . In the present study, the mRNA amount of CSN7A in gregaria decreased during the isolation, however, the mRNA level remained constant during the crowding of solitaria (Fig. 3). It suggests that CSN7A may be only involved in one direction transition from gregaria to solitaria rather than in its reverse process.
It is the first time to disclose the role of CSN in behavior plasticity of animals. CSN has been reported to be involved in neural development, and regulates dendritic morphogenesis in Drosophila brain through Cullin-mediated protein degradation 53 . More and more evidences revealed that CSN plays an important role in protein degradation through Cullin-ubiquitin-proteasome pathway [54][55][56] . Therefore, CSN might be involved in the phase transition of locust by mediating ubiquitin-dependent proteolysis. Further studies need to be carried out to explore the detailed mechanism of CSN in the regulation of phase transition.
In conclusion, a total of 1,387 proteins were identified in the locust head in the present study, and a large proportion of proteins are involved in post-translational modification, especially in protein folding, phosphorylation and ubiquitylation. Ninety proteins were identified to differentially express between two phases in the head of the migratory locust. Gregaria reveals higher expression in proteins related to structure formation, melanism and energy metabolism, whereas solitaria owns more abundant proteins related to digestion, absorption and chemical sensing. This is consistent with their differentiation in morphology and physiology. JHBP, hexamerin, Hsp, CSP and takeout are suggested to play a role in behavior formation according to their differential expression profiles between two phases. The most interestingly, RNAi of CSN7A in gregaria made the behavior shift towards solitaria within 24 h. It is the first time to disclose the role of CSN in behavior plasticity of animals. These results provide important information for further exploration of the complex mechanism of locust phase transition, as well as for the study of behavior plasticity of animals.  The mRNA expression dynamics were examined by qRT-PCR in the brain during phase transition. To make the gregarious behavior change towards solitaria, the 4 th instar gregarious nymphs were individually reared at the same condition as solitary ones. After 2, 4, 8, 16 and 32 h of isolation, the brains were dissected and sampled. Similar procedure was used to convert solitary individuals towards gregaria. The sampling and mRNA level detecting methods were as same as the isolation of gregaria. The untreated gregarious and solitary locusts were used as controls. Differences between each treatment and the corresponding control (untreated gregaria or solitaria) were compared by Student's t-test, and two levels of significance (P < 0.05 or 0.01) were adopted to judge the significance of difference.

Methods
Sample preparation and iTRAQ labeling. When the locusts developed into the second day of 4 th instar, the heads of 3 to 5 gregarious or solitary nymphs were collected and thoroughly homogenized in 500 μ L cold PBS buffer including 1 mM PMSF, 2 mM EDTA and 10 mM DTT. The samples were centrifuged for 20 min at 25,000 × g, and the supernatant was collected. A total of 100 μ g of protein per sample was reduced, alkylated, and then digested by adding 2 μ g trypsin (1 μ g/μ L) at 37 °C overnight. The digested samples were lyophilized and re-suspended in 100 μ L of 0.5 M TEAB (triethylammonium bicarbonate). The method of isobaric tags for relative and absolute quantitation (iTRAQ) was adopted for sample labelling according to the protocol of iTRAQ ® Reagents-4plex Applications Kit (AB Sciex Pte. Ltd., Foster City, USA). Each sample was labeled with an isobaric tag. The iTRAQ-labeled peptide mixtures were pre-separated by strong cation exchange (SCX) column. For SCX chromatography, the LC-20AB HPLC Pump system (Shimadzu Corporation, Chiyoda-ku, Tokyo, Japan) was used, the peptide sample was reconstituted with 4 mLbuffer A (25 mM NaH 2 PO 4 in 25% ACN, pH2.7) and then loaded onto a 4.6 × 250 mm Ultremex SCX column containing 5-μ m particles (Phenomenex, Torrance, CA, USA). The peptides was eluted at a flow rate of 1 mL/min with a gradient of buffer A for 10 min, 5-35% buffer B (25 mM NaH 2 PO4, 1M KCl in 25% ACN, pH2.7) for 11 min, 35-80% buffer B for 1 min. The system was then maintained in 80% buffer B for 3 min before equilibrating with buffer A for 10 min prior to the next injection. Elution was monitored by measuring absorbance at 214 nm, and fractions were collected every 1 min. The eluted peptides were pooled as 12 fractions, desalted by Strata X C18 column (Phenomenex, Torrance, CA, USA) and vacuum-dried. Each fraction was resuspended in certain volume of buffer A (2% ACN, 0.1% FA).

LC-MS/MS Analysis.
A total of 5 μ g of the above solution was loaded on a LC-20AD nanoHPLC (Shimadzu Corporation, Chiyoda-ku, Tokyo, Japan) equipped with a 2 cm C18 trap column, and the peptides were then eluted onto a resolving 10 cm analytical C18 column. The MS data acquisition was performed with Triple TOF 5600 System (AB SCIEX, Concord, ON) fitted with a Nanospray III source (AB SCIEX, Concord, ON) and a pulled quartz tip as the emitter (New Objectives, Woburn, MA). Data was acquired using an ion spray voltage of 2.5 kV, curtain gas of 30 PSI, nebulizer gas of 15 PSI, and an interface heater temperature of 150 °C. The MS was operated with a resolving power of greater than Three key behavioral parameters (attraction index, total distance moved, and total duration of movement) were compared between the dsCSN7A and dsGFP populations (C). Isolation of gregaria referred to the method in Fig. 3, and the behavioral assay was then performed in a rectangular arena monitored by EthoVision system. Eleven behavioral parameters (such as attraction index, total distance moved, total duration of movement, etc.) were collected to calculate the possibility of gregaria (P greg ), which were used for criterion of phase type. The behavioral data were analyzed by the Mann-Whitney U test. The phase difference between two populations was highly significant (P Mann-Whitney U test = 1.61 × 10 −12 ). Differences between treatments were compared by Student's t-test, and two levels of significance (P < 0.05 or 0.01) were adopted to judge the significance of difference. Individual numbers of gregaria and solitaria were marked directly on top of the figure.
Scientific RepoRts | 5:12542 | DOi: 10.1038/srep12542 or equal to 30,000 FWHM (full width at half maximum). The MS/MS data collection and processing was done on Analyst ® software (version 1.6, AB SCIEX, Concord, ON) with the method of Information Dependent Acquisition (IDA) according to the manual.
Database searching for protein identification. The resulting MS/MS spectra were searched against the locust protein database generated from the newly assembled genome 7 with MASCOT software (Matrix Science, London, UK; version 2.3.02). The carbamidomethylation of cysteine was considered a fixed modification, and the conversion of N-terminal glutamine to pyroglutamic acid and methionine oxidation were considered variable modifications. The minimal peptide length was seven amino acids, and a single missed cleavage maximum was used. A peptide mass tolerance of 10 ppm was allowed for intact peptide masses and 0.05 Da for fragmented ions. A stringent 0.01 false discovery rate (FDR) threshold was used to filter the candidate peptide and protein. Two thresholds were set up to filter the candidate proteins whose abundances were significantly different from others: < 0.05 for a two-tailed P-value test and > 1.5 (or < 1/1.5) for the fold-change. For gene ontology (GO, http://www.geneontology. org/) mapping, BLAST2GO software (version 2.5.0, http://www.blast2go.org) was employed to deal with the BLASTx results and then to perform the functional annotation by GO vocabularies, enzyme classification codes, KEGG metabolism pathways 57 . The default settings of BLAST2GO were used in every annotation step.
Quantitative real-time PCR (qRT-PCR). Total RNAs were extracted from the whole head and dissected brain tissues, respectively using an RNAeasy mini kit (QIAGEN, Hilden, Germany). Three heads or eight brains were used for each RNA isolation, and five biological repeats were performed during sampling. PCR reactions were performed in a 20 μ L volume and the final concentration of primers was 250 nM. PCR amplification was conducted on a Roche Light Cycler ® 480 system (Roche Applied Science, Penzberg, Germany) using SYBR green master mix (Roche Diagnostics Ltd. Shanghai, China). The PCR was initiated with a 10-min incubation at 95 °C, followed by 45 cycles of 10 s at 95 °C, 20 s at 58 °C and 20 s at 72 °C. Five biological replicates were performed for each sample. The standard curves for target genes and reference genes (ribosomal protein 49, RP49) were generated with serial (10× ) dilutions of plasmid DNAs. Efficiency of qRT-PCR and correlation coefficients were determined for the primers of each gene. The relative expression level of each target gene was normalized against RP49. The specificity of amplification was ensured by both melting curve analysis and sequencing of PCR product. The primers for qRT-PCR were listed in Table 3.
RNAi. Double-strand RNA (dsRNA) of the target gene and a negative control gene (green fluorescent protein, gfp) were prepared using the T7 RiboMAX Express RNAi system (Promega, Madison, USA) according to the manufacturer's instruction. The primers for dsRNA preparation were listed in Table 3. A total of 35 ng dsRNA was injected directly into eight brains of the 4 th instar nymphs using Nanoject II nanoliter injector (Warner Instruments, Hamden, CT, USA). Twenty four hours later, the effects of RNAi on mRNA level were detected by qRT-PCR and behavioral assay. Four biological repeats were performed for qRT-PCR. For behavioral assay, the same injection was carried out in gregaria, and 30 and 36 individuals were used for dsGFP and dsCSN7A, respectively. Phase Transition. To make gregarious behavior change towards solitaria, the 4 th instar gregarious nymphs were individually reared at the same condition as solitary ones. After 2, 4, 8, 16 and 32 h of isolation, the brains were dissected and immediately placed in RNAlater Solution (Ambion, Austin, USA) for qRT-PCR analysis. The gregarious nymphs maintained in normal situation (high population density) were used as controls. To avoid the influences of circadian rhythm and sexual difference, all samples were collected at the same time point of a day with a sex ratio of 1:1. Each treatment included five biological replicates. To make a reverse phase transition (solitaria towards gregaria), ten solitary nymphs were marked and moved into an optic perplex-made box (10 × 10 × 10 cm 3 ), and 20 gregarious individuals were then added to maintain high population density. The sampling, mRNA level detecting and other methods were as same as the isolation of gregaria.
Behavioral assay. The behavioral assay was performed in a rectangular arena (40 × 30 × 10 cm 3 ). The wall of the arena is opaque plastic and the top is clear. One of the separated chambers (7.5 × 30 × 10 cm 3 ) contained 20 4 th instar gregarious locusts as the stimulus group, and the other end of the chamber with the same dimensions was kept empty. Both ends of the chamber were illuminated equally to prevent the formation of mirror images. The floor of the open arena was covered with filter paper during the behavioral assay. The locust nymphs were gently transferred by a tunnel to the arena. Each individual was recorded for 6 min using EthoVision system (Noldus Inc. Wageningen, the Netherlands). Eleven behavioral parameters (such as attraction index, total distance moved, total duration of movement, etc.) were collected to calculate the possibility of gregaria (P greg ), which was used for criterion of phase type. Detailed information can refer to the early reported methods 3,6 . Statistical analysis. Differences between mRNA levels were compared by Student's t-test. The relative mRNA levels were presented as mean ± SEM (standard error of the mean). Behavioral data were analyzed by the Mann-Whitney U test. Two levels of significance (P < 0.05 or 0.01) were adopted to judge the significance of difference. All the statistics was analyzed using SPSS 15.0 (SPSS Inc., Chicago, USA).