The homotetramerization of a GPCR transmits the 20-hydroxyecdysone signal and increases its entry into cells for insect metamorphosis

In mammals, estrogen binds to its nuclear estrogen receptors (ERα and ERβ) to regulate gene transcription (Dahlman-Wright et al., 2006). However, a G protein-coupled receptor (GPCR), GPR30/GPER, was identified as a cell membrane receptor for estrogen (Revankar et al., 2005). Similarly, in insects, the steroid hormone 20-hydroxyecdysone (20E) binds to the nuclear receptor (ecdysone receptor, EcR) to form a transcription complex with their heterodimeric partner ultraspiracle (USP) to initiate gene transcription (Riddiford et al., 2000). However, 20E also triggers a rapid intracellular Ca²⁺ increase via GPCRs in the Bombyx mori anterior silk gland (Manaboon et al., 2009). A GPCR, dopamine/ecdysteroid receptor (DopEcR), is considered as a 20E cell membrane receptor because it binds 20E to regulate non-genomic effects of ecdysteroids in Drosophila melanogaster (Srivastava et al., 2005). Two GPCRs, ecdysone responsive (Er)GPCR-1 (Cai et al., 2014) and ErGPCR-2 (Wang et al., 2015) transmit the 20E signal via the nongenomic pathway in Helicoverpa armigera. ErGPCR-2 and DopEcR from H. armigera can bind 20E and function as 20E cell membrane receptors (Kang et al., 2019). Therefore, multiple GPCRs are likely to function as 20E receptors; however, whether any other GPCRs function in the 20E pathway and their associated mechanisms is unknown.

GPCRs can form homo- or hetero-oligomers to exert their functions (Cheng et al., 2003; Cheng and Miller, 2001). For example, a family B GPCR is present as a constitutive homodimer in the plasma membrane of pancreatic cancer cell lines (Ding et al., 2002). Oligomerization of GPCRs can be constitutive or ligand-dependent (Ding et al., 2002; Kroeger et al., 2001; Pfleger and Eidne, 2005). However, whether a GPCR functions as an oligomer in steroid hormone signaling, and the regulatory mechanism and outcomes, remain unclear.

Animal steroid hormones are lipophilic hormones that enter cells by passive diffusion across the plasma membrane (Gorski and Gannon, 1976; Plagemann and Erbe, 1976). However, some studies have suggested that steroid hormones require protein transporters for their cellular entry (Milgrom et al., 1973; Pietras and Szego, 1977). Ecdysteroid is imported into cells by a transporter, Ecdysone Importer (EcI), instead of by passive diffusion in D. melanogaster (Okamoto et al., 2018). Our previous work has shown that a GPCR, ErGPCR-2, controls 20E entry into cells in H. armigera (Wang et al., 2015); however, the mechanism and outcomes are unclear.

In the present study, a new ecdysone-responsive GPCR in H. armigera, named ErGPCR-3, was identified as a cell membrane receptor of 20E that transmits the 20E signal and promotes metamorphosis. ErGPCR-3 could bind 20E in the cell membrane and after being isolated from the cell membrane. 20E binding induced homotetramerization of ErGPCR-3. The ErGPCR-3 homotetramer transmits 20E signaling, and increases the entry of 20E into cells. This is the first example of a GPCR functioning as a homotetramer in steroid hormone signaling, and it increases our understanding of GPCR-mediated 20E signaling and supports the facilitated diffusion model of steroid hormones in lepidopterans.

Thirteen genes encoding GPCRs, which were identified from the transcriptome of the H. armigera epidermal cell line (HaEpi), were examined in the sixth instar 24 h larval epidermis using quantitative real-time reverse transcription PCR (qRT-PCR) under 20E induction (500 ng/larva) for 20 h to screen for 20E-upregulated GPCRs. Six GPCR genes (GPCR-3, -49, -24, -5, -55 and -56) were upregulated by 20E induction, in which HR3 was used as a positive control and dimethyl sulfoxide (DMSO) was the solvent control (Fig. S1A). The six 20E-upregulated GPCR genes were knocked down by injecting double-stranded RNA (dsRNA) into the H. armigera sixth instar 6 h larval hemocoel to identify the GPCRs that are involved in 20E-directed gene expression. 20E induced the expression of ecdysone nuclear receptor EcRB1, the heterodimeric partner USP1, and transcription factors HR3 and BrZ7 in the epidermis; however, knockdown of ErGPCR-3 repressed these functions of 20E compared with knockdown of the other GPCR genes (Fig. S1B-G), suggesting that ErGPCR-3 plays a role in the 20E pathway. Phylogenetic analysis indicated that H. armigera ErGPCR-3 (MN150685) and its orthologs in insects are class B receptors (Secretin receptor Mth2-like) (Fig. S2).

To study the function of ErGPCR-3 in H. armigera development, the developmental expression profiles and hormonal regulation of ErGPCR-3 were examined. ErGPCR-3 was expressed in the epidermis, midgut and fat body. The protein levels of ErGPCR-3 increased at the fifth instar molting (5 M) and metamorphic stages (sixth instar 72 h to 120 h) compared with the 5th instar feeding larvae (5 F), as assessed using western blotting with rabbit polyclonal antibodies recognizing ErGPCR-3 (Fig. 1A,A′). The expression of ErGPCR-3 in the epidermis was increased by 20E injection into the sixth instar 6 h larvae (Fig. 1B,B′). ErGPCR-3 expression was proven to be upregulated by 20E via the EcRB1-USP1 transcription complex via knockdown of EcRB1 and USP1 in HaEpi cells, the identification of an ecdysone response element (EcRE)-containing sequence (5′-56-GCGTTCATAGAACTT-42-3′ before the ATG codon) in the 5′-upstream genomic DNA region of ErGPCR-3, and using chromatin immunoprecipitation (ChIP) (Fig. S3). Immunohistochemistry showed that ErGPCR-3 was abundant and localized in the larval midgut of sixth instar 96 h larvae (Fig. 1C), suggesting that ErGPCR-3 might play an important role in the 20E pathway.

To examine the function of ErGPCR-3 in 20E-regulated metamorphosis, we knocked down ErGPCR-3 via injection of dsErGPCR-3 into sixth instar 6 h larvae. 20E promotes earlier pupation; however, after knockdown of ErGPCR-3, 20E did not promote pupation, but appeared to cause pupation delay, pupation failure and emergence failure, compared with larvae injected with dsGFP plus 20E (Fig. 2A,B). 20E injection accelerated the initiation time of pupation to 28 h on average. However, 20E could not accelerate pupation after injection of dsErGPCR-3 plus 20E, and pupation time was delayed by 38 h compared with that in larvae injected with dsGFP plus 20E injection (Fig. 2C). In the animals showing delayed pupation after dsErGPCR-3 plus 20E treatment, the rates of normal pupation, pupation failure and emergence failure were 43%, 26% and 21%, respectively (Fig. 2D). After ErGPCR-3 was significantly knocked down in the midgut, 20E could not upregulate the expression of the key genes in the 20E pathway, including EcRB1, USP1, HR3 and BrZ7 (Fig. 2E). Hematoxylin and Eosin (H&E) staining showed that the larval midgut did not separate from imaginal midgut after the first injection of dsErGPCR-3 plus 20E for 96 h, compared with the larvae injected with the control dsGFP plus 20E for 96 h. The fat bodies of the dsErGPCR-3 plus 20E-treated larvae were still closely arranged; however, the dsGFP plus 20E control larvae showed initial fat body degradation (Fig. 2F). These results suggest that ErGPCR-3 participates in 20E-regulated metamorphosis and tissue remodeling.

dsRNA-mediated knockdown of ErGPCR-3 was performed in sixth instar larvae to determine further the function of ErGPCR-3 during H. armigera growth and metamorphosis. Compared with that in the dsGFP injection control group, the larvae injected with dsErGPCR-3 showed abnormal pupation, pupation delay and pupation failure (Fig. S4A,B). The pupation of the dsErGPCR-3 injection group was delayed by about 29 h on average (recorded from sixth instar 0 h) compared with that in the dsGFP injection control group (Fig. S4C). The mRNA levels of EcRB1, USP1, HR3 and BrZ7 were downregulated after ErGPCR-3 was significantly knocked down in the midgut (Fig. S4D). The efficiency of RNAi of ErGPCR-3 was confirmed by western blotting (Fig. S4E). The midgut did not turn red after injection of dsErGPCR-3 for 96 h and the larval midgut did not separate from the imaginal midgut compared with the larvae injected with the control dsGFP injection. Similarly, the fat body of the dsGFP control larvae showed initial degradation, whereas the fat body of the dsErGPCR-3-treated larvae was still closely arranged (Fig. S4F-H). These results suggest that ErGPCR-3 is necessary for metamorphosis.

The involvement of ErGPCR-3 in 20E-induced Ca²⁺ influx and intracellular cAMP levels was detected in HaEpi cells to address its role in 20E signaling. 20E induced the rapid release of intracellular Ca²⁺ and extracellular Ca²⁺ influx. However, ErGPCR-3 knockdown (the RNAi efficiency of ErGPCR-3 is shown in Fig. S5A) decreased the 20E-induced Ca²⁺ release and influx (Fig. 3A). Similarly, the 20E-induced increase in cAMP levels was decreased after ErGPCR-3 knockdown (Fig. 3B). These data suggest that ErGPCR-3 participates in the 20E-induced rapid cellular responses.

GPCRs transmit extracellular signals by activating different heterotrimeric G proteins (Bockaert and Pin, 1999). The s and q subunits of Gα stimulate the effector molecules adenylyl cyclase (AC) and phospholipase C (PLC), thereby increasing the intracellular concentration of cAMP and Ca²⁺, respectively (Marinissen and Gutkind, 2001). To address the mechanism by which 20E increases intracellular Ca²⁺ and cAMP levels via ErGPCR-3, the protein interactions between ErGPCR-3 and Gαq or Gαs were examined by co-overexpression of ErGPCR-3-RFP-His and either Gαq-GFP-His or Gαs-GFP-His in HaEpi cells, followed by co-immunoprecipitation (Co-IP) via the fused tags. Gαq-GFP-His was co-immunoprecipitated with ErGPCR-3-RFP-His in 20E-induced cells using the anti-RFP antibody when they were co-expressed in cells (Input), compared with the negative IgG control (Fig. 3C). Similarly, Gαs-GFP-His was co-immunoprecipitated with ErGPCR-3-RFP-His using the anti-RFP antibody in 20E-induced cells (Fig. 3D). These data suggest that ErGPCR-3 transmits signals via Gαs and Gαq.

20E induces rapid protein phosphorylation, including that of cAMP response element-binding (CREB) and protein kinase A catalytic subunit 1 (PKAC1) via cAMP signaling (Jing et al., 2016), and USP1 (Liu et al., 2014b) and cyclin dependent kinase 10 (CDK10) (Liu et al., 2014a) via calcium signaling. Therefore, to address the involvement of ErGPCR-3 in 20E-induced protein phosphorylation, we co-overexpressed these four proteins (with tags for antibody recognition), with or without ErGPCR-3 knockdown. 20E induced the phosphorylation of these proteins; however, knockdown of ErGPCR-3 repressed the 20E-induced phosphorylation of the four proteins (Fig. 3E-H). These data suggest that ErGPCR-3 participates in 20E-induced rapid protein phosphorylation.

To clarify the response of ErGPCR-3 to 20E, the phosphorylation and subcellular location of ErGPCR-3 were analyzed. ErGPCR-3 was internalized into the cytoplasm after 20E treatment for 15 min, and the findings compared with those in the DMSO treatment control (Fig. 4A). Western blotting analysis showed that the molecular mass of the internalized protein was higher than that in the cell membrane, and lambda protein phosphatase (λPPase) treatment decreased the apparent molecular mass (Fig. 4A′), suggesting that 20E induced ErGPCR-3 phosphorylation and internalization.

G protein-coupled receptor kinases (GRKs) deactivate GPCR signaling by phosphorylating the C-terminus of the receptor (He et al., 2017; Wang et al., 2015). To identify the relationship between the phosphorylation and internalization of ErGPCR-3, ErGPCR-3-GFP and a C-terminal deletion (amino acids 479-534) of ErGPCR-3 (ErGPCR-3^{ΔC−terminal}-GFP) were overexpressed in HaEpi cells, and GFP was overexpressed as a control. GFP was detected throughout the cell. ErGPCR-3-GFP-His appeared in the cell membrane under DMSO treatment, and was internalized by 20E induction; however, ErGPCR-3^{ΔC−terminal}-GFP-His remained in the cell membrane under both DMSO and 20E treatment (Fig. 4B). Western blotting showed that ErGPCR-3-GFP-His was phosphorylated after 20E induction; however, ErGPCR-3^{ΔC−terminal}-GFP-His remained unphosphorylated after 20E induction (Fig. 4B′). These results suggest that 20E-induced phosphorylation of the C-terminus of ErGPCR-3 determines the internalization of ErGPCR-3.

GPCR internalization is regulated by GRK phosphorylation, which promotes its interaction with β-arrestin (Benovic et al., 1989; Lohse et al., 1990). Clathrin participates in GPCR endocytosis (Chen et al., 2014) and chlorpromazine (CPZ) inhibits the recirculation of clathrin to the plasma membrane, thereby inhibiting clathrin-mediated endocytosis (Bhattacharyya et al., 2010). Therefore, the internalization of ErGPCR-3 was examined by knockdown of the genes encoding these proteins or by inhibiting clathrin using its inhibitor CPZ to address the mechanism of ErGPCR-3 internalization. Immunocytochemical analysis showed that ErGPCR-3 was internalized after 15 min of 20E induction in the dsGFP control cells. However, ErGPCR-3 was not internalized in the dsGRK2, dsβ-arrestin-1 (the RNAi efficiency of GRK2 and β-arrestin-1 are shown in Fig. S5B,C), or CPZ-treated cells (Fig. 4C), suggesting that ErGPCR-3 was internalized by 20E via GRK2, β-arrestin-1 and clathrin. Western blotting analysis further confirmed that ErGPCR-3 was phosphorylated and internalized into the cytoplasm by 20E induction in the dsGFP control cells; however, ErGPCR-3 was not phosphorylated or internalized by 20E induction after knockdown of GRK2 or β-arrestin-1, or CPZ treatment (Fig. 4C′), suggesting that 20E induced ErGPCR-3 phosphorylation and internalization via GRK2, β-arrestin-1 and clathrin.

Co-IP using the anti-RFP antibody further showed that β-arrestin-1-GFP-His could be co-immunoprecipitated together with ErGPCR-3-RFP-His under 20E induction, when they were co-overexpressed in the cells (Fig. 4D). Similarly, GRK2-GFP-His was co-immunoprecipitated with ErGPCR-3-RFP-His (Fig. 4E). These results suggest that 20E induces the internalization of ErGPCR-3 via GRK2 and β-arrestin-1.

Chloroquine (CQ; Sangon Biotech), a lysosome inhibitor (Gonzalez-Noriega et al., 1980), was used to analyze the fate of ErGPCR-3 in cells. ErGPCR-3 was localized in the cell membrane in the DMSO control, but localized in the cytoplasm after 20E induction. After the withdrawal of 20E, ErGPCR-3 levels decreased in the cytoplasm; however, ErGPCR-3 accumulated in the cytoplasm after the addition of CQ. The subcellular localization and degradation of ErGPCR-3 were not well observed in cells (Fig. 4F); however, western blotting after subcellular fractionation showed that ErGPCR-3 was localized in the cell membrane in the DMSO control and was internalized into the cytoplasm by 20E. The level of ErGPCR-3 in the cytoplasm decreased after the removal of 20E for 2 h in CQ-free medium. However, the level of ErGPCR-3 in the cytoplasm was maintained after removing 20E in the presence of CQ (Fig. 4F′). These results suggest that ErGPCR-3 is degraded in the lysosome after internalization.

To determine whether ErGPCR-3 is a 20E receptor, a binding experiment was performed. Surflex-Dock (SFXC) in the SYBYL X 2.0 software was used to conduct a computational docking of 20E to ErGPCR-3. The model showed that 20E bound ErGPCR-3 near the transmembrane helix. 20E formed two hydrogen bonds with residues M240 and L331 of ErGPCR-3 (Fig. 5A). To prove that ErGPCR-3 binds 20E, ErGPCR-3-GFP and its mutant ErGPCR-3-M-GFP (with mutations of the 20E binding sites of ErGPCR-3-GFP; Table S2) were overexpressed in Sf9 cells. GFP in the control was distributed throughout the whole cell, and both ErGPCR-3-GFP and its mutant ErGPCR-3-M-GFP were localized in the cell membrane, with wheat germ agglutinin (WGA) staining indicating the cell membrane (Fig. 5B). A binding assay using 20E Enzyme Immunoassay (20E-EIA) showed that the amount of cell membrane-bound 20E in ErGPCR-3-GFP-overexpressing cells was significantly higher than that in ErGPCR-3-M-GFP- and GFP-overexpressing control cells (Fig. 5C). These results showed that ErGPCR-3 could bind 20E in the cell membrane.

The dissociation constant (Kd) of ErGPCR-3 binding 20E was further determined using a saturation binding curve detected using the 20E-EIA method with isolated ErGPCR-3. The purity of the isolated GFP, ErGPCR-3-GFP and ErGPCR-3-M-GFP was examined using sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) with Coomassie Brilliant Blue staining (Fig. S6). The saturable specific binding of ErGPCR-3-GFP to 20E had a Bmax=9.06±0.43 pmol/mg protein and a Kd=20.84±2.28 nM (mean±s.d.). However, ErGPCR-3-M-GFP decreased the saturable specific binding, with a Bmax=8.69±1.47 pmol/mg protein and a Kd=84.68±21.23 nM. GFP showed lower binding to 20E (Fig. 5D). The smaller the Kd value, the stronger the binding ability. The 20E-EIA assay was based on competition between the unlabeled 20E (20E bound to GPCR) and acetylcholinesterase-labeled 20E (Tracer) for the limited-specific rabbit anti-20E antiserum; therefore, an inhibition or competitive curve was not produced. HaEpi cells overexpressing the binding sites mutant, ErGPCR-3-M-GFP, had lower intracellular cAMP levels than those overexpressing ErGPCR-3-GFP under 20E incubation (Fig. 5E). Similarly, ErGPCR-3-M-GFP decreased the 20E-induced Ca²⁺ influx (Fig. 5F). These results suggest that the ErGPCR-3 binds 20E to transmit 20E signaling.

To address whether ErGPCR-3 functions as a monomer or an oligomer, ErGPCR-3 was overexpressed in HaEpi cells in various combinations, including ErGPCR-3-RFP-His and ErGPCR-3-GFP-His, ErGPCR-3-RFP-His and ErGPCR-1-GFP-His, ErGPCR-3-RFP-His and ErGPCR-2-GFP-His, ErGPCR-3-RFP-His and DopEcR-GFP-His. Co-IP using antibodies recognizing RFP resulted in the Co-IP of ErGPCR-3-GFP-His together with ErGPCR-3-RFP-His in both DMSO-treated cells and 20E-induced cells when they were overexpressed equally as Input, with IgG as the negative control (Fig. 6A), suggesting that ErGPCR-3 existed as a homodimer regardless of 20E induction. In the tag-control, RFP-His and GFP-His were not co-immunoprecipitated (Fig. 6B). In addition, ErGPCR-1-GFP-His, ErGPCR-2-GFP-His and DopEcR-GFP-His were not co-immunoprecipitated with ErGPCR-3-RFP-His (Fig. S7). These data suggest that ErGPCR-3 exists as a homodimer constitutively.

To confirm the homodimeric nature of ErGPCR-3 in vivo, suberic acid bis sodium salt [3-sulfo-N-hydroxysuccinimide ester, bis (sulfosuccinimidyl) suberate, BS3] was used to detect ErGPCR-3 oligomers in the larval epidermis using western blotting according to their molecular mass. BS3 can crosslink the protein molecules when they are associated together to prevent them dissociating during SDS-PAGE sample treatment (Sargiacomo et al., 1995). The endogenous monomer of ErGPCR-3 (60 kDa) was detected in both DMSO- and 20E-treated larval epidermis by western blotting without BS3 treatment. Surprisingly, a homotetramer (∼240 kDa) of ErGPCR-3 was detected abundantly in the 20E-treated sample after BS3 treatment. In contrast, the homodimer (∼120 kDa) of ErGPCR-3 was detected abundantly in the DMSO treatment control after BS3 treatment, although some homotetramer was detected that was likely formed by 20E in vivo (Fig. 6C). These data not only confirm that ErGPCR-3 exists as a homodimer in vivo, but also indicate that 20E induces ErGPCR-3 to form a homotetramer.

To address the mechanism of homotetramer formation, we examined the variation of homotetramer levels by overexpressing ErGPCR-3-GFP, ErGPCR-3^{ΔC−terminal}-GFP (truncated mutation of the phosphorylation region at the C terminal) and ErGPCR-3-M-GFP (multiple sites mutation of the 20E binding sites of ErGPCR-3-GFP) in HaEpi cells. Antibodies against GFP were used to detect the oligomers after BS3 treatment. The homodimer (∼190 kDa) was detected abundantly in the ErGPCR-3-GFP sample under DMSO treatment. The homotetramer (∼380 kDa) was induced in the ErGPCR-3-GFP samples after 20E treatment. ErGPCR-3^{ΔC−terminal}-GFP still formed a homotetramer after 20E treatment; however, ErGPCR-3-M-GFP could not form a homotetramer after 20E treatment (Fig. 6D). These results suggest that the 20E binding sites were essential for 20E-triggered formation of the ErGPCR-3 homotetramer.

The role of the ErGPCR-3 homotetramer in the entry of 20E into cells was examined in HaEpi cells and larval tissues. The 20E titer increased in the nuclei of cells when ErGPCR-3 was overexpressed in the cells. Moreover, the 20E titer was even higher after knockdown of GRK2, which normally arrests ErGPCR-3 in the cell membrane (Fig. S8), compared with the dsRFP control (Fig. 6E). However, the 20E level decreased in the nuclei of cells overexpressing ErGPCR-3-M-GFP, the binding site mutant, compared with that for ErGPCR-3-GFP under dsGRK2 and 20E treatment (Fig. 6F), in which conditions ErGPCR-3-M-GFP formed fewer homotetramers (∼380 kDa) than ErGPCR-3-GFP (Fig. 6G). These results suggest the ErGPCR-3 homotetramer correlates with 20E entry into cells.

To find evidence that ErGPCR-3 forms a homotetramer in vivo and its relationship with the 20E titer in tissues, we examined the wing and midgut at the sixth instar 96 h (wandering stage), at which point the wing is growing and the larval midgut is undergoing remodeling. The 20E titer was higher in the midgut than in the wing disk (Fig. 6H). Consistently, the amount of homotetramer (∼240 kDa) was higher in the midgut than that in the wing disk (Fig. 6I), presenting evidence that ErGPCR-3 forms a homotetramer and is related to 20E cell entry in vivo.

To understand the physiological significance of controlling the amount of 20E entering cells, the effect of different 20E concentrations on cell fates was examined. The proliferative signal, p-Histone H3, was detected in 16% of 20E-treated cells using 20E at a final concentration of 0.5 μM, compared with the DMSO-treated cells. In contrast, the p-Histone H3 signal decreased significantly in response to a high concentration 20E (5 μM) compared with the cells treated with DMSO (Fig. S9A,A′). Flow cytometry analysis showed that 0.5 μM 20E did nod induce apoptosis; however, higher concentrations of 20E (2–5 μM) induced apoptosis compared with DMSO treatment (Fig. S9B,B′). These data suggest that low concentrations of 20E promote HaEpi cell proliferation, whereas high concentrations promote apoptosis.

Several GPCRs are proposed to transmit steroid hormone signaling as cell membrane receptors in mammals and in insects; however, the mechanism is unclear. The present study found that ErGPCR-3 functions as a homotetramer to transmit 20E signal and increase 20E entry into cells (Fig. 7).

ErGPCR-3 is the fourth GPCR identified to bind and transmit 20E in H. armigera, following ErGPCR-1 (Cai et al., 2014), ErGPCR-2 (Wang et al., 2015) and DopEcR (Kang et al., 2019). Some differences have been observed in these GPCRs, e.g. DopEcR belongs to class A (Rhodopsin-like), whereas ErGPCR-1, ErGPCR-2 and ErGPCR-3 belong to class B (secretin receptor methuselah-2 family, Mth). ErGPCR-1 induces a Ca²⁺-PKC pathway, whereas ErGPCR-2 and DopEcR induce both GPCR-cAMP-PKA and GPCR-Ca²⁺-PKC pathways (Jing et al., 2016). ErGPCR-1 does not bind 20E; however, both ErGPCR-2 and DopEcR bind 20E (Kang et al., 2019). ErGPCR-3 also binds 20E to activate both PKA and PKC pathways. These data suggest that several GPCRs could serve as 20E cell membrane receptors by binding 20E, or serve as 20E signal transmitters without binding.

One possible reason that several GPCRs transmit the 20E signal and function as 20E cell membrane receptors is that Mth proteins are believed to have arisen from one ancestral gene and were subsequently duplicated in insects (Li et al., 2013). The physiological significance of their overlapping and/or redundant functions might be explained by the evolutionarily redundant replication of the 20E membrane receptor-related genes. Another reason might be the universally conserved positions in the barcode of G proteins that allow the receptors to bind and activate G proteins in a similar manner. Different receptors recognize the unique positions of the G-protein barcode through distinct residues, such as multiple keys (receptors) with non-identical shapes opening the same lock (G protein) (Flock et al., 2017). One ligand using multiple receptors is common in organisms. For example, in humans adrenaline uses several GPCRs (e.g. α1-adrenoceptors, α2-adrenoceptors and β-adrenoceptors) as its receptors (Alberts, 1993; Goldie et al., 1984; Ruffolo et al., 1995). The question of why many GPCRs are involved in same pathway, redundant or specific, might be understood after we obtain a detailed understanding the different functions of GPCRs.

After transmitting a signal, some GPCRs are internalized to desensitize them (Weinberg and Puthenveedu, 2019). GRKs and β-arrestin play important roles in GPCR internalization (Gurevich and Gurevich, 2019). Most GPCRs are phosphorylated by GRKs and internalized by interacting with β-arrestin (Ritter and Hall, 2009) and clathrin (Kang et al., 2014; Oakley et al., 1999); therefore, GPCRs are inhibited from further interactions with G proteins (Murga et al., 2019). ErGPCR-3 desensitization depends on a similar mechanism. This characteristic is quite similar to ErGPCR-2 (Wang et al., 2015), but different from either ErGPCR-1, which is desensitized by β-arrestin-1 binding without endocytosis (Zhang et al., 2015), and DopEcR, which is not internalized (Kang et al., 2019).

Class A and class B GPCRs can transmit signal as a single molecule; however, they are observed to exist as homo- or hetero-oligomers, and the biological roles of oligomerization are unclear (Gurevich and Gurevich, 2018). Ligand binding is able to induce dimers and oligomers of Class A GPCRs (Milligan et al., 2019). In the present study, we found that ErGPCR-3 forms a constitutive homodimer and is induced to form a homotetramer via 20E binding. This is the first example of a GPCR functioning as a homotetramer in steroid hormone signaling to transmit the signal and increase the cell entry of the hormone.

20E requires a cell membrane transporter, EcI, to enter the receiving cells from circulation, which illustrates that the steroid hormone enters the cell not by simple diffusion, but is facilitated by a transporter (Okamoto et al., 2018). EcI contains 12 transmembrane domains and is thought to transport ecdysone through a single positively charged pore; however, EcI is required for ecdysone import, but not for downstream ecdysteroid signaling (Kalliokoski and Niemi, 2009). Whether a GPCR homotetramer functions as a membrane transporter for 20E or not cannot be proven yet because of technical limitations. ErGPCR-3 signaling might affect the expression and/or subcellular localization of EcI, which in turn promotes 20E entry into the cells. This work presents evidence to support the facilitated diffusion of the steroid hormones in the lepidopteran H. armigera, in addition to the known cell membrane transporter EcI in D. melanogaster. The relationship between ErGPCR-3 and EcI, facilitating diffusion of 20E independently or interactively, is an intriguing question that needs further study.

A low concentration of 20E stimulated rapid cell proliferation in imaginal disks in B. mori (Koyama et al., 2004), and a high titer of 20E suppressed proliferation in imaginal disks and caused cessation of growth in peripheral tissues (Edgar, 2006). A high 20E concentration (5 μM) is necessary to initiate apoptosis in H. armigera by triggering activation of caspase-3 via increasing the calcium level and the cleavage of autophagy-related gene 5 (ATG5) (Li et al., 2016). The 20E titer varies from 0.1 μM during larval growth to over 9 μM during metamorphosis in H. armigera (Di et al., 2020; Kang et al., 2019). The 20E titer is similarly high in the hemolymph during metamorphosis; however, the wing disk is proliferating, and the larval midgut is experiencing programmed cell death (PCD) during this stage. Our results suggest a GPCR-mediated mechanism of 20E cell entry, which provides important implications for controlling wing disk and midgut development during metamorphosis.

Cotton bollworms (H. armigera) were reared in our laboratory on an artificial diet (Zhao et al., 1998) at 27°C under a 14 h light/10 h dark photoperiod and 40-50% relative humidity. HaEpi (Shao et al., 2008) was established and maintained at 27°C in Grace's medium supplemented with 10% fetal bovine serum (FBS; Invitrogen). The Sf9 cell line (Thermo Fisher Scientific) was cultured in ESF921 medium (Expression Systems) at 27°C without any serum or additives, unless mentioned. All cultured cells were monitored regularly for bacteria and mycoplasma contamination.

GPCRs were identified from the transcriptome of HaEpi cells treated with 2 μM 20E (Sigma-Aldrich) for 12 h. The mRNA levels of GPCRs in larvae were examined using qRT-PCR after injection of 20E or dsRNA, with an equal amount of diluted DMSO as a solvent control. The functional domains and motifs of the obtained proteins were predicted using SMART (http://smart.embl-heidelberg.de/). The molecular characterization of proteins was performed using the ExPASy ProtParam tool (https://web.expasy.org/protparam/). Multiple amino acid sequence alignment was performed using ClustalX and phylogenetic trees were constructed using the Neighbor-Joining (NJ) method in MEGA 7.0. The ligand binding site predictions of ErGPCR-3 were simulated using I-TASSER COACH (https://zhanglab.ccmb.med.umich.edu/COACH/) and PyMOL-v1.7 was used to visualize the three-dimensional structure and to label important structural features.

20E powder was dissolved in DMSO from the stock solution at 20 mM and diluted in phosphate-buffered saline (PBS) [140 mM NaCl, 10 mM sodium phosphate (pH 7.4)] when needed. 20E was injected into the sixth instar 6 h larval hemocoel. An equal amount of diluted DMSO was used as a control.

The detailed method for H. armigera dsRNA synthesis was described in a previous report (Chen et al., 2017). DNA fragments – 5′-649 bp-1102 bp-3′of ErGPCR-3, 5′-691 bp-1205 bp-3′ of USP1 (EU526832.1), 5′-127 bp-474 bp-3′ of EcRB1 (EU526831.1), 5′-1093 bp-1636 bp-3′ of GRK2 (KT364485.1) and 5′-740 bp-1171 bp-3′ of β-arrestin-1 (KP027422.1) – were amplified as the template for dsRNA synthesis using the primers RNAiF and RNAiR, which incorporated the T7 promoter (Table S1). The dsRNA was synthesized according to the method described in the MEGAscript RNAi Kit (Ambion). The phenol-chloroform method was used to purify the dsRNA. The purity and integrity of the dsRNA were examined using agarose gel electrophoresis, and then quantified using microvolume spectrophotometers (GeneQuant; Amersham Biosciences). The dsRNA for ErGPCR-3 was injected into the sixth instar larval hemocoel from the side of the front abdomen three times at 6, 24 and 48 h using a micro-syringe. Each injection comprised 500 ng/larva of dsRNA. The dsGFP was used as a control. After the third dsRNA injection, 500 ng of 20E/larva was injected into the larval hemocoel. An equal amount of DMSO was injected as a control. Thirty larvae were injected for each treatment and three independent biological replicates were performed. The phenotypes and developmental rates of the individuals were recorded daily. Total RNA and protein were extracted to verify the effects of RNAi 2 days after 20E injection.

HaEpi cells were transfected with dsRNA when the cells were cultured to 70% density. Transfection was performed using the Quick Shuttle-enhanced transfection reagent (Beijing Biodragon Immunotechnologies) according to the manufacturer's instructions. A mixture of 2 μg of dsRNA and 5 μl of transfection reagent was transfected into the cells in 1 ml of Grace's medium. The equivalent amount of dsGFP was used as the control. 20E at 2 μM was added to the Grace's medium for 6 h after 48 h of dsRNA transfection. The controls were treated with the equivalent volume of DMSO. Total RNA was then isolated and reverse-transcribed for qRT-PCR analysis. For the qRT-PCR analysis, ACTB (encoding β-actin) was used as a control to normalize the gene expression and the data were analyzed from three biological experiments using the 2^−ΔΔCt method: 2^−ΔΔCt=2^{−[(CTtreat of gene−CTtreat of ACTB)−(CTcon of gene−CTcon of ACTB)]} where CTcon was the cycle threshold (CT) value of the control group and CTtreat was the CT value of the experimental group.

A fragment encoding a portion of ErGPCR-3 (amino acids 82-621) was amplified using primers ErGPCR-3ExpF and ErGPCR-3ExpR (Table S1). The PCR product was inserted into the pET-32a(+) expression vector (Promega) to produce the recombinant protein in Escherichia coli rosette cells by 0.1 mM isopropyl β-D-thiogalactoside (IPTG) induction. The target protein was purified using an Ni²⁺-NTA affinity column (GE Healthcare). Then, 500 μg of the purified recombinant protein (in 1 ml) and 1 ml of complete Freund's adjuvant (Sigma-Aldrich) were mixed completely and injected subcutaneously into a rabbit's back. Three weeks later, 500 μg of purified recombinant protein was mixed with 1 ml of incomplete Freund's adjuvant (Sigma-Aldrich) and injected into the same rabbit according to the previous method. Then, 500 μg of purified protein was directly injected into the leg muscles of the same rabbit after an additional 2 weeks. The serum was collected after two weeks, and the specificity of the antiserum was determined using western blotting via RNAi (Fig. S4E).

Total proteins were extracted from cells or larval tissues using PBS supplemented with 1 mM PMSF, followed by centrifugation at 10,000 g at 4°C for 10 min. The protein concentration was determined using Bradford's method (Bradford, 1976). Total proteins (20 μg) from each sample were subjected to 7.5-12.5% SDS-PAGE and then electrophoretically transferred onto a nitrocellulose membrane. The membrane was incubated with blocking buffer containing 5% skim milk in Tris-buffered saline [TBS; 10 mM Tris-HCl and 150 mM NaCl (pH 7.5)] for 1 h at room temperature and then incubated at 4°C overnight with the primary antibodies diluted 1:100 for rabbit polyclonal antibodies or 1:1000 for mouse monoclonal antibodies in blocking buffer. Rabbit polyclonal antibodies against H. armigera ErGPCR-3 were produced in our laboratory. Rabbit polyclonal antibodies against β-actin and mouse monoclonal antibodies against His, GFP, and RFP were purchased from ABclonal (AC026, AE003, AE012, and AE020). Antibodies were diluted in 5% bovine serum albumin (BSA) at 1:5000-1:10,000. The nitrocellulose membrane was washed three times for 10 min each with TBST (0.02% Tween in TBS), followed by incubation for 2 h at room temperature with the alkaline phosphatase-labeled goat anti-rabbit or anti-mouse IgG secondary antibodies (Sangon, D110072 and D110089, respectively) diluted 1:5000 in blocking buffer. The membrane was washed three times for 10 min each with TBST and then washed three times for 5 min each with TBS. The target bands were visualized in 10 ml TBS containing 45 μl of 5% p-nitro-blue tetrazolium chloride (NBT) and 35 μl 5% 5-bromo-4-chloro-3-indolyl phosphate (BCIP) in the dark for 10 min. The densities of the bands on the western blotting were acquired using ImageJ software (National Institutes of Health, http//imagej.nih.gov/ij/download.html).

The larval midgut or fat body was isolated and fixed with PBS containing 4% paraformaldehyde at 4°C overnight. The fixed tissues were dehydrated gradually. The tissues were then embedded in melted paraffin and sliced into 7-μm sections using a paraffin-slicing machine (Leica RM2245). The sections were adhered to gelatin-coated glass slides immediately. The slides were dried at 42°C overnight and subsequently dewaxed. The midgut sections were rehydrated gradually and digested with 20 mM proteinase K at 37°C for 10 min. The midgut sections were stained using an H&E staining kit (Sangon) according to the manufacturer's instruction. Positive signals were observed using an Olympus BX51 fluorescence microscope.

HaEpi cells were transfected with dsRNA for 24 h as previously described. The cells were incubated with 3 μM acetoxymethyl (AM) ester Calcium Crimson™ dye (Invitrogen) in Dulbecco's phosphate-buffered saline (DPBS) [137 mM NaCl, 2.7 mM KCl, 1.5 mM KH₂PO₄ and 8 mM Na₂HPO₄ (pH 7.4)] for 30 min at 27°C. The cells were then washed with DPBS three times and exposed to 1 μM 20E in DPBS for 2 min to detect the intracellular calcium release. Next, 1 mM calcium chloride was added to the medium to induce extracellular calcium influx. A laser scanning confocal microscope (Carl Zeiss LSM 700) was used to detect the fluorescence at 555 nm every 6 s for 420 s. The data were analyzed using the Image Pro-Plus software (Media Cybernetics).

HaEpi cells were cultured in six-well plates to 70% confluence, and then Grace's insect medium was replaced with DPBS supplemented with 0.5 mM 3-isobutyl-1-methylxanthine (Sigma-Aldrich). After 30 min, the cells were treated with 2 μM 20E for 30 min. DMSO-treated cells served as a solvent control. The medium was removed and the cells were incubated with 500 μl of 0.1 M HCl for 10 min until the cells were completely dissolved. The mixture was centrifuged at room temperature for 5 min at 600 g. The supernatant was harvested and the cAMP concentration was measured using a cAMP enzyme-linked immunosorbent assay (ELISA) Kit (Beijing Biodragon Immunotechnologies) according to the manufacturer's instructions.

The open reading frames (ORFs) of genes from H. armigera were amplified using the corresponding primers (Table S1) and then inserted into pIEx-4-GFP/RFP-His vector (Invitrogen) fused with green fluorescent protein (GFP) or red fluorescent protein (RFP) through double enzyme digestion to construct the vector in which the C-terminal of the protein of interest is fused with GFP/RFP and histidine tags. The recombinant vectors were verified via direct DNA sequencing. The reconstructed plasmids were transfected into cells using the QuickShuttle-enhanced transfection reagent (Beijing Biodragon Immunotechnologies) until the cells reached 70-80% confluence. After culturing in Grace's medium supplemented with 10% FBS for 48 h at 27°C, the cells were used in subsequent experiments. The pIEx-4-GFP/RFP-His vector was transfected as a negative control. The overexpressed proteins were detected by western blotting using mouse monoclonal antibody against the corresponding tag (ABClonal, AE012 and AE020, 1:1000). After vectors were transfected into the HaEpi for 48 h, the cells were treated with 2 μM 20E for 1 h, and DMSO was used as the control. Proteins were subsequently extracted using the radioimmunoprecipitation assay (RIPA) buffer [0.1 M Tris-HCl buffer containing 0.2 M NaCl and 1% NP-40 (pH 8)] (Beyotime) with protease and phosphatase inhibitors (Roche Diagnostics) and the supernatant was collected via 4°C centrifugation at 12,000 g for 10 min. The supernatant supplemented with protein A resin was incubated with rotation for 5 min at 4°C and collected by centrifugation (1000 g for 2 min) to eliminate nonspecific binding. The antibodies were incubated with the supernatant for 4 h at 4°C. The protein A resin was then added to the protein-antibody complex and shaken gently for 2-4 h at 4°C to form the protein-resin-antibody complex. The resin was subsequently collected via centrifugation and washed three times with RIPA buffer. The collected resin was treated with SDS-PAGE loading buffer and boiled for 10 min. The samples were loaded into SDS-PAGE for western blotting using mouse monoclonal antibodies against RFP, GFP, and His (ABClonal) to detect the target proteins, respectively.

HaEpi cells grown on cover slips in a cell culture plate were washed with PBS twice and fixed with PBS containing 4% paraformaldehyde in the dark for 10 min. The plasma membrane was stained with WGA (Sigma-Aldrich; 1 μg/ml in PBS) in the dark for 5 min and then washed with PBS three times. The midgut sections and the fixed cells were blocked with 2% BSA in PBS for 1 h and then incubated with rabbit polyclonal antibodies against ErGPCR-3 as the primary antibody (1:100 dilution in 2% BSA) overnight at 4°C. The midgut sections and the cells were washed and subsequently incubated with Alexa Fluor 488-labeled goat anti-rabbit IgG secondary antibodies (diluted 1:1000 in 2% BSA) for 1 h at 37°C. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; Sangon; 1 μg/ml in PBS) in the dark for 10 min and then washed with PBS three times. Fluorescence was detected using a laser scanning confocal microscope (Carl Zeiss, LSM 700).

The molecular model of ErGPCR-3 binding to a steroid was created using the I-TASSER server (https://zhanglab.ccmb.med.umich.edu/I-TASSER/). The ligand 20E was docked into the active site of ErGPCR-3 using the Surflex-Dock (SFXC) function in the SYBYLx2.0 software (Tripos). Figures were prepared using the PyMOL program (Seeliger and de Groot, 2010).

pIEx-4-GFP-His, pIEx-4-ErGPCR-3-GFP-His and the ErGPCR-3 mutant [pIEx-4-ErGPCR-3-M-GFP-His: leucine 449 to alanine (L449A), leucine 474 to alanine (L474A), glycine 662 to alanine (G662A), and leucine 669 to alanine (L669A)], which was obtained by site-directed mutagenesis of ErGPCR-3 on possible steroid binding sites (Table S2) based on the I-TASSER prediction, were transfected into HaEpi cells, separately. The detailed method of detecting 20E bound by the cell membranes of HaEpi cells has been previously reported (Kang et al., 2019). GPCRs-GFP-His and GFP-His were overexpressed in Sf9 cells and isolated, and then a 20E-EIA kit was used for binding saturation curve analysis, according to a previously published method (Kang et al., 2019).

Midguts and wing disks (10 mg) were dissected from around five larvae of sixth instar 96 h and then freeze-dried overnight. The powders were dissolved in 200 μl of 80% methyl alcohol and ground in a cold mortar. After centrifugation at 10,000 g for 10 min, the supernatant was air dried at room temperature. The precipitates were completely dissolved with 100 μl of EIA buffer. Each sample was diluted 1000 times with EIA buffer before detection of the 20E concentration. A 50 µl sample was used to detect 20E using a 20E-EIA kit according to a previously published method (Kang et al., 2019).

dsGRK2 and dsErGPCR-3 were transfected into HaEpi cells in a six-well cell culture plate for 48 h. dsRFP was the control. pIEx-4-ErGPCR-3-GFP-His or pIEx-4-ErGPCR-M-3-GFP-His and dsGRK2 were co-transfected into HaEpi cells in a 25-cm² cell culture bottle for 48 h. pIEx-4-GFP-His and dsRFP were used as the control. 20E was then added to the medium at 2 μM and incubated for 30 min at 27°C. To detect the intracellular 20E, the nuclei of cells were isolated using a Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime) and then freeze-dried overnight. The powders were dissolved in 200 μl of 80% methyl alcohol and ground in a cold mortar. After centrifugation at 10,000 g for 10 min, the supernatant was air dried at room temperature. The precipitates were completely dissolved with 100 μl of EIA buffer. The intracellular 20E levels were detected using the 20E-EIA kit according to a previously published method (Kang et al., 2019).

BS3 (Sigma Aldrich) possesses a charged group and is used to cross-link cell surface proteins (Angers et al., 2000; Friedrichson and Kurzchalia, 1998). The assay was performed according to the manufacturer's protocol. HaEpi cells and tissues were collected and washed three times with ice-cold DPBS and then incubated with DPBS containing 1 μM 20E or the equivalent concentration of DMSO. The resuspended HaEpi cells and the tissue homogenates were incubated with 5 mM BS3 and the reaction mixture was incubated at room temperature for 1 h. Tris-HCl (pH 7.5) was then added into the mixture at a final concentration of 20 mM to quench the crosslinking reaction. The crosslinking-treated HaEpi cells and the tissue homogenates were homogenized and subjected to western blotting.

HaEpi cells were treated with 0.5-5 μM 20E for 48 or 72 h. DMSO was used as a control. The p-Histone H3 (Ser10) antibody (rabbit polyclonal antibodies, 9701, Cell Signaling Technology, 1:1000 dilution in 2% BSA) was used to detect cell proliferation. The nuclei were stained with DAPI (10 μg/ml) for 10 min at room temperature in the dark, and observed using a laser scanning confocal microscope (Carl Zeiss, LSM 700). Annexin-V and propidium iodide (PI) were used to detect cell apoptosis using an Annexin V-FITC Apoptosis Detection kit (GK3603, Genview) and flow cytometry (Amnis). Annexin V-FITC stained the earlier apoptotic cells by binding to the membrane phosphatidylserine, and PI indicated the later apoptotic cells and dead cells by entering the cells.

The experimental data were analyzed using SPSS 23.0 (IBM) and generated using GraphPad Prism 5. All data are mean±standard deviation (s.d.) of at least three biologically independent experiments. Two group datasets were analyzed using two-tailed Student's t-test, *P<0.05, **P<0.01, ***P<0.001. One-way ANOVA was used to analyze the differences among three or more groups using Duncan's multiple comparison test at P=0.05.

We thank Dr Naoki Yamanaka (University of California, Riverside, CA, USA) for advice and comments on improving the manuscript.