Subunit P60 of phosphatidylinositol 3-kinase promotes cell proliferation or apoptosis depending on its phosphorylation status

The regulatory subunits (P60 in insects, P85 in mammals) determine the activation of the catalytic subunits P110 in phosphatidylinositol 3-kinases (PI3Ks) in the insulin pathway for cell proliferation and body growth. However, the regulatory subunits also promote apoptosis via an unclear regulatory mechanism. Using Helicoverpa armigera, an agricultural pest, we showed that H. armigera P60 (HaP60) was phosphorylated under insulin-like peptides (ILPs) regulation at larval growth stages and played roles in the insulin/ insulin-like growth factor (IGF) signaling (IIS) to determine HaP110 phosphorylation and cell membrane translocation; whereas, HaP60 was dephosphorylated and its expression increased under steroid hormone 20-hydroxyecdysone (20E) regulation during metamorphosis. Protein tyrosine phosphatase non-receptor type 6 (HaPTPN6, also named tyrosine-protein phosphatase corkscrew-like isoform X1 in the genome) was upregulated by 20E to dephosphorylate HaP60 and HaP110. 20E blocked HaP60 and HaP110 translocation to the cell membrane and reduced their interaction. The phosphorylated HaP60 mediated a cascade of protein phosphorylation and forkhead box protein O (HaFOXO) cytosol localization in the IIS to promote cell proliferation. However, 20E, via G protein-coupled-receptor-, ecdysone receptor-, and HaFOXO signaling axis, upregulated HaP60 expression, and the non-phosphorylated HaP60 interacted with phosphatase and tensin homolog (HaPTEN) to induce apoptosis. RNA interference-mediated knockdown of HaP60 and HaP110 in larvae repressed larval growth and apoptosis. Thus, HaP60 plays dual functions to promote cell proliferation and apoptosis by changing its phosphorylation status under ILPs and 20E regulation, respectively.

Introduction Phosphatidylinositol 3-kinases (PI3Ks) play very important roles in various pathways of cellular responses to the extracellular signals of insulin and insulin-like growth factor (IGF) [1]. PI3Ks are composed of different P110 catalytic subunits and different P85 (P60 in insects) regulatory subunits in mammals [2]. In insulin signaling, the insulin receptor (INSR) is autophosphorylated by insulin binding-induced conformational change, which in turn phosphorylates insulin receptor substrate (IRS) and P85, thus resulting in the recruitment of the P85-P110 complex to the plasma membrane [3], during which P110 is autophosphorylated [4]. P110 phosphorylates phosphatidylinositol 4,5-bisphosphate (PIP2) to form phosphatidylinositol 3,4,5-trisphosphate (PIP3) in the cell membrane, which attracts protein kinase B/AKT to the cell membrane, where it is phosphorylated by the cell membrane located phosphoinositidedependent kinase (PDK) [5]. The phosphorylated AKT promotes cell proliferation in physiological and oncological conditions and phosphorylates forkhead box protein O (FOXO) to prevent it from entering the nucleus to play a transcription factor role in apoptosis [6]. Phosphatase and tensin homolog (PTEN) reverses the function of PI3K by dephosphorylating PIP3 [7]. The regulatory subunits (P85α, P85β, and P55γ in mammals) can bind and stabilize P110s [8]. However, regulatory subunits have different functions in different cells. The regulatory subunit P85β is a tumor driver in melanoma and breast cancer, whereas P85α functions oppositely as a tumor suppressor in breast cancer [9]. P55γ is necessary for ovarian cell proliferation [10]; however, P55γ suppresses vascular smooth muscle cell proliferation [11]. The mechanism of the contradictory functions of regulatory subunits is unclear.
Insect growth and metamorphosis are mainly regulated by insulin-like peptides (ILPs) and the steroid hormone 20-hydroxyecdysone (20E) in different insects [12,13]. ILPs regulate the growth rate by promoting cell growth and proliferation, and 20E determines growth termination and promotes apoptosis for metamorphosis [14]. ILPs, via insulin-like receptor (InR), promote cell proliferation and body growth [15,16]. Meanwhile, it promotes prothoracic gland (PG) growth to produce more ecdysone [17], which is converted to 20E as a systemic signal to end the larval growth and initiate metamorphosis [18]. 20E enters its target cells through ecdysone importer (EcI), a membrane transporter, and binds to ecdysone receptor (EcR)ultraspiracle (USP) transcription complex to activate transcription of multiple genes that are involved in molting and metamorphosis [19][20][21][22]. Although the previous studies showed that 20E counteracts insulin/IGF signaling (IIS) activity during development [23], the mechanism by which 20E regulates PI3K to antagonize IIS is poorly understood.
In H. armigera, a lepidopteran insect and a serious agricultural pest, class IA PI3K consists of a single regulatory subunit gamma, named HaP60, showing an equal degree of similarity to mammalian P85α, P85β, and P55γ; and a single catalytic subunit delta, named HaP110, being homologous to mammalian class IA PI3Ks. This is consistent with Drosophila melanogaster [24], which presents a simple PI3K model to address the roles and regulatory mechanism of P60 by ILPs and steroid hormone. Therefore, using H. armigera, we demonstrated that HaP60 functions as a phosphorylated form to promote cell proliferation by interacting with HaP110 under ILPs regulation, and functions as a non-phosphorylated form to promote apoptosis by interacting with HaPTEN under 20E regulation.

HaP60 and HaP110 show variations in phosphorylation and expression levels during insect development
To investigate the function of HaP60 in H. armigera development, we detect its developmental expression profiles using rabbit polyclonal antibodies against H. armigera HaP60 (anti-HaP60 antibodies) prepared in our laboratory, and the specificities of the antibodies were examined (S1A Fig). Western blotting showed that HaP60 appeared as a single band in the epidermis, but HaP60 appeared as two bands in the midgut and fat body (S1B Fig), and the band in the epidermis and the upper band in the midgut and fat body were identified as a phosphorylated form of HaP60 using lambda protein phosphatase (λPPase) treatment (S1C Fig). The phosphorylated HaP60 was increased at the sixth instar feeding stages and decreased at the metamorphic stages, accompanying increased expression levels at the metamorphosis compared with the fifth instar feeding stages and sixth instar feeding stages in the midgut and fat body (Fig 1A-Aii). Hematoxylin and eosin (HE) staining and immunohistochemistry showed that HaP60 was localized in both the larval midgut and imaginal midgut ( Fig 1B). These results suggested that HaP60 plays roles in larval and adult tissues as a phosphorylated form in the feeding stages and as a non-phosphorylated form during metamorphosis.
To address the hormonal regulation of HaP60 phosphorylation and expression, insulin or 20E were injected into the hemocoel of the sixth instar 6 h (6th-6 h) larvae, separately. Insulin promoted HaP60 phosphorylation, with little effect on its expression, in a dose-and timedependent manner, compared with that in the phosphate-buffered saline (PBS) control. In contrast, 20E induced HaP60 dephosphorylation and expression in a dose-and time-dependent manner, compared with that in the dimethyl sulfoxide (DMSO) control (Fig 1C-1F). These results revealed that ILPs induce HaP60 phosphorylation and 20E induces HaP60 dephosphorylation but increases its expression.
The expression profiles of HaP110 were also examined to determine its role in insect development using rabbit polyclonal antibodies against H. arimgera HaP110 (anti-HaP110 antibodies) prepared in our laboratory (S1 Fig). HaP110 appeared to be phosphorylated and showed high expression levels at the growth stage, with increased expression during the early pupal stage in the midgut. Furthermore, HaP110 was phosphorylation and had high expression levels at the growth stages and later pupal stage in the fat body (Fig 2A-Aii). HE staining and immunohistochemistry showed that HaP110 was mainly localized in the larval midgut at the 6th-48 h stage (Fig 2B). Insulin increased HaP110 expression and phosphorylation in a dose-and time-dependent manner, whereas 20E decreased HaP110 expression and phosphorylation in a dose-and time-dependent manner (Fig 2C-2F). Thus, HaP110 plays roles in different tissues during larval growth, and its expression and phosphorylation were regulated by ILPs and 20E, counteractively.

20E promotes dephosphorylation of HaP60 and HaP110 by upregulating HaPTPN6 expression
To explore the mechanism of 20E-promoted dephosphorylation of HaP60 and HaP110, the protein translation inhibitor Cycloheximide (CHX) and phosphatase inhibitor (PPI) were added to HaEpi cells (a Helicoverpa armigera epidermal cell line), followed by insulin and 20E induction, respectively. The results showed that insulin induced HaP60 and HaP110

PLOS GENETICS
phosphorylation; however, 20E repressed HaP60 and HaP110 phosphorylation, and CHX and PPI maintained HaP60 and HaP110 phosphorylation, suggesting that 20E relies on a phosphatase to dephosphorylate HaP60 and HaP110 (Fig 3A). From the literature survey, the tyrosine phosphatase, protein tyrosine phosphatase non-receptor type 6 (PTPN6), was known to act with P85 to inhibit PI3K activity in mammalian cells [25]. We identified HaPTPN6 by BLAST (basic local alignment search tool) of the genome of H. armigera using mammalian PTPN6 (S2 Fig). Therefore, the specific inhibitor of PTPN6, tyrosine phosphatase inhibitor 1 (TPI-1), was used to examine the role of HaPTPN6 in 20E-mediated dephosphorylation of HaP60 and HaP110. BN82002, an inhibitor of the cell division cycle 25 (CDC25), as control of TPI-1. TPI-1 treatment repressed 20E-mediated dephosphorylation of HaP60 and HaP110 in HaEpi cells and the larval fat body (Fig 3B and 3C), but BN82002 did not. After knockdown of HaPtpn6 in HaEpi cells, 20E could not induce dephosphorylation of HaP60 and HaP110 (Fig 3D). Quantitative real-time reverse transcription PCR (qRT-PCR) showed HaPtpn6 was knocked down in the HaEpi cells without off-target effects on another protein phosphatase gene, HaPtpn11 (S3 Fig). Knockdown of HaPtpn6 in larvae by feeding third instar larvae with Escherichia coli HT115 (E. coli HT115) that expressed dsHaPtpn6 also maintained HaP60 and HaP110 phosphorylation in the larval fat body at the 6th-72 h stage, compared with dsGFP ( Fig 3E). Meanwhile, Phenotype analysis showed HaPtpn6 knockdown resulted in 53% early pupation than 9% early pupation in dsGFP, and advanced pupation time for 13 h (S4A-S4C Fig

20E blocks ILP-induced cell membrane translocation and decreases the interaction of HaP60 and HaP110 by dephosphorylation of HaP60 and HaP110
To address the consequence of HaP60 and HaP110 dephosphorylation, we observed the subcellular localization of HaP60 and HaP110. Antibodies detection showed that HaP60 and HaP110 were localized in the cytosol in the PBS control in HaEpi cells. Insulin induced HaP60 and HaP110 to transfer to the cell membrane from the cytosol; however, 20E blocked the insulin-induced cell membrane translocation of HaP60 and HaP110, compared with the DMSO control, nevertheless, 20E hardly blocked the subcellular localization of HaP60 and HaP110 upon TPI-1 treatment (Fig 4A and 4B). The Co-IP assay showed that insulin increased the phosphorylation and interaction of the HaP60 and HaP110, but whereas 20E decreased the interaction between HaP60 and HaP110 when they were dephosphorylated (Fig 4C), suggesting that 20E inhibits the interaction between HaP60 and HaP110, and cell membrane translocation, by dephosphorylating HaP60 and HaP110 via HaPTPN6. 20E also repressed HaP110 expression; therefore, the signal for HaP110 was weak under 20E treatment.

HaP60 promotes cell proliferation under ILP regulation
To investigate the function of HaP60 in IIS, the HaP110 subcellular localization and subsequent protein phosphorylation cascade were explored via HaP60 knockdown followed by insulin treatment. The results showed that HaP60 knockdown partially blocked the insulininduced cell membrane translocation of HaP110 ( Fig 5A). Western blotting showed that HaP110 was not phosphorylated by insulin induction after HaP60 knockdown ( Fig 5B). HaP60 knockdown decreased insulin-induced HaAKT and HaFOXO phosphorylation, which was similar to the results of HaP110 knockdown (Fig 5C). HaP60 and HaP110 knockdown blocked the insulin-induced HaFOXO cytosolic localization, resulting in HaFOXO locating to the nucleus, compared with dsRFP ( Fig 5D and 5Di). HaP60 and HaP110 knockdown decreased insulin-induced cell proliferation (Fig 5E and 5Ei). Western blotting and qRT-PCR showed that HaP60 and HaP110 were knocked down effectively at the protein and mRNA levels (S6 Fig). These results suggested that HaP60 is necessary for the IIS cascade and ultimate cell proliferation.
Given that HaP60 was upregulated in the metamorphic stages by 20E, we characterized the role of HaP60 in apoptosis by overexpressing HaP60-red fluorescent protein (RFP)-His and HaP110-RFP-His in HaEpi cells, with RFP-His as a tag control. Higher CASP3 activity was detected in the HaP60-RFP-His overexpressing cells; however, less CASP3 signal was detected The anti-GFP antibody (α-GFP) immunoprecipitated overexpressed HaPTPN6-GFP-His in the cell lysate, co-immunoprecipitated HaP60 and HaP110 were detected using anti-HaP60 antibodies (α-HaP60) and anti-HaP110 antibodies (α-HaP110), respectively. Mouse IgG (α-IgG) was used as a negative control for antibody. Fi. Statistical analysis for F. G. 20E rarely promoted the interaction of HaP60 and HaP110 with HaPTPN6. The anti-GFP antibody (α-GFP) immunoprecipitated overexpressed HaPTPN6-GFP-His in the cell lysate, co-immunoprecipitated HaP60 and HaP110 were detected using anti-HaP60 antibodies (α-HaP60) and anti-HaP110 antibodies (α-HaP110), respectively. Mouse IgG (α-IgG) was used as a negative control for antibody. Gi. Statistical analysis for G. � P < 0.05 using two-tailed Student's t-test or different lowercase letters indicate significant differences (P < 0.05) using one-way ANOVA. The bars indicate the mean ± SD.
https://doi.org/10.1371/journal.pgen.1009514.g003  (Fig 6G and 6Gi), when RFP-His, HaP60-RFP-His, and HaP110-RFP-His were expressed equally (S7 Fig). Flow cytometry analysis demonstrated that HaP60-His increased the number of apoptotic cells (Fig 6H and 6Hi). The P85α-PTEN complex causes G2/M phase arrest and apoptosis [27]; therefore, HaP-TEN-GFP-His and HaP60-RFP-His were overexpressed in HaEpi cells to detect the interaction between HaPTEN and HaP60, and to identify the mechanism by which HaP60 induces apoptosis. The results showed the HaPTEN-GFP-His and HaP60-RFP-His were immunoprecipitated using anti-GFP antibodies in DMSO treatment, and 20E increased the amount of HaP60-RFP-His, compared with that induced by DMSO (Fig 6I and 6Ii). RFP and GFP did not interact in these experiments (Fig 6J and 6Ji). These data confirmed that HaP60 promotes apoptosis by interacting with HaPTEN under 20E regulation.

Both HaP60 and HaP110 are necessary for larval growth and metamorphosis
To determine the role of HaP60 in larval growth and metamorphosis, we knocked down HaP60 by feeding third instar larvae with E. coli HT115 that expressed dsGFP or dsHaP60, respectively. Phenotype analysis showed the larvae produced 33% small pupae after feeding with dsHaP60, which was significantly higher than the 9% small pupae induced by feeding with dsGFP (Fig 7A and 7B). Meanwhile, pupation time was delayed for 16 h, and the average weight of the pupa was decreased, compared with those in larvae fed with dsGFP (Fig 7C and  7D) when HaP60 was confirmed as being knocked down according to its mRNA and protein levels (Fig 7E and 7F). The midgut appeared red at 10 d post-feeding dsGFP, whereas, the larval midgut was not red and did not separate from the imaginal midgut at the same time after feeding with dsHaP60 (Fig 7G and 7H). Moreover, the larval fat body retained complete after feeding with dsHaP60 at 10 d, whereas the larval fat body started to decompose in the dsGFP-fed control ( Fig 7I). Western blotting analysis showed that HaP110, HaAKT, and HaFOXO phosphorylation decreased after HaP60 knockdown ( Fig 7J). Furthermore, the 20E titer was decreased at 10 d after HaP60 knockdown (Fig 7K). These results suggested that HaP60 is necessary for larval growth and 20E production, and apoptosis of larval tissues during metamorphosis.
To address the role of the dephosphorylation of HaP60 in the metamorphosis, dsHaP60 was injected into the 6th-72 h larva that HaP60 was dephosphorylated to knock down HaP60, dsGFP as control. Phenotype analysis showed HaP60 knockdown caused 67% delayed pupation, which was dramatically higher than 8% dsGFP (Fig 8A and 8B), and the pupation time was delayed approximately for 19 h after HaP60 knockdown (Fig 8C), and the pupa weight was no difference between dsHaP60 and dsGFP when HaP60 was knocked down in the mRNA and protein levels (Fig 8D-8F). Meanwhile, the PCD of the midgut and fat body was also delayed after HaP60 knockdown (Fig 8G-8I). Furthermore, cleaved-HaCASP3, an activated form of HaCASP3 and a marker of PCD, was reduced after HaP60 knockdown (Fig 8J and  8K), suggesting non-phosphorylated HaP60 is necessary for apoptosis of larval tissue and metamorphosis.

PLOS GENETICS
HaP110 knockdown caused 47% abnormal pupae, which was higher than the 6% abnormal pupation induced by feeding with dsGFP ( Fig 9A and 9B), suggesting a different role of the catalytic subunit HaP110 from that of the regulatory subunit HaP60. Other effects were similar to those induced by HaP60 knockdown, including delayed pupation time for 27 h ( Fig 9C) and decreased pupa weight ( Fig 9D) when HaP110 was knocked down at the mRNA and protein levels in the fat body (Fig 9E and 9F). The PCD of the larval midgut and fat body was also delayed after feeding dsHaP110, compared with dsGFP feeding (Fig 9G-9I). Western blotting analysis showed that HaAKT and HaFOXO phosphorylation decreased after HaP110 knockdown ( Fig 9J). Moreover, the 20E titer was reduced at 10 d after HaP110 knockdown ( Fig 9K). These results revealed that HaP110 promotes larval growth and 20E production to initiate metamorphosis.

Discussion
PI3Ks play pivotal roles in insulin and other cell proliferation pathways via the incorporation of subunits P110 and P85/P60. Steroid hormones, including 20E, counteract ILPs activity; however, the mechanism by which 20E regulates PI3K is poorly understood. The present study determined that HaP60 plays dual roles in cell proliferation and apoptosis in the IIS and 20E pathways. ILPs induce phosphorylation of HaP60 and HaP110; however, 20E induces dephosphorylation of HaP60 and HaP110 via HaPTPN6 to block the IIS, which ceases HaFOXO to localize to the nucleus. 20E upregulates HaP60 expression via HaFOXO, which interacts with HaPTEN to induce apoptosis.

HaP60 functions in cell proliferation and apoptosis by changing its phosphorylation status
There are many isoforms of PI3K in mammals. For example, class IA PI3Ks consist of P110 catalytic subunits (P110α, P110β, and P110δ, encoded by genes PIK3CA, PIK3CB, and PIK3CD, respectively) and P85 regulatory subunits (P85α, P85β, and P55γ, encoded by PIK3R1, PIK3R2, and PIK3R3, respectively). Other regulatory isoforms of P50α and P55α are generated by splicing of P85α transcripts [2]. Five mammalian regulatory subunit isoforms of class IA PI3Ks can act as positive and negative regulators of PI3K activity in different cell types and response to different stimuli [2]. Each regulatory subunit of PI3K is essential for the cell membrane translocation and phosphorylation of P110, the subsequent AKT and FOXO phosphorylation, and the cytosolic localization of FOXO, finally resulting in cell proliferation [6]. However, there is also a report that P85α reduces PI3K activity and insulin signaling [28]. P50α/P55α represses insulin sensitivity [29] and P55γ suppresses vascular smooth muscle cell proliferation [11]. These contradictory roles of P85/P60 to promote cell proliferation and apoptosis are confusing.

PLOS GENETICS
P60 and P110 are unique forms of class IA PI3K in D. melanogaster [30], which presents a simple model compared with the complicated composition of PI3Ks in mammals to address the regulatory mechanism of P85/P60 functions. Insect larval growth is regulated mainly by ILPs and metamorphosis is regulated by 20E [31]. The IIS is conserved from insects to humans [15]. 20E, a steroid hormone ecdysone produced in insects, is known to oppose ILPs function at high levels [32]. According to the variation of HaP60 phosphorylation and expression levels during H. armigera development from larval growth to metamorphosis, we revealed that HaP60 promotes cell proliferation as a phosphorylated form under ILPs regulation for larval growth, but HaP60 promotes apoptosis as a dephosphorylated form under 20E regulation for metamorphosis. This finding explains the mechanism by which HaP60 promotes both cell proliferation and apoptosis.

20E upregulates HaPTPN6 to dephosphorylate HaP60 and HaP110 to block IIS and stop larval growth
How animals stop the growth and maintain their body size has been studied for a long time and has remained poorly understood until now. The known mechanisms include the coordination of ILPs and steroid hormones to determine body size in insects [33]. The PGs grow as the body grows to produce more 20E [34]. The 20E titer greatly increases during metamorphosis and competes for dopamine to bind the receptor, DopEcR, to stop feeding behavior [35]. 20E suppresses dMyc activity, which is involved in IIS and TOR (the target of rapamycin) pathway, and global growth at the larval-pupal transition [36]. Moreover, 20E signaling increases Imaginal morphogenesis protein-Late 2 (Imp-L2), a Drosophila homolog of insulinlike growth factor-binding protein 7 (IGFBP7), which circulates in the hemolymph to bind and antagonize ILP, thereby attenuating peripheral IIS and body growth [37]. Furthermore, 20E reduces microRNA miR-8 expression, which activates PI3K, thereby facilitating fat cell proliferation and organism growth, whereas 20E upregulates u-shaped (USH), a target of miR-8, and inhibits PI3K activity and blocks cell growth [38,39]. However, the mechanism of the regulation of 20E on post-translational modification of PI3K to antagonize IIS is unclear. PTPN6 is reported to show high expression in insulin target tissues in obesity, in which it interacts with P85, resulting in P85 dephosphorylation and reduced AKT phosphorylation in the control of liver metabolism [40,41]; however, the mechanism is unclear. In the present study, we further revealed that 20E upregulates HaPTPN6 expression via the ErGPCRs-and EcR-signaling axis, which interacts with HaP60 and results in HaP60 dephosphorylation in H. armigera. The dephosphorylation of HaP60 blocks the IIS from the upstream part of the pathway, thus repressing the subsequent phosphorylation of HaP110, HaAKT, and HaFOXO, which finally results in HaFOXO entering the nucleus to initiate gene transcription for apoptosis. Our study revealed that steroid hormone 20E upregulation of HaPTPN6 expression to HaPTEN-GFP-His was immunoprecipitated with anti-GFP antibody (α-GFP) and the coprecipitated HaP60-RFP-His was examined by western blotting using anti-RFP antibody (α-RFP). Mouse IgG (α-IgG) was used as a negative control for the antibody. Ii. Statistical analysis for I. J and Ji. RFP-His and GFP-His tag control for I. � P < 0.05 using two-tailed Student's t-test for two pairs analysis, or different lowercase letters indicate significant differences (P < 0.05) using one-way ANOVA for multiple analysis. https://doi.org/10.1371/journal.pgen.1009514.g006

PLOS GENETICS
dephosphorylate HaP60 counteracts the IIS and is one of the mechanisms by which animals stop growing.

20E promotes HaP60 expression and utilizes HaP60 to induce apoptosis
Overexpression of P60 and a P60 mutant that lacked the inter-SH2 domain for binding to P110, caused growth arrest in the first instar for as long as 2 weeks in D. melanogaster, equivalent to a phenotype of PTEN overexpression [42]; however, the mechanism was unclear. 20E promotes HaPTEN expression, which reverses PI3K activity, and results in cell apoptosis in H. armigera [43]. Our research further demonstrated that 20E, via ErGPCRs-, EcR-and HaFOXO-signaling promotes HaP60 expression directly by HaFOXO binding to FOXOBE in the HaP60 promoter. The highly expressed HaP60 interacts with HaPTEN under 20E mediation, which further represses the IIS and results in HaFOXO localizing to the nucleus to induce apoptosis.
Our results demonstrated that HaP60 plays dual roles in the IIS and 20E pathways. HaP60 is necessary for HaP110 phosphorylation and cell membrane translocation, and the subsequent cascade of protein phosphorylation, resulting in cell proliferation. 20E promotes HaP60 and HaP110 dephosphorylation by upregulating HaPTPN6 expression, thereby blocking cell membrane translocation and decreasing the HaP60 and HaP110 interaction. 20E, via the ErGPCRs-, EcR-, and HaFOXO-signal axis, promotes the expression of HaP60, which interacts with HaPTEN to reverse HaP110 action and bring about cell apoptosis at the metamorphic stages (Fig 10).

Insects
Helicoverpa armigera were cultivated in an insect laboratory with a stable temperature of 27˚C and a 14 h light/dark 10 h cycle. The larvae were fed with artificial feed according to a previous study [44].

Preparation of HaP60 and HaP110 antibodies
Partial fragments of HaP60 or HaP110 were amplified using primers shown in S1 Table and fused with pET-30(a) to construct the expression plasmid. Genbank numbers of genes were shown in S2 Table. The plasmid was then transformed into the Escherichia coli rosetta strain grown for 3 h at 37˚C. Expression was then induced by adding 0.5 mM isopropyl β-D-Thiogalactoside (IPTG; A600168, Sangon Biotech, Shanghai, China) overnight at 25˚C. The bacterial cells were collected, suspended with phosphate-buffer saline (PBS), broken using ultrasound, Larvae fed with E. coli HT115 that expressed dsHaP60 led to small pupa and delayed pupation. A. Phenotype after feeding dsHaP60 for 10 d and 12 d. Bar, 0.5 cm. B. Statistical analysis of the phenotype in A using Student's t-test based on three repeats. Thirty larvae for each repeat. C. The pupation time from 6th-6 h to pupa 0 d. D. The average body weight statistics of insects from the third instar feeding (3F) larvae to pupae 2 days (P-2 d) by Student's t-test based on three repeats. Thirty larvae for each repeat. E and F. Efficiency analysis of HaP60 knockdown using qRT-PCR and western blotting at the mRNA and protein levels. G. Morphology of midgut after feeding dsRNA for 10 d. MG: midgut. Bar, 0.5 cm. H and I. HE staining of midgut and fat body after feeding dsRNA for 10 d. LM: larval midgut; IM: imaginal midgut. Bars, 20 μm. J. HaP60 knockdown decreased HaAKT and HaFOXO phosphorylation after feeding dsRNA for 8 d. K. 20E titer detection in the hemolymph after feeding dsRNA for 10 d. � P < 0.05 and �� P < 0.01 using two-tailed Student's t-test. The bars indicate the mean ± SD.
https://doi.org/10.1371/journal.pgen.1009514.g007 Fig 8. HaP60 knockdown by injecting dsRNA at the wandering stages delayed metamorphosis. A. Phenotype analysis after first injecting dsRNA for 60 h and 80 h. Bar, 0.5 cm. B. Statistical analysis of the delayed pupation in A by Student's t-test based on three repeats. Thirty larvae for each repeat. C. Statistical analysis of pupation time from 6th-0 h to pupa 0 d after HaP60 knockdown. D. The average weight of pupa after HaP60 knockdown. Thirty larvae for each repeat. E and F. Efficiency analysis of HaP60 knockdown using qRT-PCR and western blotting at the mRNA centrifuged at 10000 × g for 20 min, and the pellet was retained. The protein was denatured using 8 M urea, pH 8.0, purified using a His-tag column, and renatured by reducing the urea concentration step by step in the dialysate. The protein was then mixed with adjuvant and injected into a rabbit to produced anti-HaP60 antibodies or anti-HaP110 antibodies, according to a previously described method [45].

Western blotting
Western blotting analysis was conducted according to our previous method [46]. Epidermis, midgut, and fat body were extracted using 40 mM Tris-HCl, pH 7.5, with 1 mM PMSF, followed by centrifugation at 15000 × g at 4˚C for 15 min. The supernatant was obtained and the protein content was determined using Bradford's method. Proteins (50 μg) were separated using SDS-PAGE (10% SDS-PAGE for HaP60 and other proteins detection, 6% SDS-PAGE for HaP110 detection) and transferred onto nitrocellulose membranes. Non-specific binding was blocked using 5% non-fat milk in TBST (Tris-buffered saline tween; 150 mM NaCl, 10 mM Tris-HCl, 0.05% Tween-20, pH 7.5) for 1 h and then the membranes were incubated with anti-HaP60 antibodies (diluted in a ratio of 1:100) or anti-HaP110 antibodies (diluted in a ratio of 1:100) overnight at 4˚C. Then, the membranes were washed with TBST three times for 5 min each and incubated with the secondary antibody (AP-conjugated goat anti-rabbit IgG, 1:5000 in the blocking buffer) using our previously reported method. HaACTB, as a protein quality control, was detected using anti-HaACTB antibodies. Anti-GFP, anti-RFP, and anti-His (1:2000 in the blocking buffer) antibody were used to detect the GFP tag, RFP tag, and His tag protein, together with the same secondary antibody (AP-conjugated horse anti-mouse IgG, 1:5000 in the blocking buffer). and protein levels. G. Morphology of midgut after first injecting dsRNA for 60 h. Bar, 0.5 cm. H and I. HE staining of midgut and fat body after first injecting dsRNA for 60 h. LM: larval midgut; IM: imaginal midgut. Bars, 20 μm. J and K. The western blotting analysis of cleaved-HaCASP3 after first injecting dsRNA for 60 h. � P < 0.05 and �� P < 0.01 using two-tailed Student's t-test. The bars indicate the mean ± SD. https://doi.org/10.1371/journal.pgen.1009514.g008

Immunohistochemistry
The tissues were fixed in 4% paraformaldehyde (PFA) and submitted to a professional company (Servicebio, Wuhan, China). Anti-HaP60 antibodies (diluted in 1:50) and anti-HaP110 antibodies (diluted in 1:50) were used to detect the locations of HaP60 and HaP110 in the Larvae fed with E. coli HT115 that expressed dsHaP110 resulted in abnormal pupa and delayed pupation. A. Phenotype after feeding dsHaP110 for 10 d and 12 d. Bar, 0.5 cm. B. Statistical analysis of the phenotype in A by Student's t-test based on three repeats. Thirty larvae for each repeat. C. The pupation time from 6th-6 h to pupa 0 d. D. The average body weight statistics of insects from 3F larvae to P-2 d pupae by Student's t-test based on three repeats. Thirty larvae for each repeat. E and F. Efficiency analysis of HaP110 knockdown using qRT-PCR and western blotting at the mRNA and protein levels. G. Morphology of midgut after feeding dsRNA for 10 d. MG: midgut. Bar, 0.5 cm. H and I. HE staining of midgut and fat body after feeding dsRNA for 10 d. LM: larval midgut; IM: imaginal midgut. Bars, 20 μm. J. HaP110 knockdown decreased HaAKT and HaFOXO phosphorylation after feeding dsRNA for 8 d. K. 20E titer detection in hemolymph after feeding dsRNA for 10 d. � P < 0.05 using two-tailed Student's t-test. The bars indicate mean ± SD. https://doi.org/10.1371/journal.pgen.1009514.g009

Hormonal treatment in the larvae
Different concentrations (1, 2, 5, and 10 μg/larva) of human insulin (P3376, Beyotime Biotechnology) or 20E (50, 100, 200, and 500 ng/larva; H5142, Sigma-Aldrich, St. Louis, MO, USA) diluted with PBS and then 5 μL of insulin or 20E were injected into the hemolymph of the 6th-6 h larva for 6 h, individually. Equally diluted PBS for insulin and DMSO for 20E were used as controls. Insulin at 5 μg/larva or 20E at 500 ng/larva was injected into the hemolymph of the 6th-6 h larva from 1 h to 24 h; PBS and DMSO were used as controls. Each treatment used three larvae, from which the fat body was extracted using 40 mM Tris-HCl for western blotting.

Co-immunoprecipitation (Co-IP)
For the HaPTEN and HaP60 Co-IP assay, 4 μg HaPTEN-GFP-His and HaP60-RFP-His plasmids were co-transfected into HaEpi cells for 72 h using the Quick Shuttle-enhanced transfection reagent (KX0110042, Biodragon Immunotechnologies, Beijing, China), followed by 20E treatment for 6 h, with DMSO as the control. Anti-GFP antibodies (1:100) were used to immunoprecipitate HaPTEN-GFP-His. Then, the cells were lysed using lysate buffer (P0013, Beyotime Biotechnology) containing phosphatase inhibitor and protease inhibitor, and the supernatant was collected by centrifugation at 10000 × g for 15 min at 4˚C. The supernatant was added with 50 μL of protein A (SA012005, Smart-Lifesciences, Changzhou, China), and shaken at 4˚C for 1 h to remove non-specific binding. Cell lysate (20 μL) was extracted as input. The antibody was added into cell lysate overnight at 4˚C, and then 50 μL of protein A was added to the cell lysate, incubated for 4 h, and then washed with lysate buffer. Thereafter, 40 μL of lysate buffer was used to resuspend the protein A, which was then analyzed using western blotting. For the HaPTPN6 and HaP60 Co-IP assay, 4 μg HaPTPN6-GFP-His plasmid was transfected into HaEpi cells for 72 h, which were then treated with 5 μg/mL insulin for 1 h and 5 μM 20E for 6 h or only treated with 5 μM 20E for 6 h, DMSO as control, and CNBr-activated Sepharose 4B conjugated with anti-GFP antibody to immunoprecipitate HaP-TPN6-GFP-His. For HaP60 and HaP110 Co-IP assay, cells were added with 5 μg/mL insulin for 1 h, followed by 5 μM 20E for 6 h. Then, the cells were lysed using a lysate buffer containing phosphatase inhibitor and protease inhibitor, and the supernatant was collected by centrifugation 10000 × g for 15 min at 4˚C and incubated with CNBr-activated Sepharose 4B conjugated with anti-HaP60 antibodies overnight at 4˚C to immunoprecipitate HaP60. The CNBr beads were washed with wash buffer (0.1 M Tris-HCl, 0.1 M NaCl, pH 8.0), and the complex conjugated by CNBr beads were eluted using buffer (0.1 M glycine, pH 2.5) followed by neutralization, before being subjected to western blotting.

CASP3 activity detection
HaEpi cells were transfected with 4 μg RFP-His, HaP60-RFP-His, or HaP110-RFP-His for 72 h, respectively, or treated with 5 μM 20E for 72 h, and then incubated with 2 μM NucView 488 caspase-3 assay kit (30029, Biotium, Fremont, CA, USA) for 30 min at room temperature. The cells were then fixed in 4% PFA, washed with PBS, and the nuclei were stained with DAPI for 10 min. Finally, the CASP3 signal was observed under an Olympus BX51 fluorescence microscope.

Chromatin immunoprecipitation (ChIP) assay
For EcRE detection, 4 μg EcR-RFP-His were transfected into HaEpi cells for 72 h, which were then treated with 5 μM 20E for 6 h. For FOXOBE detection, 4 μg HaFOXO-GFP-His were transfected into HaEpi cells for 72 h, which were then treated with 5 μM 20E for 6 h. The cells were treated according to the instruction of the ChIP assay kit (P2078, Beyotime Biotechnology). Anti-His antibody was used to detect EcR-RFP-His, anti-GFP antibody was used to detect HaFOXO-GFP-His, and IgG was used as a negative control.

Flow cytometry detection of apoptosis
Apoptosis was detected using flow cytometry with an Annexin-FITC apoptosis detection kit (C1062M, Beyotime Biotech). HaEpi cells were transfected with 4 μg HaP60-His and His respectively for 72 h or treated with 5 μM 20E for 72 h, with DMSO as a control for 20E. The cells were washed with PBS, lysed with 0.5% trypsin, and collected by centrifugation at 1000 × g for 5 min at 4˚C and then treated according to Annexin-FITC apoptosis detection kit to detect apoptosis by flow cytometry.

E. coli HT115-expressed dsRNA feeding to larvae
Gene fragments of dsHaP60, dsHaP110, dsHaPtpn6, and dsGFP were amplified, fused with vector L4440, provided by Dr. Marek Jindra (Biology Center ASCR, Czech Republic), and then transformed into E. coli DH5α. The constructed plasmids were purified and then transferred into E. coli HT115, provided by Dr. Marek Jindra. HT115 cells expressing dsHaP60, dsHaP110, dsHaPtpn6, and dsGFP were inoculated into 300 mL of medium (pH 7.4), shaken for 3 h at 37˚C, and then induced using IPTG for 5 h. Subsequently, the bacterial culture was centrifuged at 10000 × g for 5 min at 4˚C. The bacteria were resuspended in 500 μL PBS, pH 7.4. PBS with bacteria expressed dsHaP60, dsHaP110 or dsHaPtpn6 was applied to the surface of forage that was fed to larva every day up to the 6th-72 h stage. Bacteria expressing dsGFP were fed at an equal amount at the same time. For HaPtpn6 knockdown in the larva, 2 μg dsHaPtpn6 were injected into the 6th-72 h larva after feeding E.coli expressing dsHaPtpn6 at the last day for 24 h, and the second injection was performed at the 6th-96 h, and then the last injection was performed at the 6th-120 h. The interference efficiency was checked using qRT-PCR and western blotting.

RNAi knockdown by injecting dsRNA into larva
For HaP60 knockdown in the larva, 2 μg dsHaP60 were injected into the 6th-72 h larva for 24 h, dsGFP as a control. The second injection was conducted at the 6th-96 h, and then the last injection was performed at the 6th-120 h to observe phenotype and analyze the pupation time and pupal weight. The total RNA or protein was extracted after the second injection for 12 h to detect the RNAi efficiency of HaP60 knockdown.

20E titer detection
Hemolymph (100 μL) from more than three larvae was frozen in the liquid nitrogen overnight and freeze-dried for 6 h. The samples were then ground in 500 μL of 80% methanol and centrifuged at 10000 × g at 4˚C for 10 min. The supernatant was air-dried at room temperature for 3 h and dissolved in 100 μL of enzyme immunoassay (EIA) buffer to detect the 20E titer using a 20E Enzyme Immunoassay kit (A05120, Bertin Pharma, Montignyle-Bretonneux, France).

Phylogenetic analysis
The phylogenetic analysis was conducted requiring four steps by MEGA 5.0, a program for phylogenetic analysis in a single environment. We identified and acquired Homo sapiens, Drosophila melanogaster, Bombyx mori, Danaus plexippus plexippus, and H. armigera's protein tyrosine phosphatases non-receptor (PTPNs) protein sequences linked to gene accession numbers by BLAST search, and aligned those sequences, then estimated a tree from the aligned sequences and showed the tree in an image way to convey the relevant information to others.

Statistical method
All the assays were repeated at least three times. Statistical analysis was conducted using Student's t-test for pairwise comparison or analysis of variance (ANOVA) for multiple comparisons. In the figures, asterisks represent significant differences ( � P < 0.05, and �� P< 0.01), and different lowercase letters show significant differences (P < 0.05). The values are the mean ± SD. Data were analyzed using Image-J software (NIH, Bethesda, MD, USA) for western blot and immunofluorescence, and generated using GraphPad 7.0 (GraphPad Inc., La Jolla, CA, USA). Average weight curves were analyzed using GraphPad 7.0 with thirty larvae each time and repeated three times. H. armigera tyrosine phosphatase corkscrew-like isoform X1 (HaPTPN6), labeled with red, was adjacent with D. plexippus tyrosine-protein phosphatase non-receptor type 6 (PTPN6), and "100" in the branch presents the higher genetic relationship in evolution, therefore we named H. armigera tyrosine phosphatase corkscrew-like isoform X1 as H. armigera PTPN6 (HaPTPN6)  Validation: Xiao-Fan Zhao.