Dual Effect of Raptor on Neonatal β-Cell Proliferation and Identity Maintenance

Absolute or relative deficits in functional β-cell mass underlie the onset of diabetes (1). Approaches to maintain or replenish both function and number of β-cells are promising strategies that would benefit the prevention and treatment of diabetes (2). Dynamic change of β-cell mass is controlled by a complicated balance of proliferation, neogenesis, apoptosis, and transdifferentiation from other cell types (3–7). Genetic tracing studies have shown that self-replication of preexisting β-cells is the major source for new β-cells during adult life (8,9). Mature β-cells are mostly static, and even under metabolic stress, their proliferative potential remains limited (1∼2%) (10). On the contrary, neonatal β-cells are highly proliferative (10∼20%), and their expansion largely determines adult β-cell mass. It is widely accepted that establishment of adequate β-cell mass during the neonatal period (from prenatal to early postnatal) is essential for effective glucose control in adulthood (11,12). Importantly, the neonatal period is also critical for immature β-cells to undergo functional maturation and acquire their glucose-responsive insulin-producing phenotype (13). Therefore, elucidating the biological basis of these highly proliferative β-cells and searching for targets that can trigger endogenous expansion and functional maturation of newly formed β-cells might identify new strategies for diabetes treatment.

Numerous studies have identified intracellular and extrinsic mitogenic factors that regulate mature β-cell proliferation (14–19). Given that both the proliferation rate and the regulatory mechanisms are different between neonatal and mature β-cells (20,21), these well-established cognitions of mature β-cell proliferation cannot be simply applied to neonatal β-cells. Goodyer et al. (22) reported that calcineurin/NFAT signaling promoted neonatal β-cell proliferation. Moreover, platelet-derived growth factor receptor (Pdgfr) signaling was found to promote juvenile β-cell proliferation through enhancer of zeste homolog 2 (Ezh2) (23,24). These studies have shed some light on the mechanisms of neonatal β-cell proliferation, but physiological pathways governing both cell cycle progression and identity maintenance remain elusive.

Mechanistic target of rapamycin (mTORC1) is a nutrient sensor that can transmit nutrient signals to downstream targets to regulate multiple cellular processes (25). mTORC1 is sensitive to rapamycin and contains mTOR, Deptor, Raptor, and PRAS40 (26). We and others have reported that mTORC1 regulates functional maturation, growth, and survival of pancreatic β-cells (27,28). However, inconsistent findings exist with regard to proliferation. In different β-cell–specific transgenic mice targeting mTORC1 components, some groups have found a positive impact on mature β-cell proliferation, whereas others did not observe any influence (18,29–33). Moreover, little is known about whether and how mTORC1 participates in neonatal β-cell proliferation. By using Ngn3RapKO mice, we found that mTORC1 was dispensable for endocrine cell lineage formation but specifically regulated both proliferation and identity maintenance in neonatal β-cells. Loss of Raptor caused neonatal diabetes as a result of inadequate establishment of functional β-cell mass at postnatal day 14 (P14). Unlike in mature β-cells, Raptor regulated both G1/S and G2/M phase cell cycle transition, thus permitting a high proliferation rate of neonatal β-cells. The mTORC1/Ezh2 pathway was identified to mediate G2/M phase transition and proliferation in neonatal β-cells. Our findings unraveled the dual effect of mTORC1 in neonatal β-cells and thus highlighted its role in determining postnatal functional β-cell mass.

Ngn3-Cre mice that express the Cre recombinase gene under control of Ngn3 gene promoter were a gift from W.Z. The Ngn3-Cre;Raptor^flox/flox (Ngn3RapKO) mice were generated by crossing Raptor^flox/flox mice (purchased from The Jackson Laboratory) with Ngn3-Cre mice. Ngn3-Cre;Raptor^+/+ (wild-type [WT]) mice were used as their littermate controls. All mice were housed in the animal facility on a 12-h light/dark cycle. Normal chow and water were available ad libitum. In the current study, male mice were used in all the experiments, unless otherwise stated. All animal experiments were approved by the Animal Care Committee of Shanghai Jiao Tong University.

Mouse insulinoma cells (MIN6 cell line) purchased from CAMS Cell Culture Center (Beijing, China) were grown in DMEM containing 25.0 mmol/L glucose, 15% FBS, 100 IU/mL penicillin, 100 μg/mL streptomycin, 10.2 mmol/L L-glutamine, and 2.5 mmol/L β-mercaptoethanol at 37°C in a humidified 5% CO₂ atmosphere. At 70–80% confluence, each well of cells was treated with 25 nmol/L rapamycin (Cell Signaling Technology) or DMSO (Sigma-Aldrich) for 48 h. At the end of the culture, cells were harvested for protein or RNA extraction, flow cytometry, and further analysis.

Pancreata were dissected, fixed, and processed as described previously (34). Primary and secondary antibodies used in this study are listed in Supplementary Table 1. For immunostaining analysis, entire pancreatic tissues were then continuously sectioned at 5-μm thickness. Immunochemistry was performed on continuous sections (selected every 150 μm apart, 10–12 sections per animal) to obtain representative β-cell mass information of the whole pancreas. Immunochemistry staining of insulin for β-cell mass analysis was performed using a diaminobenzidine peroxidase substrate kit (Vector Laboratories, Burlingame, CA), counterstaining with eosin. Digital images of whole pancreas were captured by an MZ 100 microscope (Nikon, Tokyo, Japan). Total pancreatic and insulin⁺ areas of each section were measured using MetaMorph version 7.1 (Molecular Devices, Sunnyvale, CA).

Mice were fasted for 6 h, and blood glucose concentrations were measured by glucometer from the tail vein. Plasma proinsulin and insulin concentrations were determined by ELISA kit (ALPCO).

Cells and freshly isolated islets were lysed, quantified, and blotted as described previously (34). Primary antibodies were as follows: rabbit anti-RAPTOR (1:1,000; Cell Signaling Technology), rabbit anti-PLK1 (1:1,000; Cell Signaling Technology), rabbit anti-EZH2 (1:1,000; Cell Signaling Technology), rabbit anti-pS6 (Ser240/244) (1:1,000; Cell Signaling Technology), rabbit anti-p57Kip2 (1:1,000; Cell Signaling Technology), and mouse anti-tubulin (1:10,000; Sigma-Aldrich). Tubulin or GAPDH was used as an internal control to normalize band intensity.

Islet total RNA was extracted using an RNeasy Micro Kit (QIAGEN) according to the manufacturer’s protocol. Total cell RNA was extracted using TRIzol reagent (Invitrogen). Reverse transcription and quantitative real-time PCR (RT-PCR) were performed as previously described (3). PCRs were performed in duplicate. The expression levels were normalized to individual β-actin. Primers used in this study are listed in Supplementary Table 2.

One hundred nanograms of control or EZH2 3′-untranslated region psiCHECK-2 plasmid were transiently cotransfected with or without 25 nmol/L rapamycin using Lipofectamine 2000 (Invitrogen). Cell lysates were harvested 48 h after transfection and subjected to the Dual-Glo Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions. The firefly luciferase activity was normalized to the Renilla luciferase activity and presented as relative luciferase activity.

Adenoviral vector encoding the mouse Ezh2 gene (AdEzh2) and negative control (AdGFP) was constructed, packaged, purified, and titrated at GeneChem Co., Ltd. MIN6 cells were transfected with purified adenovirus at 50 multiplicity of infection for 48 h. After infection, experiments were implemented as described above.

siRNA oligonucleotides for Raptor and control were designed and synthesized from GenePharma (Shanghai, China). Lipofectamine 3000 transfection reagent (Invitrogen) was used according to the manufacturer’s instructions. For Ezh2 overexpression experiments, culture medium was replaced after 24 h of siRNA transfection and then adenoviral infection with AdEzh2 or AdGFP for another 48 h. At the end of culture, cells were collected for further experiments.

A total of 1 μg qualified RNA per sample was used as input material for the library preparation. The sequencing libraries were generated using the VAHTS mRNA-seq V2 Library Prep Kit for Illumina (Vazyme) according to the manufacturer’s recommendations. The library preparations were sequenced on an Illumina HiSeq X Ten platform. Cuffdiff v1.3.0 was used to calculate fragments per kilobase of exon model per million mapped reads (FPKM) for coding genes in each sample. Genes with a corrected P < 0.05 and the absolute value of log₂ (fold-change) ≥1.5 were assigned as significantly differentially expressed.

All data are presented as mean ± SEM. Statistical analyses were performed with unpaired Student t test for two groups or two-way ANOVA for multiple groups. P < 0.05 was considered statistically significant.

Data generated in this study were deposited to the National Center for Biotechnology Information (SRP159433 for Ngn3RapKO RNA sequencing, GSE84404 for βRapKO).

To investigate whether mTORC1 signaling determines lineage of embryonic endocrine cells, we used Ngn3-Cre to delete Raptor from the developing endocrine pancreas (Ngn3RapKO). Successful knockout of Raptor was confirmed at both transcription and protein levels (Fig. 1B and C). Compared with WT controls, the mRNA expression of Raptor decreased >75%, and RAPTOR protein dramatically reduced in islets isolated from P14 Ngn3RapKO mice (Fig. 1B and C). Moreover, Ngn3⁺pS6⁺ cells could be easily found in embryonic day 15.5 (E15.5) WT pancreas, whereas pS6 staining was selectively absent in Ngn3⁺ cells in age-matched Ngn3RapKO pancreas (Fig. 1A).

At E17.5, we did not detect significant changes in either total number or proliferation rate of Ngn3⁺ cells (Fig. 1D and E), suggesting that Raptor was dispensable for the size of endocrine progenitor pool. We then calculated endocrine cell numbers and found that the absolute α- and β-cell numbers at E17.5 were comparable between the two groups (Fig. 1F). Surprisingly, several days later at birth (P0), reductions in both β- and α-cell numbers were observed (Fig. 1G and H). On the contrary, Raptor deletion did not affect δ- and PP-cell numbers at birth (Fig. 1H). These results indicate that Raptor is dispensable for lineage determination of different endocrine cell types but determines α- and β-cell number at birth.

It is known that the establishment of functional β-cell mass is determined by embryonic lineage formation, neonatal proliferation, and functional maturation (35). We first compared mTORC1 activity and proliferation rate in intact neonatal β-cells during E17.5 to P14 (Fig. 2A). The proliferation rate of β-cells reached maximum at E17.5, gradually decreased after birth, and then dropped rapidly at P14 and was kept at a very low level after β-cell maturation (Fig. 2A and Supplementary Fig. 1A). Interestingly, mTORC1 activity, evidenced by pS6 expression, was strongly induced between E17.5 and P4 (Fig. 2A and B); at P14, only a few mature β-cells showed staining for pS6. We then calculated pS6⁺Ki67⁺ β-cells at E17.5 and P4 and found ∼80% of Ki67⁺-proliferating β-cells were pS6⁺ (Fig. 2B). The maximal activation of mTORC1 during the critical β-cell proliferation period raised the hypothesis that Raptor might control neonatal β-cell proliferation. Deletion of Raptor did not affect E15.5 β-cell proliferation but significantly reduced β-cell proliferation rates at E17.5 and P0 (Fig. 2C). The reduction was maintained until P14. In addition, α-cell proliferation was diminished: the proportion of Ki67⁺/glucagon⁺ cells in mutant mice decreased by 45.0% and 54.1% at E17.5 and P0, respectively (Fig. 2D).

Ngn3RapKO mice began to display overt hyperglycemia at P14 (Fig. 3A). This was accompanied by increased plasma proinsulin concentrations but reduced insulin and glucagon levels at P14 (Fig. 3C–E). At 4 weeks, mutants developed severe hyperglycemia and had significant weight loss (Fig. 3A and B), having 72% reduced 6-h fasted plasma insulin levels and 81% decreased pancreas insulin content (Supplementary Fig. 2A and B). Ngn3RapKO mice eventually died between 5 and 9 weeks (mean life span 7 weeks) as a result of severe and sustained hyperglycemia (Fig. 3F). Female mutants developed even more pronounced hyperglycemia than male mutants (Supplementary Fig. 2C). Histological analysis on P14 Ngn3RapKO pancreas revealed that both β-cell mass and β-cell number per section (Fig. 3H and I) were significantly reduced and accompanied by half reductions in islet number (Fig. 3J) and islet size (Fig. 3K and L). Interestingly, unlike most diabetic models, diabetic Ngn3RapKO islets had very few α-cells at the periphery (Fig. 3G). Indeed, we detected significant decreases in α-cell mass, α-cell number per section, and plasma glucagon levels (Fig. 3E, H, and I) in the mutants. We did not observe apparent changes in the number of δ and PP cells in P14 Ngn3RapKO islets (Fig. 3I and Supplementary Fig. 2D). At 4 weeks old, Ngn3RapKO mice displayed further attenuated insulin immunostaining inside islets (Supplementary Fig. 2E).

To investigate the underlying mechanisms, we performed RNA sequencing analysis on WT and mutant islets at P14. Comparing gene expression profiles between the two groups, we detected significant differences in the expression levels of 2,327 genes (fold change >1.5, P < 0.05), among which 1,308 were upregulated and 1,019 downregulated. Gene ontology analysis showed that genes involved in system development that stimulate response, cell proliferation, mitotic cell cycle, and ion transport signaling pathways related to insulin secretion were significantly altered in P14 Ngn3RapKO islets (Fig. 4A).

Importantly, critical transcriptional factors involved in maintaining β-cell identity were significantly downregulated (i.e., Ins1, Pdx1, NeuroD1, MafA, Nkx2.2) (Fig. 4B). Moreover, genes involved in glucose metabolism, insulin secretion, and functional maturation were also preferentially decreased (i.e., Gck, Atp4a, Atp5c1, Cox6a2, Slc2a2, Ucn3) (Fig. 4B). Raptor depletion also affected genes important for α-cell identity and glucagon secretion (i.e., Gcg, Arx, Slc38a5, Fbp1) (Fig. 4C). By RT-PCR and immunostaining, we confirmed the downregulation of several important transcription factors (i.e., Pdx1, MafA, Nkx2.2, Arx) in P14 mutants (Fig. 4D and E). The proportion of Pdx1⁺/insulin⁺, MafA⁺/insulin⁺, Nkx2.2⁺/insulin⁺, and Arx⁺/glucagon⁺ cells in mutant mice decreased by 33.5%, 58.0%, 51.2%, and 41.4%, respectively, at P14 (Fig. 4F–I). These results demonstrate that mTORC1 signaling is critical for maintaining neonatal β- and α-cell identity and function.

To obtain an overall view of the cell cycle regulated by Raptor, we further performed principal component analysis (PCA) on the basis of cell cycle genes. In the PCA plot, cell cycle genes in the two groups were completely separated (Fig. 5A). On the basis of a database of enriched cell cycle genes in proliferating α- and β-cells (20), we found that Raptor regulated 30% (88 in 281) of genes involved in G1/S phase progression and 59% (126 in 199) of genes involved in G2/M phase progression (Fig. 5B). Importantly, we checked these significantly altered genes in 8-week-old βRapKO islets: only three of them were changed (Supplementary Fig. 3A). This observation was in line with our previous finding that Raptor was not required for basal proliferation in mature β-cells. Our data demonstrate that Raptor specifically controls proliferation and cell cycle progression in neonatal β-cells.

Further analysis revealed that Raptor regulated proliferation through a variety of factors involved in cell cycle transition (both G1/S phase and G2/M phase) (Fig. 5C and D and Supplementary Table 3). We detected significantly reduced expressions of multiple G1/S genes (i.e., Prim1, Dna2, Cdk2, Cdk4, Mcm2, Mcm3, Mcm5, Mcm6) (Fig. 5D) and confirmed their downregulation in isolated P14 mutant islets by RT-PCR (Fig. 5E).

Interestingly, we found more than one-half of the factors involved in G2/M phase transition were also significantly changed by Raptor in P14 mutants (Fig. 5B–D). Using RT-PCR, we detected reduced mRNA expression of Foxm1, Cdk1, CyclinA2, CyclinB1, and Plk1 in P14 mutant islets (Fig. 5F). Plk1-mediated G2/M phase transition was reported to be critical for adaptive expansion of mouse and human β-cells (17). Immunofluorescent staining on both E17.5 and P14 pancreas sections showed that PLK1 expression was diminished in Ngn3RapKO β-cells (Fig. 5G). Moreover, we observed reduced PLK1 protein abundance in isolated P14 Ngn3RapKO islets (Supplementary Fig. 4A and B). In addition, we found that mTORC1 inhibitor rapamycin could directly reduce Plk1 mRNA (Fig. 5H) and PLK1 protein expression (Fig. 5I) in MIN6 cells, suggesting a direct link between mTORC1 and Plk1. In conclusion, our data demonstrate that mTORC1-mediated cell cycle control in neonatal β-cells involves both G1/S phase and G2/M phase regulation.

Cell cycle inhibitors were reported to be equally important in regulating β-cell proliferation (36–38). In our study, most factors that promote cell cycle progression were downregulated: only four cell cycle inhibitors were significantly upregulated (p57Kip2, Tgfb2, Tgfb3, Gadd45b) (Fig. 5D). We then performed RT-PCR analysis on these cell cycle inhibitors in both P14 islets and 48-h rapamycin-treated MIN6 cells: only p57Kip2 expression was significantly upregulated in both preparations (Fig. 6A and B). p57Kip2 is a potent inhibitor of cyclin/cyclin-dependent kinase (CDK) complexes and is negatively regulated by Ezh2 (39,40). Indeed, Ezh2 mRNA expression was significantly downregulated in both P14 mutant islets and 48-h rapamycin-treated MIN6 cells (Fig. 6C and D).

It has been suggested that cell cycle regulation differs between neonatal and mature β-cells (20). We performed immunostaining against several essential Raptor-mediated cell cycle regulators on both E17.5 and P14 Ngn3RapKO and WT pancreas. We first checked two cyclin D family proteins—cyclin D1 and cyclin D2—which have been reported to be regulated by mTORC1 signaling in mature β-cells (16,41). We observed decreased expression of both cyclin D1 and cyclin D2 in Raptor-deficient β-cells at P14 (Fig. 6E). However, this was not the case at E17.5: cyclin D1 expression was completely absent in both WT and mutant β-cells, while the expression level of cyclin D2 was unaffected by loss of Raptor (Fig. 6E). We then checked Ezh2 and p57Kip2 expression in E17.5 and P14 Ngn3RapKO pancreas sections. Ezh2 was abundantly present in both E17.5 and P14 insulin⁺ cells, and its expression levels were diminished in age-matched Raptor-deficient β-cells (Fig. 6E). In parallel, immunostaining of p57Kip2 was exclusively present in E17.5 WT, and its expression was upregulated in Ngn3RapKO (Fig. 6E). Accordingly, Raptor-deficient islets from P14 Ngn3RapKO mice also showed a significant decrease in EZH2 protein abundance compared with WT controls (Fig. 6F). These results suggest that Ezh2/p57Kip2 might facilitate mTORC1-mediated cell cycle progression in neonatal β-cells.

We then examined Ezh2 expression and mTORC1 activity in β-cells during the physiological neonatal period by performing immunofluorescence staining for Ezh2 and pS6 on P0, P4, and P14 pancreatic buds. We observed a similar pattern of Ezh2 expression and mTORC1 activity: highly activated mTORC1 in β-cells coincided with abundant expression of Ezh2, both of which reached the maximum at P4 (Fig. 6G). Taken together, our data indicate that mTORC1/Ezh2 might play an important role in neonatal β-cell replication.

To examine whether mTORC1 directly regulates Ezh2 expression, we treated MIN6 cells in vitro with the mTORC1 inhibitor rapamycin (Fig. 7A–D). Forty-eight-hour 25 nmol/L rapamycin treatment significantly reduced Ezh2 promoter activity (Fig. 7A) and Ezh2 mRNA expression (Fig. 6D). Moreover, 48-h rapamycin treatment decreased EZH2 protein abundance in MIN6 cells, as evidenced by immunofluorescence staining and Western blot analysis (Fig. 7B–D). We further performed coimmunoprecipitation assays and found that EZH2 was readily coprecipitated with endogenous RAPTOR but not with control IgG (Fig. 7E), indicating a possible binding between RAPTOR and EZH2 protein in mammalian cells. Taken together, our results suggest that mTORC1 directly regulates Ezh2 expression through transcriptional modification and/or protein interaction.

We then explored whether mTORC1 regulated β-cell proliferation through Ezh2. Rapamycin treatment reduced EZH2 protein abundance in MIN6 cells (Fig. 7F). This was accompanied by a 20% decrease in proliferation rate in rapamycin-treated MIN6 cells (Fig. 7G). More importantly, overexpression of Ezh2 to the basal state successfully reversed rapamycin-mediated p57Kip2 upregulation and partially prevented rapamycin-induced proliferation inhibition in MIN6 cells (Fig. 7F and G and Supplementary Fig. 5A and B). We further performed cell cycle analysis using flow cytometry. Rapamycin treatment induced G1 phase arrest, manifested as a 40% increase of cells in the G1 phase and a 30% decrease of cells in the G2/M phase (Supplementary Fig. 5B–D). Interestingly, overexpression of Ezh2 did not reverse rapamycin-induced G1 phase arrest (Supplementary Fig. 5B–D), but it increased the percentage of cells in the G2/M phase and thus accelerated the completion of cell cycle progression into the mitotic stage (Supplementary Fig. 5B–D). Rapamycin treatment for 48 h significantly reduced mRNA levels of Pdx1, Nkx6.1, and Nkx2.2 but did not alter MafA expression (Fig. 7H). Unexpectedly, overexpression of Ezh2 could completely restore the downregulation of these transcription factors (Fig. 7H). To further confirm the specific effect of mTORC1 on Ezh2 expression and β-cell proliferation, we silenced Raptor, a key component for maintaining mTORC1 activity. The knockdown effect of different short hairpin Raptor and siRaptor sequences were verified at protein levels (Supplementary Fig. 6A), and then siRaptor-1# sequence was chosen for further experiments. Knockdown of Raptor caused a 43% reduction of Ezh2 protein expression (Fig. 7I), without alternating mTORC2 activity (Supplementary Fig. 6B). Restored Ezh2 expression after Raptor silence partially reversed the inhibitory effect of mTORC1 inhibition on β-cell proliferation (Fig. 7J and K), transcriptional factor (Pdx1/Nkx6.1/Nkx2.2) expression (Fig. 7L), and cell cycle progression (Fig. 7M and N).

To evaluate the hypothesis that mTORC1-Ezh2 regulates proliferation in neonatal β-cells, we isolated P14 WT and mutant islets and then treated with AdGFP or AdEzh2 virus for 48 h. Successful overexpression of Ezh2 in P14 Ngn3RapKO islets restored its expression to normal levels (Fig. 7O) and partially reversed proliferation failure of β-cells in Ngn3RapKO islets (to ∼68.3% of normal state) (Fig. 7P).

Molecular pathways governing β-cell proliferation have been under intensive investigation in the hope of identifying strategies for stimulating β-cell expansion (2). However, our knowledge of native signaling pathways facilitating neonatal β-cell proliferation is limited. Although findings from β-cell–specific transgenic animal models of mTORC1 components have provided insights into mTORC1’s multiple roles in β-cells, it remains unclear whether and how it regulates proliferation during the β-cell life span (27). Combined with our previous observations, the present study extends our understanding of mTORC1 in β-cells. First, mTORC1 has a dual effect on both proliferation and identity maintenance in immature β-cells. Second, cell cycle progression regulated by Raptor in neonatal β-cells is completely different from that in mature β-cells. Third, although mTORC1 is not involved in proliferation of unstimulated mature β-cells (3), it facilitates β-cell compensatory proliferation under metabolic stress (34). Our observation in Ngn3RapKO mice is consistent with Raptor^PANCKO mice, in which deletion of Raptor in Pdx1⁺ pancreatic progenitors reduced pancreas size with reduced β-cell number (18). However, another research group did not observe any islet morphological deformities in Ngn3RapKO mice, and the authors attributed the impact of mTOR on islet mass to mTORC2 (33). In the present study, we used the same Raptor^flox/flox mice as we previously reported (3) and provided novel data on the dual effect of mTORC1 in regulating both identity and proliferation in neonatal β-cells, as outlined in Fig. 7Q. After β-cell formation, mTORC1 hyperactivation permits high proliferation capacity of immature β-cells to expand and reach adequate β-cell mass. At the same time, mTORC1 is highly expressed to ensure that neonatal β-cells undergo functional maturation and maintain cell identity. After the β-cell matures, physiological mTORC1 activity remains at low levels, which is sufficient for the maintenance of β-cell function. Upon stimulation, β-cells can be recruited to a hyperactive state to synthesize/secrete insulin and, later, even to proliferate to cope with increased metabolic demand (34,42–45). The activity of mTORC1 adaptively changed in different developmental stages and/or various metabolic statuses, emphasizing its sensitivity and importance in metabolic control.

Adult pancreatic β-cells are largely quiescent, while nutrients and growth factors are known to promote G1/S phase transition to facilitate proliferation (14,46). G2/M phase transition was found to be equally critical for β-cell replication (17). For the first time to our knowledge, we have provided an overall picture of mTORC1 signaling in regulating both G1/S and G2/M phase transition in neonatal β-cells. First, loss of mTORC1 in neonatal β-cells reduced expression of a cluster of G1 phase factors. This was consistent with previous findings that mTORC1 can modulate synthesis and stability of cyclin D2 and CDK4 to regulate β-cell proliferation (34,41). Second, mTORC1 inactivation decreased the expression of several CDK-MCM family members, which blocked DNA replication initiation and hindered S phase DNA biosynthesis (47). Third, we detected that 126 of 199 genes involved in G2/M phase transition were downregulated. In late G2 phase during normal cell cycle, Plk1 activation depends on cyclin A2–Cdk1 activity (48–50) and serves as a critical component for β-cells to progress through the G2 phase to mitosis (17). In vivo and in vitro, we showed that loss of Raptor significantly reduced Plk1 mRNA and PLK1 protein levels, providing a link between mTORC1 and G2/M transition. Taken together, Raptor not only influences G1 phase but also regulates S phase DNA biosynthesis and G2 phase to mitotic entry and thus controls cell cycle progression in neonatal β-cells.

Interestingly, we identified Ezh2 as one of the downstream targets of mTORC1 in controlling neonatal β-cell proliferation (23). Ezh2 expression was highly induced during the physiological neonatal period and was then reduced in adult β-cells, which was in parallel with the mTORC1 expression pattern. Moreover, overexpression of Ezh2 rescued mTORC1-mediated G2/M phase transition and β-cell proliferation. We also found that Ezh2 downstream target p57Kip2 was the only upregulated CDK inhibitors in Raptor-deficient islets. p57Kip2 is a potent inhibitor of several G1 phase and G2/M phase factors (51). It was reported that p57Kip2 expression in mouse β-cells is temporarily limited in the embryonic and early postnatal period, but declines rapidly during β-cell maturation, suggesting its specific role in modulating neonatal β-cell proliferation (20). Collectively, our data have shown that mTORC1 can modulate Ezh2/p57Kip2 to exert identical effects on G2/M phase transition and neonatal β-cell proliferation. Previous studies have shown that genetic and pharmacological inhibition of mTORC1 signaling would lead to remarkable restoration of mTORC2-AKT activity (31,52). In our model, we did not detect compensatory changes of pAKT expression after silencing Raptor. Given the fact that mTORC2 is known to positively regulate β-cell proliferation, our main conclusion is that loss of Raptor compromises neonatal β-cell proliferation. We believe that the significant inhibitory effect on neonatal β-cell proliferation is primarily due to loss of Raptor rather than to the compensation of mTORC2. However, we could not rule out the possibility of mTORC2 compensation in other biological processes in Ngn3RapKO islets.

In addition to its impact on neonatal β-cell proliferation, mTORC1 maintained neonatal β-cell identity and function. Newborn immature β-cells are sensitive to nutrients and proceed to functional maturity during the neonatal period. Animal models with maternal malnutrition have been shown to induce permanent changes of β-cell function in offspring (53–55). mTORC1 has been proposed to mediate β-cell dysfunction in offspring with a maternal low-protein diet (53). Our results showed that loss of Raptor in neonatal β-cells reduced multiple genes important for β-cell identity maintenance and downregulated key factors involved in glucose metabolism and insulin secretion. Interestingly, overexpression of Ezh2 rescued rapamycin-suppressed Pdx1, Nkx6.1, and Nkx2.2 expression; the underlying mechanisms need to be further clarified. Taken together, our findings highlighted that before weaning, newly formed β-cells required mTORC1 signaling to fulfill both number expansion and functional maturation.

The effect of Raptor on proliferation and identity maintenance not only is restricted in β-cells but also affects neonatal α-cell phenotype. A previous study by Bozadjieva et al. (56) using αRaptorKO showed that mTORC1 signaling was dispensable for α-cell development but essential for α-cell mass and function. Interestingly, in contrast to most diabetic models with an increased α/β-cell ratio in islets, our Ngn3RapKO mice developed severe hyperglycemia under the condition of absolute α-cell loss and reduced plasma glucagon levels. In addition to promoting hepatic glycogenolysis and gluconeogenesis, glucagon also plays a role in differentiation of early β-cells and islet microcirculation (57–59). It would be interesting to find out whether such abnormal islet architectures with reduction in both α- and β-cells exist in certain subpopulations of patients with diabetes, especially neonatal diabetes. We currently do not know whether ablation of Raptor in multiple cells within the islet could affect one another’s cell fate; it would be ideal to perform single-cell RNA sequencing analysis to clarify the identical regulatory mechanisms in individual β- and α-cells in Ngn3RapKO mice. Meanwhile, given that Ngn3⁺ progenitors also give rise to enteroendocrine cells, the interaction between enteroendocrine cells including glucagon-like peptide 1–secreting L cells and endocrine cell number should be considered (60).

In summary, the present study emphasizes that absence of mTORC1 signaling in a critical developmental window of neonatal rodent β-cells would impair physiological establishment of functional β-cell mass and effective glycemic control in adulthood. The dual effect of mTORC1 signaling on both expansion and functional maturation of neonatal β-cells provides a potential theoretical basis for replenishing functional β-cell pools.

Acknowledgments. The authors thank Dr. Kunio Kitamura (Mitsubishi Kagaku Institute of Life Sciences) for providing the anti-Arx antibody (61).

Funding. This work was supported by National Natural Science Foundation of China grants (81670700, 81870527).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.W., J.S., Q.N., A.N., Y.G., and S.W. performed experiments and analyzed the data. Y.W., G.N., W.W., and Q.W. designed the project, supervised the research, and coordinated the execution of the experimental plan. Y.W. and Q.W. wrote the manuscript. W.Z. contributed to the data discussion. Q.W. is the guarantor of this work and, as such, has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.