Rb and p53 Execute Distinct Roles in the Development of Pancreatic Neuroendocrine Tumors

Pancreatic neuroendocrine neoplasms (PanNEN) are the second most common epithelial neoplasms in the pancreas, and the number of people affected is gradually increasing (1). In 2017, the World Health Organization (WHO) classified human PanNENs into two groups: well-differentiated PanNENs, called PanNETs, and poorly differentiated PanNENs, called PanNECs. PanNECs, which include small and large cell carcinomas, have an extremely poor prognosis and are thought to be biologically distinct from pancreatic neuroendocrine tumors (PanNET; refs. 2–5). PanNETs are further subclassified into grades (G) 1 to 3 on the basis of their proliferative activity assessed by the Ki67 labeling index and mitotic rate. Although the prognosis of patients with PanNETs correlates with the tumor grade, the underlying mechanisms of PanNET formation and grade progression are unclear.

Previous gene analyses of PanNETs revealed major mutations in MEN1, DAXX, ATRX, and genes in the mTOR pathway, but extremely rare mutations in the retinoblastoma (RB) gene or TP53 gene (6–8). In contrast, a recent report analyzed by IHC revealed a loss of Rb expression in 54.5% of G3 PanNEC but not in G3 PanNET cases (5). However, in various human neoplasms, the RB gene is functionally inactivated not only by mutations, but also by altered expression of upstream regulators (9). Indeed, a recent study demonstrated that increased expression of CDK4/6 via copy number abnormalities leads to Rb phosphorylation in 46% to 68% of human PanNETs, resulting in inactivation of the Rb pathway (10, 11). Likewise, aberrant activation of MDM2, MDM4, and WIP1 suppresses the p53 signaling pathway in approximately 70% of PanNETs (12). The involvement of Rb and p53 in the tumorigenic process in PanNETs is further suggested by findings from genetically engineered mouse models. For example, the RIP1-Tag2 mouse, in which transgenic expression of the simian virus 40 (SV40) large T antigen is under control of the rat insulin promoter (RIP), develops aggressive insulinomas through suppression of both the Rb and p53 pathways (13). Similarly, preproglucagon promoter-driven expression of SV40 large T antigen results in moderate to aggressive glucagonomas (14, 15), and homozygous deletion of Rb and p53 in renin-expressing cells leads to the development of aggressive glucagonomas in the pancreas (16). These models highlight the importance of simultaneous inactivation of both the Rb and p53 pathways in the development of aggressive PanNETs in mice. Whether these aggressive tumors develop from indolent tumors, however, remains unknown. Furthermore, no studies have investigated the individual roles of Rb and p53 in PanNET formation from pancreatic cells.

In this study, we investigated the roles of Rb and p53 in the development of PanNETs utilizing a genetically engineered mouse model. We provide the first reported evidence that pancreas duodenum homeobox protein 1 (Pdx1) Cre-dependent pancreas-specific deletion of Rb gene per se induces indolent PanNETs in islet cells. Although pancreas-specific induction of a p53 mutation alone had no effect on pancreatic tissue, it markedly accelerated the progression of PanNETs in combination with Rb deletion. These data suggest that Rb and p53 have distinctive roles in the development of PanNETs.

We used Pdx1-Cre mice (17), Rosa26R mice (18), Rb flox mice (19), and LSL-Trp53^R172H mice (20, 21), which were previously described. Nonrecombinant littermates were used as controls. All mice were maintained in a specific-pathogen-free facility at the Kyoto University Faculty of Medicine (Kyoto, Japan). All experiments were approved by the institutional animal ethics committee and performed according to the guidelines of the animal ethics committee of Kyoto University.

Mouse tissues were fixed with 10% formalin for 24 hours, embedded in paraffin, and dissected in 5-μm sections, which were stained with hematoxylin and eosin for histologic analysis. For IHC analysis, additional sections were deparaffinized, rehydrated with ethanol, and treated with 0.3% hydrogen peroxide and methanol for 15 minutes to inhibit endogenous peroxidase activity. Heat-mediated antigen retrieval was performed with 10 mmol/L citrate buffer (pH 6.0), and samples were incubated in a serum-free protein block (X0909; DAKO). Sections were incubated with primary antibody overnight at 4°C and with biotinylated secondary antibody for 1 hour at room temperature. The specimens were then incubated with avidin–biotin–peroxidase complex (Vectastain ABC Kit, Vector Laboratories) at room temperature for 30 minutes, stained with diaminobenzidine substrate (DAKO), and counterstained with hematoxylin. For immunofluorescence analysis, sections were incubated with primary antibody overnight at 4°C and incubated with fluorophore-conjugated secondary antibody (Invitrogen) for 1 hour at room temperature. Primary antibodies used in this study were mouse anti-cytokeratin (1:100; clone AE1/AE3, M3515; DAKO), rabbit anti-pancreatic alpha amylase (1:300, ab21156; Abcam), rabbit anti-chromogranin A (1:1000; ab15160; Abcam), mouse anti-synaptophysin (1:200; M7351; DAKO), guinea pig anti-insulin (1:400; A0564; DAKO), rabbit anti-glucagon (1:300; A0565; DAKO), rabbit anti-somatostatin (1:500; ab22682; Abcam), goat anti-pancreatic polypeptide (PP; 1:400; ab77192; Abcam), rat anti-Ki67 (1:300; M7249; DAKO), rabbit anti-p53 (1:500; clone CM5, VP-P956; Vector Laboratories), and rabbit anti-phospho-S6 ribosomal protein (1:400; Ser235/236, #2211S; Cell Signaling Technology). Phospho-S6 ribosomal protein was scored by applying a semiquantitative immunoreactivity method (H-score) as described previously (22).

Mice were anesthetized and briefly perfused intracardially with ice-cold fixative solution (PBS containing 4% paraformaldehyde, 30% sucrose, 0.5 mmol/L EGTA, 2 mmol/L MgCl₂, and 1% glutaraldehyde). Fixed pancreas tissues were mounted and dissected in 8-μm sections. β-Galactosidase substrate [5 mmol/L K₃Fe(CN)₆, 5 mmol/L K₄Fe(CN)₆, and 1 mg/mL X-galactosidase in rinse buffer (PBS containing 2 mmol/L MgCl₂, 0.02% NP-40, and 0.1% sodium deoxycholate)] was added to the sections, which were then incubated in the dark overnight at room temperature. The sections were washed twice with rinse buffer for 5 minutes, fixed with 10% formalin for 5 minutes, and counterstained with nuclear fast red (KPL).

Islets of Langerhans were isolated from Pdx1-Cre, Pdx1-Cre;Rb^f/f, and Pdx1-Cre;Trp53^R172H;Rb^f/f mice at 4 to 8 months of age using the collagenase digestion technique (23). Briefly, Hank's Balanced Salt Solution (HBSS) containing 0.5 mg/mL collagenase P (#11213865001; Roche) was injected into the mouse pancreas via the bile duct. The pancreas was subsequently removed and further digested in HBSS containing 0.5 mg/mL collagenase P for 30 minutes at 37°C. The digested pancreas was washed twice with ice-cold Krebs–Ringer bicarbonate (KRB) buffer (129.4 mmol/L NaCl, 5.2 mmol/L KCl, 2.7 mmol/L CaCl₂, 1.3 mmol/L KH2PO4, 1.3 mmol/L MgSO₄, and 24.8 mmol/L NaHCO₃ (equilibrated with 5% CO₂/95% O₂, pH 7.4)] containing 2.8 mmol/L glucose, suspended in 4 mL of histopaque 1.119 (#11191; Sigma), and transferred to a clean glass tube; 2 mL of histopaque 1.077 (#10771; Sigma) and 2 mL of histopaque 1.050 (prepared by mixing two volumes of histopaque 1.077 with one volume of distilled water) were then sequentially overlaid to perform a density gradient separation. After 800 × g centrifugation for 10 minutes at room temperature, the islets observed in the interphase between histopaque 1.050 and histopaque 1.077 were collected and washed twice in ice-cold KRB buffer. The resulting islets were transferred to a large dish filled with approximately 50 mL of ice-cold KRB buffer and hand-picked into a 1.5-mL tube before total RNA preparation. Total RNA was prepared from the isolated islets with an RNeasy Mini Kit according to the manufacturer's instructions (#74104; Qiagen), and the RNA concentration was measured using a Qubit RNA HS Assay Kit (#Q32855; Thermo Fisher Scientific).

Single-stranded cDNA was prepared with Superscript III (Invitrogen), and qRT-PCR was performed with a LightCycler FastStart DNA Master SYBR Green 1 Kit (Roche Diagnostic). Values are expressed as arbitrary units relative to the expression of GAPDH. Primers for quantitative PCR are as follows: Rb gene, Rb-forward (5′-TACACTCTGTGCACGCCTTC) and Rb-reverse (5′-TTCACCTTGCAGATGCCATA); Pten gene, Pten-forward (5′-TGCACAGTATCCTTTTGAAGACC) and Pten-reverse (5′-GAATTGCTGCAACATGATTGTCA); and Tsc2 gene, Tsc2-forward (5′-GAGCCCCCAAACAAGGCCTGA) and Tsc2-reverse (5′-AGGCTGGCGCTCGTAAGGGAT).

The quality of RNA extracted from the isolated islets of Pdx1-Cre and Pdx1-Cre:Rb^f/f mice was examined with an Agilent 2100 Bioanalyzer (Agilent Technologies). The RNA samples were labeled with a Sure Tag Complete DNA Labeling Kit (Agilent Technologies) and hybridized to the SurePrint G3 Mouse GE 8 × 60K Microarray Kit (Agilent Technologies) with a Gene Expression Hybridization Kit (Agilent Technologies). The raw data were quantified in Agilent Feature Extraction software (Agilent Technologies). Quantified data were normalized by Gene Spring 12.5 software (Agilent Technologies). The pathway analysis of the gene expression data was performed with DAVID 6.8. Full microarray data have been uploaded to the Gene Expression Omnibus (GEO) under accession number GSE152582.

For measurement of serum insulin and blood glucose levels, the mice underwent a 4-hour fast before collection of whole blood via orbital bleeding or a tail cut. Blood glucose was measured with a glucose meter (Glutest Every; Sanwa Kagaku Kenkyusho Co.). Serum was collected through centrifugation, snap-frozen, and stored at −80°C. Serum insulin levels were measured with a Mouse Insulin ELISA Kit (Mercodia) according to the manufacturer's instructions.

Data are presented as mean ± SE, and were analyzed with two-tailed independent-sample Student t test, as appropriate. P values < 0.05 were considered statistically significant (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

To investigate the role of the Rb gene in the development of pancreatic tissue, we generated pancreas-specific Rb-deleted mice by crossing Pdx1-Cre mice with Rb flox mice (Pdx1-Cre;Rb^f/f; Fig. 1A). After confirming in Pdx1-Cre;Rosa26R mice that the Pdx1 promoter induces Cre recombinase expression in the acinar, duct, and islet cells, as reported previously (Supplementary Fig. S1A), we analyzed the Pdx1-Cre;Rb^f/f mice at 2, 4, 6, and 8 months of age. Histopathologic analysis revealed normal exocrine glandular components in both the Pdx1-Cre;Rb^f/f mice and Pdx1-Cre control mice (Supplementary Fig. S1B). IHC analysis revealed normal amylase staining and pan-cytokeratin staining in the acinar cells and duct cells, respectively, in Pdx1-Cre;Rb^f/f mice (Supplementary Fig. S1B). The number of Ki67-positive acinar cells and duct cells did not differ significantly between Pdx1-Cre;Rb^f/f mice and Pdx1-Cre control mice (Supplementary Fig. S1C and S1D). These findings suggested that pancreas-specific Rb-deletion has no effect on the development of exocrine cells.

In islet cells in Pdx1-Cre;Rb^f/f mice, there were no morphologic abnormalities, but the ratio of α- to β-cells was significantly lower than that in Pdx1-Cre control mice (Fig. 1B and C). The lower α-cell/β-cell ratio appeared to be due to an increase in the β-cell mass, but the blood glucose level, body weight, and pancreas/body weight ratio were comparable between Pdx1-Cre;Rb^f/f mice and Pdx1-Cre control mice (Fig. 1D; Supplementary Fig. S1E and S1F). The number of Ki67-positive islet cells was higher in Pdx1-Cre;Rb^f/f mice than in Pdx1-Cre control mice at 2 months of age, but equivalent at 4, 6, and 8 months of age (Fig. 1B and E). The ratios of δ-cells and PP cells in the islet cells were comparable between Pdx1-Cre;Rb^f/f mice and Pdx1-Cre control mice (Supplementary Fig. S1G–S1I). Thus, pancreas-specific Rb gene deletion does not affect exocrine cells, but exclusively decreases the α-cell/β-cell ratio in islet cells.

We also examined the effect of p53 gene mutation on pancreatic tissue by crossing Pdx1-Cre mice with LSL-Trp53^R172H mice (Pdx1-Cre;Trp53^R172H; Supplementary Fig. S2A). The Pdx1-Cre;Trp53^R172H mice exhibited no changes in the morphology (Supplementary Fig. S2B), α-cell/β-cell ratio (Supplementary Fig. S2C), number of Ki67-positive islet cells (Supplementary Fig. S2D), ratio of δ-cells and PP cells in islet cells (Supplementary Fig. S2E and S2F), blood glucose level (Supplementary Fig. S2G), body weight (Supplementary Fig. S2H), or pancreas/body weight ratio (Supplementary Fig. S2I) compared with control mice. In conclusion, p53 gene mutation does not affect pancreatic development and maintenance.

To assess the long-term effect of pancreas-specific Rb gene deletion, we compared Pdx1-Cre;Rb^f/f mice, Pdx1-Cre control mice, and Pdx1-Cre;Trp53^R172H mice at 18 to 20 months of age. No histologic abnormalities were present in the acinar and duct cells in Pdx1-Cre;Rb^f/f mice, Pdx1-Cre control mice, or Pdx1-Cre;Trp53^R172H mice. In contrast, well-circumscribed PanNETs with a trabecular architecture, dilated abnormal vessels, and apoptotic bodies formed in Pdx1-Cre;Rb^f/f mice (N = 4/5). No tumors were found in Pdx1-Cre (N = 8) or Pdx1-Cre;Trp53^R172H (N = 8) mice (Fig. 2A). IHC analysis revealed that these tumors were positive for neuroendocrine markers such as chromogranin A and synaptophysin, confirming that the tumors were well-differentiated PanNETs (Fig. 2A). Of seven tumors, 57% (4/7) were positive for PP, 29% (2/7) were positive for glucagon, and 14% (1/7) was positive for somatostatin (Fig. 2A and B). Because of the slow progression and lack of distant metastasis, tumors that formed in the Pdx1-Cre;Rb^f/f mice were thought to be low grade. These results suggested that pancreas-specific Rb deletion, but not p53 mutation is sufficient to induce low-grade PanNET initiation. Given that the α-cell/β-cell ratio decreased in islet cells before the development of low-grade PanNETs in Pdx1-Cre;Rb^f/f mice (Fig. 1B and C), we presumed that the alteration of the α-cell/β-cell ratio in islet cells was a dysplastic state. The PanNETs in Pdx1-Cre;Rb^f/f mice were highly penetrant at 18 to 20 months of age (Supplementary Table S1).

To identify major pathways associated with the development of low-grade PanNETs in the presence of Rb gene deletion, we performed microarray analysis on pancreatic islets isolated from Pdx1-Cre mice and Pdx1-Cre;Rb^f/f mice. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that the p53 signaling pathway as well as the cell cycle and DNA replication were among the top five upregulated pathways in pancreatic islets of Pdx1-Cre;Rb^f/f mice compared with Pdx1-Cre control mice (Fig. 3A). We then focused on the p53 signaling pathway and performed immunohistochemical analysis of pancreatic tissues from Pdx1-Cre;Rb^f/f mice and Pdx1-Cre mice at 18 to 20 months of age. Expression of p53 was detected in both the islet cells and PanNETs of Pdx1-Cre;Rb^f/f mice, but not in the islet cells of Pdx1-Cre control mice (Fig. 3B). These results indicated that pancreas-specific Rb gene deletion induces p53 activation in islet cells, suggesting the contribution of tumor suppressor p53 to the indolent phenotype of PanNETs formed in Pdx1-Cre;Rb^f/f mice.

To examine the role of p53 in pancreas-specific Rb-deleted mice, we induced p53 mutation alleles in the pancreas of Pdx1-Cre;Rb^f/f mice by crossing them with LSL-Trp53^R172H mice (Pdx1-Cre;Trp53^R172H;Rb^f/f). As observed in Pdx1-Cre;Rb^f/f mice, the α-cell/β-cell ratio in islet cells was also significantly lower in Pdx1-Cre;Trp53^R172H;Rb^f/f mice than Pdx1-Cre control mice at 2 and 4 months of age (Supplementary Fig. S3A and S3B), whereas the exocrine pancreas showed no morphologic changes. In contrast to Pdx1-Cre;Rb^f/f mice, in Pdx1-Cre;Trp53^R172H; Rb^f/f mice, the number of Ki67-positive islet cells was higher than that in Pdx1-Cre control mice at both 2 and 4 months of age (Supplementary Fig. S3A and S3C). Moreover, Pdx1-Cre;Trp53^R172H;Rb^f/f mice developed slightly enlarged islets with cytological atypia, so-called microadenomas in human, at 2 months (Supplementary Fig. S3D), well-differentiated PanNETs at 4 months, and full penetrance over 6 months (Fig. 4A; Supplementary Table S1). Notably, some Pdx1-Cre;Trp53^R172H;Rb^f/f mice had PanNETs with invasive lesions into surrounding tissues (Supplementary Fig. S3E and S3F) and had multiple metastatic lesions in the liver (13.3%, 2/15) at 9 months of age (Fig. 4B). Immunohistochemically, these tumors were positive for neuroendocrine markers such as chromogranin A and synaptophysin, confirming the diagnosis of well-differentiated PanNETs as well as tumors in Pdx1-Cre;Rb^f/f mice (Fig. 4A). Regarding hormone production, insulin and glucagon were positive in 76% (26/34) and 24% (8/34; Fig. 4C) of tumors, respectively, but there were no PP-positive or somatostatin-positive tumors. Consistent with the high prevalence of insulinomas, the Pdx1-Cre;Trp53^R172H;Rb^f/f mice, compared with control mice, had significantly higher plasma insulin levels and lower glucose levels after the age of 6 months (Fig. 4D and E). Likely a result of the hypoglycemia, Pdx1-Cre;Trp53^R172H;Rb^f/f mice had lower body weight (Supplementary Fig. S3G) and a shorter life-span (Fig. 5A) than control mice, whereas the pancreas/body weight ratio did not differ between them (Supplementary Fig. S3H). Although the PanNETs that developed in Pdx1-Cre;Trp53^R172H;Rb^f/f mice were histologically well-differentiated, they were thought to be aggressive tumors due to the rapid tumorigenesis (Supplementary Table S1), invasiveness to surroundings and formation of liver metastases. Indeed, the PanNETs in Pdx1-Cre;Trp53^R172H;Rb^f/f mice were relatively larger (Fig. 5B and C), and the number of Ki67-positive cells was significantly greater (2.7% vs. 24.7%, Fig. 5D and E) than those in Pdx1-Cre;Rb^f/f mice. In summary, p53 mutation alone does not affect pancreatic tissue, but in the presence of Rb gene deletion, it accelerates tumorigenesis and promotes progression from indolent to aggressive PanNETs.

The mTOR pathway is a central pathway in the development of PanNETs in both human and mouse models, including RIP1-Tag2 mice (24). To assess the activation of the mTOR pathway in our multistep PanNET progression mouse model, we performed IHC analysis of the phosphorylation of S6 ribosomal protein (pS6), which is downstream of mTORC1 (Fig. 6A). We observed a stepwise increase in pS6 from Pdx1-Cre control islets to Pdx1-Cre;Rb^f/f PanNETs and Pdx1-Cre;Trp53^R172H;Rb^f/f PanNETs, suggesting the strong involvement of the mTOR pathway in our mouse model (Fig. 6B). Because p53 suppresses the mTOR pathway through its negative regulators (25) (Fig. 6A), we next examined the mRNA expression of phosphatase and tensin homolog (Pten) and tuberous sclerosis 2 (Tsc2) by qRT-PCR in normal islet cells isolated from Pdx1-Cre, Pdx1-Cre;Rb^f/f, and Pdx1-Cre;Trp53^R172H;Rb^f/f mice at the age before forming PanNETs. Expression of Rb mRNA was markedly lower in Pdx1-Cre;Rb^f/f and Pdx1-Cre;Trp53^R172H;Rb^f/f than in Pdx1-Cre mice, confirming that these isolated islet cells reflect each genetic background (Fig. 6C). Under these conditions, we found that the expression of Pten mRNA and Tsc2 mRNA in islet cells of Pdx1-Cre;Trp53^R172H;Rb^f/f was significantly lower than that in Pdx1-Cre or Pdx1-Cre;Rb^f/f mice (Fig. 6C). These data suggested that p53 mutation might promote the progression of indolent to aggressive PanNETs, at least in part through mTOR pathway activation via the inhibition of negative regulators.

PanNETs are biologically distinct from PanNECs according to the WHO 2017 classification (2–5). Detailed information on precancerous lesions of PanNETs or the mechanisms of their progression, however, has been lacking. In this study, by manipulating the Rb and p53 genes, we established a multistep progression model from dysplastic islets to indolent PanNETs and aggressive metastatic PanNETs in mice.

A previous report demonstrated that the β-cell mass is increased and the α-cell/β-cell ratio is decreased in islet cells in Pdx1-Cre;Rb^f/f mice, leading to the acquisition of resistance to diabetes (26). In that report, the authors reported that pancreas-specific Rb ablation affected α-cell differentiation and converted the α-cell/β-cell ratio during their observation period of 5.5 months (26). We observed almost the same phenotype in our Pdx1-Cre;Rb^f/f mice (Fig. 1); however, in our long-term observation of 20 months, we observed the development of histologically confirmed indolent PanNETs with a Ki67 index of 2.7% in Pdx1-Cre;Rb^f/f mice, at high penetrance (Fig. 2). To the best of our knowledge, this is the first report demonstrating the formation of PanNETs by Rb gene deletion alone. On the basis of our sequential and long-term observations, the early phenotype of a decreased α-cell/β-cell ratio after Rb gene deletion might be part of the dysplastic state before the formation of microadenomas or PanNETs. Interestingly, although the genetic manipulation was present in all pancreatic epithelial cells, the Rb gene deletion exclusively affected islet cells and not exocrine cells. This observation contrasts with the previous finding that pancreas-specific induction of a Kras gene mutation affects only exocrine cells and not islet cells (17). Pancreatic epithelial cells may be susceptible to oncogenes in a cell context-dependent manner. Thus, we concluded that pancreas-specific Rb gene deletion is sufficient for the development of PanNETs in mice. This concept was further supported by reports that Rb is in the same molecular pathway of cell-cycle regulation and tumor suppression as Men1 (27, 28), which is involved in PanNETs formation in both humans and mice.

In contrast to the Rb gene deletion, the induction of a pancreas-specific p53 mutation alone did not affect pancreatic cells, even after long-term observation. This is compatible with the previous reports of PHLDA3, a target gene of p53 and a suppressor of Akt-mTOR pathway; PHLDA3-deficient mice develop hyperplastic islets but do not develop PanNETs (29). However, wild-type p53 activation was observed at the mRNA and protein levels in islet cells of Pdx1-Cre;Rb^f/f mice, suggesting that activated p53 suppresses tumor progression in response to Rb gene deletion (Fig. 3). This hypothesis was confirmed by the observation that Pdx1-Cre;Trp53^R172H;Rb^f/f mice developed well-differentiated, but aggressive PanNETs with a high Ki67 index of 24.7% within only 4 months (Figs. 4 and 5). Furthermore, these mice had liver metastases in 13.3%, although the incidence was lower than in previous reports of RIP1-Tag2 mice (30). In these mice, we reasoned that PanNETs initiated by Rb gene deletion were released from tumor suppression through the induction of a p53 mutation and then progressed to aggressive tumors. This observation corresponds to the finding that PanIN initiated by a Kras mutation progresses to invasive pancreatic ductal adenocarcinoma through the induction of a p53 mutation in Pdx1-Cre;Trp53^R172H;Kras^G12D (so-called KPC) mice. Thus, p53 itself does not affect tumor initiation in pancreatic tissue but appears to be involved in the progression of endocrine tumors and exocrine tumors initiated by the Rb and Kras genes, respectively.

In human PanNETs, genes in the mTOR pathway are frequently mutated, and thus inhibitors of mTOR significantly improve survival. Loss of heterozygosity for the aforementioned PHLDA3 has been reported in 72% of human PanNET (29). Moreover, the expression of mTOR pathway components in various human NETs is predictive of the prognosis. For example, the activation of downstream targets of mTOR pathway such as p-RPS6KB1 or p-RPS6 is associated with a poor prognosis (31). Decreased expression of TSC2 or PTEN, which are negative regulators of the mTOR pathway, is significantly associated with shorter disease-free survival and overall survival (32). Notably, in our new mouse model, the mRNA expression of Pten and Tsc2 in the islet cells of Pdx1-Cre;Trp53^R172H;Rb^f/f mice was significantly decreased, and the expression of pS6, a downstream component of the mTOR pathway, was upregulated in a stepwise manner in Pdx1-Cre;Rb^f/f mice and Pdx1-Cre;Trp53^R172H;Rb^f/f mice compared with control mice. From these observations, we speculate that activation of the mTOR pathway through the regulation of TSC2 and PTEN by p53 plays a major role in switching the progression of indolent to aggressive PanNETs.

Genetic alterations in RB and TP53 have been demonstrated to be uncommon in large-scale genome analysis of human PanNETs (6, 7), rather RB loss and KRAS mutation have been shown in 54.5% and 48.7% of human PanNECs, respectively (5). Therefore, it is worth discussing why Rb gene deletion does not reflect human disease and forms PanNET in our mouse model. There may be several reasons for this. Given the high prevalence of KRAS mutations in human PanNECs, Kras mutations may be required in addition to Rb deletion to form PanNECs in mice. Because of cells of origin can be intrinsically different between PanNEC and PanNET (33, 34), mouse PanNEC may only be reproduced when the Kras and Rb abnormalities are introduced in the proper cells in the proper order. On the other hand, there are possible causes for the formation of PanNETs in mice by Rb abnormalities. For example, there have been reports of copy-number alteration (35), mutation (8), and upstream abnormalities of Rb pathway including CDK4/6 (10, 11), p27 (36, 37), RABL6A (38, 39) in human PanNETs. Thus, abnormalities of Rb pathway other than Rb gene alteration may be involved in PanNETs. Recently, the CDK4/6 inhibitors, palbociclib, commonly used in breast cancer, are trialed in PanNETs (40). Combination therapies including CDK4/6 inhibitors that is based on preclinical studies in other cancers (41) may be an effective treatment option in the future. In this point of view, further development of clinical trials evaluating Rb pathway including CDK4/6 inhibitors is eagerly awaited. As for TP53, its genetic alterations are not common in human PanNETs but still observed in a few cases (6, 7, 42). Quite interestingly, it has been suggested that TP53 mutations are more likely found in G3 PanNETs than in G1/2 PanNETs (7, 42) and may be associated with metastasis (43). Together with the results of our mouse model, these previous reports may suggest that p53 mutations are involved in tumor progression of PanNETs as a late event. Thus, the mouse model we have newly established has several genetic aspects that do not correspond to those of humans. However, this model recapitulates the stepwise progression to G3 PanNET, which could not have been sufficiently studied so far, and may provide new insights into the pathogenesis of PanNETs. This model is expected to open the way to new diagnostic and therapeutic approach for PanNETs and to be a useful tool for future preclinical studies.

In summary, our novel mouse model demonstrated that inactivation of Rb in the pancreas affects islet cells exclusively and is sufficient for the development of dysplastic islet cells and the subsequent formation of indolent PanNETs. In contrast, p53 mutation itself has no effect on PanNET initiation, but plays a crucial role in the progression from indolent to aggressive PanNETs through activation of the mTOR pathway in mice. On the basis of these findings, we conclude that Rb and p53 have distinct roles in the initiation and progression, respectively, of PanNETs.

No potential conflicts of interest were disclosed.

Y. Yamauchi: Conceptualization, resources, data curation, software, formal analysis, validation, investigation, visualization, methodology and writing-original draft. Y. Kodama: Conceptualization, funding acquisition, writing-original draft, project administration, writing-review and editing. M. Shiokawa: Data curation. N. Kakiuchi: Data curation. S. Marui: Conceptualization. T. Kuwada: Conceptualization. Y. Sogabe: Conceptualization. T. Tomono: Conceptualization. A. Mima: Conceptualization. T. Morita: Conceptualization. T. Matsumori: Conceptualization. T. Ueda: Conceptualization. M. Tsuda: Conceptualization. Y. Nishikawa: Conceptualization. K. Kuriyama: Conceptualization. Y. Sakuma: Conceptualization. Y. Ota: Conceptualization. T. Maruno: Conceptualization. N. Uza: Conceptualization. A. Masuda: Resources. H. Tatsuoka: Investigation. D. Yabe: Conceptualization. S. Minamiguchi: Investigation. T. Masui: Resources. N. Inagaki: Conceptualization. S. Uemoto: Resources. T. Chiba: Supervision and funding acquisition. H. Seno: Supervision and funding acquisition.

We wish to thank Yuta Kawamata for excellent technical support. This work was supported by the Japan Society for the Promotion of Science (JSPS) and the Ministry of Education, Culture, Sports, Science and Technology (MEXT) KAKENHI Grant Nos. JP15J05143, JP17H06803, and JP16K09395; AMED Project for Development for Innovative Research on Cancer Therapeutics (P-DIRECT).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.