Reduction of Ptf1a Gene Dosage Causes Pancreatic Hypoplasia and Diabetes in Mice

Ptf1a^cre/+ (8), Ptf1a^cbll/+ (14), and ROSA26r mice (15) were crossed in the C57BL/6 background and interbred to produce Ptf1a^cre/+;ROSA26r, Ptf1a^cre/cbll; ROSA26r, Ptf1a^cre/cre;ROSA26r, and Ptf1a^cbll/cbll mice. All animal experiments were performed in accordance with the guidelines for animal experiments of Kyoto University.

Pancreatic rudiments were dissected and submerged in RNA Later (Qiagen, Tokyo, Japan). Total RNA was extracted using an RNeasy Mini Kit (Qiagen). First-strand cDNA synthesis was performed using Superscript III first-strand synthesis systems (Invitrogen). Quantification of Ptf1a and Pdx1 gene expression was performed by quantitative real-time RT-PCR using the 7300 Real-Time PCR System (Applied Biosystems). The Ptf1a and Pdx1 probes were TaqMan Gene Expression Assay nos. Mm00479622_m1 and Mm00435565_m1 (Applied Biosystems), respectively. Data were normalized in relation to the expression of GAPDH.

Tissue processing, X-gal staining, and permeabilization were performed as described previously (8,9). The following primary antibodies were used: guinea pig anti-insulin (DAKO), rabbit anti-amylase (Sigma-Aldrich), rabbit anti-glucagon (DAKO), rabbit anti-somatostatin (DAKO), rabit anti-pancreatic polypeptide (DAKO), goat anti-Pdx1, goat anti-glut2 (Santa Cruz Biotechnology), mouse anti-cytokeratin (DAKO), and rabbit anti-phosphorylated histone H3 (Upstate). The immunofluorescent secondary antibodies were Cy3-conjugated antibody to guinea pig IgG (Cy3-conjugated antibody to rabbit IgG; Chemicon), Alexa Fluor 488-conjugated antibody to rabbit IgG (Molecular Probes), Alexa Fluor 488-conjugated antibody to mouse IgG (Molecular Probes), and Alexa Fluor 488-conjugated antibody to goat IgG (Molecular Probes). Transferase-mediated dUTP nick-end labeling (TUNEL) assays were performed using the DeadEnd Fluorometric TUNEL system (Promega). Images were visualized using a Zeiss Axioskop 2 microscope. Quantification of total β-cell number was performed by serially sectioning the entire pancreas followed by staining and counting every 10th section for embryonic day (e)14.5 embryos or every 20th section for e18.5 embryos (n = 3) in each genotype. The ratios of endocrine cell types were calculated from eight randomly chosen sections costained with hormones per animal (n = 3) in each genotype. Homogenized stool samples were centrifuged, and the supernatant was stained with Oil Red O solution (Sigma).

Glucose tolerance tests were performed by injecting d-glucose (2 mg/kg i.p.) after overnight fasting. For measurements of serum insulin, d-glucose (2 mg/g body wt i.p.) was injected and 15 min later blood was collected into tubes containing aprotinin (100 KIE/ml blood; Wako) and centrifuged to obtain serum. Plasma insulin levels were measured using the Mouse Insulin ELISA kit (u-type) (Shibayagi, Gunma, Japan).

Data are presented as mean ± SE. Data were subjected to unpaired t test or one-way nonrepeated measurements ANOVA with Bonferonni tests. P < 0.05 was considered significant.

A hypomorphic allele of Ptf1a (Ptf1a^cbll) has a 313-kb genomic deletion located 60 kb downstream of the Ptf1a locus, but the Ptf1a locus itself is intact (14). The Ptf1a^cre allele is null through the replacement of the Ptf1a protein-coding region with that of a nuclearly targeted Cre recombinase (8). We used quantitative real-time RT-PCR to evaluate the expression level of Ptf1a in pancreata from Ptf1a^cre/+, homozygous Ptf1a^cbll/cbll, and compound trans-heterozygotes of Ptf1a^cre and Ptf1a^cbll (Ptf1a^cre/cbll) compared with that in wild-type mice at two stages: e12.5, when Ptf1a is expressed in the multipotent progenitor phase of pancreas development (12), and postnatal day 1 (P1), when Ptf1a is expressed in exocrine cells in pancreas. Ptf1a RNA levels in pancreatic rudiments from Ptf1a^cre/+, Ptf1a^cbll/cbll, and Ptf1a^cre/cbll mice were substantially reduced to 55 ± 8, 32 ± 10, and 14 ± 4%, respectively, compared with wild-type mice at e12.5, and they were reduced to 60 ± 7, 48 ± 4, and 30 ± 6%, respectively, at P1 (Fig. 1A).

Hypomorphic Ptf1a^cre/cbll and Ptf1a^cbll/cbll mice were born at the expected Mendelian ratio. Ptf1a^cre/cbll mice appeared small and dehydrated at birth. Some Ptf1a^cre/cbll mice survived to P30 (the longest survival up to P45), but the rest died before ∼P14. On the other hand, Ptf1a^cre/+, Ptf1a^cbll/cbll, and wild-type mice were almost indistinguishable from each other at birth and were viable. As shown in Fig. 1B–D, postnatal growth retardation and small body size persisted into the weaning stage in Ptf1a^cre/cbll and Ptf1a^cbll/cbll mice. We crossed Ptf1a^cre/+ and Ptf1a^cre/cbll with Gt(ROSA)26Sor^tm1Sor (R26r) mice, which carry a floxed lacZ gene driven by cell type–independent ROSA promoter (15), enabling us to trace the fate of all Ptf1a-expressing cells and their progeny (8). Macroscopically, the pancreas of hypomorphic Ptf1a^cre/cbll;R26r mice was dramatically smaller than that of Ptf1a^cre/+;R26r mice at birth (data not shown), at P7, and at P30, and Ptf1a^cbll/cbll mice exhibited moderate pancreatic hypoplasia (Fig. 1E–J).

Pancreas development can be divided into four phases in mice: The first phase is the specification and formation of dorsal and ventral pancreatic buds from foregut endoderm (specification). During the second phase, extensive proliferation and branching of the pancreatic epithelium occur (branching morphogenesis). The third phase involves differentiation of large numbers of endocrine, exocrine, and duct cell lineages (cytodifferentiation). The fourth phase consists of maturation of the islet architecture and function in the postnatal period (islet morphogenesis). Developmentally, as shown in Fig. 2A–H, pancreas was small from the beginning of pancreatic bud formation at e10.5 in Ptf1a^cre/cbll;R26r mice compared with Ptf1a^cre/+;R26r littermates, and this size difference became more apparent as development proceeded.

In our previous study, lineage-labeled β-galactosidase (β-gal)-positive cells were observed in the duodenum in both the dorsal and ventral regions in Ptf1a-null (Ptf1a^cre/cre; R26r) mice (8). A subset of β-gal–positive cells were found to protrude from the gut tube in the ventral region at e12.5, but the cell fate of these cells was not identified (8). In hypomorphic Ptf1a^cre/cbll;R26r mice, at the dorsal side, ectopic Ptf1a lineage-labeled areas were observed in the duodenum at P7, similar to Ptf1a^cre/cre;R26r mice at e17.5 (Fig. 1G and Fig. 2J). Notably, at the ventral side, not only the duodenum but also the CBD contained ectopic Ptf1a lineage-labeled cells in Ptf1a^cre/cbll;R26r at e17.5 and P7 (Fig. 2H and L) and in Ptf1a^cre/cre;R26r mice at e17.5 (Fig. 2K). Histologically, substantial proportions of Ptf1a lineage-labeled cells were misspecified and adopted a Cdx2 (16)-positive duodenal epithelial cell fate or cytokeratin-positive CBD epithelial cell fates including cells of the peribiliary glands (17) in Ptf1a^cre/cbll;R26r mice (Fig. 2M–Q). These results indicate that a threshold level of Ptf1a expression is required for the acquisition of a pancreatic cell fate in the primitive foregut endoderm and that reduction of Ptf1a dosage results in misspecification of pancreatic progenitors into the CBD and duodenum. Thus, together with reduced pancreatic bud size at early developmental stages, reduction of Ptf1a dosage results in a decrease in the final cell population that adopts a pancreatic cell fate.

Next, we investigated whether reduction of Ptf1a dosage affects the growth of pancreatic precursor cells. Quantification revealed that significantly more pancreatic precursor cells are proliferating in Ptf1a^cre/+ mice (12.0 ± 2.1% of the total Ptf1a lineage-labeled pancreatic epithelial cells) than in Ptf1a^cre/cbll mice (5.9 ± 1.9%) (n = 4, five to seven sections analyzed per embryo) at e10.5, as detected by the mitosis maker phosphorylated histone H3 (Fig. 3D). At e12.5, although the difference was not significant (P = 0.071), more pancreatic precursor cells were proliferating in Ptf1a^cre/+ mice (6.3 ± 1.2%) than in Ptf1a^cre/cbll mice (3.8 ± 1.2%) and Ptf1a^cre/cre mice (2.3 ± 0.6%) (Fig. 3A–C and E). On the other hand, TUNEL staining revealed that apoptotic cells are only rarely observed in both Ptf1a^cre/+ and hypomorphic mutant mice at e10.5 and e12.5 (data not shown), suggesting that apoptosis does not contribute to the hypomorphic pancreas of the mutants.

As reported previously, Ptf1a-null mice completely lack exocrine cytodifferentiation (5,8). In Ptf1a^cre/+;R26r mice, extensive branching morphogenesis of the pancreatic epithelial tree was observed as early as e12.5, and amylase-positive cells were detected at e14.5 (Fig. 4A.C). In Ptf1a^cre/cbll;R26r mice, almost no branching morphogenesis of the pancreatic epithelial tree was seen, even at e14.5 (Fig. 4B and E). Exocrine cytodifferentiation was also markedly delayed and reduced (Fig. 4H, K, N, and Q–T). Amylase-positive cells began to appear at P1 in small areas in Ptf1a^cre/cbll mice, but they were morphologically immature in that the nucleus-to-cytoplasm ratio in Ptf1a^cre/cbll mice at P1 was as large as that in Ptf1a^cre/+ mice at e14.5 (Fig. 4C and K). At P7, pancreatic acinar structures were observed in restricted ventral regions, but the majority of the cells resembled immature duct-like epithelial structures that phenotypically corresponded to those at ∼e14.5 in Ptf1a^cre/+ mice (Fig. 4N and Q). In the adult stage, amylase-positive acinar structures were detected in the ventral pancreas and proximal dorsal pancreas, but the majority of dorsal pancreas was composed of immature duct-like epithelial cells that rarely expressed amylase (Fig. 4R–T). Persistent Pdx1 expression was observed in the immature duct-like epithelial cells at P1, P7, and P30, while normal duct cells do not express Pdx1 at these stages (Fig. 8E–G and I and data not shown). Thus, the reduced dosage of Ptf1a falls below the threshold required to elicit the proper pancreatic exocrine differentiation program, and many pancreatic duct-like epithelial cells appear to remain in an undifferentiated state until the adult stage. These results also indicate regional (ventral vs. dorsal and distal vs. proximal) differences in the exocrine differentiation program of the dorsal pancreas related to the Ptf1a dosage. Ptf1a^cbll/cbll mice showed intermediate delay in branching morphogenesis and exocrine cytodifferentiation during embryogenesis, but exocrine cytodifferentiation was fully restored by adulthood (Fig. 4D, G, J, M, and P). Actually, the body weights of Ptf1a^cbll/cbll mice were almost similar to those of wild-type mice by ∼3 months of age. As expected from the histological findings, Ptf1a^cre/cbll mice showed exocrine pancreas dysfunction at P22, demonstrated by the abundant Oil red O–stained lipid droplets in stool smears (Fig. 4U). A small amount of lipid droplets was also observed in stool smears from Ptf1a^cbll/cbll mice, but it was very rare. Thus, we hypothesize that exocrine pancreas dysfunction is, at least in part, one of the causes of growth retardation as well as impaired metabolic homeostasis in Ptf1a^cre/cbll mice (see below).

During embryogenesis, the timing of pancreatic endocrine cytodifferentiation was indistinguishable between Ptf1a^cre/+, Ptf1a^cbll/cbll, and Ptf1a^cre/cbll mice (Fig. 5A–L). However, analyses of endocrine cell types at e18.5 revealed that the ratio of α- to β-cells is significantly higher in Ptf1a^cre/cbll and Ptf1a^cbll/cbll mice and the ratio of δ- to β-cells is significantly higher in Ptf1a^cre/cbll mice compared with Ptf1a^cre/+ mice (n = 3) (Fig. 5M). Moreover, total number of β-cells was significantly lower in Ptf1a^cre/cbll and Ptf1a^cbll/cbll mice compared with Ptf1a^cre/+ mice at both e14.5 and e18.5 in a Ptf1a dosage–dependent manner (n = 3) (Fig. 5N and O). At postnatal stages (at P1 and at P30), islet architecture was disorganized in both Ptf1a^cre/cbll and Ptf1a^cbll/cbll mice (Fig. 5D, H, L, and P-R). The proportion of glucagon-producing cells was increased, and many of them were abnormally intermingled with islet β-cells (Fig. 5P–R). Pancreatic polypeptide–and somatostatin-expressing cells were normally localized in the periphery of islet but increased in number in both Ptf1a^cbll/cbll and Ptf1a^cre/cbll mice (Fig. 5P–R). Nuclear Pdx1 and cell-membranous Glut2 expressions in individual β-cells were similar to those in wild-type mice at P30 (Fig. 5S–X), suggesting their maturation.

To assess pancreas endocrine function, mice (23- to 30-day-old: wild-type, n = 4; Ptf1a^cre/+, n = 16; Ptf1a^cbll/cbll, n = 33; and Ptf1a^cre/cbll, n = 6) were subjected to intraperitoneal glucose tolerance tests. Strikingly, Ptf1a^cre/cbll mice showed dramatically elevated blood glucose levels 30 min after injection and remained hyperglycemic during the entire testing period compared with wild-type and Ptf1a^cre/+ mice (Fig. 6A). Ptf1a^cbll/cbll mice exhibited mild glucose intolerance (Fig. 6A). Serum insulin levels at 15 min after intraperitoneal glucose injection were inappropriately low in Ptf1a^cre/cbll (39 ± 0.6 pg/ml, n = 5) and Ptf1a^cbll/cbll mice (142 ± 46 pg/ml, n = 9) compared with wild-type mice (453 ± 107 pg/ml, n = 5) (Fig. 6B).

Pdx1 is essential for pancreas organogenesis and mature pancreatic β-cell function in mice and humans (18–20). A recent study has shown that Ptf1a activates Pdx1 expression in vitro and that endogenous Ptf1a binds the regulatory sequences of Pdx1 in vivo (21,22). Consistent with these findings, Pdx1 expression was substantially reduced in pancreatic precursor cells in Ptf1a^cre/cbll mice at e10.5, e11.5, and e12.5 and in Ptf1a^cre/cre mice at e12.5 compared with Ptf1a^cre/+ or wild-type mice (Fig. 7A–G, cf. signal intensity in the distal stomach vs. dorsal pancreas). Pdx1 RNA levels in pancreatic rudiments from Ptf1a^cbll/cbll and Ptf1a^cre/cbll mice at e12.5 were significantly decreased to 63 ± 2 and 17 ± 5%, respectively, compared with wild-type mice (Fig. 7H). Interestingly, Pdx1 expression and Pdx1 RNA levels in Ptf1a hypomorphic and null mutants were elevated and reached the same level as those in Ptf1a^cre/+ and wild-type mice at e14.5 and 17.5 (Fig. 8A–D, H, J, and K and data not shown). Considering that Ptf1a dosage is still low at P1 in each mutant, other positive regulators of Pdx1 would be responsible for Pdx1 reactivation at the mid-to-late stages of pancreatogenesis.

In the present study, combining hypomorphic and null alleles of Ptf1a with Cre-loxP–based lineage tracing, we have revealed dosage-dependent, crucial roles of Ptf1a in pancreatic development and function, including 1) determination of pancreas size throughout embryogenesis until the weaning stage, 2) cell fate determination of the pancreas versus CBD or duodenum in the developing foregut endoderm, 3) growth with branching morphogenesis of the epithelial tree and exocrine cytodifferentiation, 4) total β-cell number and balance of endocrine cell types, and 5) islet morphogenesis and endocrine function in the postnatal period. These findings are implicated in possible pathogenesis of pancreatic hypoplasia and neonatal diabetes in humans.

We have observed dosage-dependent pancreatic hypoplasia in Ptf1a hypomorphic mutant mice. Accounting for hypoplastic pancreas, we have revealed that reduction of Ptf1a dosage results in a decrease of the final cell population that adopts a pancreatic cell fate. Furthermore, we have shown that subsequent growth of early pancreatic precursors is also reduced in Ptf1a hypomorphic mutants. Recently, it was reported that Pdx1⁺ Ptf1a⁺ cMyc^high Cpa1⁺ multipotent progenitor cells are located in the tip of the branching pancreatic epithelial tree, while the trunk region of branches is composed of endocrine and duct cell precursors (12). In Ptf1a hypomorphic mutants, it is conceivable that reduced branching morphogenesis might result in the decrease in the number of “multipotent tip cells.” In addition, consistent with the recent studies (21,22), we have demonstrated that Ptf1a dosage is crucial for the activation of Pdx1 in early pancreatic progenitors in vivo. Considering that Pdx1 inactivation results in early growth arrest (19) and that Pdx1 positively regulates Ptf1a in the early pancreatic buds (4), it is concluded that Pdx1 and Ptf1a mutually activate each other in early pancreatic precursors, establishing the pancreatic domains in the primitive endoderm and promoting subsequent growth of pancreatic precursors. It is noteworthy that Ptf1a dosage is one of the “intrinsic” determinants of pancreatic size, which is relevant to a recent study suggesting that the pancreas size is determined by an intrinsic factor that is not amendable to growth compensation (23). It remains possible that substantial pancreatoneogenesis is still occurring at P1, although there is currently no evidence for multipotent progenitor cells that express Ptf1a at P1 and afterward.

It is interesting that the cells expressing reduced Ptf1a adopted different cell fates, including the CBD, duodenum, and pancreas in Ptf1a^cre/cbll mice. Considering that the expression level of Ptf1a per cell appears to not be same in pancreatic buds at the early developmental stage (12,24), it is likely that the cells that express Ptf1a above a certain threshold level are specified to pancreas and those cells expressing Ptf1a below this level are specified to the CBD or duodenum. It is possible that the cells most likely to be misspecified are those at the duodenal junction, since they could potentially still be receiving some kind of gut fate-inducing signals that are stronger than in the more distal cells. On the other hand, it is still an open question as to how pancreatic endocrine versus exocrine cell fate specification is determined. In the recent study, one striking observation is that Cpa1 becomes downregulated in the cleft region when the branching tips divide and Cpa1-expressing cells become incorporated into the trunk region (12). In this study, we have found that pancreatic exocrine cytodifferentiation is markedly delayed and reduced in Ptf1a hypomorphic mutants, whereas endocrine cytodifferentiation is less affected. Together with the report from Zhou et al. (12), our findings lead to an intriguing hypothesis that Ptf1a dosage might be a determinant of cell fate specification of pancreatic exocrine versus endocrine lineage: The pancreatic progenitor cells that express Ptf1a at a higher level are specified to exocrine lineage, whereas those that express Ptf1a at lower level are specified to endocrine lineage. It will be of considerable interest to test this hypothesis by chimeric mouse experiments using Ptf1a hypomorphic alleles.

The present study shows that Ptf1a dosage affects total β-cell number and the balance of endocrine cell types as well as their relative spatial relationships in late-stage endocrine cell formation, although Ptf1a is not essential for endocrine cytodifferentiation. It has been suggested that the surrounding environment might be important for proper islet formation (5). In Ptf1a hypomorphic mutants, appropriate acinar tissues are not formed during embryognenesis and the neonatal period; therefore, it is possible that lack of production of the appropriate acinar tissue affects islet formation. Alternatively, it is possible that downregulation of Pdx1 in early pancreatic multipotent progenitors affects a proportion of endocrine cell types in the mutants. Reducing Pdx1 expression by deleting promoter elements of Pdx1 gene results in disrupted proportion of each islet cell type and its distribution (4). Other mouse models of disorganized islet architecture, including transgenic overexpression of hepatocyte nuclear factor (HNF)-6 (25) and dominant-negative HNF-1α (26), also result in glucose intolerance. In all of these models, Glut2 is absent or largely reduced; therefore, β-cell dysfunction, including glucose sensing, is a main cause of diabetes. In Ptf1a hypomorphic mice, however, insulin-producing cells express normal amounts of Pdx1 and Glut2, suggesting their functional maturation. We consider the main cause of glucose intolerance to be the inadequate number of total β-cells. From the observations during embryonic stages, two mechanisms might be responsible: First, as a consequence of reduced Ptf1a expression, substantial proportion of foregut epithelial cells are misspecified to the duodenum and CBD, resulting in the reduction of the final cell population that adopts a pancreatic cell-fate. Second, reduced proliferation during early pancreatogenesis causes inadequate expansion of endocrine precursor pools in Ptf1a hypomorphic mutant mice. Supporting this notion, preliminary analyses showed that Ngn3 expression in pancreatic rudiments are reduced in Ptf1a hypomorphic mutants (50 ± 13% in Ptf1a^cre/cbll mice and 50 ± 6% in Ptf1a^cbll/cbll mice) compared with wild-type mice at e12.5 (n = 3), although the expression level of Ngn3 per cell is unclear. Reduced Pdx1 expression might affect Ngn3-expressing cell formation during early pancreatigenesis in Ptf1a hypomorphic mutants because loss of Pdx1 results in no Ngn3-expressing cell formation in mice (27). At the same time, islet dysmorphology itself (disturbed ratio of each cell type and its localization) might be associated with decreased function. Mixing of peripheral cell types with β-cells might disrupt the gap junctional or other coupling that is involved in an efficient insulin release (28).

We have already reported that hypomorphic Ptf1a^cre/cbll and Ptf1a^cbll/cbll mice exhibit cerebellar agenesis similar to that of Ptf1a-null mice (14). The levels of Ptf1a transcripts in Ptf1a^cbll/cbll embryos are dramatically reduced in the cerebellum (26.3-fold reduction), whereas they are more mildly affected in the hindbrain (4.5-fold reduction) and pancreas (3.1-fold reduction) (Fig. 1 and ref. 29). The precise mechanism that accounts for this difference is currently unknown; however, our previous report and current findings suggest that threshold levels to acquire specific cell fates differ among different organs. A nonsense mutation in Ptf1a has been recently identified as a cause of autosomal recessively inherited permanent neonatal diabetes associated with pancreatic and cerebellar agenesis in humans (13). Although hypomoprphic Ptf1a alleles have not been identified so far, our data predict a possible involvement of hypomorphic Ptf1a alleles in congenital pancreatic hypoplasia and diabetes in humans either by mutations in the Ptf1a coding region or deletions or single nucleotide polymorphisms in its regulatory sequence that are pivotal in pancreatic Ptf1a expression. In addition, our findings are also implicated in possible islet regeneration therapy for diabetes, since controlling gene dosage or levels/duration of gene expression might be one of the important factors for producing pancreatic β-cells in vitro.

This work was supported by the Japan Society for the Promotion of Science.

We thank Dr. P. Soriano for R26r mice, Dr. Maurean Gannon for critical reading of the manuscript, and Drs. Yoshio Fujitani, Toshihiko Masui, and Yuval Dor for helpful discussion.