Diabetes Caused by Elastase-Cre-Mediated Pdx1 Inactivation in Mice

Endocrine and exocrine pancreas tissues are both derived from the posterior foregut endoderm, however, the interdependence of these two cell types during their formation is not well understood. In this study, we generated mutant mice, in which the exocrine tissue is hypoplastic, in order to reveal a possible requirement for exocrine pancreas tissue in endocrine development and/or function. Since previous studies showed an indispensable role for Pdx1 in pancreas organogenesis, we used Elastase-Cre-mediated recombination to inactivate Pdx1 in the pancreatic exocrine lineage during embryonic stages. Along with exocrine defects, including impaired acinar cell maturation, the mutant mice exhibited substantial endocrine defects, including disturbed tip/trunk patterning of the developing ductal structure, a reduced number of Ngn3-expressing endocrine precursors, and ultimately fewer β cells. Notably, postnatal expansion of the endocrine cell content was extremely poor, and the mutant mice exhibited impaired glucose homeostasis. These findings suggest the existence of an unknown but essential factor(s) in the adjacent exocrine tissue that regulates proper formation of endocrine precursors and the expansion and function of endocrine tissues during embryonic and postnatal stages.

buds, developing antral stomach, duodenum and the lower bile duct. During mid-to-late embryonic stages, its expression is increased in β cells, but decreased in exocrine acinar/duct cells and epithelial cells of the common bile duct and rostral duodenum. Gene inactivation studies have demonstrated pivotal roles for Pdx1 during embryogenesis, as global Pdx1 knockout results in pancreatic agenesis, a lack of Brunner's glands and malformation of the major duodenal papilla [13][14][15] .
Since Pdx1 is indispensable for the formation of pancreatic exocrine and endocrine cells during development, we expected that exocrine-specific inactivation of Pdx1 would be an ideal way to generate exocrine-lacking or hypoplastic mutants in which we could test if exocrine tissue is required for proper endocrine formation and function. For this purpose, we performed Elastase-Cre-mediated Pdx1 inactivation and analyzed the pancreatic phenotype and function. We demonstrate that the mutant mice showed not only exocrine defects, but also fewer endocrine precursors and endocrine cells with less proliferation and delayed maturation, resulting in impaired glucose homeostasis. These findings support the notion that the exocrine pancreas is required for proper endocrine development and function, and that normal development of the pancreas occurs in an interactive, coordinated manner between the two tissues.

Results
Elastase-Cre-mediated Pdx1 inactivation causes pancreatic hypoplasia and growth retardation. First, we evaluated the specificity and efficiency of the Elastase-Cre-mediated recombination. Lineage tracing of Elastase-Cre-expressing cells in control mice (Pdx1 +/+ ;Elastase-Cre;ROSA26r and Pdx1 loxP/+ ;Elastase-Cre;ROSA26r mice) showed that most progeny of Elastase-Cre-expressing cells differentiated into acinar cells at P1, but a subpopulation was detected in small ductal structures contiguous with the acini (Fig. S1). The proportion of lineage-labeled cells in Cytokeratin(+ ) duct cells tended to decrease during late embryonic stages in Pdx1 loxP/+ ;Elastase-Cre;ROSA26r mice (approximately 10% at E16.5, 3% at P1 and 4% at P7), suggesting that lineage-labeled cells in the terminal ducts retained the ability to differentiate to acinar cells even at late embryonic to neonatal stages. Extremely few endocrine cells were labeled at P1 (Fig. S1): 0.44-1.30% and 0.57-0.80% in Pdx1 +/+ ;Elastase-Cre;ROSA26r and Pdx1 loxP/+ ;Elastase-Cre;ROSA26r mice, respectively (n = 3; at least 40 islets and 4000 endocrine cells were counted per mouse). These observations indicate satisfactory efficiency and specificity of the Elastase-Cre-based recombination in the exocrine lineage and negligible Cre expression in the endocrine lineage.
Newborn pups of Pdx1cKO mice (Pdx1 loxP/− ;Elastase-Cre;ROSA26r or Pdx1 loxP/loxP ;Elastase-Cre;ROSA26r mice) were indistinguishable from control littermates, but the body size of the mutant mice was clearly reduced by P7 (Fig. 1A). Mutants also had smaller pancreas and showed pancreatic hypoplasia at P7 based on histology (Fig. 1B). They exhibited growth retardation (Fig. 1C), and half did not survive to weaning. Pancreatic hypoplasia persisted in the survivors (Fig. 1D), and the pancreata of mutant mice were even smaller than expected from the reduced body size (Fig. 1E); the percentages of pancreas weight/body weight in the control mice were 0.61%, 0.39%, and 1.01% at P1, P7 and P28, respectively, while those in the mutants were 0.14%, 0.11% and 0.39% at P1, P7 and P28, respectively. Increased fat content in the stool of Pdx1 loxP/− ;Elastase-Cre;ROSA26r mice was detected, suggesting exocrine dysfunction and an associated decrease in lipid absorption by the small intestine. We could not detect X-gal(+ ) endocrine cells in Pdx1cKO mice at P1, strongly suggesting agenesis or the elimination of Pdx1-null endocrine cells during embryonic stages.
Elastase-Cre-mediated Pdx1 inactivation results in impaired exocrine differentiation during organogenesis. Developmentally, X-gal(+ ) cells became detectable at E12.5 in the pancreatic epithelia of control and Pdx1cKO mice ( Fig. 2A,E,I,M). Although some X-gal(+ ) cells in Pdx1cKO mice retained Pdx1 immunoreactivity at E13.5, Pdx1 expression was no longer detected in X-gal(+ ) cells at E14.5 or E16.5 (Fig. S2), showing that the Elastase promoter-driven Cre recombinase successfully activated the β -galactosidase gene at the ROSA26 locus and inactivated the Pdx1 gene by E14.5. The expression of Pdx1 in some lineage-labeled cells at E13.5 may be explained by the relatively stable nature of the Pdx1 protein translated prior to gene recombination.
Until E14.5, there were no apparent macroscopic differences between the pancreata of control and Pdx1cKO mice ( Fig. 2A,B,I,J), but the pancreata of the mutant mice was clearly smaller at E16.5 and did not recover afterwards (Fig. 2C,K). Histologically, exocrine development in the control mice resembled that of wild-type mice, with normal branching of the ductal tree at E14.5 and many amylase-expressing acinar cells detected at E16.5 (Fig. 2F,G). At P1, fully mature acinar cells were observed with basal nuclei, abundant cytoplasmic secretory granules and a low nucleus:cytoplasm ratio (Fig. 2H). In contrast, there was abnormal exocrine development in Pdx1cKO mice: despite no apparent macroscopic differences with control pancreata, the ductal tree was poorly branched and had a slightly dilated lumen at E14.5 (Fig. 2N). In addition, only few amylase-expressing cells were detected at E16.5 ( Fig. 2O-inset) and acinar cytodifferentiation was extremely impaired, as acinar cells were much smaller and cuboidal, had few cytoplasmic granules, and a centrally located nucleus (Fig. 2P).
Elastase-Cre-mediated Pdx1 mutant mice exhibit impaired glucose homeostasis with reduced insulin secretion. To test the hypothesis that disturbed exocrine development causes abnormal endocrine development and/or function, we compared Pdx1cKO mice (Pdx1 loxP/loxP ;Elastase-Cre mice) to Pdx1 +/+ ;Elastase-Cre mice controls, because a dosage effect of Pdx1 in endocrine function was previously reported 16,17 . Interestingly, the mutant mice were mildly hyperglycemic, and IPGTT demonstrated a significant impairment in glucose homeostasis in Pdx1cKO mice at P28 (Fig. 3A). Plasma insulin 15 min after glucose injection was significantly lower in mutants (Fig. 3B).

Reduced proliferation and accelerated apoptosis in the developing ductal tree in
Elastase-Cre-mediated Pdx1 mutant mice. To gain insights on the mechanism responsible for impaired glucose homeostasis in Pdx1cKO mice, we analyzed embryonic pancreas formation. At E14.5, β -gal(+ ) cells were detected in the tip of the branching ductal tree in both control and Pdx1cKO mice (Fig. 4A). The number of lineage-labeled cells was significantly reduced in Pdx1cKO mice, while that of non-labeled cells remained the same as that in control mice at this stage (Fig. 4B). Consistent with the macroscopically smaller pancreas at E16.5, PHH3 and TUNEL staining revealed that the progeny of Pdx1-inactivated cells (β -gal(+ ) cells) in mutant mice proliferated less (Fig. 4C) and had increased apoptosis at E14.5 (Fig. 4D). Interestingly, while the percentage of PHH3(+ ) cells in the non-lineage labeled cells remained the same between control and mutant mice, we observed increased cell death of the β -gal(− ) cells in mutant mice, suggesting non-cell autonomous effects within the developing ductal tree (Fig. 4C,D).

Impaired tip/trunk patterning in Elastase-Cre-mediated Pdx1 mutant mice. It was previously
reported that the developing ductal tree gradually obtains tip/trunk patterning through which regional control of endocrine and exocrine differentiation is determined; endocrine cells originate from the Nkx6.1-expressing trunk domain, whereas exocrine acinar cells are differentiated from Ptf1a-expressing tip domain after E14 18-20 . Considering possible non-cell autonomous effects on cell survival within the developing ductal tree (Fig. 4D), we speculated that conditional inactivation of Pdx1 by the Elastase-Cre transgene disturbs the tip/trunk patterning of the ductal tree to reduce the formation of endocrine precursors in Pdx1cKO mice. As shown in Fig. 5A, the majority of lineage-labeled and non-labeled cells in control mice at E14.5 were Ptf1a(+ )Nkx6.1(− ) tip cells and Ptf1a(− )Nkx6.1(+ ) trunk cells, respectively, suggesting that Elastase-Cre-mediated recombination occurred predominantly in the tip region. However, in mutant mice, not only was the percentage of Ptf1a-expressing cells reduced in the lineage-labeled tip region, but so too was the percentage of Nkx6.1(+ ) trunk cells. Reflecting this phenotype, the percentage of Ptf1a(− )Nkx6.1(− ) cells was increased in the mutants. In addition, we found abnormal cell differentiation in the trunk region. At E14.5, ductal epithelium in the trunk region is composed of Hnf1β -expressing cells 21 . However, while most Hnf1β -expressing cells coexpressed Nkx6.1 in control mice, they did not in Pdx1cKO mutant mice (Fig. 5B). Thus, Ptf1a(− )Nkx6.1(− ) cells, which were increased in the mutants, are suspected to be mainly composed of Hnf1β (+ ) cells. Finally, epithelial cells had a reduced percentage of Ngn3-expression and fewer β cells in mutant mice at E14.5 (Fig. 5C,D).

Reduced proliferation and delayed maturation of endocrine cells is accompanied by islet disorganization in
Elastase-Cre-mediated Pdx1 mutant mice. At P1, Pdx1cKO mice formed smaller islets and the total endocrine area was 17% that of control (Fig. 6A,B). Between P1-P28, approximately 5.45-fold expansion in pancreatic endocrine cell content occurred in control mice, but only 3.57-fold in Pdx1cKO mice  N with F). At E16.5, in control mice, the pancreatic ductal tree was well branched and many amylase-expressing cells were detected (G), but in Pdx1cKO mice branching of the ductal tree was reduced and amylase-expressing cells were fewer (O) (insets in (G,O) colabeled for amylase (red) and cytokeratin (green)). Hypoplastic pancreas persisted at P1 (compare L with D). In control mice, fully mature acinar cells were well developed, contained abundant exocrine granules with polarized subcellular localization of the nucleus on the basal side and had a low nucleus:cytoplasm ratio (H). In contrast, Pdx1cKO mice showed impaired acinar cytodifferentiation, and their acinar cells were smaller in size, had less polarized nuclear localization and a high nucleus:cytoplasm ratio (P). Scale bars, 50 μ m.
Scientific RepoRts | 6:21211 | DOI: 10.1038/srep21211 (Fig. 6A,B). Consequently, the endocrine content of mutant pancreata corresponded to approximately 11% that found in control mice at P28. It should be noted that the reduction in endocrine cell content was even more severe than predicted from the reduced body weight (approximately 81% that of control mice) and pancreas weight (approximately 31% that of control mice) at P28 (Fig. 1C,D). Immunolabeling for PHH3 revealed reduced proliferation of endocrine cells in the mutant pancreata at P1 (Fig. 6C), while TUNEL assays showed no significant difference (Fig. 6D).
Finally, in Pdx1cKO mice, along with the reduced pancreatic endocrine cell number, the maturation of β cells was delayed at P1, according to the reduced expression of GLUT2 and MafA. However, the production of these proteins was restored to control levels by P28 (Fig. 6E). Islet architecture was also impaired, and glucagon-producing cells and insulin-producing cells were intermingled even within the few relatively large islets that were present at P28 (Fig. 6F).

Discussion
The pancreas is an unusual organ in that two functionally independent tissues co-exist. Accordingly, investigators have sought potential interactions between exocrine and endocrine tissues. Here, we assessed the effects exocrine tissue has on endocrine development. Previous reports have shown that adult exocrine-driven factor(s) can regulate endocrine tissue. For example, Reg1 is expressed in the exocrine pancreas but stimulates β cell proliferation in a rat regeneration model 22 . More recently, Xiao et al. showed that exocrine tissue extracts from normal adult mice or from a pancreatic-duct-ligated model (PDL model) induced Ngn3 expression in cultured β cells 23 . In addition, loss of exocrine-derived Sostdc, a dual BMP and Wnt inhibitor, enhances insulin secretion under metabolic stress in adult pancreas 24 . Considering that embryonic mechanisms are sometimes reactivated in the regeneration process of an injured organ, we speculated that similar exocrine-to-endocrine effects play a role in normal pancreatogenesis.
To test this theory, we needed a mouse model whose exocrine tissue is severely hypoplastic or lacking during development. We created a new mutant mouse in which Pdx1, an indispensable gene for pancreas formation, is depleted in an exocrine-specific manner. As expected, Elastase-Cre-mediated Pdx1 inactivation caused severe exocrine defects after complete depletion of Pdx1 protein by E14.5, which is consistent with a report by Hale et al. 25 . However, they did not deplete Pdx1 in an exocrine-specific manner and instead used a tetracycline-responsive transactivator to temporally control the inactivation of Pdx1. Furthermore, they reported that Pdx1 depletion at E13.5 resulted in impaired pancreas formation characterized by immature acinar cell differentiation and dilated duct-like structures that resembled the exocrine defects in our Pdx1cKO mice. We believe the experimental results described in the present study are the first demonstration that selectively abrogating exocrine formation during embryonic stages reduces the proliferation and delays the maturation of endocrine cells along with altering the islet structure to cause insufficient insulin release in mice.
It has been reported that the lineage specification of pancreatic exocrine/endocrine precursor cells is regionally controlled within the developing branches of the epithelial tree. Zhou et al. used Cpa1-CreER-mediated time-specific lineage tracing to show that Pdx1 + Ptf1a + cMyc high Cpa1 + multipotent progenitor cells are located at the tips of the epithelial branches at early-to-mid embryogenesis (approx. E13-14) while endocrine/duct bipotent precursors are located within the trunk epithelium 19 . Consistent with this, we previously showed that a reduced dose of Ptf1a caused extreme impairment of branch formation throughout embryogenesis and glucose intolerance after birth 6 . Our study also suggested that the formation of a poorly branched ductal tree during early pancreatic development caused a reduction in the number of multipotent tip cells by E14.5, resulting in a smaller endocrine precursor pool in Ptf1a hypomorphic mice 6 . While our Elastase-Cre-mediated Pdx1 mutant mice exhibited reduced branching of the ductal tree and a smaller percentage of Ptf1a-expressing cells, the reduced branching was less severe than in Ptf1a hypomorphic mice. Pdx1 depletion occurred relatively late, at around E13.5-14.5, which coincides with when tip cells lost their multipotency. We found that inactivation of Pdx1 by the Elastase-Cre transgene, which occurred in the tip region, caused abnormal cell differentiation, less proliferation and elevated cell death in the trunk domain, resulting in less formation of Ngn3(+ ) endocrine precursors. Thus, we propose that the reduced endocrine cell number in Pdx1cKO mice is not endocrine-autonomous, but due to an unidentified factor(s) provided by the exocrine tissue that stimulates endocrine differentiation and expansion during embryonic and postnatal development. Alternatively, the reduced branching at mid-to-late developmental stages could have resulted in fewer niche locations for endocrine progenitors to form in Elastase-Cre-mediated Pdx1 mutant mice. We also provide evidence that β cell maturation is delayed in the mutant mice, as demonstrated by the decreased immunoreactivity of Glut2 and MafA in perinatal islets. Although both proteins were restored to normal levels by P28, we observed a long-lived altered islet structure. Islet dysmorphogenesis is thought to relate to islet dysfunction, as peripheral cell types intermingling with β cells might disrupt gap junctions or other intercellular coupling involved in efficient insulin release 26 . Cell death within the islets can also impair islet architecture 27 . Thus, the formation and elimination of Pdx1-inactivated endocrine cells during specific developmental stages could have contributed to islet dysmorphogenesis. These results warrant future studies on the mechanisms underlying the formation and maintenance of normal islet structure, especially in terms of exocrine effects on endocrine development.
Genotyping. Genomic DNA from either mouse embryonic heads or postnatal tail tips was genotyped by PCR using the primer sets listed in Table S1.
Tissue Preparation for X-gal Staining and Paraffin Sections. Tissue preparation, X-gal staining and paraffin sections were performed as previously described 5 . Immunohistochemistry and immunofluorescence. After rehydration, slides were washed and incubated for 30 min at room temperature with Protein Block (Dako) followed by overnight incubation at 4 °C with primary antibodies (Table S2). The sections were washed in phosphate-buffered saline (PBS) and incubated for 60 min with secondary antibodies (Table S3). Images were taken with a BX51 microscope (Olympus) or BZ-9000E HS All-in-one Fluorescence Microscope (Keyence).
PHH3 staining and TUNEL assays. Mitotic activity was analyzed by immunolabeling with rabbit anti-phospho-Histone H3 (Ser10) (Millipore). TUNEL assays were done using the DeadEnd Fluorometric system (Promega) according to the manufacturer's instructions.
Cell Counting. Cell numbers were counted at × 200 magnification in three randomly selected sections per pancreas, and the proportion of the desired cell types to all epithelial cells was calculated. To evaluate the relative endocrine area, the whole pancreas was sectioned in 3 μ m intervals, and the number of Chromogranin-A(+ ) cells was counted in every hundredth slide.

Intraperitoneal glucose tolerance test (IPGTT).
After overnight fasting, we measured blood glucose before-and 15, 30, 60 and 120 minutes after D-glucose injection (2 mg/g body weight, intraperitoneally) by a Glucocard DIA meter (GT1641) (Arkray). To evaluate insulin release 15 min after glucose challenge, blood was centrifuged in the presence of aprotinin (100 kIE/ml blood; Wako Chemicals), and serum insulin was measured with a mouse insulin ELISA kit (U-type) (Shibayagi).
Statistical Analysis. All indices were analyzed using the t-test (two-tailed).