Regulation of Pancreatic β-Cell Regeneration in the Normoglycemic 60% Partial-Pancreatectomy Mouse

Sixty percent Px was performed on 8-week-old male C57BL/6 mice (n = 20–25 per group) (Taconic), as described previously for rats (17) and mice (18). Mice were anesthetized with an intraperitoneal injection containing a mixture of ketamine (50 mg/kg) and xylazine (5 mg/kg). The surgery involved excision of the portion of the pancreas bordered by the spleen and stomach, extending to the small flap of pancreas attached to the pylorus, by gentle abrasion using cotton applicators. Control mice underwent a sham surgery involving laparotomy and gentle rubbing of the tissue. For all these studies, the guidelines set forth by the University of Vermont Institutional Animal Care and Use Committee were strictly followed.

Fed whole-blood glucose levels were measured with a glucose meter (Freestyle) and the plasma subsequently analyzed for insulin utilizing a sensitive insulin radioimmunoassay kit (Linco Research).

Islets were isolated by collagenase digestion and separated by histopaque density gradient centrifugation as described (19). Rat common pancreatic duct (CPD) was isolated in situ using a modified islet isolation procedure (20). Since both mouse and rat regenerating ducts show a similar expression pattern of insulin signaling intermediates, rat duct protein was used for immunoblotting to circumvent low abundance of protein from mouse ducts (17).

Tissue processing for paraffin embedding was performed as described (21). For the detection of labile signaling intermediates, pancreata were rapidly dissected and then paraformaldehyde fixed for 1.5 h, washed, equilibrated in 30% sucrose/PBS, and embedded in optimal cutting temperature medium (Miles Scientific).

Paraffin sections (5 μm) throughout the entire pancreas were mounted as ribbons on microscope slides to facilitate section counting. Two or three slides (200-μm apart) from the broadest pancreatic sections were analyzed for BCM measurement (n = 5–10 for each group and time point). These sections were found to be representative of the whole pancreas. Sections were immunostained for insulin, as described (21).

β-Cell fractional area was determined by digitally scanning entire sections using a microslide scanner (Nikon Super CoolScan 9000). Image files were processed in Adobe Photoshop and analyzed using NIH Image J (version 133μ), tabulated in pixel values converted to squared micrometers, and entered into Microsoft Excel for statistical analyses. Since the scanner afforded a maximal resolution of ∼40 μm²/pixel, we counted only insulin⁺ cell clusters of >10 pixels (∼400 μm²) or approximately three to four β-cells to exclude background “noise” in our measurements. BCM was estimated for each animal by determining the fractional β-cell surface area per animal multiplied by the pancreatic weight.

The relative surface area and number of β-cell clusters and islets (>400 μm²) were tallied for each animal using the same sections used for the BCM measurements. β-Cell cluster size was initially categorized into one of six classes: 400–599, 600–3,999, 4,000–7,999, 8,000–11,999 12,000–15,999, and >16,000 μm². β-Cell clusters (400–599 μm²) and small islets (600–3,999 μm²), regardless of their location, were considered to be newly formed, possibly neogenic. For measuring β-cell clusters <400 μm², we microscopically imaged 10 × 0.3 mm² fields for each animal and counted clusters of one to four cells.

At least two slides per adult pancreas were stained for the cell proliferation marker, Ki-67, and insulin as previously detailed (21). The number of Ki-67⁺ nuclei per 1,000–1,500 islet β-cells, per 300 CPD cells, and 10 × 0.3 mm² fields of acinar tissue were counted for each animal.

Frozen sections of dissected CPDs and adjoining remnant pancreas were prepared for staining as described (21). The following primary antibodies were used: cyclin D2 (1:25 dilution), phospho-S⁴⁷³Akt (1:100), phosphotyrosine (1:500), phospho-FoxO1S²⁵⁶ (1:250), and phospho-S⁴⁷³Akt (1:100), all from Cell Signaling, and IRS2 (1:100), FoxO1 (1:250; Upstate), insulin (1:500; Linco), PDX1 (1:1,000; gift from Dr. Chris Wright, Vanderbilt University), Forkhead box A (Foxa)2 (1:30; Santa Cruz), Ki-67 (1:500; BD Biosciences), and β-catenin (1:500; Zymed). All secondary antibodies contained “multiple labeling” grade anti-IgG conjugated to CY2 (1:300), CY3 (1:2,000), or CY5 (1:500) (Jackson ImmunoResearch). Images were acquired using conventional epifluorescence or with a Zeiss LSM510 confocal microscope (UVM Cell Imaging Facility). Images were acquired with the LSM software and merged and formatted on a Macintosh G5 with Adobe Photoshop.

Total-cell lysates from isolated islets and CPDs were sonicated in lysis buffer consisting of 50 mmol/l HEPES (pH 7.5), 1% (vol/vol) Nonidet P-40, 2 mmol/l activated sodium orthovanadate, 100 mmol/l sodium fluoride, 10 mmol/l sodium pyrophosphate, 1 mmol/l phenylmethylsulfonyl fluoride, and protease inhibitor cocktail (Roche). Protein content in the cell lysates was determined using the bicinchoninic acid protein assay kit (Pierce). Cell lysates were separated by SDS-PAGE and transferred to polyvinylidine flouride membranes. After blocking, membranes were incubated in appropriate primary antibody followed by horseradish peroxidase–conjugated secondary antibodies. Positive signals were visualized with chemiluminescence (Pierce). For immunoprecipitation experiments, lysates were incubated with anti-Akt1/2 overnight and immunoblots probed with phospho-activated Akt (pAkt) antibody.

Total RNA was extracted from isolated islets and collagenase digested CPDs using Trizol reagent (Invitrogen). cDNA was synthesized using the first-strand cDNA synthesis kit (Roche Applied Science). PCR analyses were conducted using a previously described protocol (20). Typically, 20–30 cycles were run to maintain the amplification within the linear range. (Primer sequences for FoxO1 and Foxa2 are available from the authors upon request.) An 18S rRNA primer pair (Ambion, Austin, TX) was used as an internal control for PCR analysis. PCR products were separated on a 6% polyacrylamide gel in Tris-borate EDTA buffer. The gel was exposed to a PhosphoImager screen and band intensities quantitated with a BioRad Molecular PhosphoImager (UVM DNA Analysis Core Facility).

Daily blood samples from 60% Px and sham-operated C57BL/6 mice were analyzed for nonfasting glucose and insulin concentrations over a period of 15 days postsurgery and were not significantly different at any time point between the groups (Fig. 1A and B). Both groups also maintained identical weights throughout the period (data not shown).

We measured BCM at 2, 6, and 15 days post-Px. Although not significant, sham-operated mice exhibited the expected increases in BCM for 8- to 10-week-old mice commensurate with age-related weight increases during the 2-week study period compared with unoperated age-matched mice (Fig. 1C). In the Px mice, BCM increased progressively from ∼46, 59, and 68% relative to shams at 2, 6, and 15 days post-Px, respectively. We next examined the basis for the increase in BCM by measuring β-cell mitotic frequencies. Although there was no significant increase at 2 days post-Px, β-cell proliferation frequency increased 2.5-fold over shams at 6 days post-Px and then returned to that of the shams by 15 days (Fig. 1D). Six days postsurgery, sham-operated mice also exhibited a small but significant increase in β-cell proliferation compared with the 2-day sham group; however, both sham and Px groups showed depressed β-cell proliferative activity at 2 days compared with untouched mice, likely due to suppressed feeding activity following laparotomy (Fig. 1D). Cyclin D2, a member of the D-type cyclin family, plays a crucial role in β-cell proliferation (4,5). We detected a marked increase in cyclin D2 protein by immunoblotting and immunoreactivity, with an increase in the number of β-cell nuclei expressing the protein in 6 days post-Px mice compared with sham mice (Fig. 1E). No differences were observed in cyclin D2 expression in the 15-day groups (data not shown).

Although controversial, β-cell neogenesis has been proposed to play an important role in BCM compensation (10,17,20–23). We therefore quantified the prevalence of β-cell clusters and islets and ranked them according to size to determine their contribution to BCM increase in C57BL/6 mice after Px. We observed an ∼2.8-fold higher incidence of β-cell clusters/small islets (1,000- to 4,000-μm² range) in 2-day Px mice compared with sham controls (Fig. 1F). In contrast, we detected no increase in the number of very small β-cell clusters (one to four cells in cross-section) between the 2 and 6 day postsurgery groups at this time point (data not shown). Proliferation was also rare in both β-cell cluster sizes with no differences detected between groups. We next quantified the prevalence of islet sizes ranging ∼4,000–8,000, 8,000–11,960, 12,000–16,000, and >16,000 μm² (Table 1). We detected an increase of about twofold in the larger 8,000–11,960 and 12,000–16,000 μm² islet size groups at 15 days postsurgery (islet diameter range ∼90–125 μm) (Table 1). Collectively, these results seem consistent with an early postmitotic differentiation of β-cell precursors, which initially appear as small β-cell clusters and then subsequently proliferate to larger islets.

To determine if the IRS2/Akt/PDX1 signaling pathway is activated during the BCM increase, we analyzed the expression of these proteins at 6 days postsurgery when β-cell proliferation is maximal. By immunostaining, no differences were observed in β-cell IRS2 levels between groups (data not shown). Immunostaining and immunoblot analyses revealed low to moderate levels of pAkt in islet β-cells of control mice (Fig. 2A and C). However, surface-oriented pAkt immunoreactivity was enhanced in islets of 6-day Px mice (Fig. 2B and C). Furthermore, pAkt expression correlated with increased levels of PDX1 expression in β-cells (Fig. 2B and C).

The forkhead transcription factor FoxO1 plays key roles in regulating apoptosis, proliferation, and glucose metabolism. The function of FoxO1 is inhibited by Akt-mediated phosphorylation and its subsequent nuclear exclusion (24). Recent studies suggest that FoxO1 regulates BCM by inhibiting Pdx1 gene transcription through Foxa2 binding (12). We therefore examined FoxO1 and Foxa2 expression in mice at 2 and 6 days post-Px. Compared with sham mice, we observed a moderate decrease in total cytoplasmic FoxO1 by immunostaining and immunoblotting and a marked decrease in phospho-FoxO1 by immunoblotting in 6-day Px mice (Fig. 2D–F). Hence, it appears that both the total and phosphorylated (presumably cytoplasmic) pools of β-cell FoxO1 are reduced in 6-day Px mice. No changes were detected in strongly FoxO1⁺ peripheral cells that were often PDX1⁺/somatostatin⁺ (Fig. 2D and E).

We next analyzed the expression of Foxa2 in 6-day Px mice by immunostaining. We detected moderate increases in β-cell nuclear Foxa2 immunoreactivity compared with sham-operated mice (Fig. 2G and H).

Our previous studies in the rat 60% Px model have shown that enhanced CPD proliferation was highest in the lateral evaginations as opposed to main epithelial cell lining (17). Accordingly, by immunostaining with Ki-67, we observed increased frequency of staining in the evaginations of 2-day Px mice (Fig. 3B and D) compared with shams (Fig. 3A and C). We quantified CPD proliferation, distinguishing between the main epithelial lining versus the evaginations. Compared with control mice, we detected a 10-fold increase in the mitotic frequency of the evaginations at 2 days post-Px (P < 0.001), which was sustained at 6 days post-Px (Fig. 3E). Although insulin⁺ cells in Px mice were associated with evaginations, their low frequency precluded a quantitative analysis.

New β-cell development, or neogenesis, is widely considered to originate from the pancreatic exocrine ducts (17,20,25–27). Hence, we next sought to examine the activity of the IRS2/Akt/FoxO1 pathway in the common pancreatic duct epithelium, since we detected transient increases in the number of small islets/β-cell clusters early in the regeneration period along with hyperproliferation of cells lining the duct evaginations. Our previous studies in the 60% Px rat model had demonstrated enhanced IRS2 and pAkt immunoreactivity in the CPD during pancreatic regeneration that correlated with β-cell neogenesis (17). Similarly, we also observed strong IRS2 staining in the CPD evaginations of 2- and 6-day Px mice that normalized by 15 days postsurgery (Fig. 4A and B). pAkt immunoreactivity was also increased at 2 and 6 days post-Px compared with sham ducts (Figs. 3A and B and 4C–E). Surface-oriented pAkt immunoreactivity correlated with increased nuclear PDX1 expression in Px mice, especially in the duct evaginations (Figs. 3D and 4D). Correspondingly, by immunoblotting, PDX1 protein levels were increased ∼1.5-fold in 2-day Px mice; however, pAkt levels were markedly enhanced at both 2 and 6 days post-Px in CPD extracts (Fig. 4E).

Since recent reports have demonstrated FoxO1 expression in pancreatic ducts and because it antagonizes Foxa2 function in islets (12), we first examined FoxO1 and Foxa2 mRNA by RT-PCR in ducts and islets of Px rats. An increase of ∼1.5-fold in FoxO1 and Foxa2 mRNA levels over sham controls was observed in isolated CPDs from 2-day Px rats, while there was a marginal but significant decrease in islets (Fig. 5A and B). By immunostaining and immunoblotting, intense cytoplasmic FoxO1 (Fig. 5C, D, and G) and nuclear Foxa2 immunoreactivity (Fig. 5E and F) were observed in the regenerating duct epithelium, especially in the evaginations. The contrasting profiles of FoxO1 mRNA and protein levels in islets and ducts suggest different regulatory mechanisms in these tissues.

To compare the status of the insulin signaling intermediates between a chronic (i.e., Px surgery) and short-term stimulus, insulin (750 mU/kg i.p.) or saline was administered to 6-day sham and Px mice and then analyzed at intervals for IRS2/pAkt/FoxO1 and PDX1 immunoreactivity in islets and ducts. At 10 min postinjection, although increased surface-oriented pAkt was observed in β-cells of saline-treated Px mice compared with shams (Fig. 6A and B), there was a striking enhancement in cytoplasmic pAkt in insulin-treated Px β-cells compared with sham islets (Fig. 6C and D). Unexpectedly, we also observed enhanced IRS2 immunoreactivity in the islets of insulin-treated Px mice compared with shams (Fig. 6E and F). FoxO1 immunoreactivity was drastically reduced in insulin-treated Px mice compared with insulin-treated sham mice (Fig. 6G and H). In contrast, increased FoxO1 immunoreactivity was observed in the CPD evaginations of insulin-treated Px mice compared with a corresponding treatment of control mice (Fig. 6I and J). These results show that the observed pattern of tissue-specific signaling intermediate expression and activation noted post-Px was augmented by exogeneous insulin.

Although the metabolic characteristics of compensatory partial regeneration of the pancreatic β-cells have been well characterized in the Px rodent model (17,28,29), a detailed study of the underlying mechanisms that are responsible for the BCM expansion have not been previously reported. Previous studies in C57BL/6 (18) and other mouse strains (30–32) following Px have highlighted the impact of β-cell proliferation in BCM homeostasis. On the other hand, pancreatic duct ligation (33) and glucose-infusion rodent models (34) have emphasized BCM contributions by neogenesis originating from exocrine tissues. However, the confounding effects of hyperglycemia in these models on BCM compensation, the signaling pathways, and β-cell gene expression could not be excluded (2). In contrast, the 60% Px rodent model offers a significant advantage in examining BCM compensation since a vigorous, but transient, regeneration process ensues despite a 60% reduction in BCM (17,18).

In this study, we have investigated the mechanisms that mediate BCM expansion in 60% Px C57BL/6 mice. There was a net increase in BCM of ∼28% versus shams by 15 days following Px due to an increase in the number of β-cell clusters/small islets, possibly from an early postmitotic differentiation of as yet unidentified β-cell precursors, followed by enhanced β-cell proliferation in preexisting islets. We have demonstrated that these consecutive processes correlate strongly with enhanced insulin receptor pathway signaling, with Akt kinase serving as a central intermediate.

Previous studies in both 60 and 90% Px rat models have shown that proliferation preceded differentiation in the epithelium of the large ducts (17,28). We detected a massive 10-fold increase in proliferation in the epithelial lining of the lateral evaginations of the CPD at 2 days post-Px. Concomitant with an increase in proliferation, we observed a transient activation of the insulin signaling pathway in the CPD, as evidenced by increased IRS2, pAkt, and FoxO1 immunoreactivity in the C57BL/6 mice post-Px in concert with enhanced PDX1 expression. Thus, in 2-day post-Px mice, PDX1, IRS2, activated Akt, FoxO1, and Foxa2 were all expressed in the CPD lining; however, they were more robustly expressed in the epithelial cells of the duct evaginations versus the lining. More importantly, strong pAkt⁺ cells correlated with intense nuclear PDX1 expression. However, since very few insulin⁺ cells were detected within the ducts of the sham and Px mice, it is unclear whether these strong pAkt⁺/PDX1⁺ cells progress to fully differentiated β-cells to contribute to the islet β-cell pool.

In parallel with a peak in duct proliferation and enhanced Akt signaling at 2 days post-Px, there was a short-lived 2.8-fold increase in the number of β-cell clusters and small islets. As the prevalence of these clusters waned, β-cell proliferation peaked at 6 days post-Px, correlating with a 50% increase in BCM. By 15 days postsurgery, increased numbers of large islets were observed in Px mice, suggesting continued growth of the islet β-cell population following the transient β-cell hyperproliferation occurring 1 week prior. Although there are currently no specific markers for newly differentiated versus newly replicated β-cells, these results suggest that in this mouse model of Px, BCM expansion involves not only the expected proliferation from preexisting β-cells but possibly from newly formed islets. The contribution of β-cell neogenesis from ducts or other pancreatic progenitor cells to both normal BCM homeostasis (1,35) and to an experimental regeneration stimulus (17,25,34) has recently been challenged by studies reporting that the primary mechanism of β-cell renewal is by proliferation of preexisting β-cells (3). Although our results strongly suggest that the early increase in the number of β-cell clusters likely contributes to BCM compensatory increase, the source of these β-cell clusters/small islets remains unknown, as we failed to detect proliferating β-cell clusters, and the frequency of insulin⁺ cells in the duct epithelium was too low to be quantified. However, increased numbers of single/double β-cells have been observed in transgenic mice overexpressing constitutively active Akt1 (9). Thus, in the current study, BCM expansion due to Akt activation may not only regulate hyperproliferation of existing β-cells but, possibly, β-cell differentiation from exocrine progenitors as well.

Studies of mice with altered gene expression suggest that the insulin signaling pathway via Akt/FoxO1/PDX1 regulates β-cell proliferation (7,12,14–16). In turn, the β-cell proliferative response is associated with increased levels of cell cycle proteins, including cyclin D1, cyclin D2, p21, and cdk4 activity mediated by Akt (4,5,36,37). Thus, whereas FoxO1 haplodeficiency restored Pdx1 expression and β-cell proliferation in IRS2^−/− and Pdk1^−/− mice (12,15), PDX1 haplodeficiency abrogated β-cell compensatory response in insulin-resistant Insr^+/−/Irs1^+/− mice and in mice lacking insulin receptor in liver through impaired β-cell–proliferation mice (14). These results suggest that FoxO1 and PDX1 may mediate proliferative signals through Akt.

A universal finding in normal rodent Px models is the several-fold increase in β-cell proliferation peaking 4–7 days postsurgery. Individual islet growth, including that following a Px stimulus, would be expected to be a product of β-cell proliferation, hypertrophy, apoptosis, and clearance (1,17,38). This was suggested by a 2.5-fold enhancement in β-cell proliferation at 6 days post-Px, concomitant with increased cyclin D2 expression. Thus, enhanced Akt signaling in β-cells from 6 days post-Px mice likely mediates this heightened replication.

Recent studies indicate that the IRS2/Akt/FoxO1 signaling pathway plays a pivotal role in compensatory BCM augmentation in insulin-resistance states, with β-cell proliferation as a primary mechanism for maintenance of this BCM (7–16,39). Rescue of diabetes in Pdk1^−/− and Irs2^−/− mice by superimposing FoxO1 haploinsufficiency, and by over-expression of PDX1 in Irs2^−/− mice, essentially restored BCM and PDX1 expression, implying a regulatory link between the insulin signaling cascade and PDX1 (12,13,15). Although the impact of this pathway in response to β-cell regeneration is not yet fully resolved, we have previously reported in the 60% rat Px model that increased IRS2 and pAkt levels correlated with enhanced neogenesis from the large ducts (17). While IRS2 levels in β-cells were not significantly changed during β-cell regeneration, possibly due to degradation (40), β-cell pAkt immunoreactivity was intense and surface oriented and, at 6 days post-Px, correlated strongly with nuclear PDX1. Akt has been ascribed roles in regulating β-cell size, survival, proliferation (9–11,41,42), and neogenesis (9,17). Many of these functions are likely manifestations of Akt signaling through the transcription factor FoxO1 and, in turn, PDX1. In islets of Px mice, cytoplasmic FoxO1 expression was decreased profoundly in β-cells of 6-day Px mice and correlated with enhanced PDX1 levels. Although not substantiated, these observations suggest that phosphorylated FoxO1 may be rapidly degraded following nuclear exclusion and PDX1 activation (12,43). In contrast to islets, increased cytoplasmic expression of FoxO1 correlated with increased nuclear PDX1 expression in the CPD evaginations. This observation was puzzling, since we anticipated decreased cytoplasmic FoxO1 immunoreactivity due to intense nuclear PDX1 and Foxa2 immunoreactivity in the CPD evaginations. Although unclear, the regulation of FoxO1 activity appears tissue specific and warrants further investigation.

We have also detected moderate increases in Foxa2 mRNA and protein levels in Px mice consistent with its role as an activator of PDX1 expression (44). As PDX1 has recently been ascribed roles in β-cell survival and proliferation (14,45), these results support our premise that enhanced BCM in Px mice involves β-cell signaling through Akt, FoxO1, and PDX1, regulating β-cell growth. Our findings that a transient stimulation of the insulin signaling pathway by insulin itself further enhanced the expression pattern of signaling elements during β-cell regeneration underscores the impact of this pathway in BCM expansion.

In conclusion, the 60% Px model serves as a useful paradigm for examining the mechanisms and signaling pathways for β-cell regeneration following pancreatic injury. The current study in the C57BL/6 mouse strain demonstrates that β-cell regeneration entails coordinated processes that involve an early increase in the prevalence of β-cell clusters with no demonstrable proliferation of these insulin⁺ cells. Subsequently, β-cell hyperproliferation occurs as a second-phase response to the β-cell regeneration stimulus. The IRS2/Akt/FoxO1/PDX1 pathway appears to mediate this β-cell growth response. We speculate that an early response involving increased insulin signaling activity and proliferation in pancreatic ducts post-Px may be related to neogenesis.

This work was supported by a CDA from the Juvenile Diabetes Research Foundation and a Research Award from the American Diabetes Association (to M.P.) and by the National Institutes of Health (grant DK-068329 to T.L.J. and grants DK-56818 and DK-66635 to J.L.L.).

We acknowledge Dr. Dhananjay Gupta for the critical reading of the manuscript. We also are indebted to Dr. Afshin Salsali who helped in the initiation of this project.