Regulated β-Cell Regeneration in the Adult Mouse Pancreas

pINS-cMycER^TAM transgenic mice were maintained in a conventional pathogen-free facility at the University of California San Francisco (San Francisco, CA) according to National Institutes of Health guidelines. The mice {originally outbred from a CBA × C57BL/6 F1 colony} were backcrossed two times to the C57BL/6 strain (Jackson Laboratories, Bar Harbor, ME). Mice were screened for the c-MycER^TAM transgene by PCR, as previously described (17). The results are representative of pancreas tissue from 4–10 mice examined for every time point indicated in the figures.

To activate cMycER^TAM in the pancreatic β-cells of adult transgenic mice, TAM (Sigma-Aldrich, St. Louis, MO) was dissolved in corn oil (10 mg/ml) and administered intraperitoneally at 1 mg · mouse⁻¹ ·day⁻¹ for 6 continual days, except where noted. Nonfasting blood glucose levels were measured throughout the course of the experiments as an assessment of normal β-cell function using a Lifescan Glucometer (as per the manufacturer's specifications).

Immunohistochemistry assays were performed on paraffin sections as described previously (18). The following primary antibodies were used: guinea pig anti-insulin (1:300), rabbit anti-glucagon (1:300; Linco Research, St. Charles, MO), guinea pig anti-glucagon (1:1,000; Linco), rabbit anti-amylin (1:1,000; Advanced Chem Tech, Louisville, KY), mouse anti-Ki67 (1:100; Novocastra, Burlingame, CA), mouse anti-GLUT2 (1:500; ADI, San Antonio, TX), mouse anti-Islet1 (Isl1) clone 39.4D5 and mouse anti-Pax6 (1:25 and 1:25, respectively; Developmental Studies Hybridoma Bank, Iowa City, IA), rat anti–pancreatic polypeptide (PP; 1:500; Chemicon, Temecula, CA), mouse anti-proinsulin and rabbit anti–C-peptide (1:500; O. Madsen, Hagedorn, Denmark), and NK6 homeobox 1 (Nkx6.1; 1:1,000) (19). Primary antibodies were detected with fluorescein isothiocyanate–and Cy3- conjugated secondary antibodies (1:200 and 1:600, respectively; Jackson ImmunoResearch Laboratory, West Grove, PA). Transferase-mediated dUTP nick-end labeling assay was performed using a TACS*XL Blue Label kit from Trevigen (Gaithersburg, MD). A Zeiss Axioskop 2 plus fluorescent microscope was used for image acquisition.

To measure β-cell mass, six insulin- and glucagon-stained pancreas sections from each mouse (n = 3), separated by at least 60 μm, were imaged using a Zeiss Axiophot 2 plus microscope. Pancreatic and islet areas were outlined and quantified using Open Lab software (Improvision, Lexington, MA) and are presented as a ratio.

To measure α-cells and PP-producing cell mass, six stained pancreas sections from each mouse (n = 3), separated by at least 60 μm, were imaged using a Zeiss Axiophot 2 plus microscope. Cell number was counted manually, and pancreatic area was outlined and quantified using Open Lab software (Improvision). The results are presented as the number of cells per millimeter squared of total pancreas area.

After a 16- to 18-h fast, mice were weighed, and a fasting blood glucose level was measured using a Lifescan Glucometer. Mice were then injected intraperitoneally with a 1 mmol/l glucose solution at 10 μl/g body wt. Blood glucose levels were then measured every 20 min for 2 h after injection. The results are shown as an average of six mice tested at day 0, 24, and 40 and five mice at day 90 of the experiment.

RNA isolation, cDNA preparation, and qPCR were performed as described previously (20). RNA expression of target genes was normalized based on comparison to cyclophilin or GUS (glucoronidase) expression. Primer sequences are included in Supplemental Table 1 (available in an online appendix at http://dx.doi.org/10.2337/db07-0013).

Activation of c-Myc induced expression of downstream targets, resulting in either cell proliferation or cell death (21,22). Pelengaris et al. (17) demonstrated that intraperitoneal administration of TAM in pINS-cMycER^TAM transgenic mice resulted in a brief phase of β-cell proliferation, characterized by a rapid, ubiquitous, and synchronous entry of β-cells into cell cycle. However, within 1 day of TAM administration, β-cells died from apoptosis, indicating that in the context of the β-cell, the proapoptotic activity of c-Myc is more prominent than its proliferative capacity.

Before TAM administration, pINS-cMycER^TAM mice displayed normal blood glucose levels of ∼100 mg/dl and no detectable signs of apoptosis in pancreatic tissue (Fig. 1A and B). After a single dose of TAM, pINS-cMycER^TAM mice had a sudden episode of hypoglycemia that was most likely a result of unregulated insulin release from apoptosing cells. However, continued administration of TAM for 6 days resulted in significant apoptosis of β-cells, paralleled by extreme hyperglycemia with blood glucose levels increasing >300 mg/dl (Fig. 1A and C). In addition, pINS-cMycER^TAM mice suffered from other diabetes-related symptoms, including polyuria, glycosuria, and significant weight loss (data not shown). The specificity of TAM for cMycER^TAM activation was confirmed by the maintenance of normoglycemia in nontransgenic mice treated with TAM (Fig. 1A). It is important to note that expression levels of cMycER^TAM in the pINS-cMycER^TAM transgenic mice were similar to c-Myc levels detected in normal mouse fibroblasts, indicating that c-Myc expression in β-cells remained within physiological levels (17). Thus, sustained administration of TAM consistently triggered β-cell apoptosis and acute diabetes in pINS-cMycER^TAM mice.

We note that the observed cell loss was specific to β-cells; representative serial sections from pINS-cMycER^TAM mice that were sacrificed 3 days after a 6-day period of continued TAM administration revealed that glucagon-, PP-, and somatostatin-positive cells were easily detected after TAM administration (data not shown). Quantification of the number of α- and PP-cells confirmed that these endocrine cell populations were not depleted on TAM administration (Supplemental Fig. 1). Loss of central β-cells resulted in collapse of islet structures by clusters of non–β-cells that were now evenly distributed throughout the islet center rather than confined to the islet periphery (Fig. 2B). Thus, c-Myc activation under control of pINS specifically abolished β-cells, whereas non–β-endocrine cells were not affected.

To determine whether insulin-positive β-cells regenerate after targeted destruction, we collected pancreata from several mice at different time points for up to 90 days post–TAM administration (Fig. 2). Examination of pancreas tissue from unadministered transgenic mice demonstrated that islet number, size, and morphology were comparable to nontransgenic animals; well-rounded islets with strong insulin expression in the center and glucagon expression around the periphery were observed throughout the pancreas (Fig. 2A and G). In contrast, dramatic changes in islet morphology were evident by day 7, 1 day post–withdrawal of TAM (data not shown), when the remaining islets were involuted and significantly smaller because of c-Myc–induced β-cell apoptosis. By day 9, 3 days post–withdrawal of TAM, only small islet remnants were detectable throughout the pancreatic tissue (Fig. 2C), identified mostly by remaining glucagon-positive cells. At 1 week post–withdrawal of TAM, the number of insulin-positive cells increased throughout the tissue samples. However, islet size was still small, and organization with respect to insulin- and glucagon-positive cells was atypical (Fig. 2D). In contrast, between days 40 and 90, pINS-cMycER^TAM islets regained their normal morphology in terms of size, number, and cell distribution (Fig. 2E and F). Morphometric analysis confirmed the immunohistochemical data and revealed a >90% reduction in β-cells at day 9 (Fig. 2G). Islet area was restored at day 90 compared with wild-type controls (Fig. 2H). However, a comparison to age-matched, vehicle-administered pINS-cMycER^TAM mice showed that islet area continued to expand in these mice, possibly a consequence of increased proliferation caused by low-level transgene expression even in the absence of TAM. Together, these results demonstrated that TAM-induced c-Myc expression leads to targeted ablation of insulin-producing β-cells, whereas inactivation of c-Myc via withdrawal of TAM revealed an astounding ability for islet regeneration.

Bonner-Weir and colleagues (23) have shown that c-Myc expression decreases insulin transcription. To confirm that we observed loss of β-cells rather than a transient reduction in insulin expression, we costained pancreatic tissue from TAM-administered transgenic mice with additional β-cell markers: islet amyloid polypeptide (IAPP) and GLUT2 (24). Unadministered adult pINS-cMycER^TAM transgenic mice displayed typical coexpression of insulin, IAPP, and GLUT2 in islets throughout the pancreas (Fig. 3A and E). After 2 days of TAM administration, β-cell marker expression was significantly downregulated (Fig. 3B and F). By day 9, 3 days after TAM withdrawal, almost no insulin/IAPP or insulin/GLUT2 double-positive cells were detected in the pancreas of pINS-cMycER^TAM mice (Fig. 3C and G), whereas regenerated β-cells displayed normal expression of mature markers at day 90 (Fig. 3D and H). These findings demonstrate that c-Myc activation destroys mature β-cells and that deactivation of c-Myc results in significant regeneration of functional β-cells.

To confirm that regeneration results in functional β-cells, insulin-positive cells were analyzed for expression of mature β-cell transcription factors (24) at day 90 (Fig. 4). Expression of all markers tested, including Isl1 (Fig. 4A and B), Nkx6.1 (Fig. 4C and D), and Pax6 (Fig. 4E and F), was comparable to unadministered tissue, indicating that newly formed cells possess the proper repertoire of mature β-cell transcription factors (quantitative PCR for these markers is shown in Supplemental Fig. 2). In addition to morphologically assessing β-cell regrowth, we examined the physiological activity of the new β-cell population over the course of regeneration. Withdrawal of TAM from pINS-cMycER^TAM mice resulted in a gradual normalization in blood glucose levels (Fig. 5A). Around day 40, when few, but typical-looking, islets were detectable (Fig. 2E), blood glucose levels were recovering, although they still were markedly above the normal level of ∼100 dl/mg. In contrast, blood glucose levels of individual mice returned close to normal in the majority of mice (Fig. 5A) by day 90, when a significant number of islets were detectable (Fig. 2F).

To more vigorously test islet function, transgenic mice were challenged with a concentrated glucose solution and examined for their ability to metabolize glucose over time. By day 90, pINS-cMycER^TAM mice exhibited a glucose tolerance profile identical to that found in age-matched vehicle-administered pINS-cMycER^TAM control mice (Fig. 5B). There was a slight decrease in the ability of the transgenic mice to regulate glucose levels compared with age-matched wild-type littermates at the 30-min time point of the analysis, but no differences were observed at the end of the assay 2 h after glucose injection. In contrast, by day 24, 18 days after TAM withdrawal, as well as on day 40, 36 days after TAM withdrawal, glucose tolerance in pINS-cMycER^TAM mice was still severely affected, with animals remaining hyperglycemic even 2 h after glucose administration (Supplemental Fig. 3). Thus, TAM-administered transgenic pINS-cMycER^TAM mice progressively improved their response proportionally to the length of the regeneration period. To determine why glucose tolerance was still impaired at day 40, a time point when substantial regeneration of β-cells has already occurred (Fig. 2E), we tested for a potential delay in insulin processing. The elevated levels of proinsulin, together with low levels of C-peptide, in day 40 administered mice suggests that newly formed β-cells still maturate after insulin expression is first observed (Fig. 5C and E). However, at day 90, β-cells seemed to be fully functional because proinsulin levels were significantly reduced and all cells expressed high levels of C-peptide (Fig. 5D and F). Thus, based on measurements of blood glucose levels with and without glucose challenge, regenerating β-cells are functional and maintain normal metabolic function over time as soon as a critical mass of fully matured β-cells is reached.

The controlled onset of β-cell ablation followed by progressive restoration of glucose homeostasis in the pINS-cMycER^TAM mice suggested that important insights into the process of mammalian β-cell regeneration can be gained from the study of these transgenic mice. Several hypotheses have been proposed to explain the regeneration of pancreatic β-cells, including neogenesis from progenitor cells located in ductal structures, spleen, bone marrow, or islets. More recently, lineage-tracing experiments have found that preexisting β-cells are the main source of new β-cells in adult mice (12). We performed several experiments designed to determine the origin of the regenerating β-cells in pINS-cMycER^TAM mice.

The increase of insulin-expressing cells in pancreatic ducts in certain animal models of pancreas regeneration or injury has served as support for the existence of pancreatic progenitor cells in ducts. Similarly, an increase in pancreatic duodenal homeobox 1 (Pdx1) expression has been observed in duct and duct-associated cells in models of pancreas regeneration (25–27). Extensive confocal microscopy analysis failed to reveal any cells displaying both insulin immunoreactivity and staining for the ductal marker lectin DBA (Dolichos biflorus agglutinin) at any of the stages analyzed in pINS-cMycER^TAM mice (Fig. 6A and B). Similarly, no Pdx1 expression was found in ducts or in close association with ducts (Fig. 6C and D), suggesting little contribution of duct cells during β-cell regeneration in pINS-cMycER^TAM mice.

To determine the extent of overall cell proliferation in pINS-cMycER^TAM mice, we performed immunohistochemical analysis for Ki67, a proliferation marker. Foci of proliferating cells were present in islets, in both involuted and well-preserved islets, and ducts adjacent to them within 6 days after TAM withdrawal (Fig. 6E and F). Active proliferating cells remained in islets at later stages during β-cell regeneration (Fig. 6G). By day 90, proliferation in islet and duct cells returned to levels similar to those found in unadministered pINS-cMycER^TAM mice.

Summarily, these results indicate that β-cell regeneration appears to be mediated by replication of remaining β-cells rather than neogenesis from ductal cells. Formal proof of this hypothesis requires the use of lineage tracing analysis. An elegant genetic lineage tracing system has recently been developed that allows the irreversible labeling of β-cells. In these mice, pINS drives the expression of a TAM-inducible form of Cre recombinase (InsCreER mice). Crossing these mice with a universal reporter line (Z/AP) allows the expression of a reporter gene encoding for the alkaline phosphatase (AP) protein on TAM administration (12). We have used this genetic system to determine the cell of origin of regenerating β-cells in pINS-cMycER^TAM mice. Unfortunately, we have not been able to achieve efficient β-cell labeling in triple-transgenic InsCreER;Z/AP;pINS-cMycER^TAM mice (data not shown), possibly because c-Myc blocks the transcription from pINS (22,23). In this scenario, c-Myc–induced inhibition of InsCreER;Z/AP impairs the expression of the AP reporter gene in β-cells.

The initiation of in vivo β-cell regeneration to repair damaged pancreatic tissues in type 1 and type 2 diabetic individuals remains a desirable and plausible, yet unrealized, therapeutic opportunity. Although growing evidence from numerous studies, including 90% pancreatectomy, pancreatic duct ligation, or induced inflammation via ectopic expression of γ-interferon (14,26,28), suggests that endocrine cells can renew upon tissue damage, the full extent of β-cell regeneration has not been assessed previously. This is in part because these models rely on global pancreas injury or administration with chemicals that also affect non–β-cells in other organs. Thus, finding a means to specifically deplete existing β-cells and to monitor subsequent β-cell regeneration in a controlled manner in vivo would be critical to elucidate a more detailed map of the latent β-cell replacement program in adult tissue.

Here we report that pINS-cMycER^TAM transgenic mice offer such a model. Repeated administration of TAM activated cMycER^TAM and resulted in the synchronous ablation of β-cells in islets throughout the pancreas. As a consequence, administered animals rapidly lost their ability to maintain glucose tolerance and became diabetic. These physiological changes were accompanied by the collapse of normal islet structures, resulting in islet remnants that were outlined by non–β-endocrine cell types unaffected by TAM administration. A previous report by Laybutt et al. (23) supports the apoptosis-promoting role of c-Myc. Constitutive activation of c-Myc in mice under control of pINS resulted in β-cell loss and postnatal death caused by hyperglycemia within 3 days after birth, a complication that prohibits the study of β-cell regeneration. In contrast, withdrawal of TAM in pINS-cMycER^TAM transgenic mice allowed gradual recovery of normoglycemia through maturation of new β-cells and restoration of islet architecture, demonstrating that the mature pancreas has the ability to fully recover from almost complete ablation of all existing β-cells.

An unexpected finding was the observation that restoration of both islet architecture and normoglycemia in pINS-cMycER^TAM transgenic mice administered with TAM gradually improved over several months. However, prolonged increase of endocrine cell mass has also been observed in other models of regeneration. Removal of 90% of pancreas tissue results in regeneration of endocrine cells, and 42% of the normal mass of β-cells are found within the pancreatic remnant 8 weeks after surgery (29). Although some of this regeneration might be caused by rapid proliferation and differentiation of ductal cells, mitotic figures of β-cells are still significantly increased 3 weeks after pancreatectomy (14,30), suggesting that regeneration occurs progressively over time. Additional support for this notion comes from another recently developed mouse model of β-cell regeneration. In these mice, the expression of diphtheria toxin, controlled by doxycycline administration, resulted in a 70–80% reduction in β-cells followed by severe hyperglycemia (31). Interestingly, withdrawal of doxycycline resulted in normalization of blood glucose levels over several months. These observations together with our results indicate that β-cell regeneration stimulated by extensive loss of existing insulin-producing cells is a relatively slow process. Furthermore, it is well known that hyperglycemia inhibits islet function and decreases expression of β-cell genes, including Pdx-1 and GLUT2 (32,33). Thus, the delayed regenerative response in pINS-cMycER^TAM transgenic mice could be explained by the inhibitive effect of hyperglycemia on the expression of genes essential for β-cell development.

Another unifying characteristic of islet regeneration is the apparent increase in general cell proliferation. Specific ablation of β-cells (27,31) or broader pancreatic injury (14,34) results in both β-cell and exocrine (including ductal tissue) cell proliferation. In agreement with these results, we also observed an increase in β-cell and duct cell proliferation in regenerating pINS-cMycER^TAM transgenic mice.

The current gold standard for cell lineage tracing in regenerating tissues is the irreversible labeling of putative progenitor cell types. Increasing evidence points to the existing β-cell as the most likely progenitor for regenerating β-cells. We have put significant effort in performing such cell lineage tracing experiments with existing transgenic mice that had previously been used for such purposes in pancreas. Unfortunately, the intrinsic mechanisms of β-cell destruction in our pINS-cMycER^TAM mice depend on strong activation of c-Myc activity, which is known to effectively block pINS. By inhibiting the activity of pINS, expression of the CreER protein required for the activation of the AP reporter is also impaired. The complete absence of AP activity is nonetheless surprising because both cMycER and CreER proteins are expressed before TAM administration and are only translocated into the nucleus on TAM binding. A possible explanation comes from the observation that even in the absence of TAM administration, a small increase in β-cell proliferation was observed in pINS-cMycER^TAM mice, suggesting precocious activity of the cMycER protein (data not shown). Thus, although our data support the notion that the major mechanism of β-cell regeneration involves expansion of the β-cells that have escaped apoptosis, we cannot unequivocally prove this point at the current time. Furthermore, the identity of putative β-cell progenitors has been controversial (35,36). Several studies have pointed to the existence of pancreatic progenitors in or associated with ducts (14,28,37). Additional pancreas cell type–specific Cre lines are currently being generated by others, including duct cell–specific Cre lines, and these could be used in future experiments to reveal the potential contribution of these cell types to regenerating β-cells.

In the future, using emerging technologies such as in vivo biophotonic imaging of cells expressing detectable markers (e.g., luciferase) may allow for more detailed monitoring of individual regenerating islets over time (38). Advances in high-resolution electron tomography will provide detailed insights into the cellular architecture of the regenerating β-cells and islets (39,40), and laser capture microscopy will provide the means for isolating and characterizing regenerating populations. From a clinical standpoint, the ultimate future success of in vivo β-cell regeneration will depend on unmasking the molecular mechanisms that underlie this process. The pINS-cMycER^TAM transgenic mouse is a highly regulated model for β-cell death and subsequent regeneration that could provide means to isolate and fully characterize “early” regenerating β-cells with the goal to understand the molecular signals that initiate and regulate β-cell expansion.

This work was supported by the Sandler Family Supporting Foundation and JDRF (Juvenile Diabetes Research Foundation) Islet Center Grant 4-2004-372. Additional support was provided by Diabetes and Endocrinology Research Center Grant P30-DK63720. I.C.R. was supported by a grant from the National Institutes of Health (5T32-A107334-13). D.A.C. was supported by a postdoctoral fellowship from the CIRM (California Institute of Regenerative Medicine).

The authors gratefully acknowledge the services provided by the University of California San Francisco Diabetes Center. We thank David Scheel, Janet Lau, Suzanne Schubbert, and Drs. Maria Wilson, Theresa Lopez, Josina Reddy, Seung Kim, and Eric Rulifson for helpful discussions and technical advice and Dr. Sapna Puri for critical reading of the manuscript. The authors also thank the Department of Developmental Biology at Stanford University for use of the confocal microscope.