Adaptive β-Cell Proliferation Is Severely Restricted With Advanced Age

All experiments with mice were performed in the animal facility at The Children's Hospital of Philadelphia according to the guidelines of the Institutional Animal Care and Use Committee (IACUC). The first set of experiments included male F1 hybrid B6129SF1/J mice (stock 101043), obtained at 1 and 8 months of age from The Jackson Laboratory (Bar Harbor, ME). The Jackson B6129SF1/J hybrid is the product of an intercross between C57BL/6J (000664) female mice and 129S1/SvImJ (002448) male mice from The Jackson Laboratory's commercial colonies. The second set of experiments used male F1 hybrid B6129SF1/J mice obtained at 1 month of age from the Taconic Farms (Hudson, NY). The Taconic B6129F1 hybrid model is the product of an intercross between C57BL/6NTac (black 6) female mice and 129S6/SvEvTac (129S6) male mice from Taconic commercial colonies.

Partial pancreatectomy was performed, as previously reported (9). The splenic portion of the pancreas was surgically removed, resulting in an ∼50% pancreatectomy. A sham operation was performed by opening the abdomen but leaving the pancreas intact.

Mice received five daily injections of low-dose (30 mg/kg) streptozotocin using established protocols (9,18,,–21). Exendin-4 was administered as previously described (9). Mice were injected daily with 24 nmol/kg body wt in the subcutaneous space daily for 21 days.

All results are reported as means ± SE for equivalent groups. Results were compared with independent Student's t tests (unpaired and two-tailed) reported as P values.

Mice were continuously labeled with 5-bromo-2′-deoxyuridine (BrdU) for 14 days after partial pancreatectomy or low-dose streptozotocin or for 21 days during exendin-4 treatment before they were killed. BrdU was administered in drinking water at 1 mg/ml as previously described (9). Paraformaldehyde-fixed, paraffin-embedded sections were stained with DAPI/insulin/BrdU or DAPI/insulin/BrdU/Ki67 as previously described (9). Images were acquired from 10–20 islets per animal and condition, which represented 900–4,000 β-cells per animal. To measure Ki67, images were acquired from more than 100–200 islets per animal and condition, which represented 4,000–8,000 β-cells per animal. Acinar cell proliferation analysis was performed by acquiring 10 random images per animal as previously described (17). In all but one group, the results represent the average from four to seven animals (the partial pancreatectomy cohort at 19 months included only three mice). β-Cell proliferation was calculated per day by dividing BrdU-positive β-cells by the labeling period (14 or 21 days) and expressed as percent total per day.

β-Cell area was quantified as previously described (9). β-Cell area was measured in all three groups of the partial pancreatectomy experiment: sham-operated, pancreatectomy-removed (the splenic portion), and residual (the duodenal portion) sections.

Partial pancreatectomy or sham operation was performed on mice of various ages. Five days later, islets were isolated from pools of at least four mice and processed into cDNA as previously described (22). Real-time quantitative dual fluorescent–labeled fluorescence resonance energy transfer (FRET) PCR (50°C for 2 min and 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min) was performed with the ABI 7900 real-time PCR thermal cycler (Applied Biosystems, Foster City, CA) to amplify triplicate samples, comparing sample values with islet log₁₀dilution curves. Relative gene product amounts were reported for each gene compared with cyclophillin and confirmed in separate studies with a large panel of control genes (for primer sequences, see supplemental Table 1, available in an online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db08-1198/DC1). Results are reported from a single experiment with technical repeats averaged and reported as means ± SEM. Representative RT-PCR data were confirmed in an independent experiment with pools of islet mRNAs derived from equivalent groups of mice.

Cells were lysed in Tween-20 buffer. Proteins were resolved on denaturing polyacrylamide gels, electrophoretically transferred to polyvinylidine fluoride (PVDF) membranes (Immobilon-P Transfer Membrane; Millipore, Bedford, MA), and blotted with mouse anti-cyclin D2 ab-4 (Lab Vision, Thermo Fisher Scientific, Fremont, CA) and α-tubulin (Santa Cruz Biotechnology, Santa Cruz, CA).

To test whether adaptive β-cell proliferation is restricted with age, we chose 50% partial pancreatectomy, a robust model to induce β-cell proliferation (9). Partial pancreatectomy modestly reduces β-cell mass while leaving sufficient β-cell function for normal control of glucose homeostasis. As expected, partial pancreatectomy was well tolerated across a wide range of ages. Blood glucose values and postoperative weight loss were equivalent in pancreatectomized and sham-operated mice (Table 1). The resected portion of pancreas comprised one-half of the total pancreatic weight at each age-group (Table 2). Thus, splenic partial pancreactomy is a reliable and well-tolerated procedure in adult mice of all ages.

We then tested whether partial pancreatectomy stimulates β-cell regeneration. To detect all proliferation events during acute β-cell regeneration, mice continuously received BrdU in the drinking water for 2 weeks after pancreatectomy and were then killed immediately. Partial pancreatectomy stimulated β-cell proliferation in mice aged 2 months (from 0.18 ± 0.05 to 1.80 ± 0.36% per day after partial pancreatectomy; P = 0.003) (Figs. 1 and 3). Partial pancreatectomy also stimulated β-cell proliferation in mice aged 8 months compared with control mice (from 0.08 ± 0.02 to 1.06 ± 0.39% per day after partial pancreatectomy; P = 0.04) (Figs. 1 and 3). Notably, the change in total number of BrdU-positive β-cells (the absolute increase in β-cell proliferation) was less in mice aged 8 months than in mice aged 2 months. Still, our results indicate that β-cell regeneration capacity is retained well into maturity, even in mice aged 8 months.

To test the hypothesis that β-cell regeneration is restricted with advanced age, we analyzed β-cell proliferation in aged mice after partial pancreatectomy. Surprisingly, partial pancreatectomy failed to stimulate β-cell proliferation in mice aged 19 months (from 0.03 ± 0.01 to 0.03 ± 0.00% per day after partial pancreatectomy; P = 0.84) (Figs. 1 and 3). This result suggests that β-cell regeneration capacity could be very limited in aged mice.

Our data indicate that partial pancreatectomy–induced β-cell regeneration capacity is restricted in mice aged 19 months. To confirm and refine this observation, we repeated our studies with middle-aged mice. Interestingly, mice aged 12 months displayed an intermediate phenotype. Partial pancreatectomy stimulated β-cell proliferation in mice aged 12 months. However, the change in total number of BrdU-positive β-cells (the absolute increase in β-cell proliferation) was very small (from 0.03 ± 0.01 to 0.25 ± 0.09% per day after partial pancreatectomy; P = 0.03) (Figs. 1 and 3). Similar to results obtained in very old mice, partial pancreatectomy failed to substantially increase β-cell proliferation in mice aged 14 months (from 0.02 ± 0.01 to 0.05 ± 0.02% per day after partial pancreatectomy; P = 0.31) (Figs. 1 and 3). Taken together, our results illustrate that partial pancreatectomy–induced β-cell proliferation capacity is retained well into maturity. However, partial pancreatectomy–induced β-cell proliferation capacity is powerfully and abruptly restricted by early middle age (12 months of age represent ∼40% of the typical mouse life span) (23).

We then considered the possibility that partial pancreatectomy might not adequately reduce β-cell mass in aged mice. In particular, we were concerned that unequal distribution of β-cells within the pancreas could influence partial pancreatectomy–mediated β-cell proliferation. If β-cell mass grew more rapidly in the head (duodenal) than in the tail (splenic) with advanced age, removal of the splenic portion of the pancreas might not efficiently reduce β-cell mass in aged mice. We quantified β-cell mass reduction after partial pancreatectomy, measuring β-cell area and β-cell mass in pancreata of sham-operated (the entire pancreas), resected (the splenic portion), and remaining (the duodenal portion) samples in each age cohort. The total β-cell area and β-cell area mass continuously increased with age in control mice (Table 2). β-Cell area and mass also continuously increased with age in resected pancreata. Similarly, β-cell area and mass continuously increased in the remaining pancreata after partial pancreatectomy. Moreover, β-cell area and mass was roughly equivalent in the resected and residual pancreata of each age-group, indicating that partial pancreatectomy reduced β-cell mass by approximately one-half. Thus, splenic partial pancreatectomy reduces β-cell mass by equivalent amounts in young and old mice. As a result, β-cell reduction by partial pancreatectomy should be a robust stimulus of β-cell regeneration in mice across all ages.

To investigate whether regeneration of other pancreatic components is also restricted with age, we tested acinar regeneration after partial pancreatectomy. Confirming previous reports, partial pancreatectomy robustly stimulated occasional patches of acinar cell regeneration in very young mice (from 0.72 ± 0.08 to 1.47 ± 0.28% per day after partial pancreatectomy at 2 months of age; P = 0.04) (Table 2) (images not shown). Partial pancreatectomy also stimulated acinar cell regeneration in very old mice aged 19 months (from 0.53 ± 0.11 to 1.07 ± 0.10% per day after partial pancreatectomy; P = 0.02) (Table 2). Similarly, partial pancreatectomy stimulated acinar cell regeneration at 8 months and 14 months of age (Table 2). These results indicate that acinar replication capacity may not be influenced by age. Thus, age-dependent restriction of regeneration capacity of β-cells is not common to all pancreatic components.

We then used the β-cell toxin streptozotocin to further test the hypothesis that β-cell regeneration capacity is restricted with advanced age. Streptozotocin is a robust stimulus of β-cell regeneration when administered in multiple low doses to adult mice (24). Low-dose streptozotocin was well tolerated and did not cause hyperglycemia or extensive weight loss. As expected, streptozotocin administration stimulated β-cell proliferation in mice aged 2 months by 0.36% per day compared with controls (from 0.18 ± 0.05 to 0.55 ± 0.07% per day after streptozotocin; P = 0.003) (Figs. 2 and 3). However, streptozotocin had little effect in aged mice: β-cell proliferation was only slightly increased in mice aged 15 months compared with controls (from 0.02 ± 0.01 to 0.04 ± 0.01% per day after streptozotocin; P = 0.17) (Figs. 2 and 3).

We hypothesized that glucagon-like peptide 1 (GLP-1)–stimulated β-cell proliferation might also be restricted in aged mice. To test this corollary hypothesis, we treated mice with exendin-4, a GLP-1 agonist, using established protocols (9,18,,–21). Exendin-4 was fairly well tolerated and was not associated with severe weight loss. Exendin-4 robustly stimulated β-cell proliferation in mice aged 2 months (0.76 ± 0.07% per day with exendin-4 treatment; P = 0.00001 compared with sham-operated controls) (Figs. 2 and 3). By contrast, exendin-4 treatment resulted in little β-cell proliferation in 14-month-old mice (0.08 ± 0.03% per day with exendin treatment; P = 0.13 compared with sham-operated controls) (Figs. 2 and 3). Notably, the change in total number of BrdU-positive β-cells' (the absolute increase in β-cell proliferation) response to exendin-4 was ∼10-fold greater in young mice than in old mice. Thus, GLP-1–stimulated β-cell regeneration is also restricted in aged mice.

The observation that adaptive β-cell proliferation is severely restricted with advanced age implies that β-cells could undergo cell cycle exit as a function of age (equivalent to cellular quiescence). To further test this hypothesis, we compared basal (nonstimulated) β-cell proliferation rates in control mice across a range of ages. Confirming our previous observations (17), basal β-cell proliferation decreased from 2 to 8 months (from 0.18 ± 0.05 to 0.08 ± 0.02% per day; P = 0.008) (Figs. 1 and 3). Basal β-cell proliferation further decreased by 12 months (to 0.03 ± 0.01% per day; P = 0.03) and remained low in 14- and 19-month-old mice (P = 0.01 at 14 months compared with 2 months; P = 0.03 at 19 months compared with 2 months). Thus, basal β-cell proliferation was fully restricted by 12 months of age, which was rapidly followed by restriction of partial pancreatectomy–induced β-cell regeneration.

Our studies on aging were initially performed with F1 hybrid B6129SF1/J mice from The Jackson Laboratory. We chose this strain because it closely approximates the mixed genetic background of laboratory knockout mice (commonly derived from SV129-derived embryonic stem cells and crossed into c57B6). However, we could not rule out the possibility that our results are unique to the F1 hybrid B6129SF1/J strain from The Jackson Laboratory. Consequently, we performed additional studies on aging in a similar but separate genetic cohort, Taconic Farms F1 hybrid c57 SV129 mice. Because the Jackson and Taconic Farms lines have significantly diverged over the years (25), the Taconic Farms F1 hybrid c57 SV129 mice represent a similar but genetically distinct lineage compared with Jackson F1 hybrid B6129SF1/J mice. Consequently, the Taconic Farms F1 hybrid c57 SV129 mice should also approximate the mixed genetic background of laboratory knockout mice. As previously indicated, partial pancreatectomy was well tolerated in the Taconic cohort at 2 and 19 months of age. Partial pancreatectomy robustly induced β-cell regeneration in young Taconic mice (from 0.24 ± 0.09 to 1.33 ± 0.26% per day after partial pancreatectomy; P = 0.009) (Fig. 4). In contrast, partial pancreatectomy had no effect on β-cell proliferation in aged Taconic mice (from 0.020 ± 0.003 compared with 0.019 ± 0.003% per day after partial pancreatectomy; P = 0.84) (Fig. 4). These results independently confirm our observations in the Jackson cohort and further illustrate that partial pancreatectomy–induced β-cell regeneration becomes severely restricted in aged mice. Moreover, these additional studies indicate that our findings may be broadly applicable to genetically engineered mice, which frequently have a mixed c57 SV129 genetic background.

Our results indicate that partial pancreatectomy–induced β-cell regeneration capacity may be severely restricted with advanced age. However, we were concerned about potential BrdU toxicity, which could theoretically reduce β-cell proliferation. Continuous low-dose BrdU administration in the drinking water is generally well tolerated and does not severely reduce β-cell proliferation compared with short-term infusions of high-dose BrdU (17). However, Hellerstein and colleagues (26) recently reported that continuous BrdU administration reduces proliferation of islet components by ∼25%, as measured by heavy water incorporation into DNA. Consequently, we tested for BrdU-associated toxicity by assessing β-cell proliferation in our cohort, as measured by the presence of Ki67. We administered BrdU in the drinking water or control water to a cohort of 1-month-old mice for 2 weeks, after which they were killed. Reassuringly, Ki67-positive β-cells were equivalent in BrdU-treated and untreated pancreata (2.29 ± 0.31 vs. 2.51 ± 0.25%, respectively; P = 0.59) (Fig. 5). Thus, prolonged infusion of BrdU does not influence β-cell proliferation, as measured by Ki67 expression. This result indicates that limited β-cell regeneration in aged mice cannot be readily explained by BrdU toxicity to proliferating β-cells.

To further characterize the role of aging in adaptive β-cell proliferation capacity, we compared gene expression in islets from young (2 months) and aged (14 months) mice. Cyclin D2, the most abundant cyclin in islets, did not change expression with age (supplementalFig. 1,A). Similarly, cyclin D2 protein expression was unaltered in aged islets compared with young islets (supplementalFig. 1,B). Although expression of many cyclins did not change, a few of the weakly expressing cyclins such as D3, E1, and E2 actually increased with age (supplementalFig. 1 A). Notably, islets from aged mice had greater basal expression of cell cycle inhibitors, including p21^Cip1, p27^Kip1, p16^Ink4a, p15^Ink4b, p15^Ink4b, and Rb (supplemental Fig. 1A). p16^Ink4a is differentially expressed in aging tissues and has been shown to restrict islet growth (27,28). Thus, differential expression of negative regulators of cell cycle such as p16^Ink4a or p15^Ink4b could potentially explain the age-dependent decline in adaptive β-cell proliferation capacity. Taken together, these results confirm the observation that partial pancreatectomy–stimulated β-cell proliferation is severely restricted in islets from aged mice.

We observe that β-cell regeneration is severely and abruptly restricted by middle age in our cohort of mice. Fifty percent partial pancreatectomy stimulated a massive amount of β-cell proliferation in young mice. However, partial pancreatectomy had little effect on β-cell proliferation in aged mice. Similarly, the β-cell toxin streptozotocin greatly increased β-cell replication in young mice but failed to stimulate β-cell regeneration in aged mice. Moreover, β-cell proliferation was stimulated by exendin-4 in young but not in aged mice. Taken together, these results reveal that adaptive β-cell proliferation is severely restricted with advanced age.

In this study, we advance the hypothesis that age is a major factor limiting human β-cell regeneration. Basal β-cell proliferation was severely reduced as a function of age in our mice, consistent with our previous observations (17). We further this observation to show that adaptive β-cell proliferation is also severely restricted with age. In support of this concept, basal replication rates in human pancreata and cultured human islets decline with donor age (29,–31). Similarly, islets from young donors have been reported to perform better when transplanted into type 1 diabetic patients (32). Notably, type 2 diabetes is typically a disease of the elderly, most commonly diagnosed during middle age or beyond. Similarly, gestational diabetes mellitus is much more frequent with advanced maternal age (33,34). This would imply that human patients of advanced age could have little regenerative capacity to increase β-cell mass. Indeed, there is indirect evidence to support this concept of an age-dependent decline in β-cell regeneration. For instance, increased rates of diabetes have been reported in patients after shock-wave lithotripsy for renal stones (35). Similarly, minimal β-cell regeneration is observed in patients after partial pancreatectomy (15). Thus, β-cell regeneration could be constrained in advanced age in humans, similar to that in the aged rodents in our study. Restricted β-cell mass expansion could have severe consequences in elderly patients with type 2 diabetes, limiting compensatory β-cell mass expansion to cope with increased insulin requirements. Interestingly, several of the recently discovered risk loci for type 2 diabetes have been implicated in cell cycle control of β-cells and could theoretically influence adult β-cell mass or alter the timing of cell cycle exit of adult β-cells (36,,–39).

Discrepancies between rodent and human β-cell regeneration capacity have confounded diabetes researchers for many years. Regeneration of β-cell function in experimental animal models has been widely observed in rodents but remains elusive and controversial in humans (13,14). Notably, islets from human cadaveric donors are typically in the 4th–6th decade of life and are therefore much more mature (32). Because our results in rodents indicate that β-cell regeneration capacity declines with age, we hypothesize that young rodents may not faithfully model the regenerative capacity of mature adult human β-cells.

Our studies reveal that the regenerative capacity of adult β-cells becomes limited by early middle age (12 months of age [∼40% of the mouse life span]) (23). As such, aging β-cells may not be comparable with hematopoietic stem cells, which gradually lose replicative capacity during the normal aging process (40). Under this schema, β-cells could have a developmental program that allows them to replicate early in adulthood to match insulin secretion capacity to peripheral insulin requirements. β-Cell replication might then become fully restricted when adult insulin requirements are established in middle age.

This work was supported by a research grant from the Juvenile Diabetes Research Foundation International (to J.A.K.). Additional support was provided by the National Institutes of Health (grants K08-DK064101, R03-DK078546, and R01-DK081469), a March of Dimes Basil O'Connor Starter Scholar Research Award, a Lawson Wilkins Pediatric Endocrine Society Clinical Scholar Award, a Charles H. Hood Foundation Child Health Research Grant, funds from The Children's Hospital of Philadelphia, and a University of Pennsylvania Diabetes and Endocrinology Research Center pilot and feasibility grant (DK19525).

No potential conflicts of interest relevant to this article were reported.