The Retinoblastoma Protein and Its Homolog p130 Regulate the G₁/S Transition in Pancreatic β-Cells

Homozygous mutant mice with disrupted p107 (p107^−/−) or p130 (p130^−/−) genes were previously prepared and described by Cobrinik and Lee (16,17). Both mutant genes function as null alleles (16,17). For generation of Rb/p130 double mutants, mice in which exon 19 of the Rb1 gene is flanked by loxP recombination sites (Rb^lox/lox) (18) were crossed with p130^−/− animals. The resultant Rb^lox/+;p130^+/− mice were intercrossed to generate Rb^lox/lox;p130^−/− mice. Concurrently, RIP-Cre transgenic mice (14,19) in which Cre expression is under the control of the rat insulin II promoter (14) were crossed with p130^−/− mice to generate RIP-Cre;p130^−/− animals. Crossing the compound mutants generated Rb^lox/+;p130^−/−;Cre mice, which were backcrossed again to Rb^lox/lox;p130^−/− mice to produce a β-cell conditional knockout (CKO) of Rb in p130^−/− mutant mice (Rb^lox/lox;p130^−/−;Cre referred to as Rb^CKO;p130^−/−). The breeding strategy is summarized in additional detail in the Figures. Mice that were either wild type for Rb (Rb^lox/lox or Rb^lox/+; Cre-negative) or heterozygous (Rb^lox/+; Cre-positive) on a p130^−/− background were also analyzed for comparison. All studies were performed on adult mice, 2- to 4-months of age, on a mixed C57/BL6–129 genetic background, with males and females being comparably represented. All studies were approved in advance by, and performed in compliance with, the University of Pittsburgh Institutional Animal Care and Use Committee.

Islets were isolated from adult mice as previously described (14,15). Briefly, islets were obtained from collagenase (Sigma, St. Louis, MO) digested pancreases after purification on a Histopaque (Sigma) gradient.

To analyze cell cycle progression, islets were dispersed into single cells with 0.05% trypsin-EDTA (Invitrogen, Carlsbad, CA), suspended in chilled 70% ethanol, and kept overnight at 4°C. Fixed cells were treated with 100 units/ml RNAase A (Sigma) in PBS and stained with 50 μg/ml propidium iodide (Sigma) at room temperature for 30 min (20). The cell cycle phase distribution of 10,000 cells was determined by flow cytometry using a BD LSR II flow cytometer (Becton Dickinson Biosciences, San Jose, CA). Analysis of data, including determination of gates, was performed using Modfit LT 3.0 software (Verity Software House, Topsham, ME) using gating shown in supplementary Figs. S1 and S4 (available in the online-only appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db08-0759/DC1).

Islet protein extractions were prepared as previously described (14,15). Immunoblotting was performed using primary antibodies for pRb (Pharmingen, 554136), p107 (Santa Cruz, C-18), p130 (Santa Cruz, C-17), p21 (Santa Cruz, F-5), p53 (Biomeda, V1003), phospho-p53 (Ser15) (Cell Signaling, 9286), Cre (Convance, MMS-106P), E2F1 (Santa Cruz, C-20), and E2F2 (Santa Cruz, TFE-25) and secondary antisera as previously published (14,15). Mice were genotyped by PCR using previously described primers (14,16,–18).

Measurements of fasting and postprandial blood glucose and plasma insulin levels, as well as intraperitoneal glucose tolerance testing were performed as detailed previously (14,15).

Six hours before being killed, mice were injected intraperitoneally with bromodeoxyuridine (BrdU; Sigma) (50 mg/kg body wt). Pancreata were fixed in Bouin's solution, embedded in paraffin, sectioned, and stained with insulin (Dako), BrdU (Abcam, Ab6326), Ki-67 (Neomarkers, SP6), or phosphohistone H3 (Millipore, 06-570) antibodies as previously described (14,15,20). For determination of β-cell death, pancreas sections were stained using the in situ cell death detection kit (Roche Diagnostics, Indianapolis, IN) for identification of transferase-mediated dUTP nick-end labeling (TUNEL) and insulin double-positive cells (20).

Islets isolated from wild-type (Rb^+/+;p130^+/+), Rb^lox/lox;p130^+/+, and Rb^lox/lox;p130^−/− mice were dispersed into single cells with 0.05% trypsin-EDTA and transduced with a replication-deficient recombinant adenovirus that expresses green fluorescent protein (Ad.GFP) or Cre recombinase (Ad.Cre) (a gift from Dr. Chris Newgard). Cells were transduced at a multiplicity of infection of 200 for 2 h at 37°C, plated on coverslips, and studied 72 h later. BrdU (1:1,000, Amersham) was added for the final 24 h of culture. Whole islets were also transduced and cultured for 72 h then frozen for immublot or PCR analysis.

All data are presented as means ± SE. Comparisons of multiple means were performed using ANOVA with SPSS 16.0 statistical software (Chicago, IL). P < 0.05 was considered significant.

Mice with inactivating mutations in p130 or p107 have previously been developed and characterized (16,17). These mice develop normally, are fertile, and apart from modest skeletal changes in p107 mutants, exhibit no obvious adult phenotypes. However, these previous studies did not include any assessment of β function or glucose homeostasis. The gene-targeting strategies used to disrupt the p130 or p107 genes (16,17) are shown in Fig. 1,A and supplementary Fig. S1A (available in the online-only appendix). Mutant mice were identified by PCR analysis of genomic DNA and confirmed as nulls by immoblotting for p130 or p107 in isolated islets (Fig. 1,B; supplementary Fig. S1B, available in the online-only appendix). Metabolic analysis of adult p130- or p107-deficient mice revealed a completely normal phenotype, similar to that seen in mice with Rb-null islets. Specifically, body weight, fasting and postprandial blood glucose, and plasma insulin values were entirely normal compared with heterozygous and wild-type littermates (Fig. 1,C–F; supplementary Fig. S1C–F, available in the online-only appendix). Glucose tolerance was also normal in mice lacking p130 or p107, compared with wild-type littermates (Fig. 1 D; supplementary Fig. S1D, available in the online-only appendix).

Consistent with normal glucose tolerance in the mutants, pancreas and β-cell mass were also normal in both p130- and p107-null animals (Fig. 2,A; supplementary Fig. S2A, available in the online-only appendix). Histomorphometric analysis of whole pancreas sections immunostained for insulin revealed that, as with pRb (14), p130 and p107 are also entirely dispensable for normal pancreas and β-cell growth (Fig. 2 B; supplementary Fig. S2B, available in the online-only appendix).

Like pRb, p130 and p107 also regulate cell cycle progression (9). Adult differentiated β-cells that have exited the cell cycle and entered a state of quiescence express both p130 and p107 (Fig. 3,B) (14,20). Therefore, we speculated that loss of p130 or p107 in β-cells might result in aberrant cell cycle reentry. Instead, loss of either p130 or p107 failed to affect β-cell replication rates (Fig. 3,A; supplementary Fig. S4A, available in the online-only appendix) and islet cell cycle phase distribution (supplementary Figs. S1 and S4B, available in the online-only appendix). This may result from functional compensation within the retinoblastoma family because there was a striking upregulation of p107 protein in p130-null islets (82.4 ± 4.0%) (Fig. 3 B). These findings demonstrate that p130 and p107 are individually dispensable for maintaining β-cell cycle arrest in adult mouse β-cells.

The preceding data suggest that in the absence of an individual pocket protein, functional redundancy among the remaining members may be sufficient to maintain cell cycle arrest in β-cells. Therefore, we generated mutant mice in which β-cells were deficient for both pRb as well as p130. Inactivation of both the Rb and p130 genes was selected because disruption of both p130 and p107 is embryonic lethal (17) and because a previous study reported that one out of 15 Rb^+/−p130^−/− mutant mice developed an insulinoma, an effect not observed in Rb^+/−p107^−/− mutant mice (21). To disrupt both Rb and p130 simultaneously, β-cell–specific pRb CKO (Rb^CKO) mice (Fig. 4,A) (14) were crossed with p130-null mice to generate compound Rb^CKO;p130^−/− double-knockout mutant mice (Fig. 4,B). In addition, mice that were wild type or heterozygous for Rb on a p130^−/− background (Rb^WT;p130^−/− and Rb^HT;p130^−/−, respectively) were also analyzed for comparison. PCR analysis of genomic DNA was used for identification of double-mutant animals based on the presence of floxed Rb alleles (283-bp), the Cre transgene, and the 320-bp mutant p130 gene (Fig. 4,C). Immunoblot analysis of pRb and p130 revealed only faint residual levels of pRb, likely reflecting expression of pRb in the few non–β-cells of isolated islets (fibroblasts, α-cells, endothelial cells, etc.) and absent p130 protein in extracts of isolated Rb^CKO;p130^−/− islets (Fig. 4 D).

The percentage of replicating β-cells was determined by immunofluorescent staining of Rb^WT;p130^−/− and Rb^CKO;p130^−/− pancreatic sections for both insulin and Ki67. Loss of both pRb and p130 resulted in an 11-fold increase in the percentage of β-cells expressing Ki67 compared with control β-cells (3.2 ± 0.6% in Rb^CKO;p130^−/− vs. 0.3 ± 0.1 in Rb^WT;p130^−/−, P < 0.05) (Fig. 5,A). β-Cell entry into S phase (BrdU⁺) also increased dramatically (14-fold) in the absence of both pRb and p130 (3.8 ± 0.7% in Rb^CKO;p130^−/− vs. 0.3 ± 0.1% in Rb^WT;p130^−/−, P = 0.004) (Fig. 5,B). Finally, there was a 30-fold increase in the number of β-cells entering mitosis as assessed using phosphorylation of histone H3 (PHH3) on serine 10 (Fig. 5 C).

To determine if increased β-cell death was associated with the accelerated cell division seen in Rb/p130 double-mutant β-cells, we examined apoptotic β-cell death (Fig. 5,D). In contrast to the very low number of apoptotic β-cells in control pancreata (0.05 ± 0.04%), β-cell death was strikingly increased in Rb/p130-deficient β-cells (1.8 ± 0.2%, P = 0.001 vs. controls) (Fig. 5,D). The mechanism underlying the increased β-cell apoptosis appears to involve a p53-dependent pathway because phosphorylation of p53 protein was increased in Rb^CKO;p130^−/− β-cells, as was protein expression of the p53 target gene, p21^CIP1 (Fig. 5 E).

In contrast to mice that lack pRB, p130, or p107 individually in their β-cells, adult double-mutant mice had significantly smaller pancreata than controls (P = 0.001; Fig. 6,B). This almost 50% reduction in pancreas mass remained significant even when adjusted for the approximate 20% reduction in body weight (P < 0.05; Fig. 6,A and B, Fig. 7,A). However, exocrine cell proliferation (BrdU⁺) was not altered in Rb^CKO;p130^−/− animals compared with controls (0.03 ± 0.05% vs. 0.04 ± 0.04%, n = 3, P = ns). In addition, there was no change in exocrine cell death (TUNEL⁺) in Rb^CKO;p130^−/− pancreata compared with control pancreata (0.01 ± 0.01% vs. 0.01 ± 0.02%, n = 3, P = ns). The reduction in pancreas mass was accompanied by a significant reduction in β-cell mass, even when adjusted for body mass (Fig. 6,C; P = 0.04 vs. Rb^WT;p130^−/−). However, the ratio of β-cell area to total pancreas area was not reduced in double mutants, suggesting a proportional reduction in both exocrine and endocrine pancreatic mass (Fig. 6 C).

Intraperitoneal glucose tolerance tests revealed that double mutants respond to a glucose challenge as efficiently as controls (Fig. 7,B). However, random sampling of postprandial blood glucose levels revealed Rb^CKO;p130^−/− mice were mildly hyperglycemic (P < 0.05) compared with controls, whereas fasting blood glucose levels were similar (Fig. 7,C). The postprandial hyperglycemia occurred despite normal levels of circulating plasma insulin in Rb^CKO;p130^−/− mice (Fig. 7 D). Thus, despite a statistically significant reduction in β-cell mass, the slightly dwarfed Rb^CKO;p130^−/− mice contained a body weight–adjusted β-cell mass and function that allows them to maintain near-normal glucose homeostasis largely in the normal range.

Off-target expression of the RIP promoter in the CNS is well described and conceivably might account for an islet phenotype. Thus, to determine whether the increase in β-cell proliferation in the double-mutant mice resulted from direct (β-cell) or indirect (CNS) effects of pRb/p130 loss, we disrupted the Rb gene ex vivo in both wild-type and p130^−/− β-cells using the Cre-LoxP method of gene recombination. Isolated β-cells from Rb^lox/lox;p130^+/+ or Rb^lox/lox;p130^−/− animals were transduced with an adenovirus expressing Cre recombinase (Ad.Cre). Expression of Cre in β-cell nuclei could be detected within 24 h after infection (Fig. 8,A). This resulted in efficient deletion of exon 19 at 72 h after transduction (Fig. 8,B). Cre-mediated deletion of pRb resulted in a reduction in pRb protein levels, associated with an increase in both p130 and p107 in Rb^lox/lox;p130^+/+ islets (Fig. 8 C). Rb^lox/lox;p130^−/− islets displayed upregulated p107 expression. This was further increased after deletion of pRb. Loss of both pRb and p130 resulted in elevated E2F2, but not E2F1, as assessed by immunoblot.

Most importantly, in agreement with the results obtained from pRb/p130 mutant mice in vivo, disruption of both pRb and p130 in β-cells ex vivo also caused a striking 19-fold increase in BrdU-incorporation (31.1 ± 5.2% Ad.Cre vs. 1.6 ± 1.3% Ad.GFP, P < 0.05) (Fig. 8,D and E). Ad.Cre transduction of wild-type or Rb^+/+;p130^−/− islets had no effect on replication (Fig. 8 D and E).

This study demonstrates that individual inactivation of the pRb homologs, p130 or p107, like pRb itself (14), is insufficient to cause abnormal β-cell cycle progression or impair the function of these cells in vivo. In contrast, we observe that simultaneous loss of two pocket proteins (pRb and p130) in the pancreatic β-cell results in strikingly increased rates of β-cell replication. This cell cycle acceleration is also accompanied by increased apoptosis in β-cells. Genetic disruption of multiple genes encoding cell cycle regulators in mice has demonstrated key roles for several proteins upstream of the pRb family (22,23). The final common pathway for enforcing cell cycle arrest has traditionally been believed to be pRb. Thus, it was surprising to learn that disruption of the Rb gene in the pancreatic β-cell is not sufficient to cause cell cycle progression (14). Based on these unanticipated findings, we sought to examine whether the two remaining pocket proteins, p130 or p107, might play a regulatory role in β-cell proliferation. Interestingly, like mice with pRb-null islets, adult p130^−/− or p107^−/− mice also displayed normal glucose homeostasis as well as quantitative β-cell histomorphometry and replication rates. From these studies, it appeared that individual pocket proteins are entirely dispensable for maintaining β-cell cycle arrest, or that the individual loss of pRb, p130, or p107 is functionally compensated by another pocket protein family member. These findings are consistent with broad compensation among the pocket protein family members (3,9,10).

Studies on mice lacking individual pocket proteins may fail to highlight the significance of any one particular family member because of complementary and overlapping roles. Perhaps the best example of this is the Rb-null mouse that does not develop retinoblastoma unless either p130 or p107 are also removed (21). In this study, we found an upregulation of p107 protein in p130-null islets (Fig. 3 B and 8C), which has also been observed in other Rb- (1,24) or p130-null (21,25) cell types. Increased levels of p107 have also been demonstrated to be responsible for enforcing cell cycle arrest in Rb-null MEFs because downregulation of p107 expression by siRNA molecules resulted in increased proliferation (24). Therefore, we speculated that in our model of unperturbed normal β-cell growth, individual or single loss of pRb, p130, or p107 may be compensated for by the redundant antiproliferative effects of the other two pocket proteins. Interestingly, in islets from p107^−/− mice, neither pRb or p130 levels were upregulated. Instead, we observed reduced levels of pRb in isolated p107^−/− islets that has also been observed in p107^−/− adult adipocyte precursors (26). The authors of that report implicated a role for the differing levels of pRb and p107 in preadipocyte differentiation. However, what role these proteins play in β-cell differentiation from their progenitors is unknown and warrants further investigation.

Several lines of evidence have suggested that overlapping functional roles for pocket proteins can be revealed when mice combinatorially lack multiple pocket protein members (25,,,–29). Double-mutant mice with tissue-specific Rb/p130 or Rb/p107 inactivation have helped uncover redundancy among family members in specific cell types such as keratinocytes (27), cardiac myocytes (28), intestinal epithelial cells (29), and others. In fact, some tissues display more aggressive phenotypes depending on which combination of pocket proteins is inactivated. For example, intestinal epithelial cells revealed a significantly more robust phenotype in Rb/p130 deficient animals than in Rb/p107 double-mutant animals (29). Further, studies from doubly deficient Rb/p130 and Rb/p107 mice revealed that the tumor spectrum in these mice does not entirely overlap (21). This is apparently true for the pancreatic β-cell as well because insulinomas arose only in Rb^+/−p130^−/− but not Rb^+/−p107^−/− mice (21). Indeed, we found significantly elevated levels of β-cell replication were present in adult mice when both Rb and p130 genes were fully inactivated.

Increases in cell cycle progression were observed using three different markers of replication (BrdU, Ki67, PHH3). Further, mutant β-cells were also capable of progressing into the late G2–early mitotic phase of the cell cycle as assessed using PHH3. The increase in β-cell proliferation was accompanied by parallel increases in the rate of cell death. This increase in cell death appeared to reflect activation of the p53 pathway because active phospho-p53 was elevated and because p21, downstream of p53, was also elevated. In agreement with these results, combined inactivation of p53 and Rb (p53^+/−Rb^+/− and p53^−/−Rb^+/−) was found to be sufficient for the development of insulinomas (30,31). The widespread apoptosis detected in pRb/p130-deficient β-cells has not been observed in other cell types deficient in pRb and p130 such as the myocardium (28) or intestinal epithelium (29). Whether loss of both pRb and p107 in the β-cell would similarly increase both cell cycle entry and apoptosis is in need of investigation. Studying β-cells deficient in all three pocket proteins will be required to reveal the collective role played by these proteins in regulating β-cell cycle progression. Unfortunately, simultaneous loss of both p130 and p107 in mice results in neonatal lethality (17), so inactivation of all three members will require the generation of a triple–conditionally null mouse.

Pancreatic histology in the double-mutant mice, both endocrine and exocrine, was interesting. In the endocrine compartment, the β-cell mass was reduced as compared with body weight, presumably reflecting a mismatch in the rates of β-cell replication and cell death. In the exocrine compartment, total pancreas weight was also decreased (Fig. 6,B), and this was disproportionate to the decline in body weight (Fig. 7 A). The reasons for the decline in body weight and pancreatic weight are unclear, but both contrasted to liver, spleen, and kidney weights, which were comparable in all three groups of animals (supplementary Fig. S4, available in the online-only appendix). Pancreatic acinar cell proliferation and apoptosis rates were normal. We speculate that the decline in body weight may have resulted from development of pituitary or hypothalamic tumors, which are known to occur with pRb loss (32). This may well have occurred in our double-mutant mice, which were globally deficient for p130 and may have been pituitary/hypothalamic deficient for pRb as a result of “leaky” expression of RIP-Cre (14,33,34).

These considerations raise the possibility that the marked increases in β-cell replication and cell death might also have resulted from CNS misexpression of RIP-Cre. Thus, we performed ex vivo studies in isolated islets using adenoviral delivery of Cre recombinase (24). These studies revealed that combined pRb and p130 loss directly in β-cells led to markedly increased β-cell replication. In agreement with our previously published report on in vivo pRb deletion in β-cells (14), we again observed no increase in proliferation after in vitro deletion of pRb alone. However, unlike the in vivo studies, we did observe a compensatory increase in both p130 and p107 levels after acute pRb ablation. On the other hand, when pRb was reduced in β-cells already deficient for p130, we observed a dramatic increase in β-cell proliferation. The increased proliferation was associated with elevated E2F2 but not E2F1 levels. E2F2 is a member of the transcriptional activator E2Fs and has been shown, along with E2F1, to play an essential role in postnatal pancreas development (35,36).

In conclusion, this study demonstrates that both Rb and p130 in combination are essential for normal pancreatic β-cell cycle control. Elucidation of therapeutic approaches that permit the stimulation of β-cell replication without induction of apoptosis may provide an approach to expand β-cells for the treatment of diabetes.

G.H. is supported by a Juvenile Diabetes Research Foundation Fellowship Award (3–2008–35). A.F.S. is supported by National Institutes of Health Grant R-01 DK-55023.

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

We thank the Don and Arleen Wagner and the Pam and Scott Kroh family foundations. We also thank Taylor Rosa and Karen Selk for assistance with the insulin radioimmunoassay and Adolfo Garcia-Ocaña and Irene Cozar-Castellano for their insightful critiques during the preparation of these studies. Finally, we thank Anton Berns for providing the Rb^lox/lox mice.