Ectopic Expression of E2F1 Stimulates β-Cell Proliferation and Function

Rat insulin promoter (RIP)CreERT mice were provided by Yuval Dor (Hebrew University, Jerusalem, Israel) (8). R26E2F1 mice harboring the Rosa26-loxP-LacZ-loxP-E2F1 conditional expression cassette (24) were from Ulrike Ziebold (Max Delbrueck Center for Molecular Medicine, Berlin, Germany). Hemizygous RIPCreERT and R26E2F1 mice were crossed to double transgenic RIPCreERT × R26E2F1. Single transgenic R26E2F1 control mice and double transgenic experimental mice were littermates. Eight-week-old male double transgenic and single transgenic mice received subcutaneous tamoxifen injections (four injections of 4 mg, every other day). All procedures were performed in accordance with the Free University of Brussels Animal Studies Committee.

Rat pancreatic β-cells were isolated from 6-week-old male Wistar rats and cultured as described (25,26). Purity was always more than 90%. Mouse islets were isolated from 8-week-old BALB/c mice or transgenic mice as described before (27). BALB/c islets produced more than 75% β-cells after partial dissociation. See supplementary Methods (available in an online appendix at http://diabetes.diabetesjournals.org/content/early/2010/03/25/db09-1295/suppl/DC1) for culture. Viral transduction was as in (28). RNA and protein were extracted from freshly isolated transgenic islets.

AdE2F1, expressing human E2F1 under control of a cytomegalovirus promoter (21), and AdAkt, expressing myristoylated hemagglutinin-tagged constitutively active Akt1 (29), were gifts of, respectively, Rudiger von Harsdorf (University of Toronto, Toronto, Canada) and Mario Pende (Inserm, Université Paris 5, Paris, France). The control virus AdNull contained no transgene. All viruses were propagated and purified as described (30). Infection of β- or islet cells was as in (28).

Mouse endocrine cells or transgenic islets were used. Immunoblotting was performed as previously described (31). Antibodies were anti-E2F1 (rabbit; Santa Cruz Biotechnology Inc., Santa Cruz, CA), anti-actin (goat; Santa Cruz Biotechnology Inc.), and anti-hemagglutinin (rabbit; Clontech, Mountain View, CA).

Real-Time PCR analysis was performed using predeveloped TaqMan assay reagents (Applied Biosystems) for mouse E2f1, Ccne1 (cyclin E1), and Casp7 (caspase 7), and for human E2f1. Expression levels were normalized to the expression of the housekeeping genes Ppia (peptidylprolyl isomerase A = cyclophilin A), Gapdh, and Actb using an adaptation of the ΔΔCt method (32). Real-time PCR was further performed as described (33).

For detection of bromodeoxyuridine (BrdU) incorporation in cells, see supplementary Methods. For immunocytochemistry, primary antibodies anti-Ki67 (rabbit; Novocastra Laboratories Ltd., Newcastle Upon Tyne, U.K.), anti-E2F1 (rabbit; Santa Cruz Biotechnology Inc.), anti-hemagglutinin for detection of recombinant Akt (mouse; Cell Signaling Technology, Beverly, MA), anti–phospho-histone H3 (rabbit; Upstate Biotechnology, Waltham, MA), and anti-insulin (guinea pig; gift of Chris Van Schravendijk, Diabetes Research Center, Vrije Universiteit Brussel) underwent incubation either for 1 h at room temperature or overnight at 4°C.

Pancreatic tissue was fixed overnight in 4% formalin solution and embedded in paraffin using standard techniques. Staining for insulin, BrdU (mouse; Cappel, Cochranville, PA), Ki67, and activated caspase-3 (rabbit; Cell Signaling Technology) was performed on 5-μm sections. Proliferation was assessed in cells costaining for insulin and BrdU or Ki67 on sections from transgenic mice intraperitoneally injected with BrdU (50 mg/kg) 16 h before killing. At least 3,000 insulin-positive cells were analyzed for each animal. Visualization and imaging are described in supplementary Methods.

Total β-cell numbers in 96-well plates were determined using the CyQuant NF Cell Proliferation Assay Kit (Invitrogen), based on measurement of cellular DNA content via fluorescent dye binding, according to the manufacturer's instructions.

To assay cell death, Hoechst 33342 (10 μg/ml) and propidium iodide (10 μg/ml) were added to the cultures. Dead, apoptotic, or living cells were identified as described (34). Three separate wells were examined (at least 400 cells per well) for each experimental condition, and the percentage of dead and living β-cells was calculated.

For cell cycle analysis and sorting of live G₀/G₁- and S/G₂/M-phase β-cells, see supplementary Methods.

Insulin content of β-cells cultured in 96-well plates, fluorescence-activated cell sorter (FACS)–sorted β-cells, or isolated pancreas of transgenic mice was determined by radioimmunoassay (33).

β-cell mass was determined on the basis of relative insulin-positive area, as measured in sections, and by pancreas weight (33). 4.7 ± 0.1% of the total pancreas volume was analyzed for each mouse.

Insulin release by AdE2F1- and AdNull-transduced rat β-cells or by islets from transgenic mice (20 freshly isolated islets per condition) was measured in static incubations. Insulin biosynthesis was measured in adenovirus-transduced rat β-cells cultured in Ham F10 medium containing 10 mmol/l glucose. For more detail, see supplementary Methods.

All data were expressed as means ± SEM. Student t test, ANOVA (Bonferroni correction), or rank test (survival analysis) was used for comparison of data. Differences were considered statistically significant when P < 0.05.

Islets from mouse pancreas containing ≥70% β-cells were dissociated, reaggregated to small cell clusters, and cultured for 24 or 48 h after infection with adenovirus AdE2F1 (multiplicity of infection [MOI] 20) expressing human E2F1. E2F1 transcript was undetectable by quantitative RT-PCR (cycle threshold [Ct] > 40, n = 3) in AdNull-infected islet cells but clearly expressed 24 h after AdE2F1 infection (Ct 21.1 ± 0.4, n = 3). E2F1 protein could be detected among immunoblotted proteins extracted from AdE2F1-infected islet cells after 24 h of culture (Fig. 1,A). When cultured for 24 and 48 h, E2F1-transduced islet cells showed a significant increase in the abundance of endogenous transcripts encoding E2F1, cyclin E1, and caspase-7, all known E2F1 targets (Fig. 1 B).

When single rat β-cells (purity ≥90%) were cultured in poly-d-lysine–coated wells and infected with AdE2F1 (MOI 20), 38 ± 3% of cells immunostained positive for E2F1 protein compared with 1.3 ± 0.5% in uninfected cultures (Fig. 1 C). These results showed that infection of murine β-cells with adenovirus AdE2F1 increased the expression of active E2F1 transcription factor.

The effects of active E2F1 on β-cell proliferation were studied on isolated, single rat β-cells cultured in poly-d-lysine–coated wells. AdE2F1 (MOI 0–20) dose-dependently increased the number of BrdU-incorporating cells at 48 h after infection (Fig. 2,A). Upon infection with AdE2F1 (MOI 20), 20 ± 2% of cells were BrdU⁺ and 42 ± 3% expressed the proliferation marker Ki67 (Fig. 2,B). E2F1 was detected in 79 ± 3% (mean ± SEM, n = 9) of BrdU⁺ cells and 78 ± 6% (n = 3) of the Ki67⁺ cells (not shown), indicating that increased E2F1 expression induced proliferation of primary β-cells. Staining with propidium iodide (labels nuclei of dead cells) revealed that E2F1 expression also caused cell death (Fig. 2,A) through induction of apoptosis (not shown). Cell cycle analysis by FACS using propidium iodide on cells that were fixed to label all nuclei showed that infection with AdE2F1 (MOI 20) induced S-phase in 20 ± 4% (n = 3) and G₂/M-phase in 3.5 ± 0.9% of the cells at 24 h after infection (Fig. 2,C). At 48 h after infection, S-phase cells still represented 20 ± 3% (n = 3) of all β-cells, and the fraction in G₂/M-phase had increased to 9.5 ± 1.5% (Fig. 2,C). At this time point, phospho-histone H3 (P-HH3)–positive mitotic β-cells were present (Fig. 2,D). P-HH3 labeling colocalized with condensed chromatin in β-cells, some of which were in anaphase (Fig. 2 E), indicating that the E2F-activated β-cells were cycling.

Purified rat β-cells were transduced with AdAkt to express constitutively active Akt/protein kinase B (Fig. 3,A). In coinfection experiments with AdE2F1 and AdAkt, the E2F1-induced β-cell death at 48 h was nearly completely prevented (Fig. 3,B). Infection with AdAkt alone did not increase β-cell proliferation (Fig. 3,B), but its coinfection with AdE2F1 increased the E2F1-induced BrdU incorporation (Figs. 2,A and 3,B). The fraction of P-HH3⁺ β-cells also enlarged upon combined expression of E2F1 and Akt, compared with E2F1 alone (Fig. 3,C). Expression of active E2F1 and Akt resulted in a net increase of β-cell numbers in vitro. The total cell number increased by 38 and 46% at 48 h, and by 39 and 72% at 96 h after infection with AdE2F1 and AdE2F1+AdAkt, respectively (Fig. 3 D). Thus, in contrast to expression of E2F1 alone, combined expression of E2F1 and Akt continued to increase the absolute number of β-cells in vitro for more than 48 h.

The effect of proliferation on insulin stores was measured in isolated rat β-cells infected with AdE2F1 or AdE2F1+AdAkt. At 48 h after infection, total insulin content of E2F1- and E2F1+Akt-transduced cells was, respectively, 44 and 79% higher than untreated cells, whereas the insulin content per cell was similar in all conditions (Fig. 4,A and B). At 96 h, the total insulin content of E2F1-transduced cells was equal to that of untreated cells, and insulin per cell was even 30% lower. In contrast, relative to untreated β-cells the total insulin content of E2F1+Akt-transduced cells increased by 113% and insulin content per cell increased by 26% (Fig. 4 A and B). Thus, ectopic expression of active E2F1 and Akt increased β-cell numbers in vitro over extended periods of time, while preserving the cellular insulin stores.

To examine whether β-cells that are engaged in DNA synthesis also increase their insulin content, 150,000 cells/well were labeled with a short BrdU pulse (1 h) at 24 h after infection, detached, and incubated with Hoechst 33342 and propidium iodide. Live propidium iodide–negative cells were further sorted on the basis of Hoechst 33342 fluorescence reflecting DNA content (Fig. 4,C) into G₀/G₁-phase and S/G₂/M-phase cell populations, the latter representing 13 ± 5% of AdE2F+AdAkt-transduced β-cells. BrdU was incorporated in the nuclei of 52 ± 4% (n = 4) of the S/G₂/M INS⁺ sorted cells versus 12 ± 2% (n = 4) in sorted G₀/G₁ INS⁺ cells (Fig. 4,C). In addition, in the S/G₂/M population, 70 ± 6% (n = 3) of INS⁺ cells stained positive for Ki67 (Fig. 4 C). This indicated that proliferating and nonproliferating β-cells could be separated by FACS sorting.

At 24 h after transduction by E2F1 or E2F1+Akt, the insulin content of sorted S/G₂/M cells was increased by, respectively, 23 and 44%, compared with sorted G₀/G₁ cells from the same cultures (Fig. 4,D). Thus, in proliferating β-cells, the progression from G₀/G₁- to S/G₂-phase is associated with an increase of the cellular insulin content. During the culture period between 20 and 24 h after infection, E2F1+Akt-transduced and control cells synthesized equal amounts of insulin (Fig. 4,E), but the release of newly formed insulin during this culture period was significantly lower in the E2F1+Akt-transduced cells (Fig. 4 E). This suggests that in vitro proliferating cells in S-phase increased their insulin content by retaining newly formed insulin, rather than through increasing synthesis.

Next, glucose-regulated insulin secretion by in vitro proliferating β-cells was examined at 48 h after infection. Glucose-induced insulin release was blocked in cells transduced with either E2F1 or E2F1+Akt (Fig. 4 F), indicating that E2F1-induced proliferation in vitro does not support regulated insulin release.

Given the capacity of E2F1 to induce β-cell proliferation in vitro, we investigated whether in vivo expression of E2F1 in mature β-cells was sufficient to induce their proliferation.

Rosa26-loxP-LacZ-loxP-E2F1 (R26E2F1) mice were crossed with RIPCreERT mice to conditionally express human E2F1 from the Rosa26 promoter upon excision of β-galactosidase encoding sequence (24), when given tamoxifen (TAM) (8). At 8 weeks of age, double transgenic RIPCreERT^+/− × R26E2F1^+/− mice and control single transgenic R26E2F1^+/− littermates (Fig. 5,A) were injected with TAM. This switched on human E2F1 in β-cells of double transgenic mice and not in β-cells of single transgenic mice (Fig. 5,B). Two weeks later, the expression level of total (mouse + human) E2F1 in double transgenic islets was 2.4-fold that of E2F1 in single transgenic islets (supplementary Fig. 1); recombinant E2F1 protein was detected in double transgenic islets (supplementary Fig. 2). The percentage of β-cells labeled with BrdU or Ki67 was increased in double transgenic compared with single transgenic mice (Fig. 5,C and D). β-cells of double transgenic mice showed a decreased level of caspase-3 activation (apoptosis) (Fig. 5,E). The β-cell mass in double transgenic mice was slightly decreased (±12%) compared with single transgenic mice (Fig. 5,F), however, their pancreas insulin content was increased by ±70% (Fig. 5,G) and their fasting blood glucose level was decreased (6 ± 0.3 mmol/l vs. 7 ± 0.4 mmol/l, n = 8, P = 0.01). In addition, double transgenic mice displayed an improved glucose tolerance in intraperitoneal glucose tolerance test (Fig. 5,H) and secreted more insulin in response to glucose (Fig. 5,I), although no significant difference in mean body weight was observed between double transgenic and single transgenic groups (Fig. 5,J). Furthermore, islets isolated from double transgenic mice secreted more insulin than single transgenic islets when exposed to 10 mmol/l glucose, whereas both groups secreted equal amounts of insulin at 2 or 20 mmol/l glucose (Fig. 5,K). This suggests a leftward shift of the glucose response curve in E2F1-expressing islets. Together, these data show that β-cells from E2F1-expressing mice have a greater propensity to proliferate and to secrete insulin. This could yield a competitive advantage under conditions of hyperglycemia. To test this, double transgenic and single transgenic littermates were subjected to a single streptozocin (STZ) dose of 200 mg/kg to induce selective loss of β-cells, and 2 days later tamoxifen injections were given as before. Blood glucose of these mice was monitored for 1 month. Kaplan-Meier survival analysis showed that significantly more double transgenic mice survived the STZ-induced diabetes (Fig. 5,L). From day 10 onward, the random-fed blood glucose levels in double transgenic mice remained significantly lower than in the single transgenic littermates (Fig. 5 M).

Adult β-cells are growth arrested in G_0/1 and known to express high levels of cyclin-dependent kinase (CDK) inhibitors (p21, p27, and p57) as well as several members of the pocket protein family including pRb, the principal brakes of the G₁- to S-phase transition (15,16). Manipulations that activate (or feed into) the pRb pathway, such as overexpression of large T antigen or CDK4/6 and/or cyclin D1 and D2, and downregulation of CDK inhibitors have been shown to stimulate β-cell proliferation (for review see [15]). In agreement with these findings, our study shows that ectopic expression of transcription factor E2F1, a direct effector of the pRb pathway, increased β-cell proliferation in vitro and in vivo. However, this finding is not trivial. First, E2F1 function provokes negative feedback loops involving ARF/p53 and repressor E2Fs (10), whereas increased E2F1/pRb dimer formation can lead to active repression of E2F target genes (35), all of which can contribute to G₁ arrest. Our findings that several transcriptional targets (36,,–39) of E2F1, such as E2F1, cyclin E, and caspase-7, were activated in β-cells and that transduced β-cells were labeled with BrdU and Ki67 suggest that overexpression of E2F1 resulted in increased levels of transcriptionally active E2F1 and activation of S-phase (35). Second, E2F1 overexpression in primary fibroblasts does not lead to S-phase entry but instead promotes senescence and apoptosis (40,41), a general observation in nonimmortal cells (10). Overexpression of E2F1 cannot enforce S-phase entry in in vitro cultures of primary cardiomyocytes in the absence of serum. Concomitant with E2F activity, expression of an anti-apoptotic factor (e.g., Bcl-2), or stimulation by the growth factor IGF-1, is necessary to achieve DNA synthesis (21). In contrast, our results show that adenoviral delivery of E2F1 ectopically stimulated traversal of S-phase within 24 h in β-cells cultured in a minimal medium without growth factors or a feeder layer (26). Moreover, the E2F1-transduced β-cells progressively populated the G₂/M-phase between 24 and 48 h after transduction of the cells, and a nonnegligible fraction of those cells underwent mitosis as shown by confocal microscopy of β-cells with phosphorylated histone H3, a marker for initial stages of chromatin condensation in late G₂ interphase until anaphase. This illustrates that overexpression of the cell cycle effector E2F1 can overcome inhibition of proliferation of primary β-cells. It is possible that autocrine stimulation by insulin helps to render cultured β-cells permissive to proliferation by ectopic E2F1 (42).

Our in vitro data suggest that the majority of β-cells that ectopically express E2F1 undergo proliferation and ultimately may undergo cell death. In view of the increased transcription of the effector caspase-7 gene in the cultured cells, part of the β-cell death may occur through p53-independent apoptosis (10). Growth factor signaling via Akt/protein kinase B has potent antiapoptotic effects in β-cells (for review see [23,43]). Furthermore, E2F1-mediated proapoptotic functions can be suppressed specifically by the PI3K/Akt pathway (44). Indeed, adenoviral coexpression of constitutively active (myristoylated) Akt reduced the E2F1-induced β-cell death to near control levels, and in addition increased E2F1-induced β-cell proliferation. Under these conditions, the β-cell number and insulin content were clearly increased compared with untreated cultures. The increased insulin stores in proliferating cells in S-phase appeared to result from reduced insulin release rather than increased synthesis; the insulin release from proliferating β-cells was found to be unresponsive to glucose.

The phenotypes of E2F1^−/− mice, and of transgenic mice overexpressing E2F1 either in the testes or in whole body, suggested that in vivo functions of E2F1 result in suppressing proliferation and/or promoting apoptosis (24,45,46). c-Myc, like E2F1, is a potent inducer of apoptosis in vitro (47). Transgenic mice activating c-Myc in mature β-cells exhibit increased β-cell proliferation accompanied by overwhelming β-cell apoptosis, which rapidly leads to diabetes (48,49). These observations do not apply to a β-cell–specific conditional expression of E2F1, which increased their proliferation activity twofold while decreasing rather than increasing their death by apoptosis. Induced double transgenic mice did not develop hyperglycemia during a period of 3 months (results not shown), also suggesting absence of long-term β-cell destruction or dysfunction. The effects on the percentage of BrdU-positive and active caspase-positive cells were however not associated with an increased β-cell mass; on the other hand, the double transgenic mice exhibited a 70% higher insulin content, which appeared responsible for a significantly higher insulin release after an intraperitoneal glucose bolus. This in vivo responsiveness is compatible with the leftward shift of the glucose dose-response curve in isolated transgenic islets. Recent findings in E2F knockout mice have also suggested that E2F is crucial for normal glucose tolerance and β-cell insulin secretion (50). Because the E2F-induced increase in pancreatic insulin reserve was not associated with an increased number of β-cells, it can be attributed to an increased hormone content per cell. Direct proof for the latter requires direct measurements at the cellular level. Our data nevertheless strongly suggest that E2F can increase the insulin content in normal β-cells, and, more importantly, that this results in a more potent glucose-induced insulin release in vivo with a more rapid normalization of hyperglycemia.

Our in vivo data suggest that E2F1-driven cell cycle activity per se is not disruptive for insulin secretion. Whereas the mRNA expression level of recombinant human E2F1 was similar to that of endogenous E2F1 in the islets of transgenic mice used in this study (supplementary Fig. 1), the expression level of recombinant human E2F1 in AdE2F1-infected cells was >1,000-fold the endogenous murine E2F1 expression of uninfected cells (−ΔΔCt = 11.9 ± 1.1, n = 3, normalized to cyclophilin A). The in vitro death and impaired insulin release seen with AdE2F1 infection may be caused by nonphysiological E2F1 expression levels highly inducing genes that interfere with β-cell survival (casp-7, this study) and function (Kir6.2 [50]), and are not necessarily caused by the high rate of β-cell proliferation. The importance of E2F1 for β-cell function as opposed to being secondary to nonphysiological overexpression is hard to determine. On the other hand, our data, both the in vitro and in vivo, support the notion that E2F1 increases cell cycle activity of β-cells.

In summary, previous loss-of-function studies in E2F^−/− mice (14,18,50) pointed out an important role for E2F1 in controlling postnatal β-cell proliferation and function. We now report a gain-of-function study and demonstrate that ectopic expression of E2F1 in adult β-cells can increase their proliferation activity in vitro as well as in vivo. Moreover, in vivo E2F1 expression was found to significantly increase the insulin content of adult β-cells that was associated with a more potent glucose-induced insulin release and subsequent correction of hyperglycemia. We propose E2F1 as a novel therapeutic target of β-cell function as it is capable of increasing physiologically regulated insulin release through elevating the hormone content of β-cells. Obviously, targeting this cell-cycle factor to increase β-cell growth/function would need to be very much controlled, to prevent oncogenesis.

This work was supported by the VUB Research Council (H.H. and M.V.D.C.), the Research Foundation–Flanders (FWO) (D.P., H.H.), the National Institutes of Health Beta Cell Biology Consortium (H.H.), and the Institute for the Promotion of Innovation by Science and Technology in Flanders (IWT) (G.G., H.H., D.P.).

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

We thank Luc Baeyens, Gabriël Schoonjans, René De Proft, Ann Demarre, Karen Sterck, Veerle Laurysens, Jan De Jonge, and Erik Quartier for technical advice and assistance.