Ku70 Functions in Addition to Nonhomologous End Joining in Pancreatic β-Cells: A Connection to β-Catenin Regulation

Mutant mice (mixed C57BL/6 and 129SV) maintenance has been previously described (8). Ku70^+/− mice (31) were crossed to p53^p/p mice (6) to generate Ku70^−/−p53^p/p mice. The protocols were approved by the Institutional Animal Care and Use Committee at The University of Texas MD Anderson Cancer Center.

For the glucose tolerance test, mice were fasted for 16 h and injected with glucose at 1.0 g/kg of body weight. For the insulin tolerance test, mice were fasted for 6 h and injected intraperitoneally with insulin 0.75 units/kg (Sigma Aldrich, St. Louis, MO). Glucose levels were measured with a glucose analyzer (Contour; Bayer HealthCare, Tarrytown, NY). Blood insulin levels were measured by ELISA as previously described (8).

Pancreata were fixed, embedded, sectioned, and stained as previously described (8). Slides were incubated with antibodies against insulin, glucagon, cyclin D1 and D2, and p53 (Cell Signaling, Danvers, MA); proliferating cell nuclear antigen (PCNA) (Dako, Carpinteria, CA); CDK4 and p16 (Santa Cruz Biotechnology, Santa Cruz, CA); Ku70 and bromodeoxyuridine (BrdU) (AbD Serotec, Raleigh, NC); β-catenin (BD Biosciences, San Jose, CA); or γH2AX and Ki67 (Abcam, Cambridge, MA). BrdU incorporation, immunohistochemistry, immunofluorescence, and microscopy were performed as previously described (8). Mouse embryonic fibroblasts (MEFs) were generated and stained as previously described (32).

A TUNEL assay kit (ApopTag; Millipore, Billerica, MA) was used to detect apoptosis.

Senescence-associated β-galactosidase (SA β-gal) activity was detected by a staining kit (Cell Signaling). Glucagon was detected by immunohistochemical staining. Slides were counterstained with Nuclear Fast Red (Vector Laboratories, Burlingame, CA) as previously described (8).

Pancreata sections were prepared as previously described (8). Each section was subjected to morphometric analysis with Image J software (National Institutes of Health, Bethesda, MD). Raw data were statistically analyzed as previously described (8).

Islets were isolated as described in Li et al. (33). Islets from at least three mice per genotype and age group were pooled together, and 40 μg of protein were loaded as in Tavana et al. (32). Antibodies were Ku70; β-actin (Sigma Aldrich); CDK4, p53, total β-catenin, and phosphorylated β-catenin (Cell Signaling); activated β-catenin (Millipore); p21 (BD Pharmingen, San Jose, CA); and GAPDH and tubulin (Epitomics, Burlingame, CA). Experiments were repeated three or more times.

Results are presented as the mean ± SEM. Differences were determined by a two-tailed unpaired Student t test with a 95% CI. The area under the curve (AUC) was calculated with Prism software (GraphPad Inc., La Jolla, CA). P < 0.05 was considered significant.

Previously, we demonstrated high Lig4 expression in isolated islets (8) and attributed it to protecting against spontaneous genomic insults caused by intrinsic metabolic agents. Therefore, we hypothesized that NHEJ is very active in pancreatic β-cells. To analyze the expression of Ku70, we performed a Western blot in purified wild-type and mutant pancreatic islets. The results showed that the Ku70 protein is expressed in the pancreatic islets and that the expression level slightly increases with age (Fig. 1A). Consistently, immunohistochemical staining reveals high expression of Ku70 in the islets (Fig. 1A).

To determine whether γH2AX foci, a prominent marker of DNA damage, were present in the Ku70^−/− and Ku70^−/−p53^p/p islets, we stained mutant pancreatic sections and compared them with littermate (Ku70^+/−p53^p/p) and age-matched wild-type controls. As shown in Fig. 1B, γH2AX foci and pan staining were abundantly present in samples from 6-month-old mutant mice but not in control sections. Additionally, pancreatic sections from young (1-month-old) mutant mice showed few γH2AX-positive β-cells, but none were observed in the control sections (data not shown). These results indicate that as in the Lig4^−/−p53^p/p mutant mice, the pancreatic islets from Ku70-deficient mice had persistent unrepaired DNA damage.

DNA damage from NHEJ deficiency triggered p53-mediated apoptosis in lymphocytes. We previously did not detect apoptosis in the Lig4^−/−p53^p/p islets (8), ruling out p53-independent apoptosis, but because Lig4 deficiency led to embryonic lethality (2), we could not accurately assess the role of p53-mediated apoptosis in β-cells after persistent DNA damage. Therefore, with use of the Ku70 deficiency, we originally hypothesized that apoptosis would occur in the Ku70^−/−p53^+/+ but not in the Ku70^−/−p53^p/p sections. Of note, a TUNEL assay detected no apoptosis in either the Ku70^−/− or Ku70^−/−p53^p/p islets and only very low levels in the wild-type sections compared with a Ku70^−/− spleen (Supplementary Fig. 1). Furthermore, after 10 Gy of γ irradiation, neither the wild-type nor the Ku70^−/− islets showed elevated levels of apoptosis after 24 h (data not shown). These data indicate that in the presence of DNA damage, no or very few pancreatic β-cells undergo apoptosis.

The accumulation of spontaneous unrepaired DNA damage in the Lig4^−/−p53^p/p β-cells triggered p53 to transactivate p21 (8). To quantitatively detect p53 and p21 levels in the wild-type, Ku70^−/−, and Ku70^−/−p53^p/p islets, Western blot analysis was performed on pooled isolated islets. As shown in Fig. 1C, mutant mice expressed higher levels of p53 than did wild-type mice at all ages, and the p53 expression was predominantly localized in the nucleus (Fig. 1D). Expression of p21 was also elevated in all the age groups of mutant mice compared with wild-type mice (Fig. 1C).

Activated p53-p21 promotes cellular senescence in Lig4^−/−p53^p/p mice as a mechanism to suppress tumorigenesis not only in the lymphoid system (7), but also in the pancreatic β-cells, which inhibits insulin production in many islets (8). We next asked whether the DNA damage-induced p53-p21 activation also triggered cellular senescence in the Ku70^−/− and Ku70^−/−p53^p/p islets. To better visualize islet-specific SA β-gal, pancreatic sections were stained with glucagon. As shown in Fig. 1E, mutant islets from older mice underwent cellular senescence, whereas no senescent cells were detected in the control pancreatic sections. These senescent islets were not driven by elevated p16 levels because the age-matched controls showed similar expression levels (Fig. 1F). Therefore, as in the Lig4^−/−p53^p/p islets, accumulation of spontaneous DNA damage triggered cellular senescence in the absence of Ku70.

Although persistent DNA damage activated p53-p21–induced cellular senescence, both Ku70^−/− and Ku70^−/−p53^p/p mice did not have increased blood glucose levels and were not diabetic. Measuring of the random nonfasting serum glucose levels in Ku70^−/− and Ku70^−/−p53^p/p mice revealed progressively decreased glycemia, which became significant starting with 4-month-old mutant mice; more specifically, 7-month-old Ku70^−/− and Ku70^−/−p53^p/p mice showed a 39 and 38% decrease, respectively, in blood glucose levels compared with age-matched controls (P < 0.005) (Fig. 2A). To measure the nonfasting serum insulin levels, we performed ELISAs in 2- and 7-month-old mice and found elevated circulating insulin levels in both groups of mutant mice. The Ku70^−/− and Ku70^−/−p53^p/p mice had significantly higher levels of insulin at 7 months than wild-type mice (P < 0.005), resulting in a 26–35% increase (Fig. 2B), which suggests an imbalance in pancreatic β-cell regulation.

Glucose tolerance tests were performed in young (1–2-month-old) (data not shown) and older (5-month-old) mice. Besides a lower fasting serum glucose level in mutant mice, no significant differences were seen between mutant mice and controls after glucose injection (Fig. 2C), indicating that mutant mice were able to clear glucose from the blood as efficiently as the controls. ELISA showed higher levels of secreted serum insulin on glucose stimulation in aged Ku70^−/− and Ku70^−/−p53^p/p mice (Fig. 2D), and AUC analysis indicated a significant difference (P < 0.05) (Fig. 2F). Among 3-month-old mice injected with insulin 0.75 units/kg (insulin tolerance test), AUC analysis showed no significant differences among genotypes (Fig. 2E and F), indicating normal insulin sensitivity. Collectively, the elevated random serum insulin levels and decreased glycemia suggest an increase in islet size and β-cell proliferation in these mutant mice.

To determine whether the increased serum insulin concentrations resulted from increased β-cell proliferation, we examined the morphological changes in the pancreatic islets by determining the ratio of islet area (immunohistochemical staining positive for insulin) divided by the total pancreas area (8). For comparison, we separated the mice into three age groups: young (1.0–2.0 months), medium (2.5–4.0 months), and old (4.5–7.0 months). As shown through representative pictures and quantification in Fig. 3A and B, mutant mice had a larger islet-to-total pancreas ratio than did the age-matched controls in all age groups (P < 0.05–0.005). The increase in mutant islets was independent of the total pancreas because there was no significant difference in the ratio of pancreas weight to body mass (Supplementary Fig. 2A). Immunohistochemical staining for glucagon, indicative of α-cells, did not reveal any abnormalities in either mutant backgrounds at any time point compared with age-matched controls (Supplementary Fig. 2B). These results indicate a progressive augmentation of pancreatic islet size in the absence of Ku70, which correlates with the increased insulin production and decreased glycemic phenotype.

The production of adult β-cells predominantly occurs through self-duplication of mature cells rather than through differentiation of their stem cell progenitors (11–13). Therefore, we hypothesized that the increased islet size observed in the mutant mice was caused by heightened β-cell proliferation.

Through dual Ki67 and insulin staining of pancreatic sections, Ki67-positive β-cells were counted and divided by the total number of insulin-positive β-cells to quantitatively assess the rate of β-cell proliferation. As shown in Fig. 3C and D, 3-month-old wild-type and littermate controls had no significant difference in β-cell proliferation; conversely, both Ku70^−/− and Ku70^−/−p53^p/p mice showed a significant increase in β-cell proliferation (P < 0.02). Furthermore, islet staining for PCNA, another proliferative marker, showed that the young mutant mice had a 59–63% increase of the β-cell–specific, hyperproliferative phenotype compared with wild-type mice (P < 0.001) (Fig. 3E and F). No elevated Ki67 or PCNA staining was observed in Lig4^−/−p53^p/p mutant sections (data not shown), indicating that the Ki67- or PCNA-positive cells observed in the absence of Ku70 did not result from accumulated DNA damage. Finally, we performed BrdU labeling of 1-month-old Ku70^−/−p53^p/p and Ku70^+/−p53^p/p littermate mice. As shown in Fig. 3G and H, Ku70^−/−p53^p/p mice had a significantly higher percentage of BrdU-positive cells in the islets than did the Ku70^+/−p53^p/p mice (P < 0.05). Collectively, these results strongly correlate with the decreased glycemic phenotype, indicating a relationship between Ku70 deficiency and increased β-cell proliferation.

Because of the paramount role that CDK4 plays in regulating β-cell proliferation (34,35), an examination of the expression pattern would further confirm the proliferative potential of Ku70^−/− and Ku70^−/−p53^p/p β-cells. Indeed, Western blot analysis of isolated pancreatic islets confirmed increased CDK4 protein expression in both mutant backgrounds at all time points compared with wild-type controls (Fig. 4B). Representative islet sections showed increased nuclear localization of CDK4 (Fig. 4A), corroborating the Western blot data (Fig. 4B). Furthermore, Western blot analysis demonstrated that cyclin D1 and D2 levels were slightly elevated in Ku70^−/−p53^p/p islets compared with those of littermate controls (Fig. 4C). Taken in concert with previous data, β-cells from Ku70^−/− and Ku70^−/−p53^p/p showed elevated proliferation, increasing the islet area over time and resulting in a decreased glycemic phenotype.

Canonical Wnt signaling regulates pancreatic β-cell proliferation. A hallmark for Wnt activation is the cytoplasmic accumulation of β-catenin, which eventually migrates into the nucleus to upregulate target genes like cyclin D1 and D2 and CDK4 (36). To determine the β-catenin levels in the islets, we performed immunofluorescent staining on pancreatic sections from Ku70^−/− and Ku70^−/−p53^p/p mice and littermate and wild-type controls. Islet β-catenin levels were comparable between 1-month-old mutant mice and wild-type mice (data not shown). However, older mutant mice (6–7 months) showed higher levels of β-catenin than did the wild-type and littermate controls (Fig. 4E), which had very low levels of β-catenin. This finding is consistent with reports of a normal downregulation of the Wnt pathway in the pancreata of wild-type adult mice (37,38). To corroborate the immunofluorescent data, we analyzed purified islet protein extracts from medium-aged wild-type, Ku70^−/−, and Ku70^−/−p53^p/p mice and observed an increase in both total and activated β-catenin levels in the absence of Ku70 (Fig. 4D). These results establish a potential link between Ku70 deficiency and increased β-catenin levels. To further demonstrate this link, we generated early passage MEFs and stained them for β-catenin. Although some nuclear localization was observed in wild-type MEFs, more cells showed stabilized β-catenin in the absence of Ku70^−/− (Supplementary Fig. 3A). Additionally, total β-catenin was elevated in both Ku70^−/− and Ku70^−/−p53^p/p MEFs, and activated β-catenin was elevated in Ku70^−/−p53^p/p samples compared with wild-type samples (Supplementary Fig. 3B and D). β-Catenin phosphorylation at sites Ser552 and Ser675 has been associated with enhanced β-catenin stabilization and signaling activity (39). As shown in Supplementary Fig. 3B, Ku70^−/−p53^p/p early passage MEFs had elevated phosphorylated β-catenin at these sites compared with wild-type MEFs. Conversely, β-catenin phosphorylation at Ser31, Ser37, Thr41, and Thr45 all indicated the initial steps of β-catenin degradation (39). As shown in Supplementary Fig. 3B and D, there was no difference in these phosphorylation sites, indicating that Ku70 did not affect the degradation of β-catenin. Furthermore, downstream Wnt transcriptional targets cyclin D2 and CDK4 were elevated in Ku70^−/− and Ku70^−/−p53^p/p MEFs compared with wild-type MEFs (Supplementary Fig. 3C); a similar pattern was seen in the mutant islets (Fig. 4B and C). Collectively, the data suggest that Ku70 can negatively regulate Wnt signaling and that Ku70 deletion stabilizes β-catenin and increases its activity. Finally, we generated Lig4^−/−Ku70^−/−p53^p/p mice to assess whether Ku70 loss could rescue the overt diabetes phenotype previously observed in Lig4^−/−p53^p/p mice. Triple-mutant mice were produced below the expected Mendelian ratio and killed early because of health problems. Nevertheless, by measuring the random blood glucose of 1-month-old mice, we observed a decrease (19%) in the blood glucose of Lig4^−/−Ku70^−/−p53^p/p mice compared with Lig4^−/−p53^p/p mice (Supplementary Fig. 4), showing that Ku70 loss could increase the proliferative capacity of pancreatic β-cells in the absence of Lig4.

In summary, the findings show that the phenotype of Ku70^−/−p53^p/p pancreatic islets resembles that of Lig4^−/−p53^p/p islets in the presence of DNA damage and cellular senescence, indicating that Ku70 plays a role in NHEJ and in maintaining genomic integrity. A remarkable finding was that Ku70^−/−p53^p/p mice were not diabetic, which contrasts with the diabetic Lig4^−/−p53^p/p mice of our previous study and indicates an NHEJ-independent role for Ku70. The data demonstrate that Ku70 prevents cellular proliferation in the presence of DNA damage, possibly through inhibition of the Wnt signaling pathway. β-Catenin regulation is essential for controlling the proliferation in pancreatic β-cells; therefore, the data reveal an important direction for future drug intervention for the treatment of certain types of diabetes.

In this study, we show that Ku70, which is best-known for its role in DNA end recognition during NHEJ, has an important function in regulating β-cell replication. More specifically, Ku70 deficiency augmented islet mass, which increased insulin production and, thus, progressively decreased glycemia. Analysis revealed a stabilization of β-catenin, which was complemented by increased markers for β-cell proliferation, Ki67, PCNA, BrdU incorporation, cyclin D1 and D2, and CDK4. This finding was in stark contrast to our previous study of Lig4^−/−p53^p/p mice, where DNA damage-induced islet senescence led to β-cell attrition and decreased insulin production, deregulating glucose homeostasis and resulting in severe diabetes (8). The similarities of these phenotypes converge on the deficiency in NHEJ, where spontaneous DNA damage progressively accumulates and induces a p53-mediated response, halting the β-cell cycle. Therefore, the most obvious difference between these phenotypes is present early in the Ku70-deficient mice, where the β-cells showed increased proliferation and escaped cell cycle inhibition. These results allow us to dissociate the function of Ku70 in the β-cell from solely a role in the conventional NHEJ-dependent DNA repair process as seen in the deficiency of Lig4. Our proposed model is depicted in Fig. 5.

Complementing the increased β-cell proliferation data, the results also show an early and steady elevation in CDK4 levels (Fig. 4). This is not surprising considering that CDK4 regulation is important for controlling β-cell proliferation (34,35,40). Of note, a constitutively active CDK4 mutant, CDK4^R24C, presented a stark increase in β-cell proliferation and augmented islet mass, leading to islet hyperplasia (34,41). This increased islet proliferation was enough to rescue previously established diabetic models (42), collectively placing CDK4 as a central β-cell regulator. Furthermore, Rane and colleagues (43) revealed that CDK4 not only promotes proliferation in quiescent adult β-cells, but also activates early β-cell progenitors in the ductal epithelium. Further elevated CDK4 levels in the embryonic pancreas, through the CDK4^R24C mutant, showed increased proliferation of early mesenchymal endocrine precursors (44). These studies underscore a pivotal role for CDK4 during embryogenesis: the regulation of pancreatic progenitors and adult β-cell regeneration by controlling the β-cell cycle. Taken together, it is tempting to speculate that the elevated CDK4 expression seen in the Ku70^−/− and Ku70^−/−p53^p/p islets contribute to increased adult β-cell proliferation.

Many reports have linked Ku70 to transcriptional regulation either directly through binding to DNA or indirectly through interacting with other proteins (26–30,45,46). Of particular interest to the current study, researchers have speculated that Ku70 helps to regulate Wnt signaling after DNA damage occurs (30). The Wnt signaling pathway has been linked to glucose homeostasis; thus, overactive Wnt signaling could drive β-cell proliferation and increase insulin secretion (15). Islet-specific activation of β-catenin has been shown to increase β-cell proliferation and elevated insulin levels (17,18), whereas β-catenin inactivation results in islet hypoplasia (20). Disrupting β-catenin activity by deleting the Wnt coreceptor LRP5 greatly impairs β-cell proliferation (47). These studies emphasize the importance of β-catenin and Wnt signaling in controlling β-cell proliferation.

Researchers have identified Ku70 as a novel inhibitor of the β-catenin/TCF4 complex; more specifically, Ku70 physically binds to and inhibits TCF4 activation (30). Furthermore, when Ku70 was abrogated, canonical Wnt signaling increased. Conversely, after DNA was damaged, Ku70 levels increased, and TCF4 was deactivated as evidenced by a decrease in putative downstream targets. Finally, this regulation was completely independent of Ku80 levels (30). These findings may explain the present data shown in Fig. 4 and Supplementary Fig. 3. It is commonly accepted that Ku70^−/− MEFs prematurely senesce because of an accumulation of DNA damage and subsequent activation of the DNA damage response (31,48). Yet in early passage MEFs, deleting Ku70 actually heightens proliferation compared with wild-type MEFs as well as rescues growth defects observed in telomere shelterin complex–deficient mice (49,50). These data combined with the present findings suggest that initial activation of Wnt signaling occurs in the absence of Ku70, which increases cell proliferation until the DNA damage accumulation becomes too excessive and prematurely activates p53-mediated cell cycle arrest, preventing tumorigenesis. Of interest, similar to many models of overactivated Wnt signaling, both Ku70^−/− and Ku70^−/−p53^p/p mice have shown stabilized β-catenin in the epithelial cells of the colon accompanied with high-grade dysplasia and adenocarcinoma (N.P.-O. and C.Z., unpublished data). Investigation is currently ongoing to elucidate the precise mechanism of Ku70 in suppressing canonical Wnt signaling.

In summary, this study highlights an unexpected and important function of Ku70: In addition to repairing DNA damage in the NHEJ pathway, Ku70 also regulates pancreatic β-cell proliferation. The connection between the Ku70 level and stabilization of β-catenin merits further investigation, which could shed light on possible ways to enhance pancreatic β-cell proliferation.

This study was financially supported in part by Robert Everett “R.E.” (Bob) Smith Educational Program Funds (to O.T.) and by The University of Texas MD Anderson Cancer Center start-up fund (to C.Z.).

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

O.T. researched data and wrote the manuscript. N.P.-O. researched data, contributed to discussion, and reviewed and edited the manuscript. J.K., M.S., and S.J. researched data. C.Z. contributed to the experimental design and discussion and wrote the manuscript. C.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank Guillermina Lozano (The University of Texas MD Anderson Cancer Center) and Frederick Alt (Harvard Medical School) for providing the mutant mice used in this study, Sherie Mudd from the histology core facility of MD Anderson, and Jill R. Delsigne from MD Anderson for her helpful comments on the manuscript.