Inactivation of arf-bp1 Induces p53 Activation and Diabetic Phenotypes in Mice^*

Generation of arf-bp1 Conditional Knock-out Mice

To construct the arf-bp1 gene targeting vector, a BAC clone (RP22-407A2) containing the 5′ portion of arf-bp1 was screened from mouse the RP22 BAC genomic library (CHORI, Oakland, CA). We decided to delete exon 11 of arf-bp1 (Ensembl), leading to a translational reading frameshift and loss of >90% of ARF-BP1 protein, including the HECT E3 ubiquitin ligase domain at the C terminus. The targeting vector was constructed by recombineering as described previously (22), which contained a 7-kb 5′ region of exon 11 and 4-kb 3′ region of exon 11 of arf-bp1. As indicated in the diagram (Fig. 1A), loxP sites were inserted in the introns flanking exon 11, to permit deletion of exon 11 upon expression of Cre recombinase. The targeting vector was linearized and electroporated into ES cells. After selection in the presence of G418, the targeted ES cell clones were identified by Southern blotting using PstI-digested ES cell genomic DNA and a 3′ probe, amplified with a 5′ primer of cctgataaattacctggagac and 3′ primer of aacagctaatggcagagggtt. The targeted ES cell clones were subsequently injected into blastocysts to generate chimera, from which the exon 11-floxed allele was transmitted through the germ line. Initially, genotyping was determined by Southern blotting (Fig. 1B). Subsequently, it was determined by PCR using primers of WT 5′ primer ctaatcacaggaagcggttacaag WT 3′ primer ctcctatagcaaagtaaaagtatag, and Neo primer gttattaggtggatccgtacgat to detect the wild-type allele and mutant allele simultaneously (Fig. 1C). The arf-bp1 conditional knock-out mice were back-crossed to C57BL/6J to produce a congenic strain for the study of ARF-BP1 functions in β-cells. The p53 conditional knock-out mice were kindly provided by Dr. Thomas Ludwig, Columbia University (23). Maintenance and experimental protocols using mice were approved by Institutional Animal Care and Use Committee (IACUC) of Columbia University.

Analysis of arf-bp1 Conventional Knock-out Mice

Breeding between arf-bp1^FL/+ female mice and Rosa26-cre homozygote male mice were set up to generate arf-bp1 knock-out mice. The resulting offspring should have equal numbers of arf-bp1^+/+, arf-bp1^+/Y, arf-bp1^+/−, and arf-bp1^−/Y, among which arf-bp1^+/+ and arf-bp1^+/Y represent wild-type embryos, arf-bp1^+/− represents heterozygote embryos, and arf-bp1^−/Y represents knock-out embryos. No viable arf-bp1 knock-out mice were obtained postnatally, indicating that arf-bp1 knock-out mice were embryonic lethal. Subsequently, phenotypic analysis was done using embryos collected from timed pregnancy at different stages of gestation. Embryos were fixed in PBS-buffered 4% paraformaldehyde overnight and embedded in paraffin. Serial 5-μm sagittal sections were collected and stained by hematoxylin and eosin according to standard procedures. The sections were also immunostained using antibodies against ARF-BP1 (ureb1; Bethyl), p53 (CM5; Novocastra), and Cleaved Caspase3 (Cell Signaling), followed by counterstaining using hematoxylin.

Histology and Immunohistochemical Analysis of Islets

Mouse pancreases were collected, fixed in 4% paraformaldehyde overnight, and embedded in paraffin. Sections of 5 μm were prepared and immunostained according to standard procedures, using antibodies against ARF-BP1 (Bethyl), insulin (Dako), and glucagon (Dako).

Determination of Blood Glucose and Insulin and Glucose Tolerance Test (GTT)

The arf-bp1 conditional knock-out mice were back-crossed to C57BL/6J at least 6 times before crossing with transgenic RIP-cre (rat insulin promoter controlled cre) mice to generate arf-bp1^FL/Y/RIP-cre mice (24, 25). Because arf-bp1 is located on the X chromosome, only the male mice (arf-bp1^FL/Y and arf-bp1FL/Y^FL/Y/RIP-cre) were used in the subsequent study. The mice were maintained at a 12-h light/dark cycle. The fasting glucose levels and the fasting insulin levels were determined after the mice were fasted overnight (16 h). The fed glucose levels were measured using mice fed at libitum. The fed insulin levels were determined after mice were first fasted overnight and then fed at libitum for 4 h to maximize insulin secretion. Blood glucose was measured using blood from tail vein with OneTouch UltraMini glucometer (LifeScan; Johnson & Johnson). The blood insulin levels were determined using an Insulin ELISA kit (Millipore) according to the manual provided by the manufacturer. The GTT was done using mice fasted overnight before injecting 2 g of glucose/kg of body weight into the peritoneal cavity. Blood glucose levels were then determined at 15-, 30-, 60-, and 120-min time points after the glucose injection.

Deletion of arf-bp1 in Mouse Embryonic Fibroblasts (MEFs) Using Adenovirus Expressing Cre Recombinase-arf-bp1^FL/Y

MEFs were prepared from E13.5 embryos and cultured in DMEM containing 10% FBS. To delete arf-bp1, 2.5 × 10⁵ of cells were mixed with either control adenovirus expressing GFP (Ad-CMV-GFP; Vector Biolaboratories) or adenovirus expressing Cre (Ad-CMV-Cre; Vector Biolaboratories) at a multiplicity of infection of 200 before being plated onto a 6-well plate. Cells were harvested after 3 days, and protein extracts were prepared in radioimmuneprecipitation assay buffer and analyzed by Western blotting. The Western blot was probed with antibodies against ARF-BP1 (Bethyl), p53 (CM5; Novocastra), p21 (Santa Cruz Biotechnology), BAX (Santa Cruz Biotechnology), and β-actin (Sigma).

Generation of arf-bp1 Conditional Knock-out Mice

The mouse arf-bp1 gene is located on the X chromosome and contains 83 exons in a span of more than 100 kb of genomic DNA (Ensembl). Because of the location and the size of arf-bp1, we decided to create a conditional knock-out allele of arf-bp1, in which exon 11 is flanked by loxP sites (floxed) to permit deletion of exon 11 upon Cre expression (Fig. 1A). After gene targeting in mouse embryonic stem cells, 2 of 200 clones screened were identified to have correct homologous recombination events by Southern blotting. Subsequently, these clones were injected into mouse blastocysts to derive chimeras, from which germ line transmission of the targeted allele was accomplished. Subsequently, homozygote conditional knock-out males (arf-bp1^FL/Y) and females (arf-bp1^FL/FL) were obtained through breeding, which were grossly normal and fertile, suggesting that floxing exon 11 has no discernible effects on arf-bp1 transcription.

arf-bp1 Knock-out Mice Were Embryonic Lethal

To understand the impacts of knock-out of arf-bp1 in vivo, we generated arf-bp1 knock-out mice (arf-bp1^−/Y) through breeding between arf-bp1^FL/+ heterozygote conditional knock-out female mice and Rosa26-cre homozygote male mice. Because of constitutive Cre expression from the Rosa26-cre allele, mice with the genotype of arf-bp1^FL/Y/Rosa26-cre were considered as arf-bp1 knock-out mice (arf-bp1^−/Y). There were 47 wild-type offspring at weaning age (Table 1), and no arf-bp1 homozygote knock-out mice were obtained postnatally (23 were expected), suggesting that arf-bp1 knock-out mice were embryonic lethal. Furthermore, the number of arf-bp1 heterozygote knock-out (arf-bp1^+/−) mice (8 were observed) was underrepresented (23 were expected), presumably due to random X chromosome inactivation of the remaining wild-type allele. To analyze the phenotypes of arf-bp1 knock-out mice, embryos from E11.5 to E17.5 were collected, from which arf-bp1 knock-out embryos were identified at all stages during development (Table 1). Although arf-bp1 knock-out embryos at E12.5 (Fig. 2D) appeared indistinguishable from wild-type embryos (Fig. 2A), all of the arf-bp1 knock-out embryos displayed hemorrhage in the abdominal region by E14.5 (Fig. 2E), followed by growth impairment, necrosis, and eventual death of the embryos (Fig. 2F). Age-matched control embryos are shown (Fig. 2, B and C). These results confirmed that arf-bp1 knock-out mice were embryonic lethal during late stage of embryonic development.

Ablation of arf-bp1 in Mouse Resulted in p53 Activation

To explore the molecular mechanism for the embryonic lethality in arf-bp1 knock-out mice, sagittal sections were prepared from the embryos collected at multiple stages from timed mating. The sections were immunostained using an anti-ARF-BP1 antibody (Bethyl), which recognizes the middle portion of ARF-BP1 protein. As expected, ARF-BP1 could be detected in nucleus and cytoplasm in the cells from wild-type embryos (Fig. 2G). In contrast, almost all of the cells in arf-bp1 knock-out embryos had lost the staining of ARF-BP1 by E13.5, indicating sufficient depletion of ARF-BP1 protein in the cells from arf-bp1 knock-out embryos (Fig. 2J). Because ARF-BP1 has been shown to regulate p53 stability (11), sections from arf-bp1 knock-out embryos were immunostained using anti-p53 antibody (Novocastra). Indeed, staining of p53 was detected in neural cells at E13.5 (Fig. 2K), compared with those of the control embryos (Fig. 2H), demonstrating that ablation of ARF-BP1 resulted in accumulation of p53 protein. In addition, Cleaved Caspase3 staining was also readily detected in these cells in arf-bp1 knock-out embryos (Fig. 2L) compared with those of the control embryos (Fig. 2I), suggesting that accumulation of p53 led to activation of apoptosis. Together, these data suggest that the embryonic lethality of arf-bp1 knock-out mice was, at least, in part, due to p53 activation in the absence of functional ARF-BP1.

Ablation of ARF-BP1 in Pancreatic β-Cells in mice Resulted in Progressive Hyperglycemia and Development of Diabetes

Although the above data indeed validate the role of ARF-BP1 in p53 regulation in vivo, with very limited ARF-BP1-null cells/tissues, the physiological role of ARF-BP1 in p53 regulation could not be determined. The difficulty in studying these arf-bp1 knock-out mice has amplified the importance of generating additional ARF-BP1 mutant mice in different organs to facilitate the analysis of the biological phenotypes in adult tissues. We chose to delete arf-bp1 in pancreatic β-cells because of the minimal effect on embryonic development by disruption of β-cell functions and the significant expression of ARF-BP1 in β-cells (Fig. 3Ai). Thus, we generated conditional arf-bp1 mutant mice with specific deletion of arf-bp1 in pancreatic β-cells, mediated by RIP-cre (24, 25). The arf-bp1^FL/Y/RIP-cre mice were born close to Mendelian ratio and appeared normal; in particular, they had similar body weight compared with the control mice at least up to 6 months of age (supplemental Fig. 1). To determine whether arf-bp1 was efficiently ablated in β-cells, sections of pancreas from 1-month-old mice were immunostained using antibodies against ARF-BP1, insulin, and glucagon. ARF-BP1 staining was present in nucleus and cytoplasm of the β-cells of the control mice (Fig. 3Ai). Some of the β-cells showed no staining of ARF-BP1, presumably due to lack of expression of ARF-BP1 in proliferating cells, as observed previously (26). This is consistent with the fact that arf-bp1 conventional knock-out embryos died at late gestation stage (E14.5), indicating critical functions of ARF-BP1 in postmitotic cells. In contrast, there was no staining for ARF-BP1 in the β-cells in the islets of arf-bp1^FL/Y/RIP-cre mice (Fig. 3Aiv), demonstrating that ARF-BP1 was sufficiently depleted in the β-cells due to RIP-cre-mediated arf-bp1 deletion. Consistent with the specificity of the RIP-cre expression, there was no difference of ARF-BP1 staining in the α-cells of the islets and the surrounding acinar cells of the pancreas between arf-bp1^FL/Y/RIP-cre mice and the control mice (Fig. 3A, iv versus i). Interestingly, there was a similar staining for insulin in the β-cells (located in the core of the islet) (Fig. 3Av) in the absence of ARF-BP1 in arf-bp1^FL/Y/RIP-cre mice compared with that of the control mice (Fig. 3Aii). There was no significant difference for glucagon staining (located in the periphery of the islet) between the control and the arf-bp1^FL/Y/RIP-cre mice (Fig. 3A, iii versus vi).

To determine the long term effects of arf-bp1 deletion on the β-cell functions, the blood glucose was monitored periodically. In addition, the proficiency of insulin response was determined by GTT at 3 and 6 months of age. Consistent with the insulin staining in islets from young mice, both fasting and fed glucose levels of arf-bp1^FL/Y/RIP-cre mice showed similar levels during the first 2 months, compared with those of the control arf-bp1^FL/Y mice (Fig. 3, B and C). However, as arf-bp1^FL/Y/RIP-cre mice aged, blood glucose levels became increasingly higher than those of the control mice (Fig. 3, B and C). For the fed blood glucose levels (Fig. 3C), in particular, the average fed blood glucose level was 209 mg/dl for arf-bp1^FL/Y/RIP-cre mice versus 167 mg/dl for the control mice at 3 months old; the differences increased to 308 versus 208 mg/dl at 6 months old and 373 versus 204 mg/dl at 9 months old (Fig. 3C). Consistently, fed insulin levels were decreased significantly in arf-bp1^FL/Y/RIP-cre mice at 6 and 9 months old of age compared with those of the control mice (Fig. 3E). There were also significant lower fasting insulin levels for arf-bp1^FL/Y/RIP-cre mice than the control mice at 9 months old (Fig. 3D). Furthermore, the insulin deficiency was confirmed by GTT. As shown in Fig. 3F, the glucose levels were significantly higher along the time course after glucose injection in arf-bp1^FL/Y/RIP-cre mice than those of the control mice at 3 months old. The differences in GTT became even bigger for 6-month-old mice, suggesting further decline of insulin response in aged arf-bp1^FL/Y/RIP-cre mice (Fig. 3G). Moreover, many of the arf-bp1^FL/Y/RIP-cre mice (Fig. 4A, right) displayed typical signs for diabetes, such as excessive urination due to high blood glucose, and upon dissection, greatly enlarged bladder (Fig. 4A, arrows), and excess food in the intestine (Fig. 4A, arrowheads). Many of the mice had fed blood glucose levels higher than 600 mg/dl (limit of the glucometer). More than half of the mice died at <1 year of age, due to severe diabetic symptoms (see Fig. 6A, red line).

Insulin Deficiency in arf-bp1^FL/Y/RIP-cre Mice Was Caused by Loss of Pancreatic β-Cells

To determine the cause of the hyperglycemia and insulin deficiency, sections of pancreas from arf-bp1^FL/Y/RIP-cre mice were collected and immunostained for insulin (β-cells) and glucagon (α-cells). At 1 month old, the islets from arf-bp1^FL/Y/RIP-cre mice had normal appearances of round shape, with center-located β-cells and periphery-located α-cells (Fig. 4B, vii and x versus i and iv, respectively). At 6 months old, the morphology of islets in arf-bp1^FL/Y/RIP-cre mice became increasingly irregular with frequent infiltration of α-cells in the middle of the islets (Fig. 4B, viii and xi versus ii and v, respectively). Disruption of the islet structure was worsened in aged mice of >9 months old, in which there was near absence of insulin staining in the islets from arf-bp1^FL/Y/RIP-cre mice (Fig. 4B, ix versus iii), suggesting dramatic loss of β-cells as mice aged. The remaining cells in the islets were α-cells because they were positive for glucagon staining (Fig. 4B, xii versus vi).

Diabetic Conditions of arf-bp1^FL/Y; RIP-cre mice Were Significantly Rescued by Concomitant p53 Ablation

Our previous study showed that ARF-BP1 plays an important role in down-regulation of p53 because RNAi-mediated knockdown of ARF-BP1 increases the basal levels (∼3-fold) of p53 proteins and enhances p53-mediated functions in human cell lines (11). Thus, it is very likely that the cause for the aging-dependent decrease of β-cell population in arf-bp1^FL/Y/RIP-cre mice is the high basal levels of p53 in these ARF-BP1-depleted β-cells and that these cells are more sensitive to normal physiological stress. To confirm the role of ARF-BP1 in regulating p53 function, we decided to delete arf-bp1 in arf-bp1^FL/Y MEFs infected with Cre-expressing adenovirus (Ad-cre), which allowed detailed analysis of the effects on p53 function by arf-bp1 inactivation. As shown in Fig. 5A, whereas ARF-BP1 was expressed as a protein of about 500 kDa in wild-type cells (lane 1), no ARF-BP1 protein was detected in arf-bp1^FL/Y cells after Ad-cre virus infection (Fig. 5A, lane 2), demonstrating depletion of ARF-BP1 protein after deletion of arf-bp1. Significantly, p53 was accumulated in arf-bp1 knock-out cells, which led to activation of its targets p21 and apoptotic gene BAX (Fig. 5A, lane 2 versus lane 1). These results showed that deletion of arf-bp1 indeed led to stabilization and activation of p53.

To determine further whether loss of β-cells in arf-bp1^FL/Y/RIP-cre mice was due to activation of p53 under physiological settings, we generated arf-bp1^FL/Y/p53^FL/FL/RIP-cre double knock-out mice, in which both arf-bp1 and p53 were deleted in β-cells. Dramatically, the hyperglycemia conditions in arf-bp1^FL/Y/RIP-cre mice were alleviated in arf-bp1^FL/Y/p53^FL/FL/RIP-cre mice. Both fasting (Fig. 5B, green bar) and fed (Fig. 5C, green bar) blood glucose levels were lowered in arf-bp1^FL/Y/p53 ^FL/FL/RIP-cre mice compared with those of the arf-bp1^FL/Y/RIP-cre mice (Fig. 5, B and C, red bar). Consistently, insulin levels in arf-bp1^FL/Y/p53^FL/FL/RIP-cre mice (Fig. 5D, green bar) were higher than that of arf-bp1^FL/Y/RIP-cre mice (Fig. 5D, red bar). Furthermore, GTT using 6-month-old mice revealed significant improvement of insulin response in arf-bp1^FL/Y/p53^FL/FL/RIP-cre mice (Fig. 5E, green line), as the glucose level at the end of the assay for arf-bp1^FL/Y/p53 ^FL/FL/RIP-cre mice was nearly the same as that of the control mice (Fig. 5E, green line versus blue line). In contrast, the glucose levels for arf-bp1^FL/Y/RIP-cre mice were significantly higher than that of the control mice at the end of the assay (Fig. 5E, red line). Notably, the double knock-out mice survived much longer than the arf-bp1^FL/Y/RIP-cre mice, due to overall improvement of the diabetic condition (Fig. 6A, green line).

Furthermore, histology analysis of pancreases revealed similar architecture of islets and the abundance of β-cells in arf-bp1^FL/Y/p53^FL/FL/RIP-cre mice (Fig. 6H) compared with the arf-bp1^FL/Y/p53^FL/FL 9-month-old mice (Fig. 6F), whereas the α-cells maintained their peripheral localization (Fig. 6, I versus G). These histological features represented significant improvement of arf-bp1^FL/Y/p53^FL/FL/RIP-cre mice compared with arf-bp1^FL/Y/RIP-cre mice, which showed dramatic reduction of insulin staining and disorganized islets (Fig. 6, D and E). Insulin and glucagon staining of 9-month-old arf-bp1^FL/Y mice was included for comparison (Fig. 6, B and C, respectively). Together, these results suggested that the diabetic phenotype in arf-bp1^FL/Y/RIP-cre mice was significantly rescued by p53 gene ablation.