Selective ablation of P53 in pancreatic beta cells fails to ameliorate glucose metabolism in genetic, dietary and pharmacological models of diabetes mellitus

Objective Beta cell dysfunction and death are critical steps in the development of both type 1 and type 2 diabetes (T1D and T2D), but the underlying mechanisms are incompletely understood. Activation of the essential tumor suppressor and transcription factor P53 (also known as TP53 and Trp53 in mice) was linked to beta cell death in vitro and has been reported in several diabetes mouse models and beta cells of humans with T2D. In this article, we set out to determine the beta cell specific role of P53 in beta cell dysfunction, cell death and development of diabetes in vivo. Methods We generated beta cell specific P53 knockout (P53BKO) mice and used complementary genetic, dietary and pharmacological models of glucose intolerance, beta cell dysfunction and diabetes development to evaluate the functional role of P53 selectively in beta cells. We further analyzed the effect of P53 ablation on beta cell survival in isolated pancreatic islets exposed to diabetogenic stress inducers ex vivo by flow cytometry. Results Beta cell specific ablation of P53/Trp53 failed to ameliorate glucose tolerance, insulin secretion or to increase beta cell numbers in genetic, dietary and pharmacological models of diabetes. Additionally, loss of P53 in beta cells did not protect against streptozotocin (STZ) induced hyperglycemia and beta cell death, although STZ-induced activation of classical pro-apoptotic P53 target genes was significantly reduced in P53BKO mice. In contrast, Olaparib mediated PARP1 inhibition protected against acute ex vivo STZ-induced beta cell death and islet destruction. Conclusions Our study reveals that ablation of P53 specifically in beta cells is unexpectedly unable to attenuate beta cell failure and death in vivo and ex vivo. While during development and progression of diabetes, P53 and P53-regulated pathways are activated, our study suggests that P53 signaling is not essential for loss of beta cells or beta cell dysfunction. P53 in other cell types and organs may predominantly regulate systemic glucose homeostasis.


INTRODUCTION
Diabetes mellitus affects more than 500 million people worldwide [1]. The main characteristics of diabetes are insufficient insulin secretion and/or disturbed cellular insulin signaling (insulin resistance), resulting in chronic hyperglycemia [2]. Type 1 diabetes (T1D), affecting approx. 5e10% of all persons with diabetes, is an autoimmune disease causing loss of beta cells and manifests mainly but not exclusively in childhood and adolescence. Approximately 90% of people with diabetes are diagnosed with type 2 diabetes (T2D), a multifactorial disease with obesity, age and genetic predisposition as main risk factors. In T2D, the extent and onset of insulin resistance and beta cell dysfunction appears to be variable, potentially in line with its recently proposed subtypes [2e5]. Post mortem analyses of pancreata from people with T2D have detected varying levels of beta cell death and reduced beta cell mass [6,7], and multiple animal 1 models of T2D show massive beta cell apoptosis [8e11]. On the other hand, processes like beta cell trans-or dedifferentiation, as well as loss of beta cell function have also been linked to T2D development [5,12,13]. Beta cells are known to experience multiple types of cellular stress during diabetes development, including but not limited to endoplasmic reticulum stress, inflammation, oxidative stress, glucolipotoxicity and DNA damage [2,14]. The transcription factor tumor protein P53 (P53, also known as transformation related protein TP53, or Trp53 in rodents) is essential for tumor prevention and regulation of cell survival and death [15,16], but also plays pleiotropic functions in cell biology and physiology [17]. Several animal and human studies link P53 activity to diabetes progression. For instance, we previously found expression of canonical P53 target genes to be upregulated in islets of obese and diabetic db/db mice and linked P53 activity to increased abundance of the pro-diabetogenic microRNA-200 family [8]. Moreover, beta cell specific ablation of P53 protected against beta cell death (but not dysfunction) induced by a rare disease-inducing glucokinase mutation [18]. In addition, whole body P53 knockout (KO) mice (after bone marrow transplantation from wildtype mice) showed increased mitophagy, resulting in maintained insulin secretion and glucose tolerance in mice with chemically induced beta cell destruction [19]. Accordingly, beta cell specific ablation of Ataxia Telangiectasia Mutated (ATM), one of several upstream regulators of P53, ameliorated chemically induced hyperglycemia [20]. In contrast, P53 action was also reported to have beneficial effects in diabetic mouse models. Secchiero et al. showed that systemic P53 deinhibition and stabilization by treatment with the small molecule Nutlin-3 reduced streptozotocin (STZ)-induced hyperglycemia, in part by acting on the immune system [21]. To summarize, most studies strongly indicate that inhibition or depletion of P53 can attenuate diabetes development potentially by preserving beta cell function and survival. However, as most of these studies were conducted with either conventional P53 KO mice or systemic P53 manipulation by virus administration or pharmacological treatments, the beta cell autonomous role of P53 in development and progression of diabetes is incompletely understood. To unravel the role of this essential protein in cellular physiology, we generated mice with beta cell specific ablation of P53 and analyzed this strain in genetic, dietary and pharmacological models of glucose intolerance, beta cell dysfunction and diabetes development.

Experimental animals
All experiments were approved by the Ethics Committee of the State Ministry of Agriculture, Nutrition and Forestry (State of North Rhine-Westphalia, Germany). The Ins1-Cre mice [22], conditional TP53 mice (also known as Trp53 mice) [23], conditional ATM mice (backcrossed onto a C57BL/6JRj background (Janvier) for at least 7 generations) [24], conventional PDX1 mice [25], and tdTomato reporter mice [26] on a C57BL/6J background have been described before. For analyses in dietary and pharmacological models of diabetes, we generated beta cell specific P53 KO mice (hence denoted as P53 BKO ) by cross-breeding mice carrying the Ins1-Cre transgene with P53 floxed mice (#008462, Jackson Laboratories, USA). For dietary and pharmacological interventions, control mice were heterozygous for the Ins1-Cre transgene but carried two wildtype P53 alleles (Ctrl). For analysis of P53 function on a background of heterozygous PDX1 deficiency, we generated mice lacking one PDX1 allele and at the same time lacking P53 specifically in beta cells (genotype Ins1-Cre tg/ wt , P53 fl/fl , PDX1 wt/KO , denoted as PDX1 wt/KO P53 BKO ) by cross-breeding P53 BKO mice with PDX1 heterozygous mice. Resulting littermates not carrying the Cre allele, but homozygous for a loxP flanked P53 allele and heterozygous for the PDX1 null allele were used as control for this genetic intervention (PDX1 wt/KO Ctrl). As second control, we also used P53 BKO mice carrying two wildtype PDX1 alleles (PDX1 wt/ wt P53 BKO ). For flow cytometric analyses, we cross-bred Rosa26-tdTomato reporter mice with P53 BKO (Tomato:P53 BKO ), ATM BKO (Tomato:ATM BKO ) or Ins1-Cre transgenic mice (genotype Ins1-Cre tg/wt , Rosa26-tdTomato fl/wt or Ins1-Cre tg/wt , Rosa26-tdTomato fl/fl , denoted as Tomato Beta ), resulting in a beta cell specific red fluorescence signal. For qPCR analyses of sorted beta and non-beta cells, also PDX1 wildtype and heterozygous mice with or without additional beta cell specific KO of P53 were cross-bred with Rosa26-tdTomato reporter mice (groups additionally denoted as Tomato reporter strains). All Tomato reporter mice were heterozygous for the Ins1-Cre knock-in allele. Male mice were used for all studies due to the inherent resistance of beta cells of female (C57BL/6J) mice against beta cell dysfunction, beta cell death and diabetes [27]. An overview of all used diabetes mouse models and their different properties is provided in Table S1. Mice were housed at three to six mice per cage (Macrolon type III) at a constant temperature of 22C and a 12 h lightedark cycle (lights on at 6 AM). Animals had free access to food and water ad libitum. After weaning at the age of 21e28 days, mice were fed either a standard laboratory chow or a high fat diet (HFD) containing 45 kcal% fat, 20 kcal% protein, and 35 kcal% carbohydrates with 4.73 kcal/g energy (D12451; Research Diets, New Brunswick, NJ) until the end of the respective study. For flow cytometric analysis of islet cell viability or gene expression analysis of pancreatic islets, age-matched mice were sacrificed at the age of 2e11 months.

Genotyping
Genotyping was performed using standard protocols using appropriate primers (Table S2) and GoTaq G2 Hot Start Green Master Mix (#M7423, Promega).
2.3. Body weight, blood glucose levels, glucose and insulin tolerance tests Body weight and blood glucose levels were determined weekly (unless stated otherwise) with an electronic scale and Contour XT glucometer (Bayer Consumer Care AG, Leverkusen, Germany), respectively. For intraperitoneal (i.p.) glucose tolerance tests (ipGTT), 16 h fasted mice were injected i.p. with 0.75e2 g/kg body weight D-glucose in PBS (as stated in each subfigure or figure legend). Glucose bolus was chosen dependent on expected maximal glucose excursions. Blood glucose levels were measured every 20 min and blood was collected before (0 min), as well as 20 and 120 min after glucose injection in heparincoated tubes. After centrifugation, plasma was stored at À20 C (short-term <1 week) or À80 C (long-term >1 week). Nonresponding mice (BG increase <20% at 20 min, potentially due to incomplete i.p. injection) were excluded from analysis (1 mouse each in Figs. S4E and S4F). For i.p. insulin tolerance tests (ipITT), mice were fasted for 4 h and then injected i.p. with 0.75 U/kg body weight insulin (Insuman Rapid, Sanofi) diluted in PBS. Blood glucose levels were measured every 20 min for 2 h.

Analysis of plasma insulin
Insulin levels were measured by ELISA (Ultra Sensitive Rat Insulin ELISA Kit; #90060, CrystalChem) according to the manufacturer's instructions in 5 ml plasma collected during ipGTTs. Mice with hemolytic plasma samples or insulin values below detection limit were excluded from analysis (2 mice in Fig. S4I). Brief Communication 2.5. Administration of STZ At week 12 of age, 6 h fasted mice of MLD-STZ or HFD-STZ cohorts were injected i.p. with 40 mg/kg STZ at 5 consecutive days (MLD-STZ model) or once with 150 mg/kg (HFD-STZ model). STZ was stored at À20 C and reconstituted immediately before injection in citrate buffer. One MLD-STZ cohort (both Ctrl and P53 BKO mice) did not respond to STZ injections (no increase in BG levels), potentially due to temperature fluctuations during STZ transport, and was excluded from analysis.
2.6. Isolation, dispersion and culture of pancreatic islets Mice were sacrificed by cervical dislocation and the pancreatic islets were isolated by ductal liberase TL (#5401020001, Merck) perfusion of the pancreas, based on a previously described method with modifications [8]. In short, pancreas was dissected and incubated for 17 min at 37 C in a shaking waterbath. Digestion was stopped with 40 ml of 10% Fetal Calf Serum (FCS, Gibco) in RPMI, followed by moderate shaking for 30 s. Cells were spun down at 300 Â g for 5 min, resuspended in 10 ml RPMI, strained through a sieve, filled up to 50 ml with RPMI and spun down again. Resulting pellet was resuspended in 3 ml 1.119 g/ml Histopaque (#11191, Merck) and overlayed with 1.083 and 1.077 g/ml Histopaque (#10831 and #10771, Merck) and RPMI, 2.5 ml each. After density gradient centrifugation at 935 Â g for 20 min with acceleration and brake active, but at lowest setting, islets accumulated between the two upper phases, were transferred and washed with RPMI containing 10% FCS. After centrifugation, islets were transferred into petri dishes containing 10 ml full islet medium (DMEM with 10% FCS, 11.11 mM glucose, 2 mM glutamax, 1 mM sodium pyruvate, 11.2 mM HEPES, 0.175 mM beta-mercaptoethanol and 100 U/ml penicillin/streptomycin) and incubated at 37 C with 5% CO 2 . Islets were harvested 3e72 h post isolation, depending on the experimental setup. For staining of dispersed islet cells ( Fig. S3L-P), islets were transferred into 15 ml tubes, washed with 10 ml PBS and trypsinized with 3 ml 0.05% Trypsin/EDTA (#25300-054, Thermo Fisher) for 12 min with additional mechanical disruption by carefully pipetting up and down three times in between. Trypsinization was stopped by adding 10 ml FCS containing islet medium. Cells were spun down at 300 Â g for 5 min, supernatant was discarded and cells were resuspended in 1 ml medium. Cells were counted and seeded in a 96well-plate (20.000 cells/well in 100 ml each) suitable for microscopy.
Cells were spun down at 300 Â g for 3 min to support cell adhesion, incubated for 6 h at 37 C, 5% CO 2 and fixed by adding 100 ml 4% PFA for 20 min at RT. Cells were washed with PBS and used for immunostaining (performed as described in 2.10 with downscaled volumes) and finally imaged at 40Â magnification using a laser scanning microscope (LSM 880, Zeiss) in confocal mode.

Immunoblotting
A minimum of 50 islets was collected in 1.5 ml tubes, spun down for 5 min at 300 Â g, washed twice with PBS and the resulting pellet lysed in Bio-Plex buffer (30 ml per 50 islets, #171304011, Bio-Rad). 4Â Laemmli buffer (#1610747, Bio-Rad) with DTT was added to samples, and after boiling for 5 min, 10 ml of protein samples were separated by SDS-PAGE using pre-cast 4e15% Stainfree gels (#4568086, Bio-Rad; allowing detection of tryptophan-containing proteins using a proprietary trihalo compound), transferred by electroblotting and PVDF membranes blocked in TBS with 0.1% Tween-20 (TBST) containing 5% nonfat-dried milk or BSA for 1 h. Membranes were incubated with primary antibodies (see 2.13) overnight at 4 C, washed thrice with TBST for 10 min, exposed to HRP-conjugated secondary antibodies for 1 h at room temperature (RT), washed thrice again with TBST for 10 min and developed using ECL Western Blotting Substrate (#170-5061, Bio-Rad). Quantification of all bands was performed using Image Lab software (Bio-Rad). The mean value of relevant control islets was set to 1 and all other values normalized to that.

Immunohistochemistry
Pancreata were fixed in 4% PFA o/n and dehydrated by transferring into 7.5, 15 and 30% Sucrose in PBS (at least 6 h each), before embedding in NEG-50 (#11912365, Fisher-Scientific). Of each pancreas, 12-18 10 mm sections with a minimum distance of 180 mm between each other were analyzed. Slides were washed for 5 min with PBS and PBS-T (PBS þ 0.2% TritonX-100) before 1 h blocking with 5% BSA in PBS-T at RT in a humidified chamber and covered with parafilm. Directly labeled or unconjugated primary antibodies were added o/n at 4 C (see 2.13). After washing with PBS twice and PBS-T once for 10 min each, secondary antibody (1:500) and Hoechst 33342 (1:5000) were added for 1 h at RT diluted in blocking buffer. Slides were washed thrice for 10 min with PBS. Fluoroshield (#F6182, Merck) was added for mounting with coverslips. Slides were imaged with a VS200 Slide Scanner (Olympus) using a 40Â objective and quantification was performed with QuPath [28] in a blinded manner, using a custom analysis macro (available upon request) and trained pixel classifier for semi-automated islet area identification and cell counting (Fig. S3). In more detail, insulin, Nkx6.1 or glucagon positive areas with a minimum size of 40 mm 2 were identified by the macro as islets. Islet were encircled automatically, but needed manual adjustments (separation of adjacent islets, deletion of islets with less than three Nkx6.1 positive cells, tightening of islet annotations). Next, Hoechst/ Nkx6.1/Nkx6.1 and Ki67 (double-) positive cells were automatically identified (by threshold) and counted. The automatic detection of Ki67 positive cells was manually controlled to avoid false positive detections due to e.g. high background signal (e.g. observed in proximity to large vessels) and if necessary, manually corrected.

Flow cytometric cell death analysis
For flow cytometric analyses, islets of wildtype mice (C57Bl/6JRj, Janvier), Cre-negative colony mates, Tomato Beta mice, Tomato:P53 BKO mice and Tomato:ATM BKO mice were used, depending on the experiment. Islets (approx. 70-100) were treated in 6-well plates as a single technical replicate, based on low interexperimental variability. Treatments were started 24 he48 h post isolation, and treatment durations were optimized to detect significant cell death (STZ 16 h, Cytokines 40 h). After treatment, the supernatant (containing single dead/dying cells) was transferred to FACS tubes. Residual islets were trypsinized (as described in 2.6) to achieve a single cell suspension, before transferring to the respective FACS tubes. Cells were spun down for 5 min at 300 Â g followed by a wash step with PBS. Cells were incubated in 500 ml FVS520 (#564407, BD Biosciences, diluted 1:1000 in PBS) for 15 min at RT. After two wash steps with PBS, cells were resuspended in 300 ml PBS and the amount of FVS520 positive (dead) and FVS520 negative (living), as well as tdTomato positive (beta) and tdTomato negative (non-beta) cells was measured using a FACSCalibur (BD Biosciences) or CytoFLEX S (Beckman Coulter). Notably, no double positive population was detectable consistent with our observations that dying/dead cells lose the tdTomato signal, likely due to membrane barrier loss. Hence, percentage of total dead (FVS520 positive ) and living (FVS520 negative ) beta (tdTomato positive ) vs non-beta (tdTomato negative ) cells was quantified using FlowJo V10 software (BD Biosciences). Single stained control samples were used for compensation and to confirm that the FVS520 signal was reproducibly stronger compared to beta cell autofluorescence.

FACS of beta-and non-beta cells
For sorting of beta-and non-beta cells, 70-350 islets of Tomato reporter mice (genotype for Ins1 and Rosa26 locus: Ins1-Cre tg/wt and Rosa26-tdTomato fl/wt or Rosa26-tdTomato fl/fl ) were trypsinized (as described in 2.6), resuspended in 300 ml islet medium, strained through a 70 mm sieve and sorted using a CytoFLEX SRT (Beckman Coulter). Gating strategy is depicted in Figure S8 No statistical method was used to predetermine sample size, but instead was based on preliminary data and previous publications as well as observed effect sizes. Due to the small sample sizes, normality testing was not performed. Each statistical test is described in the figure legends. Depending on the number of variables, a regular one-or two-way analysis of variance (ANOVA), followed by Tukey's or Sidak's multiple comparison test was used. For enhanced clarity and reduced visual cluttering, not all comparisons are shown, as stated in the figure legends. For analyses of in vivo cohort data, statistical testing was only performed between different genotypes and not between different time points. Experiments with only two groups and one variable being tested were analyzed by an unpaired two-sided Student's t-test or by multiple t-test's without correction for multiple comparisons. Statistical analyses were performed using the Graphpad Prism (GraphPad Software, La Jolla, CA, USA, Version 9) software. p-values smaller than 0.05 are shown in each figure (except if stated otherwise) and are rounded to the third decimal place.

RESULTS
3.1. HFD-feeding results in glucose intolerance independent of beta cell P53 Since P53 whole body KO mice develop early-onset cancer, immunodeficiency and show a massively reduced life span [29], we generated mice lacking P53 specifically in beta cells (P53 BKO mice). Recombination of the P53 gene locus as well as loss of P53 expression was verified on genomic DNA, mRNA and protein level in islet samples, as well as in fluorescence-activated cell sorting (FACS)-purified tdTomato positive beta and tdTomato negative non-beta cells, which confirmed near-total and specific ablation of P53 without any observable effect on beta cell marker genes such as INS1/2 (Figs. S1AeH). We initially challenged P53 BKO and Ctrl mice with a high fat diet (HFD), an established model for low to moderate beta cell stress and early T2D development (Fig. S1I). While HFD feeding does not cause beta cell failure or overt diabetes, beta cells need to undergo expansion and hypersecrete insulin to maintain near-normal glucose levels in this model [30,31]. We did not detect any significant changes between HFD-fed Ctrl and P53 BKO mice in body weight gain, glucose tolerance, insulin secretion and sensitivity, as determined by glucose and insulin tolerance tests at several ages (Figure 1AeG and S1J, K). Thus, beta cell selective ablation of P53 does not affect insulin secretion or glucose metabolism when beta cells are under HFDinduced metabolic stress.

HFD-fed PDX1 wt/KO mice develop severe glucose intolerance independent of beta cell P53
We reasoned that potentially, beta cell stress induced by HFD was insufficient to provoke a P53-dependent phenotype, since HFD feeding decreased glucose tolerance, but did not lead to overt hyperglycemia. Hence, we aimed to challenge P53 BKO mice in a model with moderate to high beta cell stress. Pancreatic and Duodenal Homeobox 1 (PDX1) is a transcription factor that is essential for pancreas development and that postnatally regulates beta cell transcription and function in humans and mice. Importantly, rare cases of PDX1 heterozygosity lead to maturityonset diabetes of the young 4 (MODY4) [32], and a partial reduction of PDX1 expression is a key observation in islets of people with T2D [33]. Loss of one PDX1 allele in mice was shown to be sufficient to impair glucose tolerance already in young animals, prevented beta cell expansion during aging, and increased spontaneous apoptosis in islets when cultured at physiological glucose levels ex vivo [31,34]. We asked if P53 ablation improves glucose tolerance and beta cell function in PDX1 wt/KO mice by generating P53 BKO additionally lacking one PDX1 allele (PDX1 wt/KO P53 BKO mice). PDX1 heterozygosity and ablation of P53 were confirmed via qPCR and immunoblotting of pancreatic islets (Figs. S2AeC). 4-week-old PDX1 wt/wt P53 BKO (used as a control with normal PDX1 expression), PDX1 wt/KO Ctrl, and PDX1 wt/KO P53 BKO littermates were fed a HFD for 16 weeks, since HFD accelerates beta cell dysfunction in this model [35]. While body weight was similar in all three groups (Figure 2A), we observed increased (non-fasted and fasted) blood glucose levels ( Figure 2B,C and S2D), strongly impaired glucose tolerance (Figure 2DeF and S2E-G) and reduced glucose stimulated insulin secretion (Figure 2GeI) in mice lacking one PDX1 allele compared to PDX1 wildtype mice, validating this model. Nonetheless, ablation of P53 in beta cells did not prevent fasting and random-fed hyperglycemia and severe glucose intolerance, and also failed to increase insulin secretion during repeated tests (Figure 2BeI and S2D-G). Notably, insulin sensitivity tended to be increased in both groups lacking one PDX1 allele compared to PDX1 wt/wt P53 BKO mice, but was not further affected by P53 deficiency (Figs. S2H and I). Since depletion of beta cell P53 did not preserve glucose homeostasis of HFDfed PDX1 wt/KO mice, we asked if beta cell counts would be altered by ablation of P53, as previously observed in a glucokinase hyperactivation mouse model [18]. Nonetheless, the number of pancreatic beta cells (as detected by Nkx6.1 staining [33], Fig. S3) and islet cells (Hoechst positive cells in glucagon, insulin and Nkx6.1 positive area) and islet area of PDX1 wt/KO P53 BKO mice was comparable to PDX1 wt/KO Ctrl mice when normalized to either the number of all pancreatic cells, or pancreas area, or islet cell number by multicolor immunoflourescence staining (Fig. 2J, S2J-L and S3). Similarly, the mean islet area, and the number of proliferating beta cells (Nkx6.1 and Ki67 double positive cells) was unchanged ( Figure 2K, L and Fig. S2M, N). When exposed to a normal diet, PDX1 wt/KO Ctrl mice and PDX1 wt/KO P53 BKO mice also showed indistinguishable weight gain, glucose tolerance, insulin secretion and insulin sensitivity (Fig. S4A-N). Gene expression analysis of pancreatic islets from all groups of animals revealed that ablation of one PDX1 allele was sufficient to reduce expression of its canonical target gene SLC2A2 (also known as glucose transporter 2, GLUT2) by approx. 70%, independent of P53 status (Fig. S4O). Notably, mRNA levels of the established P53 target gene and master regulator of beta cell apoptosis and proliferation, PHLDA3 [36], were significantly increased by ablation of one PDX1 allele, and this was completely prevented when P53 was co-ablated in beta cells, again demonstrating the efficacy of our P53 KO approach (Fig. S4O). Taken together, ablation of P53 failed to prevent beta cell dysfunction in a model of monogenetic diabetes (with or without HFD feeding).
3.3. Beta cell specific P53 ablation fails to prevent STZ-induced hyperglycemia We next asked, if P53 ablation might protect beta cells from cell death induced by a strong and acute pro-apoptotic stimulus. To induce beta cell death in vivo, we used the beta cell toxin STZ that enters beta cells via GLUT2, induces DNA fragmentation and accordingly activates the DNA damage response (DDR) including ATM [20,30,31,37,38]. When using the multiple low-dose STZ (MLD-STZ) model [8], hyperglycemia manifests within 2e3 weeks. In our study, we injected 40 mg/kg STZ on 5 consecutive days and measured blood glucose levels and body weight over the following 16 days (Fig. S5A). We found no differences in body weight or STZ mediated increase of blood glucose levels between P53 BKO and Ctrl mice (Figure 3AeD). We reasoned that potentially, activation of the DDR and a subsequent increase of proapoptotic mRNAs was mediated by proteins separate from P53, such as ATM [20]. However, we determined that mRNA levels of proapoptotic P53 target genes such as PHLDA3, BAX, NOXA and BBC3, as well as the cell cycle inhibitor CDKN1A were reduced by 25e90% in isolated islets from STZ treated P53 BKO mice compared to islets from Ctrl mice one day after the last STZ injection (Fig. 3E and S5B). Of note, BAX, NOXA and BBC3 expression was not altered in islets from untreated P53 BKO mice (Fig. S4O), confirming that STZ treatment increases these apoptotic mRNAs in a P53 dependent manner. Expression of the ATM regulated T-cell chemoattractant CXCL10 [20], which was implicated in islet destruction, was similar between Ctrl and P53 BKO islets (Fig. 3E). In addition, expression levels of beta cell markers were stable or even increased (Fig. 3E). Thus, although islet  . (E) Relative islet mRNA expression levels of indicated genes of Ctrl and P53 BKO mice at day 6 after first STZ injection. Expression levels were normalized to the housekeeping genes GUSB and 36B4, Ctrl was set to 1. Shown are mean and individual values AE SEM (n ¼ 2 independent cohorts with 11 Ctrl vs. 8 P53 BKO mice in total). (F) Nonfasted and (G) 6 h fasted BG levels of HFD-fed Ctrl and P53 BKO mice injected once with 150 mg streptozotocin (STZ) per kg BW in week 12 of age (time of injection is stated as day 1). Shown are mean values AE SEM (n ¼ 2 independent cohorts with 12 Ctrl vs. 11 P53 BKO mice in total). (H) Representative images of pancreatic sections of mice described in (F þ G), as well as HFD-fed Ctrl mice without STZ injection (see Fig. 1), stained for nuclei with Hoechst (blue), beta cells with Nkx6.1 (pink) and alpha cells with glucagon (green). Shown are Nkx6.1 and glucagon positive islets (white arrows) and pancreatic lymph nodes (yellow arrow) (scale bar ¼ 2 mm) and magnified insets of single islets indicated by red boxes (scale bar ¼ 50 mm). Note the massive loss of Nkx6.1 positive beta cells after HFD þ STZ treatment (for clear visibility, image contrast settings were optimized individually for separately stained HFD-STZ and HFD sections). Significance was determined by (A þ B, D, F þ G) two-way ANOVA followed by Sidak's multiple comparison test, (C) an unpaired two-sided student's t-test, or (E) multiple t-test's without correction for multiple comparisons. cells from P53 BKO mice are protected against STZ-induced upregulation of pro-apoptotic P53 target genes, beta cell specific P53 depletion is insufficient to prevent MLD-STZ induced hyperglycemia. We next tested if P53 ablation would protect beta cells and prevent diabetes acutely after extreme beta cell stress. To this end, we injected HFD-fed Ctrl and P53 BKO mice with one single high dose of STZ (HFD-STZ model, Figs. S5CeG) [8]. Strikingly, both P53 BKO and Ctrl mice manifested diabetes within 2 days after STZ injection ( Figure 3F,G). Accordingly, nearly all (Nkx6.1 positive ) beta cells were lost 3 days after STZ injection independent of genotype (Fig. 3H). We conclude that depletion of P53 does not protect beta cells against STZ induced cell death and subsequent diabetes.
3.4. P53 ablation does not protect against Cytokine-or STZinduced beta cell death ex vivo While our results so far indicated that P53 ablation is not able to prevent beta cell dysfunction or loss of beta cells, we wanted to directly quantify beta cell and islet cell death in a controlled ex vivo setting. To distinguish between beta and non-beta cells, we used P53 BKO and Ctrl mice additionally carrying a Cre-inducible reporter gene in the Rosa26 locus, resulting in a beta cell specific expression of a red fluorescent protein (tdTomato). Islets of these Tomato Beta and Tomato:P53 BKO mice were isolated and treated with a pro-inflammatory cytokine mix (consisting of TNF-a, IL1-b and IFN-g) widely used to stimulate beta cell death [39], or STZ. The bright viability stain FVS520 was used to identify dead (FVS520 positive ) and living (FVS520 negative ) cells by flow cytometry. Since the tdTomato signal disappeared in dying cells, most likely due to membrane leakiness, we quantified the percentage of living beta (tdTomato positive and FVS520 negative ) cells, living non-beta (tdTomato negative , FVS520 negative ) cells, and dead (FVS520 positive ) cells ( Fig. 4A and S6A-H). 1 mM STZ treatment for 16 h specifically reduced the living beta cell population, while incubation with 1.5 mM STZ caused nearly all islet cells to die ( Figure 4B,C, E, G). In contrast, Cytokine incubation for 40 h resulted in 45e50% islet cell death, with both beta cells and interestingly, non-beta cells succumbing to the proinflammatory stimulus ( Figure 4B,D, F, H). In line with our observations in vivo, islets of Tomato:P53 BKO mice were neither protected against Cytokine-nor STZ-induced cell death (Figure 4CeH). Similarly, islets from mice with beta cell specific ablation of ATM, an important DNA damage sensor and upstream regulator of P53 [16]), were not protected against STZ-induced cell death (Figs. S7AeH). To ensure that we would be able to detect a protective effect against STZ-induced beta cell death in our experimental setup, we tested the Poly (ADP-Ribose) Polymerase 1 (PARP1) inhibitor Olaparib, since PARP1 KO mice are strongly protected against STZ-induced diabetes [40]. Moreover, PARP1 and P53 directly or indirectly interact in other cell types [41e 43]. Indeed, Olaparib was able to completely prevent STZ-induced cell death in isolated islets ( Figure 4I,J) without preventing activation of the early cellular response to DNA damage (phosphorylation of Histone H2A.X, Figure S7I and J). Accordingly, our findings suggest that PARP1-induced cell death after STZ treatment in pancreatic beta cells is not dependent on ATM or P53 signaling. Overall, we show that ablation of P53 fails to protect against two major cell stress stimuli linked to beta cell loss in T1D and T2D (inflammation and DNA damage) [18,20,44].

DISCUSSION AND CONCLUSION
Pro-apoptotic processes appear to be critical for dysfunction and loss of pancreatic beta cells observed in T1D, T2D and monogenetic forms of diabetes [9,14,35,45,46]. P53 is arguably the most important cellular regulator of cell death, and activation of P53 as well as other DDR proteins has been correlated to development of beta cell failure and loss [8,18,20]. While the functional role of P53 has mainly been studied in whole body (conventional) KO mice, or by pharmacologically induced systemic inhibition or activation in models of diabetes, comprehensive studies on the beta cell specific role of P53 in vivo have been missing. To address this, we analyzed the phenotype of beta cell specific P53 KO mice in several complementary models of diabetes development. We note that the knock-in Ins1-Cre strain used in our studies is distinct from several older transgenic beta cell Cre driver strains, that appear to affect beta cell function independent of conditional alleles, in part due to unintentional expression of growth hormone (GH) from the inserted transgene [47,48]. Indeed, we confirmed that at least one inducible beta cell Cre driver strain with detectable GH expression was highly resistent against STZ-induced hyperglycemia (data not shown). Overall, our experimental approaches allowed us to investigate the effect of low, intermediate, high and extreme levels of beta cell stress and impairment in vivo and ex vivo. In contrast to our expectations, our results robustly show that P53 in beta cells is not involved in the regulation of beta cell numbers or function in multiple complementary models of diabetes in vivo. It is not immediately apparent, why the observed strongly reduced expression of pro-apoptotic mRNAs such as PHLDA3, BAX and BBC3 is insufficient to ameliorate beta cell death in P53 BKO mice. For example, conventional PHLDA3 KO mice demonstrate diminished islet cell death as well as increased islet size [36]. Our results are in line with the notion that several parallel cellular pathways, with P53 being one of them, are activated during beta cell stress, which may underlie the sparse success of existing beta cell protective drugs to decisively reverse loss of insulin secretion capacity and diabetes progression in the clinical setting. Our findings are also along the line that P53 regulated pathways and mechanisms in cell types apart from beta cells mediate its effects on systemic physiology and glucose metabolism, such as endothelial cells [49]. Fluorescent reporter protein expression also allowed us to directly assess cell death in beta cell and non-beta cell compartments in pancreatic islets. We hereby directly demonstrated that the cytokine mix of TNF-a, IL1-b and IFN-g that is widely used to induce beta cell death, as well as high dose STZ treatment, induces cell death of beta cells, but surprisingly also of non-beta cells. Future studies are needed to replicate this finding and identify the islet cell type that may be highly sensitive to inflammatory stress, since e.g. alpha cells are thought to be especially protected from Cytokine-mediated cell death [50]. In any case, since mechanistical studies on purported beta cell death in the context of T1D and T2D sometimes use methods not technically suited for cell type specific resolution of cell death (e.g. cleaved caspase immunoblots from whole islet extracts), our results clearly argue for use of more precise tools to quantify cell death. As a limitation of our study, our experimental design allowed us to test multiple, but not all relevant or potentially occurring types of beta cell stress linked to development of diabetes. Hence, we point out that some (but potentially rare) types of beta cell stress may induce loss of beta cells in a P53-dependent manner, e.g. glucokinase hyperactivation or abnormal microRNA expression [8,18]. Beta cell loss may be further influenced by the specific genetic makeup of the mouse strains involved as well as experimental and environmental conditions (for example, exact diet composition and gut microbiome). Nonetheless, our studies markedly indicate that targeting canonical P53dependent pro-apoptotic pathways appears to be an unfavourable approach for development of novel therapies for beta cell loss and

DATA AVAILABILITY
Data will be made available on request.