Elimination of Vitamin D Signaling Causes Increased Mortality in a Model of Overactivation of the Insulin Receptor: Role of Lipid Metabolism

Vitamin D (VD) deficiency has been associated with cancer and diabetes. Insulin signaling through the insulin receptor (IR) stimulates cellular responses by activating the PI3K/AKT pathway. PTEN is a tumor suppressor and a negative regulator of the pathway. Its absence enhances insulin signaling leading to hypoglycemia, a dangerous complication found after insulin overdose. We analyzed the effect of VD signaling in a model of overactivation of the IR. We generated inducible double KO (DKO) mice for the VD receptor (VDR) and PTEN. DKO mice showed severe hypoglycemia, lower total cholesterol and increased mortality. No macroscopic tumors were detected. Analysis of the glucose metabolism did not show clear differences that would explain the increased mortality. Glucose supplementation, either systemically or directly into the brain, did not enhance DKO survival. Lipidic liver metabolism was altered as there was a delay in the activation of genes related to β-oxidation and a decrease in lipogenesis in DKO mice. High-fat diet administration in DKO significantly improved its life span. Lack of vitamin D signaling increases mortality in a model of overactivation of the IR by impairing lipid metabolism. Clinically, these results reveal the importance of adequate Vitamin D levels in T1D patients.


Introduction
Type 1 diabetes mellitus (T1D) develops as a consequence of pancreatic beta-cell destruction and it is characterized by insulin deficiency, a tendency to ketosis and dependence on exogenous insulin to sustain life. Glycemic control in type 1 diabetes is of paramount importance, as it has been demonstrated to be clearly associated with long term complications. Thus, there is unquestionable evidence of a very close relationship between hemoglobin A1c (HbA1c) concentrations, maintained over the long term, and the onset or progression of microvascular and macrovascular complications [1,2]. Therefore, insulin administration is routinely used in T1D patients. There has been a dramatic increase in the types of insulin To induce Cre-mediated PTEN and VDR ablation, a single intraperitoneal injection of 25 mg/kg tamoxifen solution per 4-5 weeks old mice was given as previously described [18]. Tamoxifen (Sigma-Aldrich) was dissolved in absolute 100% ethanol and then diluted in corn oil (Sigma-Aldrich) to a 5 mg/mL concentration. All the in vivo experiments were performed 4-5 weeks after tamoxifen-induced Cre activation.
Animals were kept in a 12-h light-dark cycle at 22 • C and ad libitum access to water and regular mouse diet (Teklad Global 14% Protein 4% Fat Rodent Maintenance Diet-Envigo. Harlan Teklad, Madison, WI, USA) or high-fat diet (HFD, Paigen Diet 10 MM S9358-E030-1.25% cholesterol, 0.5% cholic acid, 15% cocoa butter, 1% corn oil). Then, 24 h before sacrifice, animals were moved to metabolic cages to quantify food intake. All surgical procedures were performed under general anesthesia with isoflurane. On the day of sacrifice, total body weights were measured. Animals were sacrificed and blood samples were collected by cardiac puncture. Organs were perfused with a saline solution through the left ventricle. Snap-frozen tissue samples were collected for molecular biology and 3 H glucose analysis and 4% paraformaldehyde-fixed samples for histological assessments.

Glucose Metabolism Analysis
Blood glucose levels were measured with a glucometer (Roche, Basel, Switzerland). To analyze glucose metabolism during fasting, glycemia was analyzed at 2 and 7 h after food removal. Overnight periods of fasting were also used. For the glucose tolerance test (GTT), experiments started 3 h after fasting (time 0) by administration of a glucose (Sigma-Aldrich, St. Louis, MO, USA) single-injection (4 g/kg; i.p.) and glucose measurements were performed at 20, 40, 60 and 120 min. For the pyruvate tolerance test (PTT), the same protocol was followed, replacing glucose with a dose of 2 g/kg sodium pyruvate (Life Technologies, Paisley, UK).

3 H Glucose Detection
After 3 h of fasting, a single i.p. bolus of a mixed solution of normal and radioactive glucose ( 3 H glucose Perkin Elmer, Westerville, Ohio, USA) 20 µCi per mouse in 3 g/kg glucose) was administered, and animals were sacrificed after 120 min. Then, different tissues were collected and homogenized with stainless steel beads (Qiagen, Hilden, Germany) using a TissueLyser LT (Qiagen). Subsequently, supernatants were mixed 1:1 with 7% HClO 4 (Sigma-Aldrich), centrifuged and neutralized for 30 min with a 2.2 M KHCO 2 (Sigma-Aldrich) solution. Then, samples were centrifuged at 14,000× g, the precipitate was discarded and supernatant was mixed with 10 mL of scintillation liquid to determine total 3 H radioactivity [22]. Radioactivity was measured using a scintillation counter (Packard 1900 TR).

Glucose Intraventricular Delivery in the CNS and Osmotic Implantation
Osmotic pumps (Alzet Mini-Osmotic pump model 2006, duration 42 days with a 0.15 µL/h pumping rate) were filled with a saturated solution of D-(+)-glucose or Dmannitol (Sigma-Aldrich) and hydrated with saline at 37 • C for 60 h before surgery. Intracerebroventricular infusion of glucose or mannitol was performed as previously reported by DeVos and Miller [23].

Serum and Urine Biochemistry
Total cholesterol (TC), HDL-C and triglycerides (TG) were measured by colorimetric methods according to standardized protocols with an AU5800 Analyzer (Beckman Coulter Inc., Fullerton, CA, USA) in the Clinical Analysis Laboratory of Arnau de Vilanova University Hospital, in Lleida, Spain. LDL-C was calculated by the Friedewald equation if TG < 250 mg/dL or by a colorimetric method if TG > 250 mg/dL. In order to determine serum c-peptide concentrations, an ELISA kit (Millipore, Bedford, UK) was used as indicated by the manufacturer. The blood urea nitrogen concentration was measured using a colorimetric assay (Spinreact, Barcelona, Spain). Serum 25-hydroxy-vitamin D (25(OH)D 3 ) and 1,25-dihydroxy-vitamin D (1,25(OH) 2 D 3 ) levels were quantified with an enzyme immunoassay (Immunodiagnostic Systems, Boldon Business Park, UK) following the manufacturer's instructions.

Histopathology Analysis
Paraffin blocks were cut at 5 µm, dried at 60 • C for 30 min and then dewaxed and rehydrated for hematoxylin (PanReac AppliChem ITW Reagents, Barcelona, Spain) and eosin (Master diagnostic MAD-109 1000) staining. For the periodic acid Schiff-alcian blue (PAS-AB) staining, after rehydration, 5 µm slices were incubated with AB for 5 min, followed by 15 min with PAS, and finally, 25 min in the Schiff solution.

Hepatic Glycogen Detection
Liver samples (100 mg) were homogenized with stainless steel beads (Qiagen) in 500 µL of homogenization buffer (50 mM TrisHCl pH 7.5, 5 mM EDTA, 1 mM DTT, 10 µL/mL PMSF, 5 µL/mL PIC, 5 µL/mL Na 3 VO 4 ) using a TissueLyser LT. Then, liver lysates were mixed with 100 µL of 50 mM TrisHCl buffer and 100 µL of 0.2 M perchloric acid and subsequently centrifuged at 14,000× g for 5 min. Supernatant was transferred to a new tube with 300 µL of 90% ethanol and maintained at −20 • C overnight. Then, samples were centrifuged and pellets containing the glycogen were let to dry. Once dried, glycogen pellets were mechanically suspended in 2 M HCl. Then, the solution was incubated at 100 • C for 20 min and subsequently the reaction was stopped by adding a 1 M NaOH and 1% 3,5-dinitrosalicylic acid solution. Finally, samples were incubated at 100 • C for 5 min and absorbance was read at 546 nm.

RNA Isolation and Quantitative Reverse Transcription PCR (qRT-PCR)
A total of 20 mg of tissue was used for total RNA isolation from liver samples using an RNA isolation kit (Macherey-Nagel, Allenton, PA, USA) following the manufacturer's instructions. RNA concentration was measured using a NanoDrop spectrophotometer and stored at −80 • C.

Statistical Analyses
All experiments were carried out at least three times. Statistical analyses were performed with GraphPad Prism 8.02 software. Values are presented as mean ± SEM. Com- parisons were assessed using one-way ANOVA followed by Tukey's test for multiple comparisons with one categorical variable and two-way ANOVA followed by the Sidak test for multiple comparisons with two categorical variables. Survival ratio analyses were performed using the Mantel-Cox test. A p < 0.05 was considered to be significant.

Lack of VDR Reduces Lifespan in PTEN Knockout Mice
To investigate the role of VDR in hypoglycemia induced by overactivation of the insulin receptor, we mated Cre-ERT:PTENfl/fl (PTEN-KO) mice with Cre-ERT:VDRfl/fl (VDR-KO) knockout mice to generate double knockout mice (DKO). At the age of two months, tamoxifen was administered, resulting in genetic excision of the floxed exons after 4-5 weeks (Supplementary Materials, Figure S1A) and a low or undetected PTEN protein, and increased AKT phosphorylation in both, PTEN-KO and DKO mice (Supplementary Materials, Figure S1B,C).
All the animals in the CNT group and in the VDR-KO group stayed alive after 65 days of Cre-induced VDR ablation ( Figure 1A), whereas animals in the PTEN-KO group showed worse survival (76.2%). Of note, DKO mice presented notable, excessive mortality, starting 20 days after Cre-induced target genes ablation, and resulting in the death of all the animals at 65 days. The patterns of the survival curves in the DKO mice were similar for both sexes ( Figure 1B).

Physiological and Biochemical Parameters
The physiological and serum biochemical parameters are shown in Table 1. All animal had ad libitum access to chow, and food intake was increased in PTEN-KO and DKO animals; however, total body weight decreased in both groups as compared with controls.

Physiological and Biochemical Parameters
The physiological and serum biochemical parameters are shown in Table 1. All animal had ad libitum access to chow, and food intake was increased in PTEN-KO and DKO animals; however, total body weight decreased in both groups as compared with controls.
Serum peptide C concentration was decreased in PTEN-KO and DKO animals as compared with the CNT and VDR-KO groups, indicating decreased insulin secretion, which is associated with lower blood glucose levels. As expected, serum 1,25(OH)2D3 levels were increased in both the VDR-KO and DKO groups due to the absence of VDR, but decreased in PTEN-KO mice. The serum concentrations of 25(OH)D3 were reduced in both PTEN-KO and DKO mice. Total cholesterol (TC) and HDL cholesterol (HDLC) were significantly reduced in PTEN-KO and DKO mice as compared with the VDR-KO group and LDL cholesterol (LDLC) showed a tendency to be reduced in DKO. Serum peptide C concentration was decreased in PTEN-KO and DKO animals as compared with the CNT and VDR-KO groups, indicating decreased insulin secretion, which is associated with lower blood glucose levels. As expected, serum 1,25(OH) 2 D 3 levels were increased in both the VDR-KO and DKO groups due to the absence of VDR, but decreased in PTEN-KO mice. The serum concentrations of 25(OH)D 3 were reduced in both PTEN-KO and DKO mice. Total cholesterol (TC) and HDL cholesterol (HDLC) were significantly reduced in PTEN-KO and DKO mice as compared with the VDR-KO group and LDL cholesterol (LDLC) showed a tendency to be reduced in DKO.

Blood Glucose Tests Reveal a Disruption of Glucose Metabolism in PTEN-KO and DKO Mice
Glucose levels were lower in the PTEN-KO and the DKO group, even with unrestricted access to food (Fed state, Figure 2A). Furthermore, after 7 h of food restriction, levels of glucose in the DKO mice had a tendency to be lower than in the PTEN-KO mice. Overnight food restriction led to a 100% mortality in DKO animals.   The results from the glucose and pyruvate tolerance tests are shown in Table 2. Intraperitoneal administration of glucose (Table 2A) produced a significant increase in blood glucose levels in both CNT and VDR-KO mice, which returned to basal values after 2 h. In PTEN-KO animals, the peak was also present, but was smaller and returned to basal values after 60 min. Animals in the DKO group showed a very small increase in glucose levels 20 min after glucose administration, which returned to basal levels at 40 min. Administration of pyruvate (Table 2B) showed similar traits in CNT and VDR-KO mice with glucose peaks that returned to normal values after 2 h. PTEN-KO mice and DKO mice showed a smaller peak in glucose after 20 min, which returned to basal values after 40 min. Therefore, all these results point to a higher utilization of glucose in DKO mice with respect to PTEN-KO mice.

Glucose Supplementation Did Not Increase Survival in DKO Mice
As the DKO group showed severe hypoglycemia, we investigated whether a sucrose supplementation in the drinking water could result in increased survival or extended lifespan. We observed a 100% mortality approximately 100 days after Cre-induced PTEN and VDR ablation, with no differences between the sucrose and vehicle group ( Figure 2B).
As severe hypoglycemia leads to functional brain failure and hypoglycemic coma, we also studied whether direct intracerebroventricular infusion of glucose could be beneficial in DKO mice ( Figure 2C). We observed that direct infusion of glucose into the cerebral ventricle did not increase animal survival.
We also studied whether glucose consumption was higher in DKO mice that in PTEN mice in a particular organ, as a way to explain the increased utilization of glucose. As shown in Figure 2D, there was a higher intake of glucose in many organs (p < 0.01 in the genotype comparison in two-way ANOVA), but the profile of each organ was not modified by the genotype (the interaction was not significant).

Faster Reduction in Glycogen Pool in Fasting DKO Mice
To determine the glycogenic hepatic capacity, total glycogen levels from mouse liver lysates were measured ( Figure 3A). In a fed state, all groups showed similar levels of liver glycogen; however, lower glycogen concentration was observed after two hours of fasting in the DKO group as compared with the CNT and VDR-KO groups at the same time point, pointing to a faster use of glycogen stores. However, after 7 h of fasting, all groups showed similar levels of glycogen in liver. Hematoxylin-eosin and PAS staining demonstrated hepatocellular ballooning in the PTEN-KO and DKO mice as compared with the CNT and VDR-KO groups after 7 h of fasting ( Figure 3B).

Delayed Gluconeogenesis in PTEN-KO and DKO Mice
In order to study the glucose metabolism in livers, we analyzed the genetic expression of PEPCK and G6PC, genes involved in gluconeogenesis. In CNT and VDR-KO mice, the expressions of PEPCK ( Figure 4A) and G6PC ( Figure 4B) were increased after 2 h of fasting and were subsequently restored after 7 h of fasting. This fact points to an early induction of glucose generation in the liver that could be already enough to re-establish glycemia. However, in PTEN-KO and DKO PEPCK and G6PC, gene expression reaches a peak after 7 h of fasting, indicating delayed gluconeogenesis.
Additionally, we analyzed the CEBPA and PGC1α gene expression, both genes that are implicated in the transcriptional regulation of PEPCK and G6PC, among other liver metabolism genes. We observed similar liver CEBPA gene expression in the CNT and VDR-KO groups in fed and fasting states, whereas in the PTEN-KO and DKO mice it was downregulated throughout fasting ( Figure 4C). PGC1α gene expression also remained similar in CNT and VDR-KO animals in fed and fasting states, however it increased by

Delayed Gluconeogenesis in PTEN-KO and DKO Mice
In order to study the glucose metabolism in livers, we analyzed the genetic expression of PEPCK and G6PC, genes involved in gluconeogenesis. In CNT and VDR-KO mice, the expressions of PEPCK ( Figure 4A) and G6PC ( Figure 4B) were increased after 2 h of fasting and were subsequently restored after 7 h of fasting. This fact points to an early induction of glucose generation in the liver that could be already enough to re-establish glycemia. However, in PTEN-KO and DKO PEPCK and G6PC, gene expression reaches a peak after 7 h of fasting, indicating delayed gluconeogenesis. approximately 3-fold after 7 h of fasting in PTEN and DKO mice, showing earlier upregulation in the DKO group ( Figure 4D).
We also measured the gene expression of GLUT2, a glucose transporter found in hepatocytes that mediates glucose diffusion across cell membranes. We observed that GLUT2 gene expression was slightly increased in fasting CNT and VDR-KO mice, whereas it remained similar in the PTEN-KO and DKO groups ( Figure 4E).

High-Fat Diet Increased DKO Survival
As fatty acid β-oxidation is used as an alternative source of energy if glucose is not available, we investigated the expression of enzymes related to that process. First, we observed a total absence of abdominal adipose tissue in PTEN-KO and DKO mice ( Figure  5A). Second, we investigated several genes involved in fatty acid oxidation. Thus, expression of PPARA was increased after 2 h of fasting, but contrary to the rest of the groups, levels decreased in DKO animals after 7 h of fasting ( Figure 5B). Levels of CPT1 were almost unresponsive to starvation in the DKO group, in contrast to the rest of the groups ( Figure 5C). Regarding ACOX1 and FGF21, the expression of the genes in the DKO group was attenuated with respect to the PTEN-KO group ( Figures 5D and 5E respectively).
As fat metabolism seems to be involved in the pathological features, we studied the effects of a high-fat diet in the survival rate of DKO mice. Thus, a high-fat diet significantly extended the DKO lifespan ( Figure 5F). Additionally, we analyzed the CEBPA and PGC1α gene expression, both genes that are implicated in the transcriptional regulation of PEPCK and G6PC, among other liver metabolism genes. We observed similar liver CEBPA gene expression in the CNT and VDR-KO groups in fed and fasting states, whereas in the PTEN-KO and DKO mice it was downregulated throughout fasting ( Figure 4C). PGC1α gene expression also remained similar in CNT and VDR-KO animals in fed and fasting states, however it increased by approximately 3-fold after 7 h of fasting in PTEN and DKO mice, showing earlier upregulation in the DKO group ( Figure 4D).
We also measured the gene expression of GLUT2, a glucose transporter found in hepatocytes that mediates glucose diffusion across cell membranes. We observed that GLUT2 gene expression was slightly increased in fasting CNT and VDR-KO mice, whereas it remained similar in the PTEN-KO and DKO groups ( Figure 4E).

High-Fat Diet Increased DKO Survival
As fatty acid β-oxidation is used as an alternative source of energy if glucose is not available, we investigated the expression of enzymes related to that process. First, we observed a total absence of abdominal adipose tissue in PTEN-KO and DKO mice ( Figure 5A). Second, we investigated several genes involved in fatty acid oxidation. Thus, expression of PPARA was increased after 2 h of fasting, but contrary to the rest of the groups, levels decreased in DKO animals after 7 h of fasting ( Figure 5B). Levels of CPT1 were almost unresponsive to starvation in the DKO group, in contrast to the rest of the groups ( Figure 5C). Regarding ACOX1 and FGF21, the expression of the genes in the DKO group was attenuated with respect to the PTEN-KO group ( Figure 5D,E respectively). Nutrients 2022, 14, x FOR PEER REVIEW 12 of 17

Discussion
In the present study, we generated inducible double KO mice for PTEN and VDR. The animals with deletion of both genes showed an increased mortality, which was partially reverted when animals were placed on a high-fat diet. As fat metabolism seems to be involved in the pathological features, we studied the effects of a high-fat diet in the survival rate of DKO mice. Thus, a high-fat diet significantly extended the DKO lifespan ( Figure 5F).

Discussion
In the present study, we generated inducible double KO mice for PTEN and VDR. The animals with deletion of both genes showed an increased mortality, which was partially reverted when animals were placed on a high-fat diet.
PTEN was identified as a tumor suppressor gene on chromosome 10q23 [25]. Vitamin D, signaling through VDR, is also known to have a protective effect against some cancers [26]. Thus, a first possibility is that the elimination of VDR could increase the susceptibility of the animals to develop some kind of cancer. The embryonic lethality of mice with biallelic excision of PTEN has limited the study of complete PTEN ablation in the development of cancer. However, the generation of PTEN conditional-KO mice has solved that problem. Mirantes et al. [18], showed that the use of CREER to delete PTEN-generated mice with a tendency to develop cancer in the thyroid gland, prostate and endometrium. In our case, no macroscopical tumors were detected in the necropsies. Furthermore, the results of the 3 H glucose uptake showed that the interaction term in the ANOVA analysis was not significant, so a different effect of the phenotype in tissular uptake of 3 H is ruled out, and an aggravation of the cancer is unlikely to be the cause of death. The differences between our results and those of Mirantes et al. may be explained by two reasons. First, our animals were used 4-5 weeks after the administration of tamoxifen, contrary to the experiments of Mirantes et al. in which they waited 8 weeks for the sacrifice of the animals. Furthermore, to generate our crossings we introduced the SWR/J background into the C57BL6;129S4 mixed background used in the experiments from Mirantes et al., which could change the susceptibility to cancer.
The first important result in PTEN-KO and DKO mice is the lower body weight and the total absence of body fat with respect to the rest of the groups. As animals in these groups had a higher food intake, the results point to an increase in energy expenditure. Previous results have shown conflicting results regarding PTEN and energy expenditure. Thus, the deletion of PTEN in the liver decreases adiposity [27], but systemic overexpression of PTEN has been also shown to increase energy expenditure and decrease adiposity [28]. Constitutive VDR-KO mice also show an increase in energy expenditure and a higher food intake with a lower body weight [29,30]. Our inducible VDR-KO mice also showed a tendency to eat more and have a lower weight. However, those differences did not reach statistical significance with respect to the controls, probably due to the fact that in our case the deletion of VDR was performed in adulthood.
In a previous study by our group, we determined that inducible PTEN-KO mice had alterations in the glucose metabolism that caused hypoglycemia [31]. In addition, vitamin D signaling has been reported to regulate insulin secretion. Thus, vitamin D deficiency inhibits pancreatic secretion of insulin [32] and it is associated with insulin resistance [33]. Furthermore, mice lacking a functional vitamin D receptor show impaired insulin secretory capacity [34] together with insulin resistance [35]. The results of the fasting experiments reported that, although not significant with respect to the PTEN-KO mice, glucose levels had a tendency to be lower in the DKO mice after 7 h of fasting. Thus, it seemed that either glucose was used faster or produced in a lower rate in those mice. The glucose tolerance test results showed that although a smaller peak of glucose could be seen 20 min after glucose administration in the DKO mice with respect to the PTEN-KO mice, the levels dropped fast, achieving similar values in both groups after 40 min. Therefore, a significant effect due to a higher uptake of glucose in the tissues of DKO mice seems unlikely, as it was also shown in the 3 H-glucose experiments. The PTT also showed no significant differences in the rate of gluconeogenesis between both phenotypes. Furthermore, and although DKO animals seemed to decrease glycogen storage at a faster rate after 2 h of fasting, the differences were not significant after 7 h. Interestingly, after an overnight fasting period, 100% of the DKO animals died. When glucose levels in blood achieve a lower threshold, symptoms and signs of encephalopathy result. The blood glucose level at which cerebral metabolism fails and symptoms develop varies, but in general, confusion occurs at levels below 30 mg/dL and coma below 10 mg/dL [36]. Thus, it is possible that hypoglycemic coma in our animals, induced by an overactivation of the insulin receptor similar to an insulin overdose, can cause unresponsiveness and inability to feed and drink, causing the death of the animals. However, supplementation with sucrose in the drinking water was also unable to decrease the mortality observed in the DKO group. Furthermore, the experiments in which glucose was infused directly into the cerebral ventricles with an osmotic pump were also unable to reduce the excess mortality in the DKO mice.
The mechanisms by which insulin-induced hypoglycemia causes sudden death are not well characterized. It has been previously shown that fatal arrhythmias and seizures are involved in this fatal complication [37]. The influence of high cholesterol levels on cardiovascular mortality has been known for decades. However, recent results point to a deleterious effect of low lipid levels on cardiovascular events, especially in atrial fibrillation (AF). Thus, a study on 15 million Chinese participants showed that low HDLC was independently associated with a higher risk of atrioventricular block, whereas high TC was a protective factor [38]. The protective effect of high levels of TC against AF has been demonstrated in many other studies in different populations in Korea, Japan, USA, China and Sweden [39][40][41][42][43]. Low HDLC has been also found to be associated with increased risk of AF [39,[44][45][46]. In our mice, lower levels of TC and HDLC were found in PTEN-KO mice after 7 h of fasting, and showed a tendency to be even lower in DKO mice. Furthermore, the maintenance of the animals on a high-fat diet was the only strategy able to increase its lifespan.
In fasting adult mammals, 60-80% of cardiac energy metabolism relies on the oxidation of fatty acids (FAs) with glucose, lactate, and ketones providing substrates for the remainder [47]. Both PTEN and vitamin D show effects on hepatic lipid metabolism. Thus, liver specific PTEN-KO mice show increased fatty acid synthesis, accompanied by hepatomegaly and fatty liver phenotype [48]. In contrast, VDR deletion induces lipid oxidation and fat consumption in hepatocytes [12]. In our inducible PTEN-KO mice, we observed that fasting-mediated induction of PPARA, a regulator of hepatic metabolism activated by fatty acids [49], was not as elevated as in CNT and VDR-KO mice. This induction was further reduced in the DKO mice, reaching basal levels after 7 h of fasting. PPARA controls gene expression levels of the rate-limiting enzymes of peroxisomal β-oxidation, including ACOX1 [49], which showed a similar profile to PPARA. Another gene implicated in fatty acid metabolism is CPT1. The protein encoded by CPT1 is responsible for the carnitine-dependent transport of fatty acids across the mitochondrial inner membrane. In our model, CPT1 expression was increased by starvation in all the groups except in the DKO animal. Finally, another PPARA-mediated target, FGF21, was increased in PTEN-KO animals but it did not show the same profile in DKO animals. The effects of FGF21 on the liver are not completely understood but reports show that it stimulates the oxidation of fatty acids [50]. Therefore, it seems that in the DKO mice, together with the alterations in glucose metabolism leading to hypoglycemia, alterations in lipid metabolism leading to delays in the fatty acid oxidation pathways can increase the mortality rate.

Conclusions
Taken together, the results shown in the present paper point to the paramount role of an adequate vitamin D signaling pathway in hypoglycemia induced by overactivation of the insulin receptor. Thus, in T1 diabetic patients, especially in the lean phenotype, maintaining correct levels of vitamin D could support proper lipid metabolism and decrease deaths induced by insulin dosing errors.