Liver-specific ablation of insulin-degrading enzyme causes hepatic insulin resistance and glucose intolerance, without affecting insulin clearance in mice

Basic Science Liver-specific ablation of insulin-degrading enzyme causes hepatic insulin resistance and glucose intolerance, without affecting insulin clearance in mice Pablo Villa-Pérez a , Beatriz Merino a , Cristina M. Fernández-Díaz a , Pilar Cidad a ,CarmenD.Lobatón a , Alfredo Moreno a , Harrison T. Muturi b,e , Hilda E. Ghadieh b,e , Sonia M. Najjar b,e , Malcolm A. Leissring c , Irene Cózar-Castellano a , Germán Perdomo d, ⁎ a Instituto de Biología y Genética Molecular, University of Valladolid-CSIC, Valladolid, Spain b Department of Biomedical Sciences, Ohio University, USA c Institute for Memory Impairments and Neurological Disorders, University of California, UCI MIND, Irvine, CA, USA d Departamento de Ciencias de la Salud, Universidad de Burgos, Burgos, Spain e Diabetes Institute, Heritage College of Osteopathic Medicine, Ohio University, USA abstractarticle info Article history: Received 23 March 2018 Accepted 6 August 2018 The role of insulin-degrading enzyme (IDE), a metalloprotease with high affinity for insulin, in insulin clearance remains poorly understood. OBJECTIVE: This study aimed to clarify whether IDE is a major mediator of insulin clearance, and to define its role in the etiology of hepatic insulin resistance. Methods: We generated mice with liver-specificdeletionofIde (L-IDE-KO) and assessed insulin clearance and action. Results: L-IDE-KO mice exhibited higher (~20%) fasting and non-fasting plasma glucose levels, glucose intolerance and insulin resistance. This phenotype was associated with ~30% lower plasma membrane insulin receptor levels in liver, as well as ~55% reduction in insulin-stimulated phosphorylation of the insulin receptor, and its downstream signaling molecules, AKT1 and AKT2 (reduced by ~40%). In addition, FoxO1 was aberrantly distributed in cellular nuclei, in parallel with up-regulation of the gluconeogenic genes Pck1 and G6pc. Surprisingly, L-IDE-KO mice showed similar plasma insulin levels and hepatic insulin clearance as control mice, despite reduced phosphorylation of the carcinoembryonic antigen-related cell adhesion molecule 1, which upon its insulin-stimulated phosphorylation, promotes receptor-mediated insulin uptake to be degraded. Conclusion: IDE is not a rate-limiting regulator of plasma insulin levels in vivo. © 2018 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http:// creativecommons.org/licenses/by/4.0/). Keywords: Insulin-degrading enzyme Hepatic insulin resistance Insulin receptor Carcinoembryonic antigen-related cell adhesion molecule 1 1. Introduction Insulin-degrading enzyme (IDE) is a 110-kDa zincmetalloendopeptidase first identified and named based on its ability to bind to and degrade insulin [1]. IDE also degrades several other intermediate-sized (b80 amino acids) bioactive peptides, including glucagon, amylin and the amyloid β-protein [2]. IDE is ubiquitously expressed at varying levels in both insulin-responsive and -nonresponsive cell types. Subcellularly, the protease is primarily cytosolic, but it has been reported to exist within a number of intracellular vesicles, and organelles, as well as being associated with membranes, and secreted into the extracellular space [1,3–6]. This wide distribution suggests a dynamic and multifunctional role for IDE. Because IDE has a high affinity for insulin (K M =~100nM),ithas been proposed as the major enzyme responsible for its catabolism [3,7]. In vivo, insulin is produced and secreted by pancreatic β-cells, which is delivered through the portal venous system to the liver, where it is primarily cleared, although kidneys and other tissues also degrade the hormone [1]. In first-pass transit, the liver removes ~50% of circulating insulin. Although the detailed molecular mechanisms of hepatic insulin clearance are not fully understood, insulin uptake and intracellular degradation is a receptor-mediated process that occurs after the binding of the hormone to its receptor (IR) [8]. The insulin- IR complex is endocytosed via clathrin-coated vesicles and is primarily delivered to endosomes. In the acidic milieu of the endosomes, insulin dissociates from the IR, at which point insulin is believed to be degraded by IDE, and the receptor is recycled back to the plasma membrane [1]. In recent years, a role for the carcinoembryonic antigenrelated cell adhesion molecule 1 (CEACAM1) in promoting receptormediated insulin uptake has emerged [9]. Upon its phosphorylation by the receptor, CEACAM1 binds to and increases the rate of the Metabolism Clinical and Experimental 88 (2018) 1–11 ⁎Corresponding author at: Universidad de Burgos, Facultad de Ciencias de la Salud, Departamento de Ciencias de la Salud, P.O. Comendadores s/n (Antiguo Hospital Militar), 09001 Burgos, Spain. E-mail address: [email protected] (G. Perdomo). https://doi.org/10.1016/j.metabol.2018.08.001 0026-0495/© 2018 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Contents lists available at ScienceDirect Metabolism Clinical and Experimental journal homepage: www.metabolismjournal.com insulin-receptor complex uptake to be targeted to the cellular degradation pathways, contributing to the regulation of systemic insulin concentrations [10]. The physiological role of IDE in insulin clearance in vivo began to be clarified with the study of mice with global null mutation of Ide (IDE-KO mice). Consistent with a functional role for IDE in hepatic insulin clearance, IDE-KO mice exhibit chronic hyperinsulinemia resulting from impaired insulin clearance [11,12]. They also exhibit marked glucose intolerance and insulin resistance [11,12]. These results spurred the notion that transient and/or partial inhibition of IDE may represent a new pharmacological approach against type 2 diabetes (T2DM) [11,13,14]. However, pharmacological inhibition of IDE in mice caused contradictory observations regarding glucose homeostasis [15–17]. Furthermore, genetic polymorphisms within or near the Ide locus have been linked to increased, rather than decreased, risk for T2DM [18–22]. Therefore, the role of IDE in insulin clearance, and how this may modulate insulin resistance remain to be clarified. To address these questions, we ablated Ide expression exclusively in hepatocytes, the major site of insulin clearance, and investigated its effect on insulin clearance and action. 2. Material and Methods 2.1. Mice Generation and Genotyping Micewerefedstandardrodentchowdietandwateradlibitum in ventilated cages under a 12:12-h light-dark cycle at the animal facility of the University of Valladolid (UVa). L-IDE-KO mice were generated from crosses between the B6.Cg-Tg(Alb-cre)21Mgn/J mouse (Alb-Cre mouse (The Jackson Laboratory, USA) that expresses Cre recombinase cDNA from the mouse albumin (Alb) promoter [23], and the Ide flox/flox mouse on the C57BL/6 J (B6) background (from Dr. Malcolm A. Leissring, University of California, Irvine). This mouse has loxP sites flanking exon 3 of the Ide gene. Cre recombinase-mediated deletion of exon 3 causes a frameshift with two stop codons in exon 4 and early termination of translation. Twenty-three independent cohorts (males and females) of the F3 generation were used for metabolic studies and tissue collection. The IDE-KO mice were kindly provided by Dr. Malcolm A. Leissring [11].TheAnimalCareandUseCommitteeoftheUVaapprovedall experiments (protocol #5003931). The Alb-Cre+ control mice did not show abnormal metabolic phenotype relative to the C57BL/6J mice that were used in backcrossing. As Supplementary Table 1 reveals, Alb-Cre+ controls manifested comparable body weight and visceral obesity to that of C57BL/6J, in addition to normoinsulinemia and normal insulin clearance (measured by steadystate molar ratio of C-peptide/insulin levels). These control mice also showed normal insulin and glucose tolerance by comparison to C57BL/6J mice (Supplementary Fig. 1). Thus, we only used Ide flox/flox; +/+ as wildtype (WT) controls in these studies. Mice genotyping was performed using genomic DNA isolated from mouse tail as described previously [24], followed by three PCRs to amplify: gapdh,Ide and Alb-Cre genes. Breeding strategy and genotyping of mice are detailed in the Supplementary Experimental Procedures. 2.2. Plasma Biochemistry Fasting (16 h) and non-fasting (fed ad libitum) blood glucose and plasma triglycerides levels were assessed as described previously [24]. Plasma mouse insulin and glucagon levels were assessed using human or mouse enzyme-linked immunosorbent (ELISA) assays following the manufacturer's instructions (Mercodia, Sweden). Plasma amylin (Phoenix Pharmaceutical, USA) and amyloid (Aβ40) (Invitrogen, USA) levels by ELISA assays following the manufacturer's instructions. Plasma C- peptide levels were assessed by ELISA (ALPCO, USA). 2.3. Hepatic Triglycerides and Glycogen Levels Hepatic triglyceride content was assessed as described previously by Perdomo et al. [25]. Liver glycogen content was assessed using the Glycogen Assay Kit (Sigma-Aldrich, USA). Briefly, liver samples (10 mg) were homogenized in 100 μL of water supplemented with protease and phosphatase inhibitor cocktails (Sigma-Aldrich, USA) on ice. Homogenates were boiled at 100 °C for 5 min to inactivate enzymes and centrifuged 5 min at 12300 ×gat room temperature to remove insoluble materials. Homogenates were diluted 1:100 in hydrolysis buffer up to a final volume of 50 μLandmixedwith2μL of Hydrolysis Enzyme Mix for 30 min at room temperature. Glycogen concentration was determined by a coupled enzyme assay, which yields a colorimetric (570 nm) product, proportional to the glycogen present in the homogenate. 2.4. Metabolic Studies To evaluate alterationsin glucose homeostasis in vivo, we performed intraperitoneal glucose tolerance tests (IP-GTTs) and insulin tolerance tests (IP-ITTs) as described previously [26]. In vivo plasma insulin clearance was performed as described by Ahrén et al. [27] with some modifications. Briefly, 3-month-old mice were fasted for 2 h, then anesthetized using a solution of medetomidine plus ketamine at 0,125 mg/100 g body weight. Afterwards, human insulin (Humulin; Lilly, USA) was injected intra-orbitally at 1 nmol/kg and blood samples were collected from the tail vein 0, 5, 10, 20, 30 and 60 min after insulin administration. Plasma human insulin levels were quantified using a human ELISA kit (Mercodia, Sweden) following the manufacturer's instructions. 2.5. Quantitative Real-time PCR Total RNA from homogenized mouse tissues was isolated using Trizol Reagent (Sigma-Aldrich, USA) following the manufacturer's protocol. After DNAse treatment 500–1000 ng of RNA was used to synthesize Ide mRNA levels (A.U.) Brain Lung Heart Liver Kidney SK muscle WAT 0.0 0.5 1.0 1.5 * L-IDE-KO WT * A B Fig. 1. Analysis of IDE expression in WT and L-IDE-KO mice. (A) Expression of Ide mRNA in 3-month-old male L-IDE-KO and control littermates. Expression levels are mean ± SEM. n= 4 per genotype. *pvalue b0.05 vs. WT by Students' t-test. (B) Hepatic IDE protein levels in WT, L-IDE-KO, and IDE-KO mice. Representative western blots of liver lysates (40 μg protein/sample) isolated from 3-month-old WT, L-IDE-KO, and IDE-KO mice using anti-IDE and anti-actin antibodies. n= 2 per genotype. 2P. Villa-Pérez et al. / Metabolism Clinical and Experimental 88 (2018) 1–11 cDNA with the Transcriptor First Strand cDNA Synthesis Kit (Roche, USA). mRNA levels were determined by real time qPCR with TaqMan® probebased assays on a Rotor-Gene 3000 instrument (Corbett Research) using the 2 −ΔΔCt relative quantification method [28]. TaqMan® Gene Expression assay references (from Applied Biosystems, USA) were as follows: Mm00473077_m1 for Ide, Mm01247058_m1 for phosphoenolpyruvate carboxykinase (Pck1), Mm00839363_m1 for glucose-6 phosphatase (G6pc), Mm01211875_m1for insulin receptor (Insr), and Mm00802831_m1 for insulin-like growth factor-1 receptor (Igf1r). Expression data were normalized to the level of the housekeeping gene of the ribosomal protein L18 (RPL18; Forward: 5′-AAGACTGCCGT GGTTGTGG-3′; Reverse: 5′-AGCCTTGAGGATGCGACTC-3′; Probe: 5′- FAM-TTCCCAAGCTGAAGGTGTGTGCA-BHQ1–3′). 2.6. Western Blot Analysis Western blot analyses on isolated mouse tissues, and subcellular localization of FoxO1 are detailed in the Supplementary Experimental Procedures. Rabbit anti-FoxO1 antibody was used as described by Qu et al. [29]. For co-immunoprecipitation experiments, polyclonal antibodies against Shc (B-9) was used to immunoprecipitate proteins from tissue lysates as previously described [30,31]. 2.7. Analysis of β-Cell Function, Pancreatic Histology and Histomorphometry To asses β-cell function, mice were fasted overnight followed by an i. p. injection of a bolus of glucose. Blood samples were drawn, and plasma insulin levels were measured during 30 min. as described above. Afterwards, mice were euthanized and pancreata dissected, fixed in 10% neutral buffer formalin, paraffin embedded, and sectioned as described previously [32]. For islets histomorphometry, sections were stained with anti-insulin antibody (Abcam, UK). β-cell mass, β-cell area, the number of islets, and the mean islets size were quantitated using the ImageJ software (NIH, USA) as described previously [32]. 2.8. Statistical Analyses Statistical analysis of data was performed using Prism v. 6.0 (GraphPad Software). Distributions were checked with the Kolmogorov-Smirnov test. Data are presented as means ± SEM. Comparisons between two groups were done using the unpaired Students' t-test. Comparisons between more than two groups were done using the one-way ANOVA (Bonferroni test for post-hoc analyses). Differences were considered significant at pb0.05. BA Blood glucose (mg/dL) 0 50 100 150 * Non-Fasting Body weight (g) 0 10 20 30 Non-Fasting Blood glucose (mg/dL) 0 20 40 60 80 100 * Fasting Body weight (g) 0 10 20 30 Fasting CD E Food Intake [kcal/bodyweight (g)] WT L-IDE KO WT L-IDE KO WT L-IDE KO WT L-IDE KO WT L-IDE KO 0.0 0.5 1.0 1.5 Fig. 2. Metabolic features of L-IDE-KO mice. Fasting (A) and non-fasting (B) blood glucose levels in 3-month-old male WT and L-IDE-KO mice. Data are mean ± SEM. n=12–13 per genotype. *pvalue b0.05 vs. WT by Students' t-test. Fasting (C) and non-fasting (D) body weight in WT and L-IDE-KO mice. Data are mean ± SEM. n = 12–13 per genotype. Food intake (E).Dataaremean±SEM.n=5–6 per genotype. 3P. Villa-Pérez et al. / Metabolism Clinical and Experimental 88 (2018) 1–11 3. Results 3.1. Generation and Molecular Characterization of Liver-specificIDEKnock- out Mice 3-month-old male and female Ide flox/flox; +/+ mice (henceforth, wildtype-WT) and Ide flox/flox; Alb-Cre/+ (L-IDE-KO) mice were studied. Hepatic Ide mRNA (Fig. 1A) and protein levels (Fig. 1B) were decreased by ~90–100% in L-IDE-KO mice, compared to controls. No differences in IDE expression were observed in other tissues, except for a ~20% reduction in Ide mRNA levels without changes in IDE protein levels in skeletal muscle (data not shown). Considering that hepatocytes make up ~85% of liver cells, it is likely that the minimal amount of IDE detected in liver of L-IDE-KO mice reflects residual expression from nonparenchymal liver cells (sinusoidal cells, endothelial, Kupffer cells, etc.) As expected, IDE-KO mice did not express IDE protein in the liver (Fig. 1B). 3.2. Hepatic IDE Deletion Leads to Glucose Intolerance and Insulin Resistance, but without Altered Insulin Metabolism At 3 months of age, male (Fig. 2A–B) and female (Supplementary Fig. 2A–B) L-IDE-KO mice showed higher fasting and non-fasting blood glucose levels, as compared to WT controls. These changes in glucose homeostasis were not correlated with an increase in body weight (Fig. 2C–D and Supplementary Fig. 2C–D) or food intake (Fig. 2E). Consistent with this metabolic phenotype, L-IDE-KO mice exhibited glucose intolerance and insulin resistance (Fig. 3A–D and Supplementary Fig. 3). The insulin resistance observed in L-IDE-KO mice was not correlated with blunted insulin-signaling in the skeletal muscle, as shown by intact phosphorylation of protein kinase B (PKB; AKT1) in response to insulin (Fig. 3E–F). Typically, insulin resistance is associated with hyperinsulinemia, but surprisingly, fasting and non-fasting plasma insulin levels were similar between L-IDE-KO and control mice (Fig. 4A–B; Supplementary Fig. AUC (Arbitrary Units) 0 10000 20000 30000 40000 * AUC (Arbitrary Units) WT L-IDE KO WT L-IDE KO 0 5000 10000 15000 20000 25000 * F p-AKT(Ser473) / AKT (A.U.) 0 5 10 15 WT L-IDE KO Insulin - * * B D A C E +- + Fig. 3. L-IDE-KO mice exhibit impaired glucose homeostasis and insulin resistance. (A) IP-GTT of 3-month-old male mice. Data are mean ± SEM. n = 12–13 per genotype. *pvalue b0.05 vs. WT by Students' t-test. (B) Area under the curve of the IP-GTT. *pvalue b0.05 vs. WT by Students' t-test. (C) IP-ITT of 3-month-old male mice. Data are mean ± SEM. n = 12–13 per genotype. *pvalue b0.05 vs. WT by Students' t-test. (D) Area under the curve of the IP-ITT. *pvalue b0.05 vs. WT by Students' t-test. (E) Lack of effect of liver-specific IDE deletion on AKT signaling in skeletal muscle. Representative western blots depicting basal and insulin-stimulated p-AKT levels in skeletal muscle. Grouping of images from different parts of the same gel have been used for the final arrangement of the figure (see Supplementary Fig. 4 for full blots). (F) Densitometric analysis of Ser-473-Akt phosphorylation. The y-axis represents the ratio of phosphorylated versus total protein in arbitrary units. Data are mean ± SEM. n= 6 per genotype. *pvalue b0.05 vs. saline treatment by ANOVA. Fig. 4. Hepatic ablation of Ide does not alter plasma insulin levels and hepatic insulin clearance. Fasting (A) and non-fasting (B) plasma insulin levels in 3-month-old male WT and L-IDE-KO mice. Data are mean ± SEM. n = 12 per genotype. (C) Plasma insulin clearance in 3-month-old male mice. Data are mean ± SEM. n=9–6 per genotype. (D) Area under the curve of the in vivo insulin clearance. (E) Fastingplasma C-peptide levelsin 3-month-oldmice fasted overnight. Data are mean ± SEM. n = 6–12 per genotype. (F) Non-fasting plasma C-peptide levels in 3-month-old mice. Data are mean ± SEM. n=10–13 per genotype. (G) Insulin clearance was calculated as the molar ratio of C-peptide to insulin in 3-month-old mice fasted overnight. Data are mean ± SEM. n=6–12 per genotype. (H) Non-fasting plasma glucagon levels in 3-month-old male L-IDE-KO and control littermates. Data are mean ± SEM. n = 6–9 per genotype. n.s.; nonstatistical significance. (I) Fasting plasma glucagon levels in 3-month-old male L-IDE-KO and control littermates. Data are mean ± SEM. n = 9 per genotype. n.s.; nonstatistical significance. (J) Fasting plasma amylin levels in 3-month-old male L-IDE-KO and control littermates. Data are mean ± SEM. n = 10–14 per genotype. (K) Fasting plasma Aβ40 levels in 3-month-old male L-IDE-KO and control littermates. Data are mean ± SEM. n = 10–15 per genotype. 4P. Villa-Pérez et al. / Metabolism Clinical and Experimental 88 (2018) 1–11 AB CD Plasma insulin levels (ng/mL) WT L-IDE KO 0.0 0.1 0.2 0.3 0.4 Fasting Plasma insulin levels (ng/mL) WT L-IDE KO 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.6 2.0 Non-Fasting Time (min) Human Insulin (mU/L) 0 102030405060 0 5 10 15 20 WT L-IDE KO AUC (Arbitrary Units) WT L-IDE KO 0 100 200 300 400 E C-peptide (pM) WT L-IDE KO 0.0 0.2 0.4 0.6 0.8 Fasting Non-fasting F C-peptide (pM) WT L-IDE KO 0 1000 2000 3000 4000 G C-peptide/insulin molar ratio WT L-IDE KO 0 10 20 30 40 50 H J I K n.s. n.s. Glucagon (pmol/L) WT L-IDE KO 0 2 4 6 8 10 Non-fasting Glucagon (pmol/L) WT L-IDE KO 0 2 4 6 8Fasting Amylin (ng/mL) WT L-IDE KO 0.0 0.5 1.0 1.5 2.0 Fasting Fasting Aβ-40 (pg/mL) WT L-IDE KO 0 20 40 60 80 100 5P. Villa-Pérez et al. / Metabolism Clinical and Experimental 88 (2018) 1–11 p-IRβ / IRβ (A.U.) 0 5 10 15 20 25 * * # IRβ / Na+/K+-ATPase (A.U.) 0.0 0.5 1.0 1.5 # & p-AKT1(Ser473) / AKT1 (A.U.) 0 2 4 6 * *# p-AKT2(Ser474) / AKT2 (A.U.) 0 2 4 6 * * # p-CEACAM1 / CEACAM1 (A.U.) 0 5 10 15 20 *# B A Insulin -+ WT -+ L-IDE-KO Insulin -+ WT -+ L-IDE-KO IGF-IR / Na+/K+ -ATPase (A.U.) 0.0 0.5 1.0 1.5 2.0 C 6P. Villa-Pérez et al. / Metabolism Clinical and Experimental 88 (2018) 1–11 5A–B). These findings cast doubt on the notion that IDE is the principal protease regulating insulin catabolism in vivo [11], and that plasma insulin clearance depends on the protease activity of IDE in the liver [1]. To directly assess the role of IDE in hepatic insulin clearance, we performed an in vivo plasma insulin clearanceassay. As shown in Fig. 4C–D, male L- IDE-KO mice (and female, Supplementary Fig. 5C–D) exhibited similar rates of plasma insulin clearance as controls. Likewise, insulin clearance measured as steady-state C-peptide/insulin molar ratio [10] was similar between L-IDE-KO and WT controls (Fig. 4E–G). Like insulin, plasma glucagon, amylin and Aβ40 levels remained unchanged in L-IDE-KO as compared to WT controls (Fig. 4H–K). 3.3. IDE Is Necessary for Hepatic Insulin Signaling To help elucidate the molecular basis underlying the insulin resistance observed in L-IDE-KO mice, we analyzed multiple components of the intracellular insulin-signaling pathway in liver tissue, a key organ involved in the regulation of whole-body glucose homeostasis. To this end, 3-month-old male L-IDE-KO and WT mice were fasted overnight, then administered an intraperitoneal injection of a bolus of insulin (0.75 U/kg) or saline solution. Ten minutes later, mice were euthanized, and tissues were dissected. Hepatic Ide deletion resulted in a mild (~30%) but significant reduction of IR protein level in the plasma membrane fractionsfrom L-IDE-KO livers, as demonstrated by Western blot analysis (Fig. 5A-B). Furthermore, its insulin-stimulated phosphorylation was reduced by ~55% in L-IDE-KO mice, compared to controls (Fig. 5A–B). Consequently, insulin-stimulated phosphorylation of its substrate, CEACAM1, was completely abolished despite no change of its hepatic basal expression by Ide deletion (Fig. 5A-B). Because L-IDE-KO mice manifested a marked reduction of CEACAM1 phosphorylation in response to insulin, we hypothesized that it would be associated with changes in the internalization of the complex insulin-IR-CEACAM1. To address this question, we carried out coimmunoprecipitation experiments to detect whether deleting Ide affects insulin-stimulated association between Shc and CEACAM1, a critical step in the insulin's internalization process [30,33,34]. As Fig. 5C shows, immunoblotting the Shc immunopellet with CEACAM1 antibody indicated 1.5-fold increase in the Shc/CEACAM1 binding in response to insulin in the controls but not the IDE null mice. On the other hand, the insulin-like growth factor-1 receptor (IGF-IR) exhibits 70% homology to IR, and in response to insulin shares some of the signaling pathways with the IR. Western blot analyses revealed that hepatic Ide deletion did not reduce IGF-IR protein levels in the plasma membrane fractions from L-IDE-KO livers (Fig. 5A-B). Likewise, Insr and Igf1r mRNA levels were similar between WT and L-IDE-KO mice in fasting and non-fasting conditions (Supplementary Fig. 7). Taken together, these results support the notion that hepatic Ide ablation alters posttranslational events of the IR in L-IDE-KO mice without affecting the related IGF-IR. To further analyze downstream effects of hepatic IDE depletion on insulin signaling, we examined AKT1 and AKT2 phosphorylation in total liver lysates. AKT1 phosphorylation, which is chiefly involved in cell survival, protein synthesis and inhibition of apoptosis, and AKT2 phosphorylation, which is exclusively involved in metabolism, were both reduced, exhibiting significant reductions of ~40–45% (Fig. 5A-B). 3.4. Hepatic Ablation of IDE Causes Up-regulation of Gluconeogenic Genes A hallmark of hepatic insulin resistance is augmented production of glucose. The forkhead box O1 (FoxO1) is a nuclear transcription factor downstream of AKT that integrates insulin signaling with gluconeogenesis in the liver, by upregulating the expression of the gluconeogenic genes Pck1 and G6pc [29]. In response to insulin, FoxO1 is phosphorylated, resulting in its nuclear exclusion and inhibition of Pck1 and G6pc gene expression. Hepatic insulin resistance reduces FoxO1 phosphorylation, which results in unleashed gluconeogenesis [35]. To determine whether the increased blood glucose levels observed in L-IDE-KO mice were associated with activation of gluconeogenesis, we prepared cytoplasmic and nuclear fractions of liver tissues from WT and L-IDE-KO mice treated with saline or insulin for 10 min. As shown in Fig. 6A, FoxO1 proteins were detected predominantly in the cytoplasmic fractions after stimulation with insulin in WT mice. In contrast, in L-IDE- KO mice, FoxO1 underwent a substantial subcellular redistribution, shifting from the cytoplasmic fraction to the nuclear fraction upon exposure to insulin (Fig. 6A). This was correlated with up-regulation of Pck1 and G6pc mRNA expression levels under either fasting or non-fasting conditions (Fig. 6B–C). In addition to FoxO1, AKT phosphorylates and deactivates glycogen synthase kinase 3 (GSK-3), leading to activation of glycogen synthase (GS) and thus, inducing glycogen synthesis [36]. To examine the effect of IDE depletion on glycogenesis, we quantified phosphorylation levels of GSK-3 from WT and L-IDE-KO mice treated with saline or insulin for 10 min. As Fig. 6D revels, insulin-mediated AKT phosphorylation inactivates GSK-3, which in turn restores GS activity and glycogen synthesis in WT mice. Unexpectedly, relative to WT mice, basal GSK-3 phosphorylation was 3.5-fold higher in L-IDE-KO mice and, moreover, insulin treatment failed to result in increased phosphorylation (Fig. 6D). To explore the functional impact of these perturbations, we analyzed glycogen levels in L-IDE-KO and WT livers. Surprisingly, hepatic Ide ablation had no effect on liver glycogen levels, under either fasting (Fig. 6E) or non-fasting (Fig. 6F) conditions. Taken together, this suggests that Ide deficiency in liver must result in other, as-yet unidentified mechanisms that compensate for the lack of the expected effect of AKT on GSK-3 phosphorylation and hepatic glycogen content. Because insulin also regulates hepatic lipid metabolism via an AKT2 phosphorylation-dependent mechanism [37], we then assessed triglycerides levels in the plasma and livers of 3-month-old WT and L-IDE-KO male mice. Hepatic triglyceride content remained similar between WT and L-IDE-KO mice under either fasting (Fig. 6G) and non-fasting conditions (Fig. 6H). Similarly, circulating plasma triglycerides levels were comparable in both mouse groups regardless of the feeding status (Fig. 6I-J). 3.5. β-Cell Function and Histomorphometry in the L-IDE-KO Mice To asses β-cell function, we monitored plasma insulin levels following an i.p. glucose injection after an overnight fasting period in WT and L-IDE-KO mice. As Fig. 7A-B shows, plasma insulin levels during IP-GTT were similar between WT and L-IDE-KO mice. Consistently, fasting and non-fasting plasma C-peptide levels remained unchanged between L- IDE-KO mice and littermate controls (Fig. 4E–F). In addition, we have assessed β-cell mass and the number of islets by histomorphometry in WT and L-IDE-KO mice. As shown in Fig. 7C–F, β-cell mass, β-cell area, Fig. 5. Hepatic Ide deficiency causes multiple changes to intracellular insulin-signaling pathways. (A) Representative Western blots depicting the effects of insulin on total and phosphorylated IR, IGF-IR, CEACAM1, AKT1 and AKT2 levels. Expression of actin and Na + /K + -ATPase were determined to ensure similar protein levels for total liver lysates and lysates from liver membrane fractions (insulin receptor) respectively. Grouping of images from different parts of the same gel have been used for the final arrangement of the figure (see Supplementary Fig. 6 for full blots). (B) Densitometric analyses of the data in panel A. The y-axis represents the ratio of phosphorylated versus total protein in arbitrary units. Data are mean ± SEM. n = 6 per genotype and condition. *pvalue b0.05 vs. saline treatment by ANOVA. # pvalue b0.05 WT + insulin vs. L-IDE-KO + insulin treatment by ANOVA. & pvalue b0.05 WT-insulin vs. L-IDE-KO-insulin treatment by ANOVA. (C) Co-immunoprecipitation of CEACAM1 and Shc. Some aliquots of the above lysates used for western blotting were subjected to immunoprecipitation with α-Shc followed by immunoblotting with α-CEACAM1 antibody (α-CC1). Upper panel, representative image of two independent experiments with identical results. Lower panel, densitometric analysis of n= 3 independent experiments. *pvalue b0.05 by ANOVA. 7P. Villa-Pérez et al. / Metabolism Clinical and Experimental 88 (2018) 1–11 B Pck1 mRNA levels (A.U.) Fasting Non-Fasting 0.0 0.5 1.0 1.5 WT L-IDE KO * * G6p6 mRNA levels (A.U.) Fasting Non-Fasting 0.0 0.5 1.0 1.5 WT L-IDE KO * * C E F A D Triglyceride levels (mg/g liver) WT L-IDE KO 0 10 20 30 40 50 Fasting Triglyceride levels (mg/g liver) WT L-IDE KO 0 100 200 300 Non-Fasting GH IJ Plasma triglyceride levels (mg/dL) WT L-IDE KO 0 20 40 60 Fasting Plasma triglyceride levels (mg/dL) WT L-IDE KO 0 50 100 150 200 Non-Fasting Fig. 6. Effects of Ide deletion on gluconeogenesis and glycogenesis. (A) Subcellular distribution of basal and insulin-stimulated FoxO1 in L-IDE-KO and WT mice. Upper panel: ratio of cytoplasmic vs. nuclear FoxO1 levels. Lower panel: representative image of cytoplasmic and nuclear fractions containing FoxO1. Data are mean ± SEM. n= 4 per genotype and condition. *pvalue b0.05 vs. saline treatment by ANOVA. # pvalue b0.05 WT + insulin vs. L-IDE-KO + insulin treatment by ANOVA. Effects of Ide ablation on hepatic gluconeogenic gene expression. Livers from 3-month-old male L-IDE-KO and WT mice under fasting and non-fasting conditions were analyzed for mRNA levels of Pck1 (B) and G6pc (C). Data are mean ± SEM. n = 3 in triplicate per genotype and condition. *pvalue b0.05 vs. WT by Students' t-test. (D, E, F) Effects of Ide deletion on glycogenesis signaling. (D) Upper panel: densitometric analyses of GSK-3α. The y-axis represents the ratio of phosphorylated versus total protein in arbitrary units. Lower panel: representative western blots depicting basal and insulin-stimulated total and phosphorylated GSK-3α. Grouping of images from different parts of the same gel have been used for the final arrangement of the figure (see Supplementary Fig. 8 for full blots). Data are mean ± SEM. n = 6 per genotype and condition. *pvalue b0.05 vs. saline treatment by ANOVA. # pvalue b0.05 WT vs. L-IDE-KO by ANOVA. Liver glycogen content under fasting (E) and non-fasting (F) conditions in 3-month-old male WT and L-IDE-KO mice. Data are mean ± SEM. n = 4 per genotype and condition. * pvalue b0.05 vs. WT by Students' t-test. (G, H, I, J) Effects of Ide deletion on triglycerides levels. Hepatic triglycerides levels under fasting (G) and non-fasting (H) conditions in 3-month-old male WT and L-IDE-KO mice. Data are mean ± SEM. n = 4 per genotype and condition. Plasma triglycerides levels under fasting (I) and non-fasting (J) conditions in 3-month-old male WT and L-IDE-KO mice. Data are mean ± SEM. n= 12 per genotype and condition. 8P. Villa-Pérez et al. / Metabolism Clinical and Experimental 88 (2018) 1–11 the number of islets, and the mean islets size were similar between both genotypes. Taken together these results indicate that β-cell function and mass are comparable in L-IDE-KO mice and WT mice. 4. Discussion Using a mouse harboring liver-specific ablation of Ide,weshowfor the first time that hepatic deficiency of Ide does not result in hyperinsulinemia, as expected from deleting a protein that has been postulated to play a major role in hepatic insulin clearance. Extrahepatic tissues, such as kidneys, are also involved in insulin clearance in vivo, albeit to a lower extent than the liver [1]. However, IDE renal protein levels were not induced in L-IDE-KO mice to compensate for the loss of IDE in liver (Supplementary Fig. 9). Therefore, normoinsulinemia in L-IDE-KO mouse rules out a major role for IDE in hepatic insulin clearance. Consistently, Durham et al. demonstrated that NTE-1, an inhibitor of IDE, did not increase plasma insulin levels in rodents [16]. By excluding IDE as a significant mediator of insulin clearance in vivo, we postulate that other proteins might be more critical in this process. This could include CEACAM1, a substrate of the IR in liver that, upon its phosphorylation, promotes the uptake of the insulinreceptor complex and its targeting to the degradation process [9,10,38]. Hepatic loss-of-function of CEACAM1 by overexpressing its dominant-negative phosphorylation-defective isoform, reduced insulin clearance to cause hyperinsulinemia followed by secondary insulin resistance and impaired glucose tolerance [10]. CEACAM1 promotes insulin clearance by inducing the rate of insulin endocytosis via the less abundant, high-affinity isoform A of the IR (IR-A) [33]. Upon its phosphorylation on Tyr488 by the IR tyrosine kinase, via a mechanism requiring insulin-stimulated phosphorylation of the IR at Tyr1316, CEACAM1 recruits one or more molecules (including Shc) that bind to Tyr960 of the juxtamembrane domain of the IR to form a stable complex with the IR, and induce its uptake into clathrin-coated pits/vesicles [33]. Hence, CEACAM1 phosphorylation is required but not sufficient to allow receptor-mediated insulin endocytosis and degradation [33]. Our results demonstrate that in the absence of hepatic IDE, IR phosphorylation is decreased, but that of CEACAM1 is disproportionally abrogated. Thus, it is plausible that IDE participates in the internalization of the insulin-IR-CEACAM1 complex. Furthermore, hepatic depletion of IDE remarkably reduced insulin-stimulated association between Shc and CEACAM1, a critical step in the insulin's internalization process. It has been proposed that IDE is a multifunctional protein, which may perform unrelated functions as a scaffold protein involved in maintaining cellular homeostasis [39]. Whether such a proposed function of IDE is dependent on its protease activity, remains to be clarified. If IDE deletion completely blunts CEACAM1 phosphorylation (and function), one would question how is circulating insulin being removed in L-IDE-KO mice? First, CEACAM1 regulates insulin clearance by modulating insulin uptake via the less abundant IR-A isoform in the liver. The contribution of the more abundant IR-B isoform in insulin clearance in L-IDE-KO mice remains to be elucidated. Second, at physiological concentrations of the hormone, hepatic insulin clearance is mediated primarily by IR in the hepatocyte. However, the contribution of Kupffer cells to total insulin clearance in the liver is ~15%. Thus, it is possible that these cells compensate for the loss of insulin clearance in hepatocytes. Third, β-cell function and mass are similar between L-IDE-KO and control mice, suggesting that plasma insulin levels are AB CD AUC (Arbitrary Units) WT L-IDE-KO 0.0 0.2 0.4 0.6 0.8 1.0 n.s. Time (min) Plasma insulin levels (ng/mL) 0 5 10 15 20 25 30 0.0 0.1 0.2 0.3 0.4 0.5 WT L-IDE-KO β -cell area (%) WT L-IDE-KO 0.0 0.5 1.0 1.5 2.0 β -cell mass (mg) WT L-IDE-KO 0 1 2 3 4 EF N° Islets / mm 2 WT L-IDE-KO 0.0 0.5 1.0 1.5 Islet size (mm 2 ) WT L-IDE-KO 0.000 0.005 0.010 0.015 0.020 Fig. 7. Effects of Ide deletion on β-cell function and mass. (A) Plasma insulin levels during IP-GTT. Data are mean ± SEM. n=9–13 per genotype and condition. (B) Area under the curve (AUC) of plasma insulin levels during the IP-GTT. n.s.; nonstatistical significance. Histomorphometry of pancreata. (C) β-cell area, (D) β-cell mass, (E) the number of islets, (F) the mean islets size. Data are mean ± SEM. n=10–13 per genotype. 9P. Villa-Pérez et al. / Metabolism Clinical and Experimental 88 (2018) 1–11