Peroxiredoxin 6, a Novel Player in the Pathogenesis of Diabetes

3210 Diabetes Volume 63, October 2014

Francesca Paciﬁci,1 Roberto Arriga,1 Gian Pio Sorice,2,3 Barbara Capuani,1 Maria Giovanna Scioli,4 Donatella Pastore,1 Giulia Donadel,1 Alfonso Bellia,1 Sara Caratelli,1,5 Andrea Coppola,1 Francesca Ferrelli,1 Massimo Federici,1 Giuseppe Sconocchia,1,5 Manfredi Tesauro,1 Paolo Sbraccia,1 David Della-Morte,1,6 Andrea Giaccari,2,7 Augusto Orlandi,4 and Davide Lauro1 Peroxiredoxin 6, a Novel Player in the Pathogenesis of Diabetes

Diabetes 2014;63:3210–3220 | DOI: 10.2337/db14-0144

Enhanced oxidative stress contributes to the pathogen- by an imbalance between reactive oxygen species (ROS) esis of diabetes and its complications. Peroxiredoxin 6 production and antioxidant defense systems. Among all (PRDX6) is a key regulator of cellular redox balance, with body tissues, pancreatic b-cells are very sensitive to oxida- the peculiar ability to neutralize peroxides, peroxynitrite, tive stress because of their low expression of antioxidant and phospholipid hydroperoxides. In the current study, enzymes like superoxide dismutase (SOD) and glutathione we aimed to define the role of PRDX6 in the pathophys- peroxidase (2). Moreover, hyperglycemia by itself induces iology of type 2 diabetes (T2D) using PRDX6 knockout IR, increasing oxidative stress injuries, which lead to overt 2 2 ( / ) mice. Glucose and insulin responses were evalu- type 2 diabetes (T2D) (3). Interestingly, a relatively new ated respectively by intraperitoneal glucose and insulin family of antioxidant proteins, the peroxiredoxins (PRDXs), tolerance tests. Peripheral insulin sensitivity was ana- is more highly expressed in pancreatic b-cells (4). Among lyzed by euglycemic-hyperinsulinemic clamp, and molec- the six members of this non-seleno peroxidase family, ular tools were used to investigate insulin signaling. PRDX6 is present in the cytoplasm and is unique because Moreover, inflammatory and lipid parameters were eval- 2/2 it has peroxidase and also phospholipase A2 activity (5). uated. We demonstrated that PRDX6 mice devel- fi oped a phenotype similar to early-stage T2D caused Several ndings demonstrate the importance of PRDX6 METABOLISM by both reduced glucose-dependent insulin secretion in maintaining redox homeostasis, as follows: lack of and increased insulin resistance. Impaired insulin sig- PRDX6, in fact, increases the susceptibility to oxidative naling was present in PRDX62/2 mice, leading to reduc- stress in different tissues (6,7). Nevertheless, data on the tion of muscle glucose uptake. Morphological and relationship between PRDX6 and the pathogenesis of IR ultrastructural changes were observed in islets of Lang- and T2D are not available (8). Therefore, we hypothesized erhans and livers of mutant animals, as well as altered that, in terms of physiological status, PRDX6 may play a role plasma lipid profiles and inflammatory parameters. In in the etiology of IR and diabetes conditions through tissue conclusion, we demonstrated that PRDX6 is a key me- redox levels. In the current study, we tested our hypothesis 2 2 diator of overt hyperglycemia in T2D glucose metabo- in a model of PRDX6 knockout mice (PRDX6 / ). lism, opening new perspectives for targeted therapeutic strategies in diabetes care. RESEARCH DESIGN AND METHODS Animal Models C57BL/6J wild-type (WT) mice weighing 18–20 g were A large body of evidence supports a pivotal role for obtained from The Jackson Laboratory (Bar Harbor, ME), 2 2 oxidative stress in the etiopathogenesis of insulin resis- while PRDX6 / mice of mixed background (C57BL6/129SvJ) tance (IR) and diabetes (1). Oxidative stress is characterized were provided by Professor Xiaosong Wang (The Jackson

1Department of System Medicine, University of Rome Tor Vergata, Rome, Italy Corresponding author: Davide Lauro, [email protected]. 2 Division of Endocrinology and Metabolic Diseases, Università Cattolica del Sacro Received 27 January 2014 and accepted 28 April 2014. Cuore, Rome, Italy This article contains Supplementary Data online at http://diabetes 3Diabetic Care Clinics, Associazione dei Cavalieri Italiani Sovrano Militare Ordine .diabetesjournals.org/lookup/suppl/doi:10.2337/db14-0144/-/DC1. di Malta, Rome, Italy 4Anatomic Pathology, Department of Biomedicine and Prevention, University of F.P. and R.A. contributed equally to this work. Rome Tor Vergata, Rome, Italy © 2014 by the American Diabetes Association. Readers may use this article as 5Institute of Translational Pharmacology, National Research Council, Rome, Italy long as the work is properly cited, the use is educational and not for profit, and 6Istituto di Ricovero e Cura a Carattere Scientifico San Raffaele Pisana, Rome, Italy the work is not altered. 7Fondazione Don Carlo Gnocchi, Milan, Italy diabetes.diabetesjournals.org Pacifici and Associates 3211

Laboratory) (6). All mice were housed in a temperature- insulin (1 unit/kg) was injected through the portal vein. controlled animal room with a 12-h light-dark cycle, and Five minutes after injection, mice were killed, and the were given free access to commercial mouse chow and gastrocnemius was excised and frozen immediately in water. Three-month-old male mice were used for all liquid nitrogen. Subsequently, gastrocnemius samples were experiments. Five animals were used in each experimen- homogenized with 1 mL cold extraction buffer containing tal group. 20 mmol/L Tris (pH 7.6), 137 mmol/L NaCl, 1.5% NP40, 1 mmol/L MgCl , 1 mmol/L CaCl , glycerol 10%, 2 mmol/L Glucose and Insulin Tolerance Tests and Insulin Levels 2 2 phenylmethylsulfonyl fluoride, 13 protease inhibitor cock- An intraperitoneal glucose tolerance test (IPGTT) was tail (Roche, Indianapolis, IN), 2 mmol/L Na VO ,and2 performed with 2 g/kg glucose injection in mice fasted 3 4 mmol/L EDTA using a dounce homogenizer. The homoge- for 16 h. Glucose levels were measured after 0, 30, 60, 90, nates were maintained on ice for 30 min and then were and 120 min using an automated OneTouch Glucometer centrifuged at 14,000 rpm for 30 min. Supernatants were (LifeScan, Milpitas, CA). Insulin concentration was quan- collected and stored at 280°C, while pellets, containing tified on plasma samples with a Mouse Insulin ELISA Kit membrane proteins fractions, were resuspended in cold (Mercodia, Uppsala, Sweden). An insulin tolerance test extraction buffer, maintained on ice for 30 min, and sub- (ITT) was performed using intraperitoneal injection of 0.75 sequently centrifuged at 50,000 rpm for 1 h. Superna- international units (IU)/kg insulin in mice fasted for 4 h, tants, containing membrane proteins, were maintained and glucose levels were measured after 0, 15, 30, and 60 at 280°C before analysis. min. Blood samples were obtained from the retro-orbital Protein concentration was determined using a Bradford sinus. Animal studies were approved by the University of assay (Bio-Rad Laboratories, Hercules, CA), loaded on Rome Tor Vergata Animal Care and Use Committee. precast 4–15% gels (Bio-Rad Laboratories), separated by Euglycemic-Hyperinsulinemic Clamp Studies SDS-PAGE, and then transferred to nitrocellulose mem- Insulin clamp studies were performed as previously de- branes using the Trans-Blot Turbo Transfer System (Bio- scribed (9). Briefly, 3 to 5 days prior to the clamp, a catheter Rad Laboratories). Afterward, membranes were incubated was inserted into the right internal jugular vein and ex- with antibodies directed against phospho-Akt-1 Ser473, tended to the level of the right atrium. At time zero, after phospho stress-activated protein kinase/Jun NH2-terminal mice were fasted for 6 h, a primed continuous infusion of kinase (JNK) (Cell Signaling Technology), phospho-Akt-2 human insulin (18.0 mU/kg/min; Actrapid 100 IU; Novo Ser474, GLUT4 (Abcam), tubulin (Sigma-Aldrich), Akt-1, Nordisk, Copenhagen, Denmark) was started simulta- Akt-2, and stress-activated protein kinase/JNK (Cell Signal- neously with a variable infusion of 20% dextrose in order ing Technology). The antigen-antibody complexes were to maintain the plasma glucose concentration constant at detected with enhanced chemiluminescence (GE Health- its basal level (80–100 mg/dL). Blood samples (;2 mL) care), followed by exposure on blotting to X-ray film. were obtained from the tail vein at 10-min intervals for Bands were quantified using Gel Doc XR+ with Image at least 2 h. Average glucose concentrations and glucose Lab Software (Bio-Rad Laboratories). infusion rates were measured during the last 30 min of Immunoprecipitation the steady-state clamp period. Insulin receptor substrate (IRS) 1, IR b-subunit, and phos- Pancreatic Islets Extraction phoinositide 3-kinase (PI3K) were immunoprecipitated Pancreatic islets extraction was performed according to using standard protocols. Briefly, antibodies against Carter et al. (10). Mice were anesthetized with Avertin IRS1 or IR b-subunit were rocked for 1 h at 4°C with 5 mg/10 g body weight and killed. Pancreata were perfused protein A and protein G, respectively (GE Healthcare), with collagenase P after the common bile duct was and centrifuged at 15,000g for 10 min. Then, 1 mg muscle clamped, and then were incubated for 8 min at 37°C in lysates were added to the complexes and were immuno- cold Hanks’ balanced salt solution. Digested pancreata were precipitated by rocking overnight at 4°C. Immunoprecipi- washed three times with G solution (Hanks’ balanced salt tates were washed three times, resuspended in 45 mL23 solution plus 1% BSA) and subsequently filtered. After cen- LDS Sample Buffer (Invitrogen, Carlsbad, CA), and heated trifugation, pellets were resuspended in 10 mL Histopaque at 70°C for 10 min. Subsequently, immunoprecipitates 1100 solution and centrifuged for 20 min at 1,200 rpm. were loaded on precast 4–12% gels (Invitrogen), separated Afterward, islets were washed three times and then were by SDS-PAGE, and subjected to immunoblot analysis as transferred into a sterile Petri dish containing 5 mL RPMI reported above. Antibodies against phosphotyrosine RC20 1640 medium with 10% FBS and penicillin/streptomycin. (BD), PY20 (Santa Cruz Biotechnology), PI3K p85 a-subunit, Islets were cultured for 24 h to allow recovery from the insulin receptor b-chain (Cell Signaling Technology), and stress associated with the isolation process. IRS1 (Millipore) were used. Western Blot Analysis of Insulin Signaling and Microscopic Evaluation of Nonalcoholic Subcellular Fractionation Steatohepatitis and Pancreatic Islets Experiments were carried out in mice fasted overnight. Formalin-fixed paraffin4-mm-thick hepatic and pancreatic Animals were anesthetized with Avertin (5 mg/10 g), and tissue sections were cut and stained with hematoxylin-eosin 3212 PRDX6 and Pathogenesis of Diabetes Diabetes Volume 63, October 2014 for microscopic examination (11). Liver microscopic fea- of pancreatic islet volume (Fig. 3). It is therefore possible tures were grouped into five broad categories for the that increased insulin production represents a compensatory microscopic evaluation of nonalcoholic steatohepatitis mechanism for increasing insulin secretion by the residual (NASH), as already described (12). The number of pancreatic pancreatic islets. islets and the percentage of the islets area were calculated Next, we sought to perform euglycemic-hyperinsulinemic on images acquired by a digital camera (DXM1200F; Nikon clamp to quantify the amount of peripheral IR (Fig. 1E). Italia, Milan, Italy) and measured using Scion Image soft- During this procedure, the insulin concentration was ware (Scion Corporation, Frederick, MD) at a 320 mag- raised to ;600 pmol/L, while the plasma glucose concen- nification. All analyses were performed in a blinded tration was maintained at ;6 mmol/L through a variable fashion by two different pathologists, with an interob- infusion of glucose. We found a significant reduction of 2 2 server variability of ,5%. peripheral whole-body glucose disposal in PRDX6 / mice compared with WT mice (P , 0.005). The decrease Gene Expression Analysis by Real-Time PCR 2 2 of the mean value in PRDX6 / mice suggests that di- Total RNA from skeletal muscle, liver, and white adipose minished insulin sensitivity could be located mainly at the tissue (WAT) was isolated using TRIzol (Invitrogen). Two skeletal muscle level, since ;80% of glucose uptake occurs and one-half micrograms of total RNA was reverse tran- in skeletal muscle during the test and PRDX6 is largely scribed into cDNA using a High-Capacity cDNA Archive Kit expressed in skeletal muscle (data not shown) (14). (Applied Biosystems, Foster City, CA). Qualitative real-time PCR was performed using an ABI PRISM 7500 System and Insulin Signaling Is Impaired in Skeletal Muscle of TaqMan reagents (Applied Biosystems). Each reaction was 2 2 PRDX6 / Mice performed in duplicate using standard conditions, and To confirm this hypothesis, we analyzed phosphorylation results were normalized with b-actin. The relative expres- and the activity of key components of the insulin cascade sion was calculated using the comparative DDCT method, 2/2 2 in skeletal muscle of PRDX6 and WT mice. After in and the values were expressed as 2 DDCT (13). vivo insulin stimulation, muscle insulin receptor was iso- Statistical Analysis lated with immunoprecipitation, and insulin-induced ty- Data were analyzed using Prism 5 (GraphPad, La Jolla, rosine phosphorylation was measured by immunoblotting CA). All data are expressed as mean 6 SE. Statistical with antiphosphotyrosine antibody. A significant increase analysis was performed by two-way ANOVA followed by in insulin receptor b-subunit phosphorylation was ob- 2 2 Bonferroni post hoc test or unpaired one-tailed Student served in PRDX6 / mice compared with WT mice (P , t test when appropriate. Values of P , 0.05 were consid- 0.05) (Fig. 2A). Conversely, a significant reduction of IRS1 ered statistically significant. tyrosine phosphorylation after insulin stimulation was 2 2 observed in PRDX6 / mice (P , 0.005) (Fig. 2B). The RESULTS IRS1-PI3K interaction was also evaluated, and a significant PRDX62/2 Mice Develop an Early Stage of Diabetes reduction in the amount of p85 was found in IRS1 immu- 2 2 Linked With Higher Levels of IR noprecipitates from the muscles of PRDX6 / mice (Fig. To evaluate the impact of PRDX6 deletion on glucose and 2C). On the contrary, we did not find any differences in 2 2 insulin metabolism, PRDX6 / and WT mice underwent phosphorylation levels of IRS2 (data not shown). an IPGTT and an ITT. As shown in Fig. 1A, after 30 min of Since JNK, when activated by oxidative stress, induced glucose administration, and up to the end of the test, serine phosphorylation of IRS1, reducing the insulin 2 2 PRDX6 / mice had significantly higher glucose values signaling pathway (15), protein and phosphorylation lev- compared with WT mice (P , 0.05). Moreover, as dem- els of JNK were evaluated. We found higher phosphory- 2 2 2 2 onstrated with the ITT, PRDX6 / mice were insulin re- lation levels of JNK in PRDX6 / mice compared with sistant, having significantly higher glucose values at 30 WT mice (P , 0.05) (Fig. 2D), suggesting that the reduc- and 60 min after insulin injection compared with WT tion of insulin signaling is, at least in part, linked with mice (P , 0.05) (Fig. 1B). These results were consistent JNK activation. with reduced peripheral insulin sensitivity, which is a con- Next, we measured phosphorylation levels of Akt-1 dition of early-stage diabetes. In addition, during the and Akt-2 at Ser473 and Ser474, respectively, the pyruvate IPGTT, plasma levels of insulin were significantly reduced dehydrogenase kinase 2 domains. Phosphorylation of Akt- 2 2 after 15 min in PRDX6 / mice compared with WT mice 1 and Akt-2 was evident in insulin-stimulated skeletal (P , 0.0005) (Fig. 1C), demonstrating an impaired insulin muscles, and a significant reduction of Akt-1 and Akt-2 2 2 secretion. phosphorylation levels was found in PRDX6 / mice 2 2 Although PRDX6 / mice have lower insulin secretion compared with WT mice (P , 0.005) (Fig. 2E and F). after glucose load, insulin mRNA was significantly in- Since it is known that Akt-2 can modulate GLUT4 2 2 creased in pancreatic islets of PRDX6 / mice compared cell membrane translocation to increase glucose uptake with WT mice (Fig. 1D), suggesting higher synthesis of in- (16,17), we quantified the amount of membrane-associated 2 2 sulin in PRDX6 / pancreatic b-cells. Interestingly, this de- GLUT4 after insulin stimulation, showing a significantly 2 2 fect of insulin secretion was also associated with a reduction lower GLUT4 plasma membrane level in PRDX6 / diabetes.diabetesjournals.org Pacifici and Associates 3213

Figure 1—PRDX6 knockout mice are glucose intolerant and insulin resistant. A: Mice IPGTT was performed on 3-month-old male WT (open circle) and knockout (PRDX62/2; ﬁlled square) mice weighing 18–20 g. Fasted (16 h) mice received 2 g/kg body weight glucose intraper- itoneally, and blood glucose levels were measured at 0, 30, 60, 90, and 120 min after glucose injection. B: ITT was performed on fasted (4 h) WT and PRDX2/2 mice. Mice received an intraperitoneal injection of insulin at 0.75 IU/kg body weight, and blood glucose levels were measured at 0, 15, 30, and 60 min after insulin administration. C: Insulin secretion levels were analyzed by ELISA at 0, 15, 30, and 120 min after glucose injection. D: Insulin gene expression was evaluated on WT mice (white bar) and PRDX62/2 mice (black bar) pancreatic islets. E: Glucose uptake in both mouse models was studied by performing euglycemic-hyperinsulinemic clamp. Values are expressed as mean 6 SE. *P < 0.05, **P < 0.005, ***P < 0.0005. Graphs illustrate one of three separate studies, all yielding similar results (n = 5 mice per group). a.u., arbitrary units; M, mean.

2 2 mice compared with WT mice (P , 0.05) (Fig. 2G). Con- PRDX6 / Mice Develop Diabetic Dyslipidemia sistent with this finding, the GLUT4 cytoplasmic fraction A dyslipidemic profile characterized by hypertriglycer- 2 2 was significantly higher in the muscle of PRDX6 / com- idemia, high levels of small dense LDL protein, and low pared with WT mice (P , 0.05) (Fig. 2H), confirming levels of HDL cholesterol is often observed in patients defective transporter translocation by insulin in the mu- with diabetes (18). To gain information on lipid me- 2 2 tant strain. tabolism in PRDX6 / mice, we measured blood levels of triglycerides, VLDL, HDL cholesterol, and hepatic 2 2 Morphological Alteration in Pancreas of PRDX62/2 enzymes. As reported in Table 1, PRDX6 / mice dis- Mice played augmented levels of VLDL and triglycerides and A morphometric study was performed to evaluate mor- a reduced level of HDL cholesterol compared with 2 2 phological alterations of pancreatic islets in PRDX6 / WT littermates (P , 0.05), which is like the condition mice. We analyzed five different tissue sections for each of diabetic dyslipidemia. Moreover, microscopic analy- 2 2 animal in the same pancreas area. In Fig. 3A, arrowheads sis of liver tissue (Fig. 4A)revealedthatPRDX6 / 2 2 indicate the islets of Langerhans in WT and PRDX6 / mice have alterations resembling human NASH, in- mice.ThebargraphsinFig.3B and C reveal a reduction cluding cell ballooning and lymphocyte infiltration 2 2 in the density and size of islets in PRDX6 / mice com- (Fig. 4B). pared with WT mice (P , 0.05), suggesting a relevant In order to better understand the defects underlying role for PRDX6 in maintaining anatomical functional is- diabetic dyslipidemia, we analyzed the expression of the let mass. principal factors regulating lipid metabolism in skeletal 3214 PRDX6 and Pathogenesis of Diabetes Diabetes Volume 63, October 2014

Figure 2—Western blotting analysis of insulin-signaling pathway. Fasted (overnight) 3-month-old male WT (white bar) and PRDX62/2 (black bar) mice weighing 18–20 g were injected with insulin at 1 IU/kg body weight into the portal vein, and skeletal muscle was collected. One milligram of protein extracts from skeletal muscle was immunoprecipitated with antibodies against IR b-subunit (A) and IRS1 (B) and, after gel separation, was immunoblotted with speciﬁc antibodies for phosphotyrosine (PY20 and RC20, respectively). C: PI3K activation was studied by analyzing the interaction between IRS1 and p85 a-subunit. Fifty micrograms of total lysates were immunoblotted with antibodies against JNK (D), Akt1 p-Ser473 (E), and Akt2 p-Ser474 (F). GLUT4 translocation from cytoplasm (H) to plasma membrane (G) was studied by immunoblotting in both mouse models after cellular fractionation. Band intensities were quantiﬁed and expressed as mean 6 SE. *P < 0.05 and **P < 0.005 by Student t test. Graphs illustrate one of three separate studies, all yielding similar results (n = 5 mice per group). a.u., arbitrary units; IP, intraperitoneal.

2 2 muscle, liver, and WAT (18) (Fig. 5 and Supplementary PRDX6 / Mice Develop a Proinflammatory Status Table 1 [arrows indicate statistical significance]). In par- Diabetes is frequently associated with a proinflammatory 2 2 ticular, we found a significant increase in PRDX6 / mice state that accompanies strong IR (19). Inflammatory sta- 2 2 of patatin-like phospholipase domain containing 2 and tus was assessed in PRDX6 / mice, analyzing the expres- fatty acid synthase in all three tissues analyzed compared sion of proinflammatory and anti-inflammatory cytokines with WT mice, which suggests a possible increase in like interleukin (IL)-6, IL-1b, tumor necrosis factor-a both lipolytic and lipogenetic processes in these animals. (TNF-a), and IL-10, in liver, skeletal muscle, and WAT. 2 2 2 2 Moreover, in the liver and WAT of PRDX6 / mice, we Interestingly, PRDX6 / mice have a marked increase observed a significant reduction of peroxisome proliferator– of IL-1b, TNF-a, and IL-10 in all analyzed tissues (Fig. 6 activated receptor g coactivator 1 a, which was not present and Supplementary Table 1 [arrows indicate statistical in the muscle. significance]), while IL-6 was significantly increased in diabetes.diabetesjournals.org Pacifici and Associates 3215

Figure 3—Comparative analysis of pancreatic islets in WT and PRDX62/2 mice. Immunohistochemical analysis of pancreatic Langerhans islets (arrowheads) in 3-month-old male WT mice (A, left panel) and PRDX62/2 mice (A, right panel) weighing 18–20 g. Magniﬁcation 320. Bar graphs show the quantiﬁcation of mean islet density (B) and size (C) in PRDX62/2 mice (black bars) compared with WT mice (white bars). Values are expressed as mean 6 SE. *P < 0.05 (n = 5 mice per group).

2 2 WAT, but not in skeletal muscle and liver (Fig. 6). Then, PRDX6 / mice develop a proinflammatory status linked we investigated whether chemokine expression levels with high levels of IR and reduced glucose-stimulated in- 2 2 were increased in PRDX6 / mice. mRNA expression of sulin secretion. chemokine motif ligand 1 was significantly increased in WAT and liver (P , 0.05), but not in skeletal muscle; in DISCUSSION contrast, chemokine (C-C motif) ligand 3 mRNA expres- Oxidative stress is considered one of the main mecha- sion was increased in WAT (P , 0.05), liver (P , 0.0005), nisms in the pathogenesis of diabetes and IR (1). Among and skeletal muscle (P , 0.005) (Supplementary Table 1 the antioxidant enzymes that are able to regulate ROS- and Supplementary Fig. 2). These results suggest that mediated injuries, PRDX6 belongs to a relative new family of proteins that is widely distributed in the tissues of the body (5). PRDX6 is abundant in b-cells (2) as well as in Table 1—Metabolic parameters associated with diabetic muscle, and its powerful antioxidant role has been dyslipidemia reported previously (6). In the current study, we demon- 2/2 P WT mice PRDX6 mice value strated that PRDX6 can affect metabolic homeostasis in Triglycerides 117 6 14.2 171 6 9.6 ,0.05 mice and represents, therefore, a candidate for being a de- Cholesterol 109 6 9.5 78 6 6NSterminant of diabetes susceptibility in humans. 2/2 VLDL 23.4 6 2.8 34.2 6 1.9 ,0.05 We clearly showed that PRDX6 mice spontaneously HDL 65 6 1526 2.6 ,0.05 develop a metabolic defect resembling early-stage T2D, AST 99 6 3.5 106 6 16.6 NS which is characterized by higher glucose levels after IPGTT and IR. These defects were accompanied by impaired insulin ALT 45.2 6 8.9 30.7 6 9.6 NS signaling in the muscle and by reduced insulin secretion in Data are presented as mean 6 SEM, unless otherwise indicated. response to glucose. Moreover, we showed a distinctive al- ALT, alanine aminotransferase; AST, aspartate aminotransferase. teration in lipid profiles and an increase in the inflammatory 3216 PRDX6 and Pathogenesis of Diabetes Diabetes Volume 63, October 2014

Figure 4—Histological evaluation of liver tissue. A: Representative micrographs showing morphological aspects of NASH as hepatocyte injury or ballooning (arrowheads) and lymphocyte inﬁltration (arrows) in the liver tissue of 3-month-old male PRDX62/2 mice (right panels) and WT mice (left panels) weighing 18–20 g (magniﬁcation 3400). B: Bar graph shows the increased NASH score in PRDX62/2 mice (black bar) compared with WT mice (white bar). Values are expressed as mean 6 SE. **P < 0.005 (n = 5 mice per group).

2 2 status of PRDX6 / mice that are associated with the pre- A previous study (27) conducted in a model of diabetic and diabetic phenotypes, which is in agreement knockout mice for SOD1 gene, a potent antioxidant en- with similar findings reported in previous studies (20). zyme, demonstrated that mice lacking SOD1 were signif- Our results are consistent with a pivotal role of PRDX6 in icantly more susceptible to the development of glucose the physiopathology of diabetes and its related complica- intolerance, with a nonsignificant reduction of peripheral 2 2 tions. To our knowledge, this is some of the first evidence glucose disposal (;20%). Differently from SOD1 / mice, 2 2 in which modifications of antioxidant defense systems may PRDX6 / mice have a significant reduction of peripheral leadtoadiabeticphenotype,evenintheabsenceofdi- glucose disposal (;31%), suggesting a specific role of rect inducers of oxidative stress. As we described above, PRDX6 in the process leading to IR. Insulin signaling PRDX6, compared with other PRDXs, has a double func- has well-defined pathways, which regulate tissue glucose tion (phospholipase A2 and peroxidase), which makes uptake (28). Interference in each step that regulates these it more efficient in the detoxification processes (5). pathways may lead to a decreased uptake of circulating However, other studies are necessary to understand glucose, as depicted in the model in Fig. 7. Mainly, the the mechanism of PRDX6 regulation of glucose metabo- lack of PRDX6 is associated with a reduction in insulin lism and inflammation. It is possible to speculate that secretion and a lowering of IRS1 activation in skeletal PRDX6isakeyenzymeinregulatingROSproduction, muscle, leading to reduced levels of Akt-1/Akt-2 phos- and the action of this enzyme is important even in basal phorylation, GLUT4 translocation, and then to reduced conditions. Indeed, in a model of mice lacking neuronal peripheral glucose uptake. Different levels of IRS1 phos- nitric oxide synthase, PRDX6 upregulation can compen- phorylation and activation are key processes that mod- sate for the absence of neuronal nitric oxide synthase in ulate insulin signaling in skeletal muscle (29). The scavenging of superoxide; this result supports our hy- molecules involved in the regulation of oxidative stress pothesis concerning the importance of oxidative stress have been previously associated with differences in IRS1 unbalance in our model (21). activation and expression (30,31). Furthermore, JNK In particular, ROS production has been demonstrated phosphorylation is increased in the skeletal muscle of 2 2 to be important in the etiology of diabetes by acting at the PRDX6 / mice, and augmented JNK activation can following different levels: 1) blocking glucose uptake (3); lead to the deregulation of IRS1, as has already been 2) impairing insulin signaling pathways (3); 3) altering reported in insulin-resistant nonobese subjects (32). Our pancreatic islet morphology (22); 4) modifying metabolic results, in agreement with those of previous studies parameters and lipid metabolism (23,24); and 5) increas- (33,34), support the hypothesis that antioxidant enzymes ing inflammatory response (25,26). In the current study, can reduce insulin-signaling activation and decrease glu- 2 2 by comparing PRDX6 / and WT mice, we investigated cose uptake. the role of the enzyme in all of the above mechanisms of The morphological changes, including reduction in the ROS-mediated metabolic derangement. size of pancreatic islet mass in mice with diabetes, may diabetes.diabetesjournals.org Pacifici and Associates 3217

Figure 5—Genetic expression of the principal enzymes involved in lipid metabolism. Expression levels of genes involved in lipid metabolism, such as patatin-like phospholipase domain containing 2 (PNPLA2), Fas, peroxisome proliferator–activated receptor g coactivator 1 a (PGC1a), and CD36, were analyzed by real-time PCR. mRNA was extracted from skeletal muscle, liver, and WAT of fasted (4 h) WT mice (white bars) and PRDX62/2 mice (black bars). A, E, and I: Increased levels of lipolysis associated with increased lipogenesis in knockout mice (C, G, and M). B, F, and L: Reduction in PGC1a gene expression in PRDX6 mutant mice. D, H, and N: A slight reduction of CD36 expression in liver of PRDX62/2 mice. Values are expressed as mean 6 SE. *P < 0.05, **P < 0.005, ***P < 0.0005 (n = 5 mice per group). a.u., arbitrary units.

result from increased b-cell death and/or birth defects injury (38). Moreover, our results are in agreement with through replication and neogenesis (35). Oxidative stress previous work, which demonstrated that transgenic di- is an important mechanism leading to these defects (26). abetic mice with overexpression of PRDX4 had reduced 2 2 We present here evidence that PRDX6 / mice have a de- NASH scores and a significant improvement of dyslipidemia creased pancreatic b-cell function as suggested by the (39). In addition, previous results using mice heterozygous following: 1) reduction of insulin secretion in response for mitochondrial trifunctional protein showed how mito- to glucose challenge with higher levels of insulin mRNA chondrial oxidation may be important in the pathogen- expression and 2) decreased area and volume of the islets esis of nonalcoholic fatty liver disease and NASH of Langerhans, probably due to higher levels of apoptotic associated with IR. This finding provides evidence of destruction of pancreatic b-cells, although this latter as- the contribution of genetic susceptibility to the devel- pect was not specifically addressed. Interestingly, the opment of nonalcoholic fatty liver disease and IR. Fur- overexpression of another PRDX, PRDX4 in mice, has thermore, the authors reported (40) that in the same been demonstrated to protect pancreatic b-cells from animal model diets high in polyunsaturated fatty acids ROS damage after high-dose streptozotocin-induced dia- can induce NASH. In conclusion, we can hypothesize betes (20). that PRDXs and genetic susceptibility can have a pri- In keeping with the clinical profile of many diabetic mary role in the progression of hepatic dysfunction to 2 2 patients (36), PRDX6 / mice have augmented levels of NASH. triglyceridesandVLDLsanddecreased concentrations of PRDX6 has been shown to play an important role HDL. This finding is also in agreement with a potential during inflammatory processes, and, according to data role of PRDX6 in cardiovascular disease, as previously from a recent study (41), topically administered PRDX6 2 2 suggested (37). Furthermore, PRDX6 / mice had higher maintained the homeostasis of corneal cells, reducing in- NASH scores and liver morphological alterations, features flammation and suppressing neovascularization and apo- very similar to those present in T2D patients with diabetic ptosis, induced by ultraviolet irradiation. Inflammation is dyslipidemia (24). An exacerbated hepatocellular injury a main factor leading to diabetes and its associated com- 2 2 was also previously demonstrated in PRDX6 / mice plications (19). A recent study (42) in b-cells demon- linked to a different model of hepatic ischemia-reperfusion strated that mRNA and protein expression of PRDX6 3218 PRDX6 and Pathogenesis of Diabetes Diabetes Volume 63, October 2014

Figure 6—Genetic expression of cytokines in insulin target tissues. Expression levels of the main cytokines involved in diabetes and NASH were analyzed by real-time PCR. mRNA was extracted from skeletal muscle, liver, and WAT of fasted (4 h) WT (white bars) and PRDX62/2 (black bars) mice. A, E,andI: Increased levels of IL-6 in WAT of knockout mice. B, F,andL: Higher genetic expression of IL-1b in PRDX62/2 mice. C, G, and M: Increased expression of the anti-inﬂammatory cytokine IL-10 overall. D, H, and N: TNF-a is highly expressed in PRDX6 mutant mice. Values are expressed as mean 6 SE. *P < 0.05, **P < 0.005, ***P < 0.0005 (n = 5 mice per group). a.u., arbitrary units.

was decreased after treatment with proinflammatory patients, likely reducing the inflammatory state and ox- cytokines such as TNF-a and interferon-g.Inthecurrent idative stress. 2 2 study, we demonstrated a different cytokine profile ex- While PRDX6 / mice represent an excellent model to 2 2 pression between PRDX6 / and WT mice, suggesting a study the general impact of oxidative stress and reduced relationship between PRDX6 and inflammatory stimuli antioxidant capacity on glucose homeostasis, we have to that would be worth further investigation. Since PRDX-6 acknowledge some limitations of the current study that can be hyperoxidized and inhibited by excess hydrogen are typical of studies using transgenic or knockout mice to peroxide (43), this enzyme may become inactivated in investigate complex disease (46). These limitations in- diabetic patients, thus contributing, in a vicious circle, to clude the following: 1) lack of the evaluation of compen- disease pathogenesis and progression. Thus, PRDX6 may satory mechanisms for the loss of individual proteins and represent a specific target for antioxidant-based pharma- 2) difficulty in distinguishing phenotypes arising from de- cological interventions; moreover, PRDX6 oxidation velopmental defects from those resulting from impaired level, for instance in blood cells, may represent a novel signaling. The results of these limitations may be the marker for monitoring oxidative/metabolic status and elevated levels of anti-inflammatory cytokines IL-6 and 2 2 response to preventive therapy. In line with these ideas, IL-10 observed in our PRDX6 / mouse model. However, PRDX6 levels were significantly higher in diabetic it is important to remark that the link of IL-6 and IL-10 patients with endothelial dysfunction compared with with IR and diabetes is still controversial. IL-10 has been control subjects, which may possibly represent a physio- considered an anti-inflammatory cytokine that improved logical adaptation against oxidative stress in patients hepatocellular injury (47); this may explain, at least in with atherosclerosis (44). Moreover, another study part, its increased levels in our animals that were found (45) conducted in diabetic patients demonstrated that to have damage from NASH. physical training was able to increase PRDXs levels Genetic variants of PRDX6 have been associated with measured in erythrocytes, counteracting the oxidative different responses to chemotherapy in patients with damage that is typical of diabetes. These results are im- breast cancer because of its antioxidant properties (48). portant in outlining the role of the PRDX family in the Assessing and comparing PRDX6 expression levels in management of glucose homeostasis and in the preven- human skeletal muscle, liver, and/or WAT specimens tion of diabetes and cardiovascular disease in diabetic obtained from control, prediabetic, and frankly diabetic diabetes.diabetesjournals.org Pacifici and Associates 3219

laboratory animals. B.C. performed the protein experiments. M.G.S. performed the histological experiments. G.D. and M.T. performed the statistical analysis. A.B. contributed to the analysis of the data and drafting of the manuscript. S.C. and A.C. performed the PCR experiments. F.F. performed the blood analysis. M.F. contributed to the writing of the manuscript. G.S. performed the histological experiments. P.S. conceived the experimental design and contributed to the writing of the manuscript. D.D.-M. helped to organize the data and contributed to the writing of the manuscript. A.G. handled the laboratory animals and contributed to the data analysis. A.O. conceived the histological protocols and contributed to the data analysis. D.L. helped to conceive the experimental design and organize the data, and contributed to the writing of the manuscript. D.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Prior Presentation. Parts of this study were presented in abstract form at the 74th Scientiﬁc Sessions of the American Diabetes Association, San Francisco, CA, 13–17 June 2014.

References 1. Styskal J, Van Remmen H, Richardson A, Salmon AB. Oxidative stress and diabetes: what can we learn about insulin resistance from antioxidant mutant mouse models? Free Radic Biol Med 2012;52:46–58 2. Tiedge M, Lortz S, Munday R, Lenzen S. Protection against the co-operative Figure 7—Schematic illustration of PRDX62/2-impairing insulin signaling pathways in skeletal muscle. PRDX62/2 impairs insulin sig- toxicity of nitric oxide and oxygen free radicals by overexpression of antioxidant naling pathways in skeletal muscle at different levels, as follows: 1) enzymes in bioengineered insulin-producing RINm5F cells. Diabetologia 1999;42: decrease of insulin secretion by b-cells; 2) reduction of IRS1 tyro- 849–855 sine phosphorylation with subsequent diminished interaction be- 3. Rains JL, Jain SK. Oxidative stress, insulin signaling, and diabetes. Free tween IRS1 and PI3K; 3) decrease of PKB/Akt phosphorylation Radic Biol Med 2011;50:567–575 4 and activation; ) lower GLUT4 translocation from cytoplasmic re- 4. Wood ZA, Schröder E, Robin Harris J, Poole LB. Structure, mechanism and serve pool to plasma membrane; and 5) less glucose uptake. regulation of peroxiredoxins. Trends Biochem Sci 2003;28:32–40 P, phosphorylation. 5. Fisher AB. Peroxiredoxin 6: a bifunctional enzyme with glutathione peroxidase and phospholipase A₂ activities. Antioxid Redox Signal 2011;15:831– 844 6. Wang X, Phelan SA, Forsman-Semb K, et al. Mice with targeted mutation of individuals in association with genetic studies evaluating peroxiredoxin 6 develop normally but are susceptible to oxidative stress. J Biol polymorphisms of the PRDX6 gene would have great im- Chem 2003;278:25179–25190 pact on further understanding the role of PRDX6 in the 7. Fatma N, Singh P, Chhunchha B, et al. Deficiency of Prdx6 in lens epithelial severity of inflammatory conditions such as diabetes. cells induces ER stress response-mediated impaired homeostasis and apoptosis. In conclusion, on the basis of the present data, the Am J Physiol Cell Physiol 2011;301:C954–C967 emerging picture is of a complex interaction between 8. Brinkmann C, Chung N, Schmidt U, et al. Training alters the skeletal muscle PRDX6 and insulin-signaling pathways involved in the antioxidative capacity in non-insulin-dependent type 2 diabetic men. Scand regulation of glucose homeostasis, lipid metabolism, and J Med Sci Sports 2012;22:462–470 inflammatory response. PRDX6, through its functional 9. Liang H, Balas B, Tantiwong P, et al. Whole body overexpression of PGC- activity against oxidative stress and inflammation, may be 1alpha has opposite effects on hepatic and muscle insulin sensitivity. Am J Physiol Endocrinol Metab 2009;296:E945–E954 indicated as a new molecular target that will be useful in 10. Carter JD, Dula SB, Corbin KL, Wu R, Nunemaker CS. A practical guide to the development of preventive strategies and novel rodent islet isolation and assessment. Biol Proced Online 2009;11:3–31 therapies for T2D and its related diseases. Further studies 11. Ferlosio A, Arcuri G, Doldo E, et al. Age-related increase of stem marker should be performed to clarify and better define the role expression influences vascular smooth muscle cell properties. Atherosclerosis of PRDX6 in diabetes and its associated metabolic 2012;224:51–57 dysfunctions. 12. Kleiner DE, Brunt EM, Van Natta M, et al.; Nonalcoholic Steatohepatitis Clinical Research Network. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology 2005;41:1313–1321 Funding. This work was supported by Research Project 2009 grant, Fondazione 13. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using Roma; PRIN 2010 and 2011 grants from the Ministero dell’Istruzione, dell’Università real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 2001; e della Ricerca (to D.L. and P.S.); Fondazione Umberto Di Mario; Società Italiana 25:402–408 Diabetologia (SID) Lazio grant; 2010 Grant from Associazione Italiana per la Ricerca 14. Thiebaud D, Jacot E, DeFronzo RA, Maeder E, Jequier E, Felber JP. The sul Cancro; and grant D.3.2-2013 from Università Cattolica del Sacro Cuore. effect of graded doses of insulin on total glucose uptake, glucose oxidation, and Duality of Interest. No potential conflicts of interest relevant to this article glucose storage in man. Diabetes 1982;31:957–963 were reported. 15. Asano T, Fujishiro M, Kushiyama A, et al. Role of phosphatidylinositol Author Contributions. F.P. and R.A. performed laboratory experiments 3-kinase activation on insulin action and its alteration in diabetic conditions. Biol and contributed to the writing of the manuscript. G.P.S. and D.P. handled the Pharm Bull 2007;30:1610–1616 3220 PRDX6 and Pathogenesis of Diabetes Diabetes Volume 63, October 2014

16. Ojuka EO, Goyaram V, Smith JA. The role of CaMKII in regulating GLUT4 34. Kobayashi H, Matsuda M, Fukuhara A, Komuro R, Shimomura I. Dysregu- expression in skeletal muscle. Am J Physiol Endocrinol Metab 2012;303:E322– lated glutathione metabolism links to impaired insulin action in adipocytes. Am E331 J Physiol Endocrinol Metab 2009;296:E1326–E1334 17. Ng Y, Ramm G, Lopez JA, James DE. Rapid activation of Akt2 is sufficient to 35. Weir GC, Bonner-Weir S. Islet b cell mass in diabetes and how it relates to stimulate GLUT4 translocation in 3T3-L1 adipocytes. Cell Metab 2008;7:348–356 function, birth, and death. Ann N Y Acad Sci 2013;1281:92–105 18. Mooradian AD. Dyslipidemia in type 2 diabetes mellitus. Nat Clin Pract 36. Arca M, Pigna G, Favoccia C. Mechanisms of diabetic dyslipidemia: rele- Endocrinol Metab 2009;5:150–159 vance for atherogenesis. Curr Vasc Pharmacol 2012;10:684–686 19. Hotamisligil GS. Inflammation and metabolic disorders. Nature 2006;444: 37. Wang X, Phelan SA, Petros C, et al. Peroxiredoxin 6 deficiency and ath- 860–867 erosclerosis susceptibility in mice: significance of genetic background for as- 20. Ding Y, Yamada S, Wang KY, et al. Overexpression of peroxiredoxin 4 protects sessing atherosclerosis. Atherosclerosis 2004;177:61–70 against high-dose streptozotocin-induced diabetes by suppressing oxidative stress 38. Eismann T, Huber N, Shin T, et al. Peroxiredoxin-6 protects against mito- and cytokines in transgenic mice. Antioxid Redox Signal 2010;13:1477–1490 chondrial dysfunction and liver injury during ischemia-reperfusion in mice. Am J 21. Da Silva-Azevedo L, Jähne S, Hoffmann C, et al. Up-regulation of the Physiol Gastrointest Liver Physiol 2009;296:G266–G274 peroxiredoxin-6 related metabolism of reactive oxygen species in skeletal 39. Nabeshima A, Yamada S, Guo X, et al. Peroxiredoxin 4 protects against muscle of mice lacking neuronal nitric oxide synthase. J Physiol 2009;587:655– nonalcoholic steatohepatitis and type 2 diabetes in a nongenetic mouse model. 668 Antioxid Redox Signal 2013;19:1983–1998 22. Wang X, Vatamaniuk MZ, Roneker CA, et al. Knockouts of SOD1 and GPX1 40. Ibdah JA, Perlegas P, Zhao Y, et al. Mice heterozygous for a defect in exert different impacts on murine islet function and pancreatic integrity. Antioxid mitochondrial trifunctional protein develop hepatic steatosis and insulin re- Redox Signal 2011;14:391–401 sistance. Gastroenterology 2005;128:1381–1390 23. Dowman JK, Tomlinson JW, Newsome PN. Pathogenesis of non-alcoholic 41. Shi H, Yu HJ, Wang HY, et al. Topical administration of peroxiredoxin-6 on fatty liver disease. QJM 2010;103:71–83 the cornea suppresses inflammation and neovascularization induced by ultravi- 24. Masuoka HC, Chalasani N. Nonalcoholic fatty liver disease: an emerging olet radiation. Invest Ophthalmol Vis Sci 2012;53:8016–8028 threat to obese and diabetic individuals. Ann N Y Acad Sci 2013;1281:106–122 42. Paula FM, Ferreira SM, Boschero AC, Souza KL. Modulation of the perox- 25. Wellen KE, Hotamisligil GS. Inflammation, stress, and diabetes. J Clin Invest iredoxin system by cytokines in insulin-producing RINm5F cells: down-regulation 2005;115:1111–1119 of PRDX6 increases susceptibility of beta cells to oxidative stress. Mol Cell En- 26. Kaneto H, Katakami N, Kawamori D, et al. Involvement of oxidative stress in docrinol 2013;374:56–64 the pathogenesis of diabetes. Antioxid Redox Signal 2007;9:355–366 43. Rhee SG, Woo HA. Multiple functions of peroxiredoxins: peroxidases, sen- 27. Muscogiuri G, Salmon AB, Aguayo-Mazzucato C, et al. Genetic disruption of sors and regulators of the intracellular messenger H₂O₂, and protein chaperones. SOD1 gene causes glucose intolerance and impairs b-cell function. Diabetes Antioxid Redox Signal 2011;15:781–794 2013;62:4201–4207 44. El Eter E, Al Masri A, Habib S, et al. Novel links among peroxiredoxins, 28. Taniguchi CM, Emanuelli B, Kahn CR. Critical nodes in signalling pathways: endothelial dysfunction, and severity of atherosclerosis in type 2 diabetic patients insights into insulin action. Nat Rev Mol Cell Biol 2006;7:85–96 with peripheral atherosclerotic disease. Cell Stress Chaperones 2014;19:173– 29. Gual P, Le Marchand-Brustel Y, Tanti JF. Positive and negative regulation of 181. insulin signaling through IRS-1 phosphorylation. Biochimie 2005;87:99–109 45. Brinkmann C, Blossfeld J, Pesch M, et al. Lipid-peroxidation and peroxiredoxin- 30. Archuleta TL, Lemieux AM, Saengsirisuwan V, et al. Oxidant stress-induced overoxidation in the erythrocytes of non-insulin-dependent type 2 diabetic men loss of IRS-1 and IRS-2 proteins in rat skeletal muscle: role of p38 MAPK. Free during acute exercise. Eur J Appl Physiol 2012;112:2277–2287 Radic Biol Med 2009;47:1486–1493 46. Plum L, Wunderlich FT, Baudler S, Krone W, Brüning JC. Transgenic and 31. Bloch-Damti A, Bashan N. Proposed mechanisms for the induction of insulin knockout mice in diabetes research: novel insights into pathophysiology, limi- resistance by oxidative stress. Antioxid Redox Signal 2005;7:1553–1567 tations, and perspectives. Physiology (Bethesda) 2005;20:152–161 32. Masharani UB, Maddux BA, Li X, et al. Insulin resistance in non-obese 47. Marra F, Bertolani C. Adipokines in liver diseases. Hepatology 2009;50: subjects is associated with activation of the JNK pathway and impaired insulin 957–969 signaling in skeletal muscle. PLoS ONE 2011;6:e19878 48. Seibold P, Hall P, Schoof N, et al. Polymorphisms in oxidative stress-related 33. Lei XG, Vatamaniuk MZ. Two tales of antioxidant enzymes on b cells and genes and mortality in breast cancer patients—potential differential effects by diabetes. Antioxid Redox Signal 2011;14:489–503 radiotherapy? Breast 2013;22:817–823