Beta-cell-specific expression of NOX5 aggravates high fat diet-induced impairment of islet insulin secretion in mice

(1)

Article

Reference

Beta-cell-specific expression of NOX5 aggravates high fat diet-induced impairment of islet insulin secretion in mice

BOUZAKRI, Karim, et al.

Abstract

NADPH oxidases (NOX-es) produce reactive oxygen species and modulate β-cell insulin secretion. Islets of Type2 diabetic subjects present elevated expression of NOX5. Here we sought to characterize regulation of NOX5 expression in human islets in vitro and to uncover the relevance of NOX5 in islet function in vivo using a novel mouse model expressing NOX5 in doxycycline-inducible, β-cell-specific manner (RIP/rtTA/NOX5 mice). In situ hybridization and immunohistochemistry employed on pancreatic sections demonstrated NOX5 mRNA and protein expressions in human islets. In cultures of dispersed islets NOX5 protein was observed in somatostatin-positive (δ) cells in basal (2.8mM glucose) conditions.

siRNA-mediated knock-down of NOX5 in human islets cultured in basal glucose concentrations resulted in diminished glucose-induced insulin secretion in vitro. However, when islets were preincubated in high (16.7mM) glucose media for 12 hours, NOX5 appeared also in insulin-positive (β) cells. In vivo, mice with β-cell NOX5 expression developed aggravated impairment of glucose-induced insulin secretion compared to control mice when [...]

BOUZAKRI, Karim, et al. Beta-cell-specific expression of NOX5 aggravates high fat

diet-induced impairment of islet insulin secretion in mice. Antioxidants & Redox Signaling, 2020

PMID : 31931619

DOI : 10.1089/ars.2018.7579

Available at:

http://archive-ouverte.unige.ch/unige:129173

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

1

Antioxidants and Redox Signaling Beta‐cell‐specific expression of NOX5 aggravates high fat diet‐induced impairment of islet insulin secretion in mice (DOI: 10.1089/ars.2018.7579) This paper has been peer‐reviewed and accepted for publication, but has yet to undergo copyediting and proof correction. The final published version may differ from this proof.

Original Research Communication

Beta‐cell‐specific expression of NOX5 aggravates high fat diet‐

induced impairment of islet insulin secretion in mice

Karim Bouzakri^1,2,Christelle Veyrat‐Durebex³,Chet Holterman⁴,Caroline Arous^1,3,Charlotte Barbieux⁵, Domenico Bosco⁵, Jordi Altirriba⁶, Mohamed Alibashe⁵, Benjamin B. Tournier⁷, Jenny E. Gunton^8,9, Sarah Mouche^3,12, William Bietiger², Alexis Forterre², Thierry Berney¹⁰,

Michel Pinget², Gerhard Christofori¹¹, Christopher Kennedy⁴ and Ildiko Szanto^12,13,§

1Department of Genetic Medicine and Development

2Centre Européen d’Etude du Diabète, Strasbourg, France

3Department of Cellular Physiology and Metabolism, University of Geneva, Geneva, Switzerland

4Department of Medicine, Division of Nephrology, Kidney Research Centre, Ottawa Hospital Research Institute, Ottawa, Canada

5Cell Isolation and Transplantation Center, Department of Surgery, Geneva University Hospitals and University of Geneva, Geneva, Switzerland

6Laboratory of Metabolism, Department of Internal Medicine, Geneva University Hospitals, Geneva, Switzerland

7Vulnerability Biomarkers Unit, Division of General Psychiatry, Department of Psychiatry, University Hospitals of Geneva,

8Centre for Diabetes, Obesity and Endocrinology, Westmead Millennium Institute, The University of Sydney, Sydney, Australia

9Diabetes and Transcription Factors group, Garvan Institute of Medical Research, Sydney, New South Wales, Australia

10Division of Transplantation, Department of Surgery, University Hospitals of Geneva, Geneva, Switzerland

11Department of Biomedicine, University of Basel, Basel, Switzerland

12Department of Internal Medicine, University Hospitals of Geneva and University of Geneva, Geneva, Switzerland

13Diabetes Center of the Faculty of Medicine at the University of Geneva, Geneva, Switzerland

Downloaded by Université de Genève from www.liebertpub.com at 01/20/20. For personal use only.

(3)

2

§Correspondence:

Ildiko Szanto,

Department of Internal Medicine

Geneva University Hospitals and University of Geneva 1 Rue Michel Servet

CH‐1211 Geneva4 Switzerland

Phone: +(41)‐22‐379‐5238 ildiko.szanto@unige.ch

Running title: NOX5‐induced beta cell dysfunction Word count: 5351, Number of references: 66 Greyscale illustrations 3; Color illustrations 4

Keywords: NOX5, NADPH oxidase, Beta cell, Insulin, Islet, Diabetes, Somatostatin

(4)

3

ABSTRACT

NADPH oxidases (NOX‐es) produce reactive oxygen species and modulate β‐cell insulin secretion. Islets of Type2 diabetic subjects present elevated expression of NOX5. Here we sought to characterize regulation of NOX5 expression in human islets in vitro and to uncover the relevance of NOX5 in islet function in vivo using a novel mouse model expressing NOX5 in doxycycline‐inducible, β‐cell‐specific manner (RIP/rtTA/NOX5 mice). In situ hybridization and immunohistochemistry employed on pancreatic sections demonstrated NOX5 mRNA and protein expressions in human islets. In cultures of dispersed islets NOX5 protein was observed in somatostatin‐positive (δ) cells in basal (2.8mM glucose) conditions. siRNA‐mediated knock‐down of NOX5 in human islets cultured in basal glucose concentrations resulted in diminished glucose‐induced insulin secretion in vitro. However, when islets were preincubated in high (16.7mM) glucose media for 12 hours, NOX5 appeared also in insulin‐positive (β) cells. In vivo, mice with β‐

cell NOX5 expression developed aggravated impairment of glucose‐induced insulin secretion compared to control mice when challenged with 14 weeks of high‐fat diet.

Similarly, in vitro palmitate pre‐incubation resulted in more severe reduction of insulin release in islets of RIP/rtTA/NOX5 mice compared to their control littermates. Decreased insulin secretion was most distinct in response to theophylline stimulation suggesting impaired cAMP‐mediated signaling due to increased phophodiesterase activation. Our data provide the first insight into the complex regulation and function of NOX5 in islets implying an important role for NOX5 in δ‐cell mediated intra‐islet crosstalk in physiological circumstances but also identifying it as an aggravating factor in β‐cell failure in diabetic conditions.

(5)

4

Introduction

Reactive oxygen species (ROS) are key modulators of islet insulin production and release.

As for the sources of these radicals, the members of the NADPH oxidase family (NOX‐es) emerged as of particular interest due to their strict intracellular compartmentalization, regulated ROS production, and their capacity to provide a link between cellular metabolism/energy state (NADPH) and ROS production (45,59). NOX enzymes are essential regulators of physiological insulin secretion but may exert harmful effects when chronically overactivated (2,18,38,42‐45). The NOX family comprises of five isoforms (NOX1‐5) and 2 dual oxidases (DUOX1‐2) (35). In rodent and human pancreatic islets the expressions of NOX1, NOX2 and NOX4, and the accessory subunits NOXA1, p22^phox, p47^phox and p67^phox have been confirmed (2,23,38,53,64). NOX5 is a particular member of the NOX family as it is expressed in diverse human tissues and cell lines but lacks a homologue in rats and mice (4,5,7,11,16,25). The presence of NOX5 mRNA in human islets has been reported by recent studies using RT‐PCR, real‐time PCR and RNA sequencing, however, these studies did not conduct any follow‐up investigations to establish the cell‐type specific expression or the pathophysiological significance of NOX5 (2,38). A possible link between NOX5 and islet function was suggested by gene expression profiling that showed elevated NOX5 mRNA levels in islets of Type2 diabetic patients compared to islets of non‐diabetic subjects (Geoprofiles GDS3882 / 220641_at).

The molecular structure and the characterictics of its enzymatic activity renders NOX5 a particularly intriguing molecule in islet function. Indeed, NOX5 contains four Ca²⁺ binding elements and several phosphorylation sites that control NOX5 activity in a concerted fashion (20,46). Calcium‐binding is essential for NOX5 superoxide production while phosphorylation of specific residues increases the Ca²⁺ sensitivity of the EF‐hands allowing NOX5 to respond to regulatory cues at physiological (10‐100nM) Ca²⁺ concentrations (4,6,20,46,48). Calcium signals, NADPH and ROS form a complex network to modulate islet glucose sensing and insulin secretion (40). Increase in glucose concentrations leads to cytosolic Ca²⁺ accumulation that triggers a biphasic wave of ‐cell insulin release. NADPH acts as one of the “amplifying factor” sustaining the second phase of insulin secretion (51).

In line with these studies, insulin secretagogues increase the NADPH/NADP ratio in rodent

(6)

5

islets and inhibition of NADPH formation reduces insulin secretion (3,27,39). Last but not least, a transient elevation in cellular ROS was identified as an essential second messenger promoting insulin secretion (45,50). NOX5 produces ROS through NADPH oxidation in a Ca²⁺‐dependent manner but its potential physiological role or pathological effect in islet insulin secretion has not yet been investigated.

The lack of substantial data concerning NOX5 in islet function urged us to i) characterize the expression of NOX5 in situ in human pancreatic sections and in vitro in isolated islets, ii) to assess its role in insulin secretion in human islets by employing siRNA‐mediated knock‐down of NOX5 and iii) to assess the potential of NOX5 role in vivo in diabetes‐

related impairment of ‐cell function using a novel mouse model. To achieve this latter goal we generated mice with doxycycline‐inducible, ‐cell specific NOX5 expression and evaluated their islet insulin secretory capacity in chow‐diet (CD)‐fed condition and after 14 weeks of a high‐fat diet (HFD) regime.

Results

Glucose‐induced upregulation of NOX5 expression in human islet ‐cells

Conventional RT‐PCR validated the presence of NOX2, NOX4 and NOX5 mRNAs in isolated whole human islets (Fig. 1A and Supplementary Fig. S1). NOX5 mRNA can undergo differential splicing giving rise to different NOX5 isoforms. The two major forms are termed NOX5 and , and have been cloned from the spleen and testis, respectively (4,20). Primers selectively recognizing either the  or  isoform demonstrated the presence of Nox5 mRNA in islets (Fig. 1B and Supplementary Fig. S1). The functionality of RT‐PCR reactions for NOX5 was confirmed by the amplicon obtained in the reaction using human spleen cDNA. The correct identity of the band obtained in islets with the NOX5 isoform‐specific primers was validated by gel purification of the band and subsequential sequencing. In situ hybridization using pancreatic sections confirmed the presence of NOX5 mRNA in the majority of islet cells with occasional cells displaying more prominent signal intensity. Exocrine pancreas was devoid of NOX5 labeling. No signal was obtained when using the negative control NOX5 sense probe (Fig. 1C). NOX5 protein expression was then evaluated by immunohistochemistry and immunofluorescence using human paraffin‐

embedded pancreatic sections. For immunohistochemistry a previously tested monoclonal NOX5‐specific antibody was employed (1,11) revealing prominent immunodetectable

(7)

6

NOX5 protein in numerous islet cells (Fig. 2A). Identification of islet cells with the highest NOX5 protein expression was then achieved by immunofluorescence employing an unrelated NOX5 antibody (HPA019362; Atlas Antibodies). The specificity of the antibody was established by Western blot and immunofluorescence comparing WT and NOX5‐

overexpressing HEK293 cells. This antibody detected a prominent band of approximately 72kDa in the NOX5‐transfected but not in WT HEK293 cells. By confocal immunofluorescence NOX5 transfected cells displayed dispersed cytoplasmic labeling while no signal was observed in WT HEK293 cells (Supplementary Fig. S2A). The specificity of the NOX5 antibody in immunofluorescence application using paraffin‐embedded tissues was ascertained on human pancreas serial sections that were incubated either with a non‐

relevant isospecific IgG (Control) or with NOX5 antibody (NOX5). Images taken in brightfield light were superposed to NOX5‐stained sections for islet identification (NOX5+Brightfield). In addition, serial sections were used for Hematoxylin‐eosin staining (HE) (Fig. 2B). Next, NOX5‐expressing cells in human islets were identified by immunofluorescence co‐staining with the aforementioned NOX5 antibody in conjuction with antibodies recognizing the three major islet hormones: somatostatin (SST), glucagon and insulin. High NOX5 expression was detected in cells that were also positive for SST.

Glucagon and insulin‐positive cells did not show NOX5‐positivity detectable by this antibody (Fig. 2C). Next, we employed immunofluorescence on dispersed whole human islets in order to investigate the regulation of NOX5 expression. Dispersed islet cells incubated under basal (2.8 mM) glucose concentrations revealed prominent NOX5 labeling in SST‐positive () cells and in colocalization with somatostatin‐containing vesicles. At this basal glucose concentration no NOX5 staining was observed in insulin‐positive () cells (Fig. 2D). Next, immunofluorescence was applied to islets subjected to 16.7mM glucose media for 12 and 24 hours. Upon high‐glucose insults, a lower intensity, punctuated NOX5 labeling was visible in insulin‐positive cells, though not in colocalization with insulin‐

containing vesicles (Fig. 2E). Glucagon‐positive () cells were devoid of NOX5 signal in both basal (2.8mM) and high glucose (16.7mM) conditions (Supplementary Fig. S2B). NOX5 expression is mainly controlled at the protein level through proteasomal degradation (15).

In line with a postranscritional regulation we did not observe any changes in NOX5 mRNA levels assessed by real‐time PCR in whole islets incubated with 16.7mM glucose for 12 and

(8)

7

24 hours. However, glucose‐induced upregulation of NOX5 expression was observed in islets subjected to high‐glucose media for 48 hours (Fig. 2F). Transcription of other NOX isoforms detected in islets (Nox2 and Nox4) remained unchanged in the same conditions while Duox2 mRNA levels showed a tendency towards a non‐significant decrease after 48 hours of high‐glucose media incubation (Supplementary Fig. S2C). To substantiate the in vivo relevance of NOX5 upregulation we compared NOX5 expression in samples of islets of a previously described cohort of control and Type2 diabetic (T2DM) subjects (24). Islets of T2DM subjects displayed significantly increased NOX5 mRNA expression compared to non‐

diabetic controls (Fig. 2G). Expressions of NOX1, 2, 3, 4 and DUOX2 were statistically not different between control and T2DM subjects (Table 1).

siRNA‐mediated knock‐down of NOX5 in human islets abrogates glucose‐induced insulin secretion

To gain further insight into the relevance of NOX5 in islet insulin secretion we performed siRNA‐mediated knock‐down of NOX5 in dispersed human islet cell cultures (Fig. 3).

Efficiency of the NOX5 siRNA was verified in cultures of primary human skeletal muscle myotubes showing a suppression of NOX5 mRNA levels of approximately 35% of scrambled RNA transfected cells (Supplementary Fig. S3). For the human islet transfection experiments, islets were dispersed and cultured in 5.6 mM glucose media overnight and transfected by NOX5‐specific or control (scrambled) siRNAs and further kept in culture for 48 hours in low (5.6 mM, LG) or high (16.7 mM, HG) glucose media. At the end of the incubation period a GIIS assay was performed using 16.5mM glucose as stimulatory agent.

Islets transfected with scrambled siRNA displayed an average 1.8‐fold increase in insulin secretion in response to glucose stimulation (S) compared to their basal (B) secretion when cultured in LG media (Fig. 3A, scrambled, LG). 48 hours preincubation in HG media led to an increase in basal secretory rate with a blunted secretory response to glucose (Fig. 3A, scrambled, HG). Islets transfected with NOX5‐specific siRNA (siNOX5) displayed a tendency towards lower basal insulin secretion and an abrogated glucose‐induced insulin secretory response compared to scrambled siRNA transfected islets (Fig. 3A, siNOX5, LG). Upon HG preincubation, siNOX5 transfected islets also responded by an increase in basal insulin secretion but failed to reach the levels observed in control transfected islets. Similar to control islets, glucose stimulation did not further increase their insulin secretion (Fig. 3A,

(9)

8

siNOX5, HG). Insulin contents were similar between scrambled and siNOX5 transfected islets (Fig. 3B).

Creation of mice with inducible ‐cell‐specific expression of NOX5

The in vitro human islet experiments suggested that NOX5 is indeed involved in the regulation of islet insulin secretion and the adaptation to high glucose conditions.

However, these experiments could not properly distinguish between the effects of the lack of NOX5 in ‐ or ‐cells and could not provide information concerning the role of NOX5 in islet function in vivo. The Nox5 gene is present in the human genome, yet rats and mice, the most frequently employed species in islet function research, are natural NOX5 gene knockouts (4). Thus, in order to study NOX5 function in vivo, diverse mouse models with cell‐specific knock‐in of Nox5 were created. These models have successfully been used to characterize the role of NOX5 in diabetic kidney disease, hypertension and in stroke (12,28,61). To distinguish the role of NOX5 between islet ‐ or ‐cells in vivo, two novel NOX5‐expressing mouse models with RIP or SST‐promoter driven knock‐in of the Nox5 gene needed to be generated. In our first set of investigations that we present in this paper, we set out to examine the direct role of NOX5 in ‐cell insulin secretion by employing mice with doxycycline‐inducible ‐cell‐specific expression of NOX5 (RIP/rtTA//NOX5) mice. For this purpose, TetO/NOX5 transgenic FVB mice were crossed with RIPrtTA C57Bl6/CBA mice that express the tetracycline‐activated transcription factor rtTA under the direction of the rat insulin 2 promoter (RIP) (41). In the presence of doxycyline (Dox) the rtTA transcriptional activator binds the TetO promoter triggering NOX5 expression in islet ‐cells. (Fig. 4A). NOX5 and rtTA transgene transmissions were verified by PCR (Fig. 4B and Supplementary Fig. S4A). Then, 5‐week‐old male mice with similar transgene intensity signals were provided drinking water with (Dox+) or without (Dox‐) 1mg/ml doxycycline for 6 weeks. Induction of NOX5 mRNA expression was verified by RT‐PCR in isolated islets. Dox+ mice displayed a strong NOX5 amplicon while no signal was detected in Dox‐ mice (Fig. 4C Supplementary Fig. S4B). Besides islet ‐cells, the RIP promoter is also known to induce gene expression in selected hypothalamic neurons (32).

In line with these data, NOX5 mRNA was also detected in the hypothalamus of Dox+ mice whereas other organs (spleen and liver) remained devoid of NOX5 (Fig. 4D and Supplementary Fig. S4C). Immunohistochemistry confirmed NOX5 expression in the

(10)

9

majority of islet cells in pancreatic sections of Dox+ mice with no signal observed in Dox‐

mice (Fig. 4E). Functionality of the NOX5 protein was validated in isolated islets by measuring superoxide production using the nitroblue tetrazolium (NBT) reduction test.

Islets derived from Dox+ mice showed approximately 2‐fold higher superoxide production than Dox‐ mice (Fig. 4F). The ROS‐producing capacity of NOX5 is enhanced by increases in calcium concentrations. Thapsigargine is a non‐competitive inhibitor of the sarco/endoplasmic reticulum Ca²⁺ATPase pump (SERCA) responsible for Ca²⁺transport into the ER. Blocking ER Ca²⁺transport leads to ER storage depletion resulting in secondary Ca²⁺

influx through the plasma membrane triggering a raise in cytosolic Ca²⁺concentrations. ER Ca²⁺store depletion leads to ER stress and the induction of Ca²⁺‐dependent apoptotic pathways in islets (29,49). In order to evaluate whether NOX5 is involved in the onset of Ca²⁺‐mediated ER stress we compared Thapsigargin‐induced expressions of two ER stress marker proteins, 78‐kDa glucose‐regulated protein (GRP78) and C/EBP homologous protein (CHOP), between islets derived from Dox‐ and Dox+ mice (Fig. 4G). Expressions of these two proteins were comparable between mice with and without islet NOX5 expression.

Metabolic characterization of RIP/rtTA/NOX5 mice

To investigate the potential role of NOX5 in relation to hypernutrition‐induced impairment of insulin secretion we studied four groups of male mice. Mice were first administered drinking water for 2 weeks with or without Dox addition, and were subsequently provided either a control chow (CD) or a high‐fat containing diet (HFD) for 14 weeks while continuing with or without Dox administration. Two weeks of Dox application had no effect on body weight gain (3.70  0.42 g and 3.30  0.33 g in Dox‐ and Dox+ mice, respectively). 14 weeks of HFD feeding resulted in elevated body weight gain and fat deposition compared to the CD‐fed groups but there was no difference between Dox‐ and Dox+ mice within the same regime group (Fig. 5A‐B). Similarly, no difference was observed in fed or starved glucose levels between Dox‐ and Dox+ mice regardless whether they were kept under CD or HFD regime (Table 3). A detailed metabolic characterization was performed in calorimetric cages. Heat production and locomotor activity was similar between Dox‐ and Dox+ mice in both diet regime (Fig. 5C). Interestingly, however, CD‐fed Dox+ mice displayed increased diurnal food (p=0.054) and water intake (p=0.031), and an

(11)

10

increase in the diurnal respiratory exchange ratio (RER) reflecting more prominent carbohydrate utilization (Fig. 5D). 24 hours cumulative food and water intake showed no differences between CD‐fed Dox‐ and Dox+ mice, pointing towards disturbed daily rhythm of feeding in NOX5‐expressing mice (Fig. 5E). HFD feeding resulted in attenuated differences between light and dark period food and water intake, and in diminished RER values indicating prominent fat utilization in Dox‐ mice. HFD‐fed Dox+ mice were similar to Dox‐ mice in respect to these parameters (Fig. 5D‐E). In order to evaluate the effect of differences in circadian rhytm on metabolic homeostasis we evaluated random fed glucose and insulin levels in CD‐fed control and NOX5 expressing mice between 2:30‐3:00 PM when their food intake and energy utilization showed the most noticeable differences.

Glycemic values and insulin concentrations between control and NOX5 expressing mice showed no differences (Fig. 5F). At the same daylight period we also evaluated the expressions of hypothalamic orexigenic (neuropeptide Y, NPY) and anorexigenic (proopiomelanocortin, POMC) peptides that showed no difference between control (C) and NOX5‐expressing (NOX5) mice (Fig. 5G). Similarly, hypothalamic mRNA levels of the circadian rhythm‐regulating Bmal and Reverba genes were were undistinguishable between these two groups of mice (Fig. 5H).

Beta‐cell‐specific NOX5 expression aggravates diet‐induced impairment of insulin secretion in vivo and in vitro

To characterize the effect of ‐cell NOX5‐expression on islet function we compared glucose‐induced insulin secretion (GIIS) between CD‐ and HFD‐fed control and NOX5‐

expressing mice both in vivo and in vitro. CD‐fed Dox‐ and Dox+ mice displayed similar serum insulin levels both in starved conditions (Basal, B) and after 5 minutes of 2 mg/kg intraperitoneal glucose injection (Stimulated, S) (Fig. 6A; left panel). Similarly, islets isolated from CD‐fed Dox‐ and Dox+ mice displayed comparable insulin secretion in basal conditions (2.8mM, Basal, B) and in response to 16.7mM glucose challenge (Stimulated, S) in vitro. Maximum secretion capacity assessed as theophillyne‐induced (Theo) insulin secretion was also unaltered between control and NOX5‐expressing mice (Fig. 5A; right panel). Islet insulin content was undistinguishable between CD‐fed Dox‐ and Dox+ mice (data not shown). After nine weeks of HFD feeding Dox‐ but not Dox+ mice displayed elevated starved serum insulin levels (0.240.01 and 0.390.06 vs. 0.230.02 and

(12)

11

0.250.01 ug/L in CD and HFD‐fed Dox‐ vs. CD and HFD‐fed Dox+ mice, respectively) (Fig.

6B; left panel). Intraperitoneal glucose injection led to a modest increase in insulin levels in HFD‐fed control mice but failed to induce any detectable increase in insulin secretion in NOX5‐expressing mice. Islet number and size quantified on pancreatic sections revealed no differences between HFD‐fed control and NOX5‐expressing mice (Supplementary Fig. S5).

Abrogated glucose‐induced insulin secretion was also confirmed in vitro using islets isolated from Dox‐ and Dox+ mice sacrificed after 14 weeks of HFD (Fig. 6C; right panel).

Indeed, islets of Dox‐ mice still retained an approx. 30% increase in insulin release upon exposure to 16.7mM glucose. By contrast, islets of Dox+ mice lacked any glucose‐induced insulin secretory response and their theophylline‐induced insulin release capacity was significantly diminished. Islet insulin content was unaltered between Dox‐ and Dox+ mice (data not shown). Enhanced lipotoxicity‐associated ‐cell secretory defect related to NOX5 expression was also implied by additional in vitro experiments. 7‐week‐old male mice receving drinking water with or without Dox supplementation for 6 weeks were sacrificed, the islets isolated and incubated in the presence of 0.5mM palmitate for 72 hours and subsequently subjected to GIIS test. Palmitate‐exposed islets displayed abrogated insulin secretory response to 16.7mM glucose challenge, with no difference between islets of Dox‐ and Dox+ mice. However, theophylline‐induced insulin secretion was significantly lower in NOX5‐expressing compared to control islets (Fig. 6C). Theophylline induces cAMP levels through inhibition of phosphodiesterase. To investigate the effect of NOX5 on cAMP‐mediated insulin secretion from the opposite angle we compared the effect of the GLP‐1 analogue exendin‐4, an activator of adenylcyclase. Surprisingly, exendin‐4 (Ex) raised insulin secretion in Dox‐ and Dox+ islets to the same extent. Total islet insulin content was similar between Dox‐ and Dox+ mice (data not shown).

Discussion

Reactive oxygen species are established bidirectional modulators of islet insulin secretion and survival exerting both beneficiary and detrimental effects depending on the amount and the source of ROS (50). This dual role of ROS in islet function have been highlighted by several studies (reviewed in (45)). Indeed, a transient increase in ‐cell ROS production after glucose administration has been attributed an essential function in insulin secretion (38,44). In contrast, sustained ROS elevation has been linked to impaired insulin release

(13)

12

(19,36). NOX enzymes are major sources of cellular ROS. NOX‐es are expressed in a tissue‐

and cell‐specific manner, regulated by particular stimuli and act as crucial mediators of various physiological cellular processes. On the other hand, chronic overactivation of NOX‐

derived ROS production has been linked to the onset of diverse pathologies (35,45).

Concerning islet function in specific, a physiological inhibitory role for NOX2 in glucose‐

induced insulin secretion was established by investigations conducted in NOX2‐knock‐out mice employing both in vitro isolated islets and in vivo experiments (18,38,44). By contrast, deregulated activation of NOX2/NOX1 was associated with hampered islet function in diabetic or inflammatory conditions (42,43). A similar dual modulatory role was also proposed for NOX4. Employing a chemical inhibitor of NOX4 Anvari et al. found that short‐

term activation of NOX4 was necessary for appropriate glucose‐induced insulin secretion, however, prolonged overstimulation of NOX4 resulted in damaged ‐cell function (2). The fact that in physiological conditions NOX enzymes can both exert antagonistic (NOX2) and agonistic (NOX4) effects on insulin secretion highlights the necessity of further exploration of their islet cell type (, , ) distribution, subcellular localization, regulation and signaling mechanisms. Most organs and cell types express several NOXes but the function of each NOX isoform seems to be unique and not replacable by other isoforms (6,16). In line with a specific regulatory role for each NOX isoform, no report has suggested a compensatory upregulation of other NOX‐es in mice rendered NOX‐deficient by classical gene recombination, siRNA‐mediated gene silencing or chemical inhibition (2,18,37,38).

Our study examined the role of one of the least investigated NOX isoforms, NOX5 in islet function. Expression of NOX5 mRNA in human pancreas and isolated islets was previously observed by RT‐PCR, real‐time PCR and RNA sequencing (2,21,38). Employing pancreatic sections we validated the presence of Nox5 mRNA and protein in human islets in situ. Of note, that in situ hybridization detected Nox5 mRNA expression in the majority of islet cells implying the presence of Nox5 mRNA in ‐cells. These data are in line with RNA sequencing analysis that reported Nox5 mRNA in human islet sorted islet ‐cells (9).

Compared to the more ubiquitous pattern of Nox5 mRNA expression immunodetectable NOX5 protein was observed in a more restricted number of cells scattered within the islet, identified by co‐immunoflourescence staining as ‐cells. This divergence between mRNA and protein detection pattern might either be related to different sensitivity between the

(14)

13

mRNA and protein detection methods, or alternatively, it might reflect a differential rate of hsp70/ubiquitin‐mediated NOX5 protein degration between ‐ and ‐cells. Expression of NOX5 is regulated at the protein level through hsp70‐induced proteasomal degradation allowing upregulation of NOX5 protein levels without concomitant alterations in mRNA transcription in diverse cell types (13‐15). In line with this well‐characterized regulatory mechanism of NOX5 expression, we showed divergent NOX5 mRNA and protein levels in human islet cell types. Indeed, our in‐situ hybridization experiments demonstrated the presence of NOX5 mRNA in the majority of islet cells implying the presence of NOX5 mRNA in ‐cells (insulin‐positive cells) that make up more than 80% of islet cells. On the contrary, NOX5 protein expression was below the detection level of the two antibodies employed on both human pancreatic sections and on dispersed human islet ‐cells but was raised above detection level once ‐cells were induced by high glucose concentrations. Of additional interest, that a low ability of ‐cells to increase HSP70 expression under high glucose/inflammatory challenges together with an increase of NOX activity has previously been proposed as a key factor in ‐cell failure (34). Diminished proteasomal degradation of NOX5 could provide an explanation for the detection of NOX5 protein in ‐cells in islets incubated in high‐glucose media and could also account for the delay in the elevation of NOX5 mRNA levels (48hours) compared to protein levels (12 hours) observed in these islets. A similar delayed elevation of Nox5 mRNA compared to the protein amount was also observed in Angiotensin II‐treated human podocytes (28). NOX enzymes and NOX5 in specific, display cell‐specific expression patterns within different organs with cell‐specific functions and related pathologies (61). In this context a cell‐specific (‐ and ‐cells) expression and regulation for NOX5 in different islet cell identities is in line with current knowledge concerning NOX5. Our current study established this dual expression pattern and then proceeded to evaluate the lack of proper NOX5 function in human islets in vitro using an siRNA approach. These experiments indicated that in human islets cultured under low glucose conditions NOX5‐mediated signals are essential for proper glucose‐induced insulin secretion. Decreased expression of islet NOX5 also attenuated upregulation of basal insulin release in response to chronic glucose exposure. Taken together, these results suggest dual roles for NOX5: a key mediator of insulin release in response to physiological cues and a mediator in the islet secretion compensatory mechanism when confronting

(15)

14

pathologically elevated glucose levels. Unbridled insulin release is a hallmark of islets of Type 2 diabetic patients. Thus, the lack of chronic high glucose‐induced increase of basal insulin secretion together with the previously observed appearance of NOX5 protein in insulin‐positive cells might suggest a negative role for NOX5 in ‐cell insulin secretion in hyperglycemic conditions. The siRNA‐mediated knock‐down approach provides a suitable mean for understanding the role of NOX5 in whole human islet insulin secretion however, lacks the capacity to differentiate between the direct contribution of NOX5 in ‐cells and the indirect regulatory effects originating from ‐cells. In addition, while these experiments highlighted the role of NOX5 in islet function in vitro, they did not provide proper information about the role of NOX5 in islet function in vivo. Therefore, in our next set of investigations presented in this paper we focused our attention on the in vivo role of NOX5 in ‐cells, the islet cell type direcly accountable for insulin secretion.

To achieve this goal we generated mice with ‐cell specific expression of NOX5. A similar transgenic approach was successfully employed to uncover the role of NOX5 in vivo the development of diverse pathologies, notably diabetic nephropathy, high blood pressure and stroke (12,28,33). Using our novel mouse model we demonstrated a deteriorating effect for excess NOX5 on insulin release under hyperglycemic/diabetogenic conditions and linked this effect to enhanced degradation of cAMP, a critical amplifying component in insulin secretion. A previous study demonstrated NOX2‐immunoreactivity in association with insulin‐secretory vesicles together with an antagonistic effect for NOX2 on insulin release involving the cAMP‐mediated pathway (38). In our investigations, we observed several critical differences between the islet effects of NOX5 compared to NOX2.

First, as described before, NOX5 protein became detectable in ‐cells only when islets were cultured in high‐glucose conditions. Second, in line with the appearance of immunodetectable NOX5 in high glucose‐incubated ‐cells, the deteriorating effect of NOX5 on insulin secretion was observed only when islets were subjected to prior diabetogenic insults; e.g. islets of mice with diet‐induced diabetes were used or when isolated islets were preincubated with palmitate in vitro. Third, NOX5 immunoreactivity was not observed in direct colocalization with insulin‐containing vesicles suggesting a distinctive signaling pathway for NOX5 compared to NOX2. This observation is in line with the high specificity of cellular localization and action for each NOX isoform (35). And last

(16)

15

but not least, the negative effect of NOX5 was most prominently observed during theophylline‐induced insulin secretion while no differences were noted upon the addition of exendine‐4, a GLP‐1 receptor agonist. Both theophylline and exendine‐4 increase cAMP levels but through different pathways: theophylline is an inhibitor of phosphodiesterases (the enzymes degrading cAMP), while exendine‐4 is an activator of adenylcyclase (the enzyme generating cAMP). cAMP is a critical amplifying signal of ‐cell insulin secretion.

NOX2‐derived ROS was shown to act as an inhibitor of physiological glucose‐induced insulin secretion by inhibiting adenylcyclase activation (38). In contrast, our results suggest that the detrimental effect of NOX5 on insulin release is relayed through activation of phosphodiesterases (PDEs) (52). In islets different PDE isoforms are present and their precise isoform‐specific implication in glucose‐induced insulin secretion is subject of intensive investigations (52). The identification of PDE isoform linked to NOX5 action by PDE isoform‐specific inhibitors was not attempted in this study. Theophyllin is also considered as an agonist of adenosine receptors that mediate their effects through modulation of intracellular cAMP levels (58). Inhibition of the A1 adenosine receptor action increases cellular cAMP levels and insulin secretion (66). The potential involvement of Gi coupled adenosine receptor activation in the observed effect of Theophyllin in our experiments cannot be excluded and should be addressed in the future. Noteworthy however, that in our mouse model NOX5‐linked modification of Theophyllin action occurs only under diabetogenic conditions that also exert an enhancing effect on ‐cell NOX5 expression in human islets in vitro. The physiological relevance of glucose‐induced NOX5 expression in ‐cells in vitro was underlined by our data demonstrating elevated NOX5 levels in Type2 diabetic islets. The latter results are in line with previously reported expression profiling data comparing control and Type2 diabetic islets (Geoprofiles GDS3882 / 220641_at). Islet insulin secretion is modulated through crosstalk with hypothalamic neurons (54). The RIP promoter employed in our mouse model is also known to induce hypothalamic gene expression (32). The results of in vivo insulin secretion assays performed in the RIP‐NOX5 mice were in line with the in vitro experiments performed on isolated islets. These results suggest (but does not completely exclude) that the effects of hypothalamic RIP expression did not directly affect islet function in our mouse model.

Hypothalamic nuclei are major regulators of food intake and RIP‐expressing neurons