Introduction

High glucose-induced generation of reactive oxygen species (ROS) is an important contributor to the pathogenesis of diabetic nephropathy [1–8]. Although superoxide can be generated from mitochondrial electron transport, uncoupled endothelial nitric oxide synthase activity or activation of the family of NADPH oxidase (NOX) enzymes [9–12], the contribution of each of these pathways to kidney injury has not been fully elucidated.

Several isoforms of NOX have been identified in the kidney, including NOX1, NOX2 and NOX4 [9]. Activation of the membrane-bound NOX isoforms, NOX1 and NOX2, is dependent on recruitment and phosphorylation of several cytosolic subunits, including p47phox [13–15], which has been implicated in the generation of superoxide in rat mesangial cells under high glucose conditions [16]. Apocynin, a non-specific antioxidant that targets p47phox, reduces albuminuria and mesangial expansion in a mouse model of diabetes [17–20]. p47phox-independent NOX4 has also been implicated in diabetes-induced oxidative stress in the kidney [21–24]. Therefore, although NOX-induced ROS may play a role in the progression of diabetic nephropathy, the role of specific NOX isoforms and the cytosolic subunits that regulate their activation remains uncertain.

Accordingly, we examined the effect of deleting p47 phox (also known as Ncf1), the gene that encodes the NOX subunit, on diabetic kidney injury in the Akita mouse model of type 1 diabetes mellitus. Our aim was to test the hypothesis that p47phox-dependent activation of NOX is an important determinant of experimental diabetic nephropathy.

Methods

Animals

Four groups of male mice were studied: (1) non-diabetic mice (Ins2 WT/WT /p47 phox+/+); (2) non-diabetic p47 phox-null mice (Ins2 WT/WT /p47 phox−/−); (3) diabetic Akita mice (Ins2 WT/C96Y /p47 phox+/+); and (4) diabetic p47 phox-null mice (Ins2 WT/C96Y /p47 phox−/−). All mice were on a C57BL/6 background and had free access to water and standard 18% (wt/wt) protein rodent chow. Blood glucose levels and body weights were measured weekly from 4 to 16 weeks of age. Albumin excretion rates were determined from 24 h urine samples in 8- and 16-week-old mice using kits (Albuwell M ELISA; Exocell, Philadelphia, PA, USA). Systolic blood pressure was measured as previously described [25, 26]. At 8 and 16 weeks, mice were killed and kidneys removed and fixed in 10% (vol./vol.) formalin or snap-frozen. Pancreases were placed in 10 ml acid−ethanol (1.5% [vol./vol.] hydrochloric acid in 70% [vol./vol.] ethanol) for 18 h at −20°C. The tissue was homogenised, incubated overnight at −20°C and centrifuged for 15 min at 200 g and 4°C. The aqueous layer was transferred into a 15 ml tube. Acid−ethanol extract (100 μl) was neutralised with 100 μl of 1 mol/l TRIS at pH 7.5. Insulin content was measured with an RIA kit (Linco Research, St Charles, MO, USA). Total protein was measured by the Bradford assay (Bio-Rad Laboratories, Hercules, CA, USA). NOX was measured as previously described [25]. All procedures were conducted in accordance with the guidelines of the University of Toronto Animal Care Committee.

Primary mouse mesangial cell culture

Mesangial cells were isolated as previously described [25]. Cells were maintained in serum-free medium for 18 h then treated with 5.6 mmol/l or 30 mmol/l d-glucose for 16 h.

Glomeruli isolation

Mice were anaesthetised at 8 weeks of age by isoflurane and perfused through the left ventricle with 10 ml PBS containing 200 μl Dynabeads M-450 (Invitrogen Dynal, Oslo, Norway). Both kidneys were removed, minced with a razor blade and incubated at 37°C for 30 min with 10 ml HBSS containing 0.01 g collagenase A (Roche Diagnostics, Indianapolis, IN, USA) and 7.5 μl deoxyribonuclease I (Invitrogen, Carlsbad, CA, USA). After incubation, samples were passed through a 100 μm cell strainer (BD Falcon, Bedford, MA, USA), washed with 10 ml ice-cold HBSS and centrifuged for 5 min at 200 g and 4°C. Glomeruli were collected with a magnetic particle concentrator and stored at −80°C until use.

Renal histology

Frozen kidney tissue sections (10 μm thick) were incubated for 1 h with dihydroethidium (2 μmol/l; Invitrogen Canada, Burlington, ON, Canada) at 37°C and scanned with a confocal laser-scanning microscope (LSM510; Carl Zeiss Canada, Toronto, ON, Canada). The image colour intensity of dihydroethidium-stained kidney sections was scored blindly on a scale of 0 to 4 (0 for dark, 4 for the strongest intensity). Formalin-fixed paraffin-embedded kidneys were sectioned and stained with periodic acid–Schiff’s reagent (PAS), Picrosirius red and Masson’s trichrome reagents as previously described [25, 26]. Mouse glomeruli (approximately 60 to 100) were scored blindly by a nephropathologist for severity of diabetic glomerulosclerosis in PAS-stained sections. Each glomerulus was given a score of 0 (normal), 1 (mild, mesangial matrix [MM] increase approximately two times the width of a mesangial cell nucleus), 2 (moderate, MM increase approximately three to four times the width of a mesangial cell nucleus) or 3 (severe, MM increase more than four 4 times the width of a mesangial cell nucleus). The mean glomerular MM score was then calculated for each animal. Glomerular volume was calculated from scanned PAS slides using ImageScope software (Aperio, Vista, CA, USA). Formalin-fixed paraffin-embedded sections were used for immunohistochemical analysis. Anti-collagen I primary antibody was from Cedarlane (Cedarlane, Burlington, ON, Canada). All slides were scanned digitally at the Advanced Optical Microscope Facility (Princess Margaret Hospital, Toronto, ON, Canada) and ImageScope software was used to quantify collagen I immunostaining.

Wilms tumour 1 staining and podocyte quantification

De-paraffinised mouse kidney slides were incubated with Wilms tumour 1 (WT-1) antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and then with anti-rabbit IgG antibody (Vectastain ABC Kit; Vector Laboratories, Burlington, ON, Canada). All slides were scanned digitally at the Advanced Optical Microscope Facility (Princess Margaret Hospital) and ImageScope software was used to count WT-1-positive nuclei in the glomerular profiles.

Quantitative real-time PCR

RNA was extracted from isolated glomeruli and cultured primary mouse mesangial cells using a kit (RNeasy Mini; Qiagen Canada, Mississauga, ON, Canada). mRNA expression levels for Nox2 (also known as Cybb) Nox4, p47 phox, p22 phox (also known as Cyba), p67 phox (also known as Ncf2), p40 phox (also known as Ncf4), collagen type I α1, collagen type I α2, fibronectin, Pai1 (also known as Serpine1), Tgfb1 and nephrin were quantified by real-time PCR (TaqMan) using a sequence detection system (ABI Prism 7900; Applied Biosystems, Foster City, CA, USA) as previously described [25]. Specific mouse primer sets were purchased from Applied Biosystems (Foster City, CA, USA).

Western blot

Western blot analysis of protein lysates from isolated glomeruli or mesangial cells was performed as previously described [25] using primary antibodies for p47phox, NOX2 and β-actin (Santa Cruz). β-Actin was used as loading control and densitometry was measured using Scion Image software (Scion, Frederick, MD, USA).

Mouse islet isolation

Under isoflurane anaesthesia, the common bile duct was clamped at the point where it enters the duodenum. A collagenase A solution (2 ml, 2 mg/ml; Roche Diagnostics) was injected into the common bile duct. The pancreas was removed and placed in a 50 ml tube, then incubated at 37°C in a water bath for 17 min. Ice-cold HBSS/HEPES solution (20 ml) was added to stop collagenase digestion, prior to centrifugation for 1 min at 450 g. The pellet was washed with ice-cold HBSS/HEPES, filtered through gauze and centrifuged for 1 min at 450 g. The pellet was then re-suspended in ice-cold HBSS/HEPES. Islets were isolated under a dissecting microscope, transferred into a 35 mm culture dish with medium (RPMI-1640, 10% [wt/vol.] FBS, 1 mmol/l sodium pyruvate, 11 mmol/l glucose, 2 mmol/l l-glutamate and 50 U/ml penicillin/streptomycin), and incubated overnight at 37°C to allow recovery.

Intraperitoneal glucose tolerance test and insulin sensitivity test

Mice (p47 phox wild-type and p47 phox-null) were fasted from 07:00 hours for 6 h. An intraperitoneal glucose tolerance test (IPGTT) was performed in accordance with the American Diabetic Complications Consortium protocol [27]. d-Glucose (100 mg/ml; BDH Chemicals, Toronto, ON, Canada) was injected intraperitoneally (10 μl/g body weight). For the insulin sensitivity test, insulin (Humulin R; Eli Lilly, Indianapolis, IN, USA) was injected intraperitoneally (0.75 U/kg body weight). Blood glucose values were measured at 0, 5, 15, 30, 60 and 120 min using tail venous blood and a device (Contour Meter; Bayer, Toronto, ON, Canada).

Statistical analysis

Results are expressed as mean ± SEM unless otherwise specified. Comparisons between multiple groups were performed by one-way ANOVA followed by Bonferroni’s post hoc test. The two-tailed Student’s t test was used for comparison between two groups. The 24 h urinary albumin excretion rates are presented as median with interquartile range. Comparisons between multiple groups were performed by the non-parametric Kruskal–Wallis test followed by Dunn’s multiple comparison test. GraphPad Prism software was used for statistical tests (GraphPad, La Jolla, CA, USA).

Results

Kidney studies in 16-week-old mice

Four groups of mice were studied: non-diabetic p47 phox wild-type (Ins2 WT/WT /p47 phox+/+), non-diabetic p47 phox-null (Ins2 WT/WT /p47 phox−/−), diabetic p47 phox wild-type (Ins2 WT/C96Y /p47 phox+/+) and diabetic p47 phox-null (Ins2 WT/C96Y /p47 phox−/−) mice. The onset of hyperglycaemia occurred between 4 and 6 weeks of age in both diabetic groups (Fig. 1a). Mean values for blood glucose diverged after 10 weeks of age and at 16 weeks were approximately 10 mmol/l greater in Ins2 WT/C96Y /p47 phox+/+ than in Ins2 WT/C96Y /p47 phox−/− mice (electronic supplementary material [ESM] Table 1). All groups of mice gained weight over the 16 weeks, although values for Ins2 WT/WT /p47 phox−/− mice tended to be higher (Fig. 2a, ESM Table 1). Both groups of diabetic mice had an increased urinary albumin excretion rate. Deletion of p47 phox in the Ins2 WT/C96Y /p47 phox−/− diabetic mice did not reduce albuminuria (Fig. 1c, ESM Table 1).

Fig. 1
figure 1

(a) Blood glucose and (b) body weight in four groups of mice followed to 16 weeks of age. (c) 24 h urinary albumin excretion in 16-week-old mice. aSome values exceeded the measurable range of the glucometer (33.3 mmol/l) and were therefore recorded as 33.3 mmol/l on the graph. *p < 0.05 vs non-diabetic groups; † p < 0.05 vs the diabetic p47 phox-null group; ‡ p < 0.05 vs the diabetic p47 phox wild-type group. DM, diabetic; WT, p47 phox+/+; KO, p47 phox−/−; white circles, non-diabetic p47 phox wild-type mice; white triangles, non-diabetic p47 phox-null mice; black circles, diabetic p47 phox wild-type mice; black triangles, diabetic p47 phox-null mice

Full size image
Fig. 2
figure 2

Sections of kidneys from 16-week-old, non-diabetic p47 phox+/+ (a) and p47 phox−/− (b), and diabetic p47 phox+/+ (c) and p47 phox−/− (d) mice were stained with dihydroethidium to detect superoxide levels. Representative images from each group are shown (magnification: ×630). (e) The intensity of emission from dihydroethidium (DHE)-stained sections was scored for each of the four groups of mice. *p < 0.05 vs the non-diabetic groups and † p < 0.05 vs the diabetic p47 phox wild-type group. DM, diabetic; WT, p47 phox+/+; KO, p47 phox−/−

Full size image

At 16 weeks of age, both diabetic groups exhibited greater kidney:body weight ratios (KW:BW) than their non-diabetic littermates (ESM Table 1). Kidney hypertrophy and KW:BW were reduced by deletion of p47 phox in diabetic mice (Ins2 WT/C96Y /p47 phox−/−). Deletion of p47 phox also attenuated diabetic glomerular hypertrophy (ESM Table 1).

Oxidative stress in glomeruli was assessed in the four groups of mice using dihydroethidium staining to detect superoxide (Fig. 2a–d). Dihydroethidium staining was increased threefold in the glomeruli of Ins2 WT/C96Y /p47 phox+/+ compared with Ins2 WT/WT /p47 phox+/+ mice, this increase being attenuated by deletion of p47 phox (Fig. 2e). Increased oxidative stress was associated with a significant increase in the MM score in the glomeruli of Ins2 WT/C96Y /p47 phox−/− compared with Ins2 WT/C96Y /p47 phox+/+ mice (Fig. 3, ESM Table 1). In parallel with the MM score, glomerular collagen I immunostaining was increased in Ins2 WT/C96Y /p47 phox+/+ mice compared with Ins2 WT/WT /p47 phox+/+ mice and reduced by deletion of p47 phox (ESM Fig. 1e). There were no differences in the number of glomerular WT-1-positive cells in the four groups of mice (ESM Fig. 1j) and no significant differences in the glomerular basement membrane thickness across the four groups (data not shown). Renal cortical nephrin mRNA expression was similar in all four groups (ESM Fig. 2).

Fig. 3
figure 3

Sections of 16-week-old mouse kidneys were stained with PAS (a–d), Picrosirius red (e–h) and Masson’s trichrome (i–l) reagents. Representative images show glomeruli from each group, i.e. (a, e, i) non-diabetic p47 phox wild-type, (b, f, j) non-diabetic p47 phox-null, (c, g, k) diabetic p47 phox wild-type and (d, h, l) diabetic p47 phox-null mice (magnification: ×600). (m) The MM score was derived from four groups of 16-week-old mice. *p < 0.05 vs the non-diabetic groups and † p < 0.05 vs the diabetic p47 phox wild-type group. DM, diabetic; WT, p47 phox+/+; KO, p47 phox−/−

Full size image

The effect of deletion of p47 phox on plasma insulin concentration and pancreatic insulin content in 16-week-old diabetic mice

Blood glucose levels were significantly lower in diabetic mice with a deletion of p47 phox than in Ins2 WT/C96Y /p47 phox+/+ diabetic mice (Fig. 4a, ESM Table 1), a difference that emerged at 10 weeks of age (Fig. 1a). Plasma insulin levels and pancreatic insulin content were significantly greater in the diabetic mice with a deletion of p47 phox, suggesting that beta cell function was better preserved in these mice at 16 weeks of age (Fig. 4b, c).

Fig. 4
figure 4

(a) Blood glucose, (b) plasma insulin and (c) pancreatic insulin content in two groups of diabetic mice at 16 weeks of age. (d) Plasma insulin, (e) intraperitoneal glucose tolerance test and (f) insulin sensitivity test in 8-week-old non-diabetic p47 phox wild-type (white bars/circles) and p47 phox-null (white bars/triangles) mice. (g) Islet NOX activity. Islets were isolated from all four groups. AU, arbitrary units. *p < 0.05 vs the p47 phox wild-type group and † p < 0.05 vs all other groups. WT, p47 phox+/+; KO, p47 phox−/−

Full size image

The effect of deletion of p47 phox on beta cell function and insulin sensitivity in non-diabetic mice

To test the hypothesis that deletion of p47 phox might improve beta cell function, we studied 8-week-old non-diabetic Ins2 WT/WT /p47 phox+/+ and Ins2 WT/WT /p47 phox−/− mice. Mice with deletion of p47 phox exhibited significantly better glucose tolerance than their wild-type littermates (Fig. 4e), albeit with slightly lower insulin sensitivity than in wild-type mice (Fig. 4f). We then isolated pancreatic islets from the four groups of mice and measured NOX activity. The Ins2 WT/C96Y mutation increased NOX activity three- to fourfold in isolated islets of Ins2 WT/C96Y /p47 phox+/+ mice. Deletion of p47 phox markedly attenuated this response (Fig. 4g).

Kidney studies in 8-week-old mice

To determine the effect of deletion of p47 phox on the kidney response to hyperglycaemia, we studied another four groups of mice at 8 weeks of age when no differences in blood glucose levels were evident (ESM Fig. 3a, ESM Table 2). Urinary albumin excretion rates were increased to a similar extent in both diabetic groups; mean values for systolic blood pressure were also similar in the four groups (ESM Table 2). There were no differences in body weight between the groups, although both diabetic groups exhibited an increase in kidney weight and KW:BW (ESM Table 2). Mean values for glomerular volume tended to be greater in the diabetic groups and there were early (but not significant) increases in MM scores in the diabetic compared with the non-diabetic groups (ESM Table 2). NOX activity was increased twofold in isolated glomeruli from Ins2 WT/C96Y /p47 phox+/+ mice and attenuated by deletion of p47 phox (Fig. 5a). Dihydroethidium staining was also increased in Ins2 WT/C96Y /p47 phox+/+ compared with Ins2 WT/C96Y /p47 phox−/− mice (Fig. 5b–e).

Fig. 5
figure 5

NOX activity and superoxide concentration of isolated glomeruli from 8-week-old mice. (a) NOX activity of isolated glomeruli measured by a lucigenin chemiluminescence method. *p < 0.05 vs all other groups. (b) Sections of kidneys from 8-week-old, non-diabetic p47 phox+/+ and (c) p47 phox−/−, and (d) diabetic p47 phox+/+ and (e) p47 phox−/− mice were stained with dihydroethidium to detect superoxide levels. Representative images from each group are shown (magnification: ×630). DM, diabetic; WT, p47 phox+/+; KO, p47 phox−/−

Full size image

NOX subunit expression in isolated glomeruli

NOX is a multiunit complex, so we looked at mRNA expression of the NOX subunits in isolated glomeruli from the four groups of mice (ESM Table 3). Hyperglycaemia-induced activation of NOX was associated with significant increases in expression of p47 phox, Nox2, p22 phox, p40 phox and p67 phox in the glomeruli of Ins2 WT/C96Y /p47 phox+/+ compared with those of Ins2 WT/WT /p47 phox+/+ mice. Mean values for Nox4 mRNA expression tended to increase, but the difference did not reach statistical significance. Interestingly, deletion of p47 phox attenuated the increase in expression of Nox2, p22 phox, p40 phox and p67 phox that was induced by hyperglycaemia, but did not affect Nox4 mRNA expression. We then related changes in mRNA expression to protein levels of p47phox and NOX2 in isolated glomeruli from the four groups of mice. Western blot analysis showed that p47phox and NOX2 abundance paralleled the changes in mRNA expression (Fig. 6a–c).

Fig. 6
figure 6

Glomeruli were isolated from 8-week-old mice and protein levels of p47phox and NOX2 determined by western blot. (a) Representative western blot of p47phox, NOX2 and β-actin. (b) Quantitative densitometry analysis of western blot for p47phox and (c) NOX2. N. D., non-detectable. *p < 0.05 vs the non-diabetic p47 phox wild-type group; †p < 0.05 vs all other groups. ND, non-diabetic; DM, diabetic; WT, p47 phox+/+; KO, p47 phox−/−

Full size image

Pro-fibrotic gene expression in isolated glomeruli

To relate these early effects of deletion of p47 phox on NOX activation to pro-fibrotic gene expression, we measured mRNA expression of collagen Iα1, collagen Iα2, fibronectin, Tgfb1 and Pai1 in isolated glomeruli. Mean values for these pro-fibrotic genes were increased in the glomeruli of Ins2 WT/C96Y /p47 phox+/+ compared with those of Ins2 WT/WT /p47 phox+/+ and Ins2 WT/C96Y /p47 phox−/− mice (ESM Table 3).

Studies of NOX activation in primary mouse mesangial cells

Our in vivo data showed that diabetic nephropathy was attenuated by deletion of p47 phox. To relate this finding to a cellular response to high glucose, we studied primary mesangial cells derived from Ins2 WT/WT /p47 phox+/+ and Ins2 WT/WT /p47 phox−/− mice. Primary mesangial cells were exposed to 5.6 mmol/l or 30 mmol/l d-glucose. At 30 mmol/l, d-glucose increased NOX activity two- to threefold in wild-type mesangial cells from Ins2 WT/WT /p47 phox+/+ mice (Fig. 7a). This effect was not due to an osmotic stimulus (Fig. 7b) and the high glucose-induced increase in NOX was attenuated in mesangial cells from Ins2 WT/WT /p47 phox−/− mice.

Fig. 7
figure 7

(a) NOX activity of p47 phox wild-type (WT) and p47 phox-null (KO) mouse mesangial cells treated with d-glucose as indicated. (b) NOX activity of p47 phox wild-type mouse mesangial cells treated with d-glucose or d-glucose with d-mannitol as indicated. Results are expressed as arbitrary units (AU); *p < 0.05 vs non-diabetic groups; † p < 0.05 vs the diabetic p47 phox wild-type group; ‡ p < 0.05 vs all other groups

Full size image

NOX subunit expression in primary mesangial cells

High glucose-induced activation of NOX was associated with an increase in mRNA expression of the NOX subunits p47 phox, Nox2, Nox4, p22 phox, p67 phox and p40 phox in primary mesangial cells, consistent with our in vivo data. Deletion of p47 phox attenuated, but did not normalise the increases in Nox2, p22 phox, p67 phox and p40 phox, while there was no effect on Nox4 expression (ESM Table 4). Western blot analysis showed that p47phox and NOX2 protein levels paralleled the changes in mRNA expression (Fig. 8a–c).

Fig. 8
figure 8

p47 phox wild-type (WT) and p47 phox-null (KO) mouse mesangial cells were treated with d-glucose as indicated and protein levels of p47phox and NOX2 determined by western blot. (a) Representative western blot of p47phox, NOX2 and β-actin. (b) Quantitative densitometry analysis of western blot for p47phox and (c) NOX2. N. D., non-detectable. *p < 0.05 vs 5.6 mmol/l d-glucose-treated p47 phox wild-type group; † p < 0.05 vs all other groups

Full size image

Pro-fibrotic gene expression in mouse mesangial cells

To relate the effect of deletion of p47 phox on high glucose-induced pro-fibrotic gene expression, we measured mRNA expression of collagen Iα1, collagen Iα2, fibronectin, Tgfb1 and Pai1 in primary mesangial cells. Mean values for these pro-fibrotic genes were increased by 30 mmol/l glucose in mesangial cells from Ins2 WT/WT /p47 phox+/+ and Ins2 WT/WT /p47 phox−/− mice, but the magnitude of the increase was significantly reduced in cells from Ins2 WT/WT /p47 phox−/− mice (ESM Table 4).

Discussion

Oxidative stress is postulated to play a central role in the pathogenesis of diabetic nephropathy [1–6]. In this report, we focussed on the role of NOX, and specifically the cytosolic subunit p47phox in the generation of superoxide in a high glucose environment. The rationale for this approach is based on in vitro studies of rat mesangial cells exposed to high glucose [16, 20] and on in vivo studies of a mouse model of diabetes by Ohshiro and co-workers [15].

Our first major observation was that deletion of the p47 phox gene attenuated diabetic nephropathy in the Akita mouse. Hyperglycaemia was associated with increased urinary albumin excretion rates, kidney and glomerular hypertrophy, and MM expansion [25, 26, 28, 29]. These diabetes-induced changes were associated with increased renal oxidative stress. Deletion of p47 phox lessened oxidative stress, kidney and glomerular hypertrophy, and MM expansion. The protective effect on diabetic kidney injury of the deletion of p47 phox observed by us was at least partially dependent on improved glycaemic control.

Our second major observation was that deletion of p47 phox also lessened the severity of diabetes, even though the onset and early phase of hyperglycaemia were similar in both groups of diabetic mice. This effect of p47 phox deletion emerged and was significant after 10 weeks of age, with the difference persisting through to 16 weeks of age. The Akita mouse (Ins2 WT/C96Y) harbours a mutation of Ins2 (Cys96Tyr), which disrupts a disulfide bond between A and B chains of the insulin molecule [30]. This mutation leads to C/EBP homologous protein (CHOP)-dependent ER stress in the beta cell, with the subsequent apoptosis leading to insulin deficiency and hyperglycaemia [20–32]. p47phox and NOX-generated ROS may play a role in ER stress-induced beta cell apoptosis [32, 33]. Accumulation of ROS has also been shown to be an initiation factor and a consequence of ER stress; it is also an important cellular response, linking protein misfolding in the ER to beta cell apoptosis [34]. Our data showing attenuation of the severity of diabetes in the Akita mouse suggest that the deletion of p47 phox may have partially protected the beta cell from ER stress-induced injury, thus sustaining beta cell function over time.

To further explore the effect of deletion of p47 phox on beta cell function, we measured plasma insulin concentrations and pancreatic insulin content in 16-week-old diabetic mice. The difference in blood glucose levels was associated with significant increases in pancreatic insulin content and circulating insulin levels in Akita diabetic mice with a deletion of p47 phox. In non-diabetic mice, the deletion of p47 phox was also associated with improved glucose tolerance compared with wild-type littermates, a finding that was independent of an effect on insulin sensitivity, suggesting that beta cell function was enhanced in non-diabetic mice by deletion of p47 phox. Finally we examined NOX activity in isolated islets from the four groups of mice at 8 weeks of age when blood glucose levels were similar in the two diabetic groups. Hyperglycaemia was associated with increased NOX activity in isolated islets from Ins2 WT/C96Y /p47 phox+/+ mice, while this response was attenuated in Ins2 WT/C96Y /p47 phox−/− mice, supporting our hypothesis that deletion of p47 phox protects pancreatic beta cells from injury by attenuating oxidative stress.

The 10 mmol/l difference in blood glucose levels between the two diabetic mouse groups could have contributed, at least in part, to the protective effect of p47 phox deletion on glomerular injury in Ins2 WT/C96Y /p47 phox−/− mice, so our next series of studies was designed to more directly test the hypothesis that p47 phox deletion attenuated the glomerular response to hyperglycaemia. We studied mice at 8 weeks of age when the blood glucose levels were similar in the two groups of diabetic mice. Our third major observation was that the deletion of p47 phox attenuated NOX activity in isolated glomeruli of diabetic mice independently of blood glucose levels.

NOX is a protein complex consisting of two membrane subunits: NOX and p22phox. There are several NOX isoforms, including NOX1, NOX2 and NOX4. The activation of NOX1 or NOX2 is dependent on recruitment of four cytosolic proteins: p40phox, p47phox, p67phox and rac GTPase to the cell membrane [9]. The phosphorylation of p47phox is a critical event in this recruitment [9]. In contrast to NOX1 and NOX2, NOX4 is constitutively active and does not require the cytosolic subunits for activation [35–37]. All of these subunits are produced in the kidney [13, 14], and our findings suggest that p47phox-dependent NOXs are important sources of superoxide in the diabetic glomerulus. We also measured the mRNA expression of NOX subunits in isolated glomeruli from the four groups of mice. It has been reported that diabetes is associated with increased abundance of NOX2 [17, 21, 38, 39] and NOX4 [21–24, 40] in kidney cortex. We found that increased NOX activity was associated with significant increases in the mRNA expression of the cytosolic subunits p47 phox, p40 phox and p67 phox, and of the membrane subunits p22 phox and NOX2 in the glomeruli of diabetic mice. NOX2 and p47phox protein levels in isolated glomeruli paralleled the changes in mRNA levels. These effects were attenuated by deletion of p47 phox, suggesting that NOX-mediated oxidative stress may exert a positive feedback loop on subunit levels in the diabetic glomerulus. Although we did not address the mechanism responsible for this effect, studies by Bondi et al [41] provide a possible explanation that ROS-induced TGF-β1 may function in an autocrine manor to increase NOX subunit levels.

Our fourth major finding was that the early increases in NOX subunit abundance and activity were associated with increased levels of the extracellular matrix proteins collagen I and fibronectin, and of the pro-fibrotic factors TGF-β1 and plasminogen activator inhibitor 1 (PAI-1) [42–46] in isolated glomeruli. These early changes preceded significant differences in the MM scores. These findings are consistent with in vivo studies of the effect of apocynin on the severity of diabetic kidney injury, although apocynin functions as a general antioxidant rather than a specific antagonist of p47phox in non-phagocytic cells [17–20]. Previous studies of protein kinase C-beta and diabetic nephropathy have also suggested a role for p47phox in the development of kidney injury [15].

One limitation of our study is that we did not measure blood pressure at 16 weeks of age. Although blood pressure values were similar at 8 weeks of age, it remains possible that potential later differences could have contributed to the attenuation of diabetic nephropathy. Fujita and co-workers showed that an antioxidant, apocynin, reduced albuminuria and MM expansion in the Akita mouse at 14 weeks of age, but did not lower blood pressure. They also found that a blood pressure reduction of 13 mmHg with a calcium channel blocker did not attenuate albuminuria or MM expansion in the Akita mouse, unless the agents blocked the renin–angiotensin system [47].

Glomerular injury in the Akita mouse has been related to effects on podocytes and mesangial cells [18, 25, 26, 28, 48]. Although no difference in the number of podocytes was detected in our four groups of mice, we sought to link our in vivo observations to a direct cellular effect of p47 phox deletion in mesangial cells. High glucose-induced activation of NOX was attenuated, but not normalised in primary mesangial cells derived from p47 phox-null mice. We observed similar changes in abundance of the NOX subunits in the in vitro and in vivo studies. Changes in extracellular matrix protein abundance and pro-fibrotic factors like TGF-β1 and PAI-1 were attenuated, but not normalised in mesangial cells from p47 phox-null mice. These findings confirm previous work in rat mesangial cells [16, 20], and together support the hypothesis that deletion of p47 phox attenuates the mesangial cell response to high glucose.

Although our studies focused on p47phox-dependent NOX activation, other pathways may also contribute to oxidative stress in diabetic glomeruli in vivo. More recently, Chacko and co-workers reported that treatment with a mitochondria-targeted ubiquinone reduced glomerular injury in diabetic Akita mice, suggesting a role for mitochondrial-derived superoxide [49]. In addition, Gorin et al delivered NOX4 antisense in vivo using osmotic mini-pumps, and showed that less oxidative stress and reduced glomerular injury occurred in a streptozotocin-induced rat model of diabetes [21]. Experimental studies of diabetic mice with deletion of Nox4 have not been reported; however, recent studies of the heart suggest that NOX4 may play a protective role [50].

In summary, our studies support the hypothesis that p47phox-dependent activation of NOX is an important determinant of glomerular injury and beta cell dysfunction in the Akita mouse model of type 1 diabetes mellitus. The renoprotective effect of p47 phox deletion on the kidney is dependent in part on improved glycaemic control.