DELETION OF PROTEIN TYROSINE PHOSPHATASE 1B IMPROVES PERIPHERAL INSULIN RESISTANCE AND VASCULAR FUNCTION IN OBESE, LEPTIN RESISTANT MICE VIA REDUCED OXIDANT TONE

Abstract

Rationale

Obesity is a risk factor for cardiovascular dysfunction, yet the underlying factors driving this impaired function remain poorly understood. Insulin resistance is a common pathology in obese patients and has been shown to impair vascular function. Whether insulin resistance or obesity, itself, is causal remains unclear.

Objective

The current study tested the hypothesis that insulin resistance is the underlying mediator for impaired nitric oxide mediated dilation in obesity by genetic deletion of the insulin-desensitizing enzyme protein tyrosine phosphatase 1B (PTP1B) in db/db mice.

Methods and Results

The db/db mouse is morbidly obese, insulin resistant and has tissue-specific elevation in PTP1B expression compared to lean controls. In db/db mice, PTP1B deletion improved glucose clearance, dyslipidemia, and insulin receptor signaling in muscle and fat. Hepatic insulin signaling in db/db mice was not improved by deletion of PTP1B, indicating specific amelioration of peripheral insulin resistance. Additionally, obese mice demonstrate an impaired endothelium dependent and independent vasodilation to acetylcholine and sodium nitroprusside, respectively. This impairment, which correlated with increased superoxide in the db/db mice, was corrected by superoxide scavenging. Increased superoxide production was associated with increased expression of NAD(P)H Oxidase 1 and its molecular regulators, Noxo1 and Noxa1.

Conclusion

Deletion of PTP1B improved both endothelium dependent and independent nitric oxide mediated dilation and reduced superoxide generation in db/db mice. PTP1B deletion did not affect any vascular function in lean mice. Taken together, these data reveal a role for peripheral insulin resistance as the mediator of vascular dysfunction in obesity.

INTRODUCTION

The prevalence of obesity and its cardiovascular complications represents a significant health concern in Western societies ^{1, 2}, but the root causes of cardiovascular dysfunction in obese individuals remain unclear. Metabolic dysfunction, notably insulin resistance, is evident in obesity ^{3, 4}. It has been speculated that insulin resistance, rather than other aspects of obesity, is the underlying cause of cardiovascular injury in obese patients ^5–9. This hypothesis has been difficult to test since insulin-sensitizing drugs have off-target effects ^{4, 10} and non-obese models of insulin resistance do not evaluate the relative importance of obesity versus insulin resistance ^11–17.

The insulin receptor is a classic receptor tyrosine kinase ¹⁸ and as such, is de-activated by protein tyrosine phosphatases, notably protein tyrosine phosphatase 1B (PTP1B) ^19–21. Deletion of PTP1B improves insulin sensitivity in mouse models of obesity ²² and putative PTP1B antagonists have been used pharmacologically to improve glucose tolerance ^23–25. Increases in the activity and/or expression of PTP1B correlate with blunted insulin signaling in a variety of tissue types ^26–28. Whether PTP1B deletion and amelioration of insulin resistance improves cardiovascular dysfunction associated with obesity remains unknown.

The current study tested the hypothesis that PTP1B deletion attenuates vascular dysfunction in a model of obesity-induced insulin resistance. Four experimental genotypes were generated through breeding of db/db^+/− and PTP1B^−/− mice to produce double knockout PTP1B null, obese mice. Metabolic profiling, insulin receptor phosphorylation, and PTP1B gene expression were used to assess insulin sensitivity in target tissues. Endothelium-dependent and -independent vascular function were determined in vitro. Molecular techniques examined the mechanism whereby deletion of PTP1B improved vascular function. Taken together, these studies critically test the hypothesis that insulin resistance in obesity is the underlying risk factor driving vascular dysfunction in obese individuals.

MATERIALS AND METHODS

Animal Models

Two parental strains of mice were used in these studies: leptin receptor mutant db/db mice bred on a C57BL/6 background (Jackson Laboratories) and PTP1B null mice bred on a BALB/c background (Michel Tremblay, Ph.D. at the Cancer Institute of McGill University (Montreal, Canada)). Since db/db mice are sterile, progeny were generated from dual heterozygotes (H_db, heterozygous for mutant leptin receptor; H_PTP, PTP1B gene deletion). Dual heterozygotes were interbred, producing obese, PTP1B gene null, and dual KO mice at 1:4, 1:4, and 1:16 ratios, respectively. In the F₄ generation, dual heterozygotes were bred to heterozygotes for the leptin receptor mutation and PTP1B gene null mice. This breeding strategy yielded obese and dual KO mice at 1:4 and 1:8 ratios, respectively. Dual heterozygous littermates were used as lean controls and littermates heterozygous for db and PTP1B gene deletion were used as lean PTP1B null controls. All experiments were conducted in male progeny. In all cases, mice are designated as H or K, indicating heterozygote or knockout. The db gene is designated first and the PTP1B second. Thus, H_dbH_PTP are heterozygous for both genes, H_dbK_PTP are lean PTP1B KO mice, K_dbH_PTP are obese mice with intact PTP1B, and K_dbK_PTP are deficient in both leptin receptors and PTP1B. Mice were genotyped by PCR of genomic DNA. Metabolic phenotyping was accomplished by assessment of glucose tolerance and plasma chemistry (Online Supplement).

Insulin signaling

To determine the effects of PTP1B deletion on insulin receptor phosphorylation, mice were subjected to an insulin stimulation protocol in vivo ²⁹. Briefly, mice were anesthetized with isoflurane, and either saline or insulin (1 mU/g) was injected into a jugular vein catheter. After 12 minutes, mice were euthanized by isoflurane overdose and samples of liver, skeletal muscle, and adipose tissue were obtained and snap-frozen in liquid nitrogen. The time period between overdose and tissue harvesting was less than 2 minutes total and samples were obtained in differing order to avoid collection bias.

Western Blotting

Tissue homogenates (20–50 μg) were separated via SDS-PAGE and transferred to Immobilon–P PVDF membranes. In order to determine the expression of relevant proteins, immunoblots were probed with antibodies for PTP1B (Upstate), actin (Calbiochem), insulin receptor-β (IRβ, Santa Cruz), endothelial nitric oxide synthase (eNOS), and phosphorylated eNOS 1177/79 (BD Transduction Laboratories).

Insulin receptor phosphorylation

An anti-IRβ antibody was used to immunoprecipitate insulin receptor proteins from 400–1000 μg of tissue lysates and phosphorylation status assess with phosphor-tyrosine antibody (PY4G10, Upstate). (Online Supplement)

In Vitro Microvessel Preparation

Small mesenteric arteries (SMA) are defined in this study as the 2^nd and 3^rd distal arcuate artery branches secondary to the conduit superior mesenteric artery ranging from 50 to 150 μm in internal diameter. SMA were dissected and segments (0.25 − 1 mm in length) were mounted in a vessel bath between two glass micropipettes (25 μm-diameter tip) and secured with 10–0 silk ophthalmic suture. SMA were then placed in a chilled, oxygenated (21% O₂, 5% CO₂, and 74% N₂) Krebs-Ringer bicarbonate solution composed of (in mM) 118.3 NaCl, 4.7 KCl, 2.5 CaCl₂, 1.2 MgSO4₄, 1.2 KH₂PO₄, 25 NaHCO₃, and 11.1 D-glucose prior to analysis to hibernate physiological activity. The lumen of the vessel was filled with Krebs buffer through the micropipette and maintained at a constant pressure of 60 mmHg. Vessels were monitored under a Nikon inverted light microscope (Melville, NY) connected to a video monitor. Internal diameter was continually measured using video calipers and expressed in micrometers. Buffer temperature was increased to 37°C, and microvessels were allowed to develop spontaneous myogenic tone. After tone was developed, vasodilator responses were measured with sequential doses of: acetylcholine (1 × 10⁻¹⁰ mol/L to 1 × 10⁻⁵ mol/L), sodium nitroprusside (1 × 10⁻⁹ mol/L to 1 × 10⁻⁴ mol/L), or papaverine (1 × 10⁻⁹ mol/L to 1 × 10⁻⁴ mol/L). Superoxide dismutase (SOD) was used to scavenge superoxide (100 U/mL). ω-Nitro-L-arginine methyl ester (L-NAME) was used to inhibit nitric oxide synthase (100 μM). Dose responses are expressed as a percentage of dilation compared to initial diameter and maximum passive diameter. One dose response curve was performed per vessel per mouse.

Assesment of Superoxide

The quantitative abundance of super-oxide was assessed using Electron Paramagnetic Resonance (EPR) Spectroscopy. Qualitative assessment of super-oxide localization was made using dihydroethidium staining (Online Supplement).

Real Time RT-PCR

Mesenteric arterial cascades were harvested from euthanized animals, removed of non-vascular tissue, and snap-frozen in liquid nitrogen. Total RNA was extracted using Trizol Plus RNA (Invitrogen) and cDNA synthesized using the iScript cDNA Synthesis Kit (Biorad). cDNA was then used to assess relative gene expression using real time RT-PCR (Bio Rad iQ SYBR Green). Primer sequences for the selected genes are described in Table ST1.

Statistics

All data are expressed as mean ± SEM. Differences among all four genotypes were compared by One Way ANOVA or by Student's t-test with Bonferroni correction test used as the post-hoc test. A P-value of less than 0.05 was considered statistically significant.

RESULTS

Weight gain

Increases in body mass are depicted in Supplemental Figure 2 (S2). Data are presented as a scatter plot of 5 male mice of each genotype. Summary data at age 12 weeks are shown in Table 1. Consistent with the inactivating mutation in the leptin receptor gene, K_dbH_PTP mice displayed morbid obesity compared to H_dbH_PTP. Deletion of PTP1B did not affect weight gain in either H_dbK_PTP or K_dbK_PTP mice compared to PTP1B intact controls.

Table 1.

Metabolic phenotyping of fasted male dual KO mice yielded marked improvements in several indices of insulin sensitivity. n > 10 for each measurement.

Parameter	HdbHPTP	KdbHPTP	HdbKPTP	KdbKPTP
Body Weight (gms)	29±1	57±2*	26±1	53±3*
Food Intake(gms/day)	2.5±0.4	5.3±0.7*	2.4±0.5	4.6±0.8*
Water Intake(ml/day)	4±1	9±2*	4±1	8±1*
Urine Output (ml/day)	1±.02	3.2±.07*	0.8±.01	3.0±.07*
Plasma Glucose (mg/dl)	108±9	170±27*	120±7	172±13*
HbA1c (%)	4.8±.01	10.2±0.2*	4.9±0.07	7.6±0.2*,**
% of glucose bolus cleared in 30 minutes (%)	68±6	24±10*	87±6	87±10**
Plasma Insulin (ng/ml)	0.48±0.07	5.8±0.8*	0.29±0.08*	5.1±0.6*
Plasma Cholesterol (mg/dl)	73±5	143±13*	78±7	124±20*
Plasma Triglyerides (mg/dl)	47±5	145±31*	35±12	51±11**
Non-Esterified Fatty Acids(mEq/ml)	0.49±0.06	0.81±0.1*	0.51±0.07	0.34±0.05**
Plasma Leptin (pg/ml)	147±38	1075±300*	80±21*	1105±215*

^*= p < 0.05 all other genotypes vs. lean controls (HdbHPTP)

^**= p < 0.05 vs. KdbHPTP

H_dbK_PTP mice had modest reductions in plasma leptin versus H_dbH_PTP mice, consistent with sensitization of the leptin receptor, which is also a substrate of PTP1B. In db/db mice, plasma leptin levels were markedly increased along with body weight and were unaffected by deletion of PTP1B (Table 1).

Food and water intake and urine output are also summarized in Table 1. H_dbH_PTP and H_dbK_PTP mice exhibited normal and similar food and water intake and urine output. K_dbH_PTP mice displayed hyperphagia, polydypsia, and polyuria, consistent with obesity. Deletion of PTP1B on the obese background did not affect food and water intake.

In sum, these data indicate that the fundamental defects in leptin signaling that drive obesity in db/db mice are not moderated by the deletion of PTP1B. Thus, the metabolic improvements arising from PTP1B deletion must be attributed to changes in insulin receptor signaling and not modification of obesity.

Glucose Metabolism

Baseline levels of plasma glucose and insulin are shown in Table 1. Fasting glucose in H_dbH_PTP mice was euglycemic and deletion of PTP1B in H_dbK_PTP mice did not alter fasting glucose. The obesity observed in K_dbH_PTP mice was associated with moderate hyperglycemia. Deletion of PTP1B in K_dbK_PTP mice did not reduce fasting blood glucose, suggesting persistent hepatic insulin resistance in the K_dbK_PTP mice. Consistent with these observations, H_dbH_PTP mice were euinsulinemic but K_dbH_PTP and K_dbK_PTP mice had persistent hyperinsulinemia. Insulin levels in H_dbK_PTP mice were similar to H_dbH_PTP mice.

In vivo clearance of a glucose bolus is shown in Supplemental Figure 3 (S3). H_dbH_PTP mice displayed rapid glucose disposal, and clearance of glucose in H_dbK_PTP mice was similar. K_dbH_PTP mice showed markedly blunted glucose clearance, but K_dbK_PTP mice showed normalization of glucose clearance despite obesity.

HbA_1c levels, an index of total glycemic load, are shown in Table 1. H_dbH_PTP and H_dbK_PTP mice showed HbA_1C levels lower than 5%, consistent with euglycemic control. In contrast, K_dbH_PTP mice showed markedly elevated HbA_1c levels, consistent with their observed glucose intolerance and fasting hyperglycemia. Although not completely normalized, HbA_1c levels were significantly reduced in K_dbK_PTP mice compared to K_dbH_PTP mice, despite equivalent food intake.

Lipid Metabolism

Plasma concentration of FFAs, triglycerides, and cholesterol are shown in Table 1. H_dbH_PTP and H_dbK_PTP mice show normal levels of all three lipid compounds. K_dbH_PTP mice had elevated fasting FFAs and increased triglyceride levels, consistent with the loss of insulin sensitivity in fat cells. K_dbK_PTP mice displayed largely normal levels of FFAs and triglycerides, suggesting a normalization of adipocyte insulin resistance by deletion of PTP1B. In contrast, total cholesterol was elevated to a similar extent in both K_dbH_PTP and K_dbK_PTP mice.

Expression of PTP1B

To determine the effects of obesity on the tissue expression of PTP1B, Western blotting was performed on extracts from liver, muscle, and fat, which are the three major targets of the metabolic actions of insulin. The results are shown in Figure 1A–C. PTP1B expression was heterogeneous with marked increases in expression in the skeletal muscle (Figure 1B) and adipose tissue (Figure 1C) of obese mice. Expression of PTP1B in the liver (Figure 1A) was not statistically different between lean H_dbH_PTP and obese K_dbH_PTP mice.

Figure 1.

The PTP1B gene was significantly upregulated in obese, insulin resistant mice compared to control in muscle (1B) and fat (1C) but not hepatic tissues (1A). PTP1B gene deletion markedly improves insulin receptor tyrosine phosphorylation in muscle (1E) and fat (1F) of dual KO animals compared to obese, insulin resistant animals. Dual KO hepatic tissue insulin resistance persisted (1D). Control animals and PTP1B KO lean animals displayed similar in insulin receptor signaling in all tissues.

[A-F: * = p < 0.05 vs. H_dbH_PTP, n > 5]

Insulin signaling

Phosphorylation of the insulin receptor was used as a molecular readout of insulin signaling capacity. In skeletal muscle, adipose, and liver tissue samples, insulin provoked a marked increase in receptor tyrosine phosphorylation that was similar in H_dbH_PTP and H_dbK_PTP mice. In contrast, in K_dbH_PTP mice, the tyrosine phosphorylation of the insulin receptor was markedly reduced, consistent with obesity-induced insulin resistance. In skeletal muscle (Figure 1E) and adipose tissue (Figure 1F), insulin receptor phosphorylation was markedly increased in K_dbK_PTP mice, suggesting that deletion of PTP1B improved insulin signaling. In contrast, insulin receptor phosphorylation remained depressed in the liver (Figure 1D) of K_dbK_PTP mice, suggesting persistent hepatic insulin insensitivity in these animals.

Vascular Reactivity

Endothelium-dependent, acetylcholine-mediated vasodilation in the SMA from all mice is shown in Figure 2A. Smooth muscle reactivity to nitric oxide (NO) was determined using sodium nitroprusside (SNP, Figure 2B). Lean mice that are PTP1B deficient (H_dbK_PTP) do not have differences in maximum dilation to acetylcholine (70% vs. 68 %, p<NS) or sodium nitroprusside (96% vs. 94%, p<NS) from H_dbH_PTP mice, indicating that PTP1B deletion does not affect endothelium-dependent or - independent vasodilation. The maximum vasodilator response to acetylcholine was reduced markedly in K_dbH_PTP compared to H_dbH_PTP (50% vs. 70 %, p<0.05), indicating impairment of endothelial function in K_dbH_PTP mice. A significant deficiency in reactivity to exogenous NO was also detected in K_dbH_PTP mice (68% vs. 94 %, p<0.05). The maximum vasodilator response to acetylcholine was markedly reduced in all mice following treatment with 100 μM L-NAME (Figure 2C), indicating NO as the primary dilator mediating the response to acetylcholine. Furthermore, the NO-independent component of acetylcholine-induced vasodilation was not different among all groups of mice. Taken together, these data indicate that vasodilator reactivity is compromised at the level of NO utilization in obese mice. In contrast to the findings observed in K_dbH_PTP mice, endothelium-dependent vasodilation in obese K_dbK_PTP was similar to that observed in lean H_dbH_PTP mice (Figure 2A). The impaired response to exogenous NO (SNP) was also restored by PTP1B deletion (Figure 2B). Vascular dysfunction was not due to loss of eNOS expression or phosphorylation as these variables were similar in all strains of mice (Figure S4). Responses to the NO-independent vasodilator papaverine were similar across all mice (Figure S5), suggesting that the vasodilation deficiency in K_dbH_PTP mice is not due to a general deficit in vascular dilation, but is confined to NO-mediated dilation.

Figure 2.

Obese, insulin resistant mice (K_dbH_PTP) demonstrated a significant decrease in endothelium dependent, acetylcholine-mediated dilation (2A) and endothelium independent, SNP-mediated dilation (2B) that is ameliorated by PTP1B gene deletion. Endothelium dependent dilation was nearly abolished by L-NAME (2C). Superoxide scavenging produced a complete restoration of acetylcholine-mediated dilation in obese, insulin resistant mice (2D). Superoxide dismutase incubation improves endothelium independent nitric oxide-mediated dilation in K_dbH_PTP (2E).

[A, B: * = p < 0.05 K_dbH_PTP vs. H_dbH_PTP, n > 5; C: * = p < 0.05 (+) L-NAME vs. (−) L-NAME for each mouse, n > 5; D, E: * = p < 0.05 (+) PEG-SOD vs. (−) PEG- SOD, n > 5]

To determine whether elevated superoxide production was a mechanism of impaired vasodilation in obesity, vascular function was assessed in the presence of 100 U/mL PEG-SOD. PEG-SOD reversed the impaired dilation to acetylcholine (Figure 2D) and SNP (Figure 2E) in K_dbH_PTP mice with no effect on vascular function in the other genotypes. To further determine if PEG-SOD was indeed restoring NO bioactivity, endothelium-dependent dilation was assessed in the presence of PEG-SOD and both the presence and absence of 100 μM L-NAME (Figure S6). All mouse vessels exhibited equivalent degrees of L-NAME resistant dilation thus confirming that the main dilation mechanism in these microvessels is NO and further that scavenging of superoxide did not improve NO-independent dilation in K_dbH_PTP mice.

Passive mechanics were assessed in a zero-Ca²⁺ Krebs solution and results are shown in ST2. Vascular architectural changes were assessed as previously described. Maximal vessel wall thickness and wall to lumen ratio (at 120 mmHg of intraluminal pressure) were similar across all genotypes. Vascular compliance, as calculated by the exponential fit of a circumferential stress-strain plot (β-coefficient), also remained similar across all genotypes. Taken together, these data indicate that neither obesity nor deletion of PTP1B produce structural changes that could account for observed deficits in vasodilator function.

Superoxide production

EPR spectroscopy was employed to semi-quantitatively measure superoxide. The relative PEG-SOD inhibitable signal was 4 times higher in mesenteric vessels from K_dbH_PTP mice versus control H_dbH_PTP mice (Figure 3A) and reversed to control levels by deletion of PTP1B in obese K_dbK_PTP. The increased superoxide signal in the K_dbH_PTP mice was nearly eliminated following acetovanillone (apocynin) incubation (Figure 3B), suggesting it derives from NAD(P)H oxidases. Taken together, peripheral insulin resistance increases vascular superoxide production that is corrected by PTP1B deletion. As a further measure of vascular ROS production, we also performed DHE staining of blood vessels. These results are in agreement with the EPR studies and demonstrate higher levels of superoxide in the blood vessels of K_dbH_PTP mice compared to controls (Figure 3C). The deletion of PTP1B in obese animals (K_dbK_PTP) decreased DHE staining to control levels, and there was no difference in superoxide production due to the deletion of PTP1B in lean animals. DHE staining was most intense in the medial layer of mesenteric microvessels in all 4 groups of mice.

Figure 3.

EPR spectroscopy revealed a 5-fold increase in superoxide signal from obese insulin resistant mice, while PTP1B gene deletion ameliorates this increase (3A). Acetovanillone (apocynin) incubation of K_dbH_PTP samples decreased the superoxide signal to levels seen in controls (3B). Dihydroethidium staining revealed broad localization of superoxide signal throughout the arteries of K_dbH_PTP mice that is absent in all other mice.

[A, B: * = p < 0.05 K_dbH_PTP vs. H_dbH_PTP , n > 6; † = p < 0.05 K_dbK_PTP vs. K_dbH_PTP , n > 6]

The source of elevated superoxide levels in K_dbH_PTP mice was addressed using real time quantitative RT-PCR. As shown in Figure 4, the expression levels for Nox1 and its novel activator and organizer (Noxa1 and Noxo1) are significantly elevated in K_dbH_PTP mice as compared to control (2^−ΔΔCt ± SEM in H_dbH_PTP vs. K_dbH_PTP; Nox1: 1.45 ± .23 vs. 3.52 ± 1.6; Noxa1: 1.17 ± .11 vs. 9.94 ± 1.5; Noxo1: 1.14 ± .23 vs. 8.81 ± 2.8). Nox2 and Nox4 are expressed similarly in all animals (Figure 4), as are p22phox, p47phox and p67phox (data not shown). Select anti-oxidant enzymes were also examined, demonstrating a statistically significant increase in SOD2 and SOD3 in K_dbK_PTP mice compared to both controls and K_dbH_PTP mice (2^−ΔΔCt ± SEM in K_dbH_PTP vs. K_dbK_PTP; SOD2: 1.00 ± .5 vs. 3.36 ± .52; SOD3: .68 ± .10 vs. 4.76 ± 1.6). A summary for all genes studied is shown in Supplemental Table 3.

Figure 4.

Gene expression of superoxide-generating and anti-oxidant defense enzymes were assessed using quantitative Real-Time RT-PCR. K_dbH_PTP mice showed major increases in Nox1, Noxa1 and Noxo1 gene expression.

[* = p < 0.05 all genotypes vs. H_dbH_PTP , n > 6; † = p < 0.05 K_dbK_PTP vs. K_dbH_PTP , n > 6]

DISCUSSION

The goal of the current study was to determine whether the metabolic consequences of obesity or the state of obesity itself results in vascular dysfunction. To test this hypothesis we generated obese mice harboring deletion of PTP1B, a tyrosine phosphatase that antagonizes insulin signaling. The key findings of this study are: 1) deletion of PTP1B in db/db mice does not affect obesity and corrects peripheral but not hepatic insulin resistance; 2) correction of peripheral insulin resistance improves NO-mediated dilation in SMA despite persistent obesity; 3) correcting peripheral insulin resistance mediates the improvement in NO dilation by decreasing superoxide levels, primarily through reduced expression of Nox1, Noxa1 and Noxo1.

Metabolic effects of PTP1B deletion in obese mice

Although PTP1B has attracted considerable attention as a target in the treatment of non-insulin dependent diabetes, the impact of obesity on the relative distribution of PTP1B expression in tissues central to insulin action is unclear. Moreover, expression of PTP1B in models of obesity and diabetes is complicated as it varies with the stage of diabetes and genetic background ³⁰. As shown in Figure 1A–C, obesity in the db/db mice used in these studies caused a differential increase in PTP1B expression, with the most prominent increases in muscle and fat and a statistically undetectable difference in the liver. The increases in PTP1B expression correlate with decreased insulin receptor phosphorylation and are reversed by PTP1B deletion.

Deletion of PTP1B did not affect weight gain in either lean or obese mice. Food and fluid intake, urine output, and plasma leptin levels also remained unchanged. These observations are consistent with those of Cheng et al. ³¹ in which ob/ob mice heterozygous for PTP1B and ob/ob mice with deletion PTP1B showed similarity in weight gain. In contrast, adenoviral delivery of PTP1B anti-sense RNA produces reductions in body weight and fat mass ³² and when ob/ob mice that lack PTP1B are compared to ob/ob mice with wild-type PTP1B expression, an approximately 15% weight difference is observed ³¹. It is important to note that in this study, deletion of PTP1B improves glycemic control in ob/ob mice when wild type (2 copies) and knockout (no copies) are compared. However, one could not determine in this setting whether the improvement in glycemic control reflects the actions of PTP1B on insulin signaling or the weight loss in ob/ob mice. In the current study, we evaluated metabolic control between mice in which body weight is identical and the leptin receptor is completely missing. Thus, differences between K_dbH_PTP and K_dbK_PTP, which are equally obese, reflect the effects of PTP1B deletion on improvements in the insulin signaling pathway.

To verify the functional importance of the increase in PTP1B expression in K_dbH_PTP mice, we used physiologic (plasma serum chemistry and glucose tolerance) and molecular (phosphorylation of the insulin receptor) measurements as indices of insulin signaling. Consistent with recent observations from Delibgovic et al ³³, the elevated expression of PTP1B in skeletal muscle of obese mice correlated with marked impairment in muscle insulin receptor phosphorylation and significant impairment in glucose tolerance and increased HbA_1c percentage. These variables were markedly improved by the deletion of PTP1B. These findings, combined with the marked improvement in insulin receptor tyrosine phosphorylation in muscle, indicate that PTP1B is a key determinant of skeletal muscle insulin sensitivity in db/db mice.

Reductions in serum triglycerides and FFAs following deletion of PTP1B, combined with the marked increase in PTP1B expression in visceral adipose tissue and improvement in insulin receptor signaling in visceral adipose tissue, indicate that PTP1B also plays a critical role in the adipose tissue of obese mice. In this study, the lack of a leptin receptor results in similar body weight between K_dbH_PTP and K_dbK_PTP mice but marked differences in triglyceride and FFA levels in the plasma, indicating an improvement of insulin signaling in fat tissue. Plasma cholesterol was elevated in K_dbH_PTP mice compared to H_dbH_PTP mice (Table 1), but was not affected by PTP1B deletion in H_dbK_PTP or K_dbK_PTP mice. Since both K_dbH_PTP and K_dbK_PTP mice retain the hyperphagic phenotype, the lack of a difference in plasma cholesterol likely indicates that elevated cholesterol in these animals reflects dietary intake or persistent hepatic insulin resistance.

Fasting blood glucose, primarily driven by hepatic gluconeogenesis, was similar in K_dbH_PTP and K_dbK_PTP mice. Plasma insulin levels are also elevated, consistent with the loss of insulin receptor function in the liver ^{34, 35}. PTP1B expression was not significantly increased in the liver and insulin receptor phosphorylation remains depressed in K_dbK_PTPmice despite deletion of PTP1B. Although previous studies have described a role for PTP1B in hepatic insulin signaling in lean mice ^{36 37} or with non-genomic methods ³⁸, our model does not reflect this outcome, likely due to the background of these mice³⁰ . Nevertheless, the lack of improvement in hepatic insulin signaling and moderate fasting hyperglycemia indicate that hepatic insulin resistance cannot explain observed impairments in cardiovascular function.

Effect of deletion of PTP1B on vasodilation

Previous studies in obese rodents have indicated that obesity is a risk factor for vascular dysfunction ^{11, 39, 40}, but the culpable component of obesity has remained elusive. As described above, the dual KO mice developed for these studies remain obese, but peripheral insulin resistance is improved when PTP1B is deleted. Moreover, the results described in these studies indicate that when PTP1B is deleted in obese mice, vasodilation to NO is improved.

The role of PTP1B in improving NO-mediated dilation could be attributed to: 1) a direct effect of PTP1B on endothelial function or 2) an improvement in vasodilation secondary to correction of peripheral insulin resistance. A direct effect of PTP1B deletion is refuted by the lack of vascular outcomes in lean H_dbK_PTP mice, consistent with previous work in which over-expression of PTP1B in cultured endothelial cells did not influence the function of eNOS ⁴¹. These observations preclude PTP1B as a direct modulator of vasodilation.

Our observations are more consistent with the hypothesis that insulin resistance is the causal factor in vessel dysfunction in obesity. To date, this hypothesis has been primarily based on studies in non-obese models of insulin resistance ^{11, 13, 42, 43} and studies with pharmacological compounds ^{4, 10, 42}. Clear interpretation of these studies is confounded by off-target effects of drugs and because they lack the physiologic context of obesity. In the current study, we have developed a novel double knockout model and characterized in detail the metabolic parameters relevant to insulin resistance. The outcome is that we have described a model in which improvement of peripheral insulin resistance improves vascular function, despite persistent obesity and modest hyperglycemia. This normalization provides strong evidence that obesity has minimal impact on vasodilation mediated by NO in the absence of insulin resistance. Since the deletion of PTP1B in this model improves peripheral but not hepatic insulin resistance, we can refine the metabolic hypothesis of vessel dysfunction beyond whole body insulin resistance to specific compartments. Impairment of vascular function correlates with markers of metabolic dysfunction in muscle and fat but not the liver. To the authors' knowledge, this is the first study to localize vascular dysfunction in obesity to peripheral insulin resistance.

Effect of PTP1B deletion on reactive oxygen species

The mechanisms by which insulin resistance impairs NO-mediated vasodilation are incompletely understood. In the current study, we present evidence that the primary mechanism is an increase in reactive oxygen species (ROS). In obese mice, the level of superoxide is increased versus lean controls (Figure 3A, 3C); vasodilation is restored by oxidant scavenging and components of the NAD(P)H oxidase pathway are increased (Figure 2D,E; Figure 4; Table 2). Blockade of NAD(P)H oxidases normalizes oxidant load. Correction of insulin resistance by deletion of PTP1B in obese mice corrects the augmented ROS levels. Scavenging of superoxide restores sensitivity to acetylcholine as does deletion of PTP1B in K_dbK_PTP mice. The L-NAME resistant component of endothelium-dependent dilation remains the same in all mice, ruling out differences in NO-independent endothelial vasodilation. Taken in toto, these data suggest that insulin resistance corrupts endothelial vasodilation by NAD(P)H oxidase-derived oxidants.

Previous studies have attributed increases in superoxide levels to an increase in NAD(P)H oxidase activity [44, 45]. Nox2 (gp91_phox) was originally identified as the primary source of pathological superoxide production in insulin resistant states, but more recently roles for Nox1 and Nox4 have been identified [46–48]. A key finding in the current study is that the expression levels of Nox1 and its regulatory enzymes Noxa1/Noxo1 in the vasculature correlate with obesity-induced insulin resistance. In obese mice, the expression of Nox1, Noxa1, and Noxo1 are markedly increased and these levels are reversed by the deletion of PTP1B. Co-expression of Nox1 with Noxo1 and Noxa1 results in the constitutive and high-level production of superoxide [49, 50]. The significance of increased Nox1 activity in insulin resistant states in not clear. The loss of Nox1 attenuates, while an increased expression of Nox1 potentiates, angiotensin-induced hypertension [51, 52]. In large conduit blood vessels, Nox1 expression is unchanged in db/db mice [53]; however, our study is the first to measure expression levels of these proteins in microvessels. The expression level of other Nox isoforms was examined in addition to Nox1, and in contrast to other studies [53] there were no changes with obesity.

In summary, these experiments provide new evidence that in the context of obesity, the underlying risk factor that impairs both endothelium-dependent and - independent NO dilation is peripheral insulin resistance. Moderate hyperglycemia or morbid obesity does not cause endothelial dysfunction when the enzyme PTP1B is absent. These data indicate that PTP1B may represent an important therapeutic target for not only the metabolic but also cardiovascular therapy of obesity.

Non-Standard Abbreviations and Acronyms

The abbreviations used are:

PTP1B

protein tyrosine phosphatase 1B

H_db

heterozygous for mutant leptin receptor

H_PTP

heterozygous for PTP1B gene deletion

K_db

homozygous for mutant leptin receptor

K_PTP

homozygous for PTP1B gene deletion

HbA_1c

Hemoglobin A1c

FFA

free fatty acids

eNOS

endothelial nitric oxide synthase

IRβ

insulin receptor-β

SMA

small mesenteric arteries

SOD

Superoxide dismutase

L-NAME

ω-Nitro-L-arginine methyl ester

CMH

methoxycarbonyl-2,2,5,5-tetramethyl-pyrrolidine

DHE

Dihydroethidium

Nox

NAD(P)H oxidase

Noxa1

NAD(P)H oxidase novel activator

NAD(P)H oxidase novel organizer