Treatment of Obese Diabetic Mice With a Heme Oxygenase Inducer Reduces Visceral and Subcutaneous Adiposity, Increases Adiponectin Levels, and Improves Insulin Sensitivity and Glucose Tolerance

Male obese (ob) mice (B6v-Lep ob/J) were purchased from Harlan (Chicago, IL) at the age of 7 weeks and used at the age of 8 weeks. Age- and sex-matched lean mice (B6.V, lean; Harlan) were used as controls. Mice were fed a normal chow diet and had free access to water and food. Body weights of ob and lean mice at the beginning of the treatment were 34 ± 5 and 26 ± 3 g, respectively. Glucose levels were 229 ± 21 and 154 ± 9 mg/dl for ob and lean mice, respectively.

Glucose monitoring was performed using an automated analyzer (Lifescan, Milpitas, CA). Beginning at 9 weeks, when all ob mice have established diabetes, cobalt protoporphyrin (CoPP) (3 mg/kg once per week) or CoPP plus stannous mesoporphyrin (SnMP) (2 mg/100 g body wt three times per week) were administered intraperintoneally for 6 weeks. Metalloporphyrins were dissolved in 10 mmol/l Tris base, and the pH was adjusted to pH 7.8 with 0.1 N HCl. A Tris/HCl solution free of metalloporphyrins was used to inject control animals. There were six groups of animals: 1) lean, 2) lean-CoPP, 3) lean-CoPP-SnMP, 4) ob control, 5) ob-CoPP, and 6) ob-CoPP-SnMP. Food intake did not change in the mice treated with various treatments. The Animal Care and Use Committee of New York Medical College approved all experiments.

When mice were killed, subcutaneous and visceral fat in the abdomen (the visible mesenteric fat, fat around the liver, fat around the kidney, and fat around the spleen) were dissected free, pooled for each mouse, weighed, and used to isolate adipocyte MSCs. Cells were frozen until needed for protein measurements. Aorta, heart, and kidney were also harvested, drained of blood, and flash frozen in liquid nitrogen. Specimens were maintained at −80°C until needed. Frozen aorta and kidney segments were pulverized, and the supernatant was isolated and used for HO activity, HO-1 and HO-2 protein (26).

HO activity was assayed in homogenates of various tissues or MSCs as previously described (27). 300 μg protein was used for CO measurement in the presence and absence of heme and NADPH-generating system (28,29).

Multiples assay kits were used for quantification of the proteins in mice serum and were done according to the manufacturer's protocol. Plates were analyzed using a Luminex 100IS analyzer (Lumunes, Austin, TX). The data were saved and evaluated as median fluorescence intensity using appropriate curve-fitting software (Luminex 100IS software version 2.3). Certain measurements, including adiponectin (high molecular weight), TNF-α, IL-1β, and IL-6, were determined in mouse serum using an ELISA assay (Pierce Biotechnology, Woburn, MA). Insulin was measured using an ELISA kit (Millipore, Billerica, MA).

After a 12-h fast, mice were injected intraperitoneally with glucose (2.0 g/kg body wt). Blood samples were taken at various time points (0–120 min), and blood glucose levels and serum insulin levels were measured. For insulin tolerance test, mice were injected intraperitoneally with insulin (2.0 units/kg). Blood samples were taken at various time points (0–90 min) and blood glucose levels were measured.

Bone marrow smears made from the tibia were stained with 0.5% oil red O in isopropanol (w/v) for 10 min, and lipid droplets were then evaluated using a light microscope digitalized with a charge-coupled device camera and an image analysis system (Imaging & Computers, Milan, Italy). Mean number of lipid droplets was calculated from six different fields.

Lean, obese, and diabetic aorta or bone marrow were placed in plastic scintillation minivials, containing 5 μmol/l lucigenin for the detection of O₂⁻, as described previously (30,31).

The bone marrow cells were collected from the mice femur and tibia of ob mice. Nonadherent cells were removed on day 4, and adherent cells were further incubated in fresh Iscove's modified Dulbecco's medium, including 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA) and 1% antibiotic-antimycotic solution (Invitrogen). The cultured cells with spindle-like morphology after three passages were defined as MSCs and were induced to differentiate into adipocyte cells in vitro. Colony forming units–fibroblasts (CFU-F) were determined using MesenCult media according to manufacturer's instruction (Stem Cell Technology, Vancouver, Canada).

MSCs were cultured in adipogenic medium (high glucose Dulbecco's modified Eagle's medium containing 10 μg/ml insulin, 0.1 mmol/l dexamethasone, 0.2 mmol/l indomethacin, 10% FBS, and 1% antibiotic-antimycotic solution). At 50% confluence, 2 μmol/l CoPP and 5 μmol/l SnMP were added, and adipogenesis was measured using the oil red O staining as described previously (32). Briefly, cells were fixed by ice-cold 10% formalin in PBS for 10 min, rinsed with distilled water, and stained with oil red O solution for 20 min. Cells were placed in absolute propylene glycol for 5 min, rinsed in 85% propylene glycol followed by distilled water, and air-dried. Oil red O stain was extracted with isopropanol, and optical absorbance was measured at 490–520 nm.

The International Society for Cellular Therapy has provided the following minimum criteria for defining multipotent mesenchymal stromal cells. MSC are normally plastic adherent under standard culture conditions and express CD105, CD73, and CD90. MSCs must lack the expression of CD45, CD34, CD14 or CD11b, CD79 or CD19, and HLA-DR and be able to differentiate into osteoblasts, adipocytes, and chondroblasts in vitro (33). The MSCs presented as a homogeneous fibroblastoid cell population. Expression of stem cell markers assessed with RT-PCR showed that after passage 2, these cells were negative for hematopoietic cell markers (CD34 and CD45) and positive for CD90, CD105, and CD166 (ALCAM), markers of MSCs. Flow cytometric analysis of passage 4 cells confirmed that cells were negative for CD34 and CD45 and that cells were positive for CD29 (1-integrin) and CD90 (Thy-1) (34).

Statistical significance between experimental groups was determined by the Fisher method of analysis of multiple comparisons (P < 0.05). For comparison between treatment groups, the null hypothesis was tested by a single-factor ANOVA for multiple groups or unpaired t test for two groups. Data are presented as means ± SE.

Western blot analysis showed a significant decrease (P < 0.05) in the ratio of HO-1 to actin in the aorta of ob compared with lean mice (Fig. 1A). CO production in the aorta of lean and obese mice was measured by gas chromatography–mass spectrometry (GCMS) and was found to be significantly (P < 0.05) lower in obese mice compared with lean mice. This reflects the less active HO system in obese mice compared with lean mice. A weekly injection of CoPP for 6 weeks resulted in a significant increase in HO-1 protein levels in both lean and ob mice (Fig. 1B).

CoPP administration increased aortic HO-1 protein in lean mice by 4.3-fold (P < 0.01) and in ob mice by 9-fold (P < 0.001) (Fig. 1B). HO activity was measured in aortas isolated from these animals (Fig. 1C). There was a significant (P < 0.05) decrease in HO activity in ob mice compared with age-matched lean controls (0.27 ± 0.02 and 0.40 ± 0.03 nmol bilirubin · mg⁻¹ · h⁻¹, respectively). CoPP administration increased bilirubin formation in the aorta of both lean and obese mice (1.9 ± 0.35 and 1.67 ± 0.28 nmol bilirubin · mg⁻¹ · h⁻¹, respectively). SnMP abolished the increase in HO activity (results not shown). This pattern of HO expression and activity was also observed in kidneys and heart (data not shown).

There was a significant decrease in serum adiponectin levels in ob mice compared with age-matched lean controls (Fig. 2); adiponectin levels in ob mice were 2.51 ± 0.33 μg/ml compared with 3.86 ± 0.46 μg/ml (P < 0.029) in lean animals. CoPP administration resulted in a marked fourfold increase (P < 0.009) in the levels of serum adiponectin in ob mice compared with untreated ob animals. Serum adiponectin increased to 10.2 ± 1.66 μg/ml after HO-1 induction compared with 2.51 ± 0.33 μg/ml in untreated ob mice. The concomitant administration of SnMP with CoPP blocked the increase in serum adiponectin.

Obese mice exhibited a significant increase in serum IL-6 levels (620 ± 6 pg/ml) when compared with age-matched lean controls (89 ± 23 pg/ml) (Fig. 3). Administration of CoPP resulted in a significant decrease in serum IL-6 levels in ob mice when compared with untreated ob mice. This decrease was blocked by administration of SnMP. IL-6 serum levels in CoPP in combination with SnMP were 600 ± 122 pg/ml compared with 257 ± 48 pg/ml in CoPP-treated ob mice. Similar results were observed with serum TNF-α and IL-1β levels (Fig. 3B and C).

The levels of O₂⁻ in aorta obtained from ob mice treated with vehicle alone were 8.0 ± 0.6 (×10⁴ cpm) compared with 1.1 + 0.3 × 10⁴ cpm (P < 0.03) in vehicle-treated lean mice (Fig. 3D). CoPP-treated ob mice showed a significant decrease in O₂⁻ levels in the aorta compared with vehicle-treated ob mice. The levels in the CoPP-treated obese mice decreased to 1.7 ± 0.4 × 10⁴ cpm (P < 0.039). Inhibition of HO activity by administration of SnMP to CoPP-treated mice increased superoxide production to 6.3 ± 0.4 × 10⁴ cpm.

As seen in Fig. 4A, CoPP treatment prevented weight gain in ob mice when compared with age-matched controls. The prevention of body weight gain was manifested by a reduction in visceral fat in obese mice. In contrast, control lean mice administered CoPP did not lose as much weight (30.33 ± 0.53 vs. 27.42 ± 0.66 g for control and CoPP-treated lean mice, respectively). The prevention of weight gain in obese mice after treatment with CoPP was partially reversed by co-administration of SnMP. Food intake in the treatment and control groups was comparable (Fig. 4B). As seen in Fig. 4C and D, fat and body appearance of obese mice confirmed the reduction in fat content and body weight loss. Visceral fat in obese mice was decreased by CoPP treatment from 6.1 ± 0.1 to 3.9 ± 0.4 g (P < 0.03). Subcutaneous fat in obese mice was decreased from 3.8 ± 0.2 to 2.5 ± 0.2 g (P < 0.01). The ability of SnMP to reverse the CoPP-induced loss of body weight and the increased adiponectin levels in ob mice suggests that the decrease in adiponectin and the resultant weight gain were dependent on decreased HO activity.

Glucose tolerance and insulin sensitivity were determined after development of insulin resistance (Fig. 5A and B). Plasma glucose levels at all time points in obese mice were higher than those in obese mice treated with CoPP. Blood glucose levels in obese mice were significantly elevated (P < 0.05) 30 min after glucose administration and remained elevated. In CoPP-treated obese mice, blood glucose levels were significantly elevated (P < 0.05) after 30 min but returned to initial levels at 120 min. Insulin administration to CoPP-treated obese mice produced a rapid decrease in glucose but not in the vehicle-treated obese mice, suggesting improved insulin sensitivity in CoPP-treated obese mice.

As seen in Fig. 6A, bone marrow cells from obese mice stained for oil red O displayed a significant increase in adipocytes (Fig. 6A, middle) compared with obese mice treated with CoPP or lean mice. Adipogenesis as measured by the number of oil red O lipid droplets was markedly decreased in ob mice treated with CoPP to levels similar to those seen in lean mice (Fig. 6B), suggesting that HO-1 induction exhibits an inhibitory effect on adipogenesis in vivo.

As seen in Fig. 6C, HO activity in the marrow from lean mice was significantly lower in ob mice compared with lean mice (0.45 ± 0.042 vs. 0.69 ± 0.05 nmol bilirubin · mg⁻¹ · h⁻¹, respectively, P < 0.05). CoPP treatment increased HO activity in both lean and obese mice to 1.93 ± 0.26 and 1.74 ± 0.24 nmol bilirubin · mg⁻¹ · h⁻¹, respectively (Fig. 6C). Superoxide levels in bone marrow increased (P < 0.05) in obese mice compared with lean. CoPP treatment resulted in a significant (P < 0.001) decrease in superoxide levels compared with obese mice (Fig. 6D).

MSCs cells were stained with CD45, CD4, CD166, CD90, and CD105 and then measured by fluorescence-activated cell sorter (FACS). Confirmation of the MSC phenotype was made by the presence of positive markers: 83.6% of the cells were positive for CD166, 76.2% were positive for CD90, and 57.26% were positive for CD105. Also, the absence of CD34, a hematopoietic stem cell marker, and CD45, a lymphocytic marker (Fig. 7A), indicate that were <0.2% cells positive for the negative markers CD45 and CD34, suggesting that the MSCs were pure and not contaminated.

Bone marrow mononuclear cells of ob fat mice (3 × 10⁵/well in six-well plate) were cultured for 14 days and stained with Gimesa. The CFU-F colony numbers increased in the presence of high levels of glucose and SnMP when compared with control. CoPP administration resulted in a decrease in CFU-F when compared with the SnMP and high glucose groups (Fig. 7B and C), and the levels were comparable with the control group (15 ± 1), indicating that inhibition of HO-1 increased CFU-F, whereas the induction of HO-1 decreased CFU-F. Inhibition of HO-1 by high glucose or SnMP resulted in increased MSC clonal efficiency. In contrast, induction of HO-1 decreased clonal efficiency.

As seen in Fig. 8A, glucose increases adipogenesis in MSCs similar to that seen in bone marrow. Induction of HO-1 inhibited adipogenesis in MSCs (P < 0.002). In contrast, inhibition of HO-1 did not significantly affect adipogenesis (Fig. 8A and B). This may be due to inhibition of HO activity by glucose (35).

Figure 8C shows representative Western blots depicting HO-1 expression in MSCs before and after glucose exposure. In MSCs exposed to glucose, levels of HO-1 protein were barely detectable; this is similar to control MSC cell homogenates (Fig. 8C, bottom). CoPP and SnMP increased HO-1 in a similar manner, markedly inducing the levels of HO-1 protein.

In obese diabetic mice, we report a decrease in HO-1 protein levels and HO activity when compared with age-matched lean control mice, and that increases in HO-1 protein levels after CoPP administration resulted in increased insulin sensitivity and decreased glucose tolerance. Furthermore, increases in HO-1 protein expression were paralleled by increases in serum adiponectin levels, decreases in visceral and abdominal fat content, and decreased plasma TNF-α, IL-6, and IL-1β levels. Induction of HO-1 also resulted in decreased aorta adipocyte and bone marrow superoxide production (Figs. 3D and 6D). Upregulation of HO-1 by CoPP treatment caused a decrease in adipogenesis in bone marrow both in vivo and in vitro in cultured MSCs and a decrease in secretion of adiponectin in the culture media. This effect was blocked by the inhibitor of HO activity, SnMP, clearly delineating the existence of an HO-1–adiponectin axis that is responsible for the beneficial changes that occur in the metabolic syndrome. This hypothesis is bolstered by the observation that the increased activity of the HO-1–adiponectin axis was associated with increased insulin sensitivity and increased glucose tolerance. An increase in adiponectin levels inhibited the diabetes-mediated increase in TNF-α, IL-1β, and IL-6. The metabolic syndrome and obesity are characterized by increases in serum levels of inflammatory cytokines such as TNF-α and IL-6, which decrease insulin sensitivity (36,37). Additionally, a decrease in serum adiponectin (ACR30) is the result of an increase in ROS, thus contributing to the pathogenesis of insulin resistance (38).

We report here a prevention of both body weight gain (Fig. 4A) and visceral fat content (Fig. 4C) that parallels the increase in adiponectin. The latter is an indicator of improvements in the metabolic syndrome which, in turn, leads to decreases in arterial diseases and heart disease and increased insulin sensitivity (3,10,37,39–41). Although increases in obesity are considered a risk factor for cardiovascular complications (37), obesity-associated improvement in the diabetic phenotype, including increases in insulin sensitivity and glucose tolerance, may occur through adipose tissue and increased adiponectin secretion (10,42). We believe that an increase in the activity of the HO-1–adiponectin axis is crucial in providing adipocyte cells with tolerance to ROS.

The effect of CoPP on body weight with normal food intake is not unexpected. Multiple low-dose regimens result in a prolonged prevention in body weight gain in genetically obese rats and mice (43,44). Low-dose regimens of CoPP are not associated with alterations in endocrine homeostasis or hepatic heme metabolism or food intake as confirmed (Fig. 4A).

The mechanism by which HO-1 is involved in increased adiponectin levels is, we believe, related to the function of HO-1 as a stress response/chaperone protein and to its ability to decrease ROS by increasing glutathione and extracellular superoxide dismutase levels (4,22,45) and by decreasing O₂⁻ production (23,24). PPARγ agonists are shown to induce both HO-1 (21) and the rate-limiting chaperone protein EroL (46,47). PPARγ agonist that increases adiponectin may do so by increasing the levels of EroL chaperone protein (46). Because PPARγ also increases HO-1 protein levels (21) and HO-1 is known as a chaperone protein, it is possible that one of the mechanisms by which HO-1 can increase adiponectin levels is through more efficient adiponectin stabilization and protection. This would confirm the report of Wang et al. (47) that showed that the chaperone protein EroL increased adiponectin. We also have previously shown that upregulation of HO-1 protein in diabetic rats provided both cardio-protection and vascular protection against ROS (24).

The seminal finding of a decrease in HO activity with a resultant decrease in CO generation (Fig. 1A) in obesity reported in this study suggests that CO may be involved as a signaling molecule in adiponectin secretion. This result is in contrast to a previous report that indicated that the metabolic syndrome increases endogenous CO production, promoting hypertension and endothelial dysfunction in obese Zucker rats (48). These investigators measured HO-derived CO by Hb-CO, but exhaled CO was determined by GCMS. In the present study, CO was measured using GCMS. The difference in results remains to be elucidated but could be explained by the use of different animal models and the high concentration of inhibitors of HO system (48).

The effect of HO-1 induction on the bone marrow fat cell is an excellent indicator of the effect on adipogenesis. HO-1 induction decreased MSC-derived adipocytes in bone marrow and may decrease circulating adipocyte stem cells in the body. As a result, it decreases adipogenesis in bone marrow, diminishes adipogenesis in visceral and subcutaneous fat, and promotes the loss of body weight even with normal food intake. This was associated with an increase in adiponectin, insulin sensitivity and improvement in glucose tolerance (Fig. 5). The mechanism by which HO-1 increases insulin sensitivity may be related to HO-1–mediated reduction in IL-1β, TNF, and IL-6 levels. The reduction in IL-1β, TNF, and IL-6 (Fig. 3) levels in vivo after HO-1 expression by CoPP treatment could also contribute to a reduction in oxidative stress and thus contribute to the increase in adiponectin levels insulin sensitivity. TNF, IL-1β, and IL-6 expression induced insulin resistance in human obesity (49,50). Because insulin acts centrally to decrease body weight, it has been suggested that the obesity state observed in ob mice is, in part, due to insulin resistance (37). The results that we present here demonstrate that CoPP treatment decreased IL-1β and improved insulin sensitivity and glucose tolerance (Fig. 5). The HO-1–mediated increase in adiponectin may play an important role in modulating glucose tolerance and insulin sensitivity in these mice. Our results support recent reports of the beneficial effect of increased levels of adiponectin in the metabolic syndrome and obesity (10,22,37). The present study is of considerable interest from a clinical and basic science perspective, clearly defining the existence of an HO-1–adiponectin regulatory axis that can be manipulated to ameliorate the deleterious effects of increased insulin resistance, obesity, and the metabolic syndrome in critical areas of cell damage associated with cardiovascular disease and diabetes.

This work was supported by National Institutes of Health grants DK-068134, HL-55601, and HL-34300.