Adipose Tissue Hypoxia, Inflammation, and Fibrosis in Obese Insulin-Sensitive and Obese Insulin-Resistant Subjects

We confirmed fat hypoxia in obese as compared to lean subjects. However, fat oxygenation was similar in obese insulin sensitive and insulin resistant subjects suggesting fat hypoxia may be simply a consequence of fat expansion.

Abstract

Context:

A substantial number of obese individuals are relatively insulin sensitive and the etiology for this variation remains unknown.

Objective:

The primary objective was to detect factors in adipose tissue differentiating obese insulin-sensitive (OBIS) from obese insulin-resistant (OBIR) individuals and investigate whether adipose tissue hypoxia is a contributing factor in the pathogenesis of insulin resistance.

Design and Setting:

This was a cross-sectional study in the general community.

Participants:

Subjects consisted of nondiabetic OBIS and OBIR subjects with similar body mass index, age, and total body fat but different insulin sensitivity index as well as lean insulin-sensitive subjects.

Intervention(s):

There were no interventions.

Main Outcome Measure(s):

We examined adipocytokines and the expression of candidate genes regulating hypoxia, inflammation, and lipogenesis in adipose tissue and adipose tissue oxygenation.

Results:

OBIS subjects had increased plasma adiponectin but similar plasma TNFα and leptin levels as compared with OBIR subjects. Genes regulating inflammation (CD68, MCP1, scavenger receptor A, and oxidized LDL receptor 1) were increased by 40%–60% (P < .05) in OBIR vs OBIS cohorts. In addition, genes involved in extracellular matrix formation such as collagen VI and MMP7 were up-regulated by 43% and 78% (P < .05), respectively, in OBIR vs OBIS. The expression of HIF1α and VEGF gene expression was increased by 37% and 52%, respectively, in OBIR vs OBIS (P < .01). Despite the differential expression in hypoxia-related genes, adipose tissue oxygenation measured by a Licox oxygen probe was not different between OBIS and OBIR subjects, but it was higher in lean subjects as compared with obese subjects.

Conclusions:

We confirmed that adipose tissue inflammation and fibrosis play an important role in the pathogenesis of insulin resistance independent of obesity in humans. Whether hypoxia is simply a consequence of adipose tissue expansion or is related to the pathogenesis of obesity-induced insulin resistance is yet to be understood.

Although there is a strong relationship between obesity and insulin resistance, a substantial number of obese individuals are protected from impaired insulin sensitivity and the metabolic consequences of insulin resistance and impaired lipid metabolism such as type 2 diabetes and cardiovascular diseases (1, 2). Obese insulin-resistant (OBIR) and obese insulin-sensitive (OBIS) subjects are not clearly defined subgroups (3) but represent the extremes of a continuum. Dysfunctional adipose tissue, characterized by inadequate angiogenesis, hypoxia, inflammation, and fibrosis, has been proposed as a key pathological process linking obesity and metabolic disease (4). Therefore, studying adipose tissue in insulin-resistant and insulin-sensitive subjects with similar levels of adiposity will shed light on potential mechanisms involved in insulin resistance independent of obesity.

Although there are limited data comparing sc adipose tissue of OBIR and OBIS individuals (5–7), these studies suggest that inflammation in adipose tissue is responsible for the pathological metabolic consequences in individuals with obesity. Recent studies have identified decreased adipose tissue oxygenation in overweight and obese individuals as compared with lean control subjects (8). It has been shown that hypoxia results in inflammation in adipose tissue and insulin resistance in vitro and in animal studies (9–12). Hence, healthy adipocyte expansion may occur with appropriate angiogenesis, whereas pathological expansion results in hypoxia, inflammation, fibrosis, and subsequent insulin resistance (4).

Most of the studies of adipose hypoxia and insulin resistance have been performed in preclinical studies, and much less is known about the role of adipose tissue expansion in the pathogenesis of obesity induced insulin resistance. In this study, we hypothesized that the expression of genes regulating the extracellular matrix, inflammation, and hypoxia will be increased in OBIR when compared with OBIS individuals, and OBIS subjects will have improved adipose tissue oxygenation as compared with OBIR subjects. A better understanding of the pathophysiology of obesity-induced insulin resistance would provide a basis for future research into the prevention of obesity related metabolic disorders.

Materials and Methods

Study population and study design

All participants signed consent forms approved by the institutional review board. Studies were performed in the Clinical Research Center of the University of Arkansas for Medical Sciences (group 1) and the University of Colorado (group 2).

Subjects with major health problems including but not limited to diabetes, heart disease, stroke, liver or renal disease, anemia, clotting disorders, and sleep apnea were excluded from both groups. The use of antiinflammatory medications such as nonsteroidal antiinflammatory drugs and steroids in the 6 months prior to the study was also one of the exclusion criteria. Diabetes was excluded using a standard 75-g oral glucose tolerance test and fasting glucose (group 1) or a glycated hemoglobin and fasting glucose (group 2). Subjects with no prior diagnosis of diabetes who had a fasting glucose of 126 mg/dL or greater or glycated hemoglobin levels of 6.5% or greater or 2-hour glucose of 200 mg/dL or greater were excluded. Pregnant women were excluded from both studies.

Insulin-modified (0.04 U/kg) frequently sampled iv glucose tolerance test (FSIGT) was performed (13) in both groups. Insulin sensitivity index (S_I) was calculated from the insulin and glucose data obtained from FSIGT using the MinMod Millennium program (13, 14). Visits from menstruating women were timed such that the FSIGT visit coincided with the follicular phase of the menstrual cycle. Body fat percentage was measured using dual-emission X-ray absorptiometry.

The participants in group 1 were selected from a cohort (n = 48) of obese, nondiabetic, healthy subjects between the ages of 21 and 59 years with a body mass index (BMI) between 31 and 40 kg/m². To better understand the differences between OBIS and OBIR subjects, individuals with an S_I in the highest and lowest quartiles of this cohort (respectively for OBIS and OBIR) were studied (n = 24). Based on this cohort, the OBIS subjects had an S_I greater than 2.75 × 10⁻⁴ min⁻¹/μU · mL (Figure 1), which was also used to define OBIS subjects in group 2. Group 2 included a total of 16 nondiabetic healthy subjects as follows: four lean (BMI < 25 kg/m²), six OBIR subjects, and six OBIS subjects between the ages of 25 and 50 years. To identify OBIS subjects, individuals without metabolic syndrome who had S_I values higher than 2.75 × 10⁻⁴ min⁻¹/μU · mL were included. The OBIR subgroup included subjects with metabolic syndrome who were matched to the OBIS subgroup in regard to BMI and age.

Figure 1.

OBIS was defined based on the top quartile of S_I and OBIR as the lowest quartile.

To eliminate effects of other variables on adipose oxygenation in this small cohort, we excluded subjects who were postmenopausal, regularly exercising (three times a week for 30 min) or had a significant weight change (>10% of weight) in the past 6 months. Due to effects of smoking and altitude on oxygenation, current smokers (more than a half-pack per day) and subjects living in altitude higher than 7000 feet for 6 months in the past year prior to the study were also excluded from group 2. Baseline characteristics of subjects are summarized in Table 1.

Table 1.

Baseline Characteristics of Subjects

	Group 1			Group 2
	OBIS	OBIR	P Value	Lean	OBIS	OBIR	P Valuea
BMI	34.4 ± 2.2	34.3 ± 2.5	.93	23 ± 1	32 ± 1b	34 ± 2b	.13
Age	40.4 ± 10.2	43.8 ± 6.0	.34	31 ± 3	36 ± 4	37 ± 3	.19
Gender, female/male	11/1	9/3	.59	3/1	4/2	6/0	.45
SI, × 10−4 min−1/μU · mL	3.46 ± 0.65	1.15 ± 0.23	<.01	4.4 ± 0.8	3.5 ± 0.3	1.8 ± 0.2b	<.01
Body fat, %	44.2 ± 2.1	43.4 ± 3.1	.55	27 ± 4	38 ± 3b	41 ± 1b	.1
LDL-C, mg/dL	128.3 ± 40.7	108.6 ± 43.0	.27	74 ± 16	76 ± 19	108 ± 10	.16
HDL-C, mg/dL	52.3 ± 9.7	47.5 ± 8.8	.23	49 ± 5	51 ± 7	43 ± 3	.32
Triglycerides, mg/dL	105.2 ± 41.5	160.8 ± 91.8	.07	79 ± 2	118 ± 25	127 ± 18b	.79

Abbreviation: C, cholesterol.

^aP value of OBIS vs OBIR subjects.

^bP < .05 as compared with lean subgroup.

To determine adipose tissue oxygenation, we measured adipose tissue partial pressure of O₂ (ATpO₂) using a combined oxygen and temperature probe (catalog number CC1.P1; Integra Life Sciences) inserted through a 3.2-cm-long 14-gauge iv catheter (Medex) to a depth of 1.5 cm in the abdominal sc adipose tissue (one-third of the distance between the umbilicus and the superior iliac crest of the left abdomen). The probe was connected to an electronic unit (LICOX CMP; brain oxygen monitoring unit), which recorded ATpO₂ and temperature after allowing for 60 minutes of equilibrium at a room temperature of 25°C. After insertion, the system was allowed to equilibrate for 30 minutes, and ATpO₂ and adipose tissue temperature were measured every 60 seconds for up to 2 hours. Recording was stopped when the measurements of ATpO2 remained stable (<1 mm Hg fluctuation) for at least 10 minutes (steady state); values from the last 10 minutes were averaged. The subjects were fasting for at least 12 hours prior to the procedure, which was performed in the morning.

An incisional sc adipose tissue biopsy was performed from the lower abdominal wall using local anesthesia.

Laboratory procedures

Insulin was measured using an immunochemiluminescent assay (MLT Assay) in the General Clinical Research Center core laboratory at the University of Arkansas (group1) and using a RIA (Millipore Corp) in the Clinical and Translation Research Center core laboratory at the University of Colorado (group 2). Plasma glucose was measured in duplicate by a glucose oxidase assay. Plasma adiponectin and TNFα were measured using ELISA method (R&D Systems) following the manufacturer's instruction. Plasma angiogenesis factors including plasma angiopoietin-like 2, hepatocyte growth factor (HGF), vascular endothelial growth factor (VEGF)-A, fibroblast growth factor 2, epidermal growth factor, and endothelin were measured using Millipore Milliplex kits following the manufacturer's instruction.

Total RNA from adipose tissue was isolated using an RNAeasy minikit (QIAGEN) per the manufacturer's instruction. The quantity and quality of the isolated RNA were determined by an Agilent 2100 bioanalyzer. A real-time PCR was conducted as described previously (15). All data were expressed in relation to 18S RNA as a control, in which the standard curves were generated using pooled RNA from the samples assayed. Therefore, the data represent arbitrary units that accurately compare each set of samples with each other but do not necessarily accurately compare samples between different assays. Primer sequences of 18S, CD68, collagen, leptin, macrophage chemoattractant protein-1, visfatin, scavenger receptor A, oxidized low-density lipoprotein (LDL) receptor 1, collagen VIa, matrix metalloproteinase-7 (MMP7), thrombospondin 1 (TSP1), hypoxia-inducible factor α (HIF1α), and HIF1α responsive genes such as VEGF-A were obtained as published previously (15–20).

Data analysis

The cross-sectional data from this study were analyzed using t tests to compare outcomes between OBIS and OBIR subjects. Analysis of covariance was used to compare group means adjusted for potential confounders found to be imbalanced between the groups. Results of comparisons between OBIS and OBIR groups are presented in graphical format with mean ± SE for each group. The adjusted P values from the t tests, Fisher's test (for categorical variables), and the analysis of covariance model are also included. The P values from tests comparing group means were not adjusted for multiple comparisons because this is a study to identify putative factors involved in the development of adipose tissue hypoxia, inflammation, and fibrosis in OBIS and OBIR subjects. Analyses were done using the R software package (http://www.r-project.org, version 2.7.2).

Results

A total of 24 nondiabetic subjects (12 in each group) were included in group 1 of this study. All of the subjects in this cohort were obese defined by a BMI greater than 30 kg/m² with a mean BMI of 34 kg/m². The OBIS and OBIR groups were defined as the top and lowest quartiles of S_I (3.46 ± 0.65 vs 1.15 ± 0.23 × 10⁻⁴ min⁻¹/μU · mL, respectively, P < .001) measured by the FSIGT (Figure 1). The baseline characteristics of the group 1 cohort are summarized in Table 1. Age, cholesterol, BMI, and total body fat were similar between the two subgroups of OBIS and OBIR in group 1 and two subjects in each subgroup were taking statins. However, OBIS subjects had an increased S_I when compared with OBIR subjects by design, and their 2-hour glucose levels were decreased when compared with the OBIR group (100.4 ± 33.7 vs 133.7 ± 30.5, P = .05, respectively).

Plasma adipokines and adipose gene expression in OBIS and OBIR subjects

Low levels of adiponectin have been indicated as an independent risk factor for the development of metabolic syndrome and diabetes (21, 22). OBIS, when compared with OBIR subjects, had increased plasma adiponectin levels (13.3 ± 2.0 vs 7.8 ± 0.8 μg/mL, respectively P = .02). However, there was no significant difference in the levels of TNFα and leptin (Figure 2A).

Figure 2.

A, Plasma levels of adiponectin but not TNFα or leptin were increased in OBIS. B–D, Expression of candidate genes in adipose tissue of OBIS as compared with OBIR subjects. All data were expressed in relation to 18S RNA as a control.

Adipose tissue expression of inflammatory markers and extracellular matrix components in OBIR and OBIS

The expression of candidate genes as markers of inflammation in sc adipose tissue (SAT) was studied. Adipose tissue macrophage accumulation has been associated with insulin resistance (15), and hence, there was a significant increase in CD68 expression in OBIR (Figure 2B). Other inflammatory markers that were elevated in the SAT of OBIR subjects (compared with the OBIS group) included visfatin, scavenger receptor A, and oxidized LDL receptor 1. Macrophage chemoattractant protein-1 expression was trending higher in the OBIR group but was not statistically significant (P = .06) (Figure 2B).

Increased expression of extracellular matrix components in OBIR compared with OBIS

The extracellular matrix of the adipocytes provides mechanical support and contributes to cell signaling. With increasing obesity and adipose tissue expansion, the extracellular matrix remodels to accommodate adipocyte growth. Consequently, an up-regulation of extracellular matrix components has been associated with adipose tissue fibrosis in insulin resistance (23, 24). We examined the expression of collagen VIa, MMP7, and TSP1, which are important components involved in modulating the adipose tissue extracellular matrix. MMP7 is involved in the breakdown of the extracellular matrix (25), whereas TSP1 is a matrix glycoprotein that binds to collagen and has antiangiogenesis property (20). OBIR subjects had a significantly increased expression of collagen VIa, MMP7, and TSP1 when compared with the OBIS group (Figure 2C).

Increase in expression of HIF1α and VEGF-A in SAT in OBIR compared with OBIS

Recent data suggest that adipose tissue hypoxia and angiogenesis are potentially important mechanisms in the development of insulin resistance in obesity (8). Therefore, we investigated the expression of HIF1α and HIF1α-responsive genes such as VEGF-A. OBIR subjects had significantly increased expression of HIF1α and VEGF-A in SAT, suggesting adipose tissue hypoxia. The expression of the CD31 gene as a marker of endothelial cells was also increased in OBIR as compared with OBIS subjects (Figure 2D).

Adipose tissue oxygenation correlated with BMI but was not different between OBIS and OBIR subjects

To further investigate adipose tissue oxygenation, we studied ATpO₂ in vivo using a combined oxygen and temperature probe inserted into the abdominal sc fat. A total of four lean and 12 obese non diabetic subjects were examined (group 2). Baseline characteristics of this cohort are summarized in Table 1. OBIS and OBIR subgroups had similar BMI and body fat percentage but significantly different insulin sensitivity index (P < .001) by study design. The lean subgroup had a lower BMI and body fat percentage as compared with the obese subgroups (P < .001) as expected but similar S_I when compared with the OBIS subgroup (P = .29).

Obese subjects had decreased adipose tissue oxygenation (ATpO₂) as compared with lean subjects (39.3 ± 1.5 mm Hg vs 53 ± 1.9, P < .001), yet ATpO₂ was similar between OBIS and OBIR groups (41.1 ± 1.2 vs 37.7 ± 2.4 mm Hg, P = .27) (Figure 3). Despite similar S_I, the OBIS subgroup had a lower ATpO2 as compared with lean controls (P < .001). Adipose tissue temperature remained similar among all groups (Figure 3). ATpO₂ negatively correlated with BMI (Figure 4), body fat, and waist circumference (r = −0.8, −0.7, and −0.7, respectively, P < .05) but did not correlate with S_I. Angiogenic markers were measured in the circulation and angiopoietin-like 2 levels were increased in OBIS subjects as compared with OBIR subjects (335.0 ± 40.7 vs 159.5 ± 33.5 pg/mL, P = .008). Plasma levels of VEGF-A, fibroblast growth factor-2, and hepatocyte growth factor were similar among groups. Epidermal growth factor and endothelin levels were not detected in the circulation.

Figure 3.

Adipose tissue oxygenation. Obese subjects had a lower ATpO₂ as compared with lean subjects (P < .01). ATpO₂ in OBIS vs OBIR subjects was not significantly different (P = .1). Adipose tissue temperature remained similar among groups.

Figure 4.

ATpO₂ correlated with BMI in lean and obese subjects.

Discussion

Previous studies have identified many abnormalities in the adipose tissue of obese insulin-resistant subjects, which include increased numbers of macrophages, increased inflammatory cytokine expression, increased extracellular matrix components, and fibrosis as well as hypoxia and a decrease in capillaries (22, 26, 27). Most of these studies, however, are not able to determine the primary abnormality that triggers these changes. Elegant studies in mice that involved a knockout of collagen VI suggested that a restrictive extracellular matrix in the face of expanding adipocytes resulted in hypoxia, inflammation and insulin resistance (23), and subsequent reviews (4) have suggested that individual subjects are capable of healthy or pathological adipose tissue expansion, depending on the ability of the tissue to allow vascular expansion without excessive restriction of the extracellular matrix. Many of these concepts were based on mouse data involving a fairly rapid adipose tissue expansion, as opposed to humans who usually develop obesity over the course of many years.

This study was designed to examine elements of adipose tissue inflammation, extracellular matrix, and hypoxia in OBIS and OBIR subjects. The OBIR and OBIS subjects had the same BMI but significantly different levels of insulin sensitivity and thus would correspond to adipose tissue with healthy vs pathological expansion. As expected, plasma adiponectin levels were 42% higher in the OBIS compared with the OBIR group and the adipose tissue from OBIS subjects demonstrated lower levels of inflammatory markers. Plasma TNFα was not different between the OBIS and OBIR groups, which is consistent with a prior study showing adipose tissue secretion of TNFα but not the plasma level of TNFα is associated with insulin resistance (28), suggesting that TNFα is more active locally than systemically. Likewise, we reported that OBIS and OBIR subjects had similar levels of plasma leptin. The OBIS and OBIR groups were not significantly different in BMI or percentage body fat, and therefore, the plasma leptin levels are consistent with studies linking plasma leptin to adiposity (29).

In addition to inflammation, the OBIR subjects demonstrated an increase in extracellular matrix markers, with increased expression of collagen VIa, MMP7, and TSP1 in OBIR when compared with OBIS subjects. Because these groups were matched for BMI, these data suggest that a change in the extracellular matrix of adipose tissue is not a simple consequence of obesity but that alterations in the extracellular matrix play a role in the development of insulin resistance. These results are consistent with other animal studies reporting the absence of collagen VI is associated with improved whole-body energy homeostasis and suggests weakening the extracellular matrix of adipocytes allows stress-free expansion during positive energy balance and results in an improved inflammatory profile and decreased fibrosis (23).

More recently, adipose tissue hypoxia has been proposed to be closely linked to an unforgiving extracellular matrix and therefore the underlying cause of the development of inflammation and cellular dysfunction in adipose tissue of obesity (30, 31). Adipose tissue hypoxia is shown to play an important role in macrophage chemotaxis, adipocytokine dysregulation, and impaired insulin signaling (8, 10–12). Recent data have suggested that the expansion of adipose tissue in obesity is associated with impaired adipose tissue microcirculation and local hypoxia (9, 10). Hypoxia in the adipocyte activates the HIF1α and nuclear factor-κβ pathways, leading to increased inflammation and stimulation of angiogenesis via increased VEGF expression and transcription of other proangiogenic genes. In agreement with previous studies (8), we showed that the gene expression of HIF1α was increased in OBIR as compared to OBIS subjects. In addition, VEGF-A, which is well recognized to be transcriptionally regulated by HIF1α, was higher in OBIR when compared with OBIS subjects.

Consistent with previous reports (8), we confirmed that obesity is associated with adipose tissue hypoxia. The ATpO₂ in the SAT of obese subjects was 26% lower than that of lean subjects. However, prior studies did not examine adipose oxygenation in obese subjects with a wide range of insulin sensitivity (ie, OBIS and OBIR subjects). Despite a significant difference in S_I, OBIS and OBIR subjects in our cohort had similar adiposity including BMI, body fat, and waist circumference. When adipose tissue oxygenation was measured using a combined temperature and oxygen probe, ATpO₂ was not significantly different between the two groups. Furthermore, ATpO₂ correlated only with makers of adiposity (BMI, body fat, and waist circumference). Similar to Pasarica et al (8), we did not find any association with ATpO₂ and insulin sensitivity (n = 16). These results suggest that adipose tissue hypoxia is likely a consequence of fat expansion and may not be a causal factor of insulin resistance. On the other hand, it is possible that a mild drop in ATpO₂ experienced by some subjects (OBIR) is able to trigger a hypoxic response and an increase in HIF1α, whereas OBIS subjects are more resistant to this effect.

This study is limited by investigating the sc fat that compromises most total body fat, although not correlated as strongly as visceral fat with insulin resistance. In addition, it is possible that conflicting findings related to tissue oxygenation could be due to differing methodologies in the use of the oxygen probe technique, suggesting the need for further investigation. Larger studies are needed to further investigate hypoxia and insulin resistance in humans. Our study was not powered to detect small differences in adipose tissue oxygenation (<10 mm Hg) between the two subgroups of OBIS and OBIR. However, we successfully enrolled a unique population of obese insulin-sensitive subjects who were similar in obesity to the typical obese insulin-resistant subjects but were as insulin sensitive as the lean subgroup. This subgroup (OBIS) had similar ATpO₂ to the OBIR subgroup but significantly lower levels of ATpO₂ than the lean controls. Furthermore, the association of ATpO₂ with obesity but not with insulin resistance suggests that adipose tissue hypoxia might not be involved in the pathogenesis of insulin resistance in humans. It should also be noted that groups 1 and 2 were slightly different due to the design of study and differences in the geographical area of recruitment. The intention of this study was to investigate adipose tissue gene expression and oxygenation in obese individuals who are protected from impaired insulin sensitivity as compared with typical obese insulin-resistant subjects with similar BMI. As planned, we successfully studied two cohorts of OBIS and OBIR subgroups who were similarly obese but had significantly different insulin sensitivity levels, but, unfortunately, we were unable to perform all studies in the same cohort.

In summary, the expression of many genes regulating inflammation, hypoxia, and extracellular matrix formation were significantly decreased in OBIS vs OBIR individuals. Our results imply that a subgroup of obese individuals who are protected from insulin resistance have a distinct adipose tissue composition and also suggest that dysfunctional adipose tissue, rather than an excess of adipose tissue, is the pathological process linking obesity and metabolic disease. We confirmed that adipose tissue inflammation and fibrosis play an important role in the pathogenesis of insulin resistance independent of obesity in humans. Whether hypoxia is simply a consequence of adipose tissue expansion or is related to the pathogenesis of obesity induced insulin resistance is yet to be understood.