Background
There is increasing evidence that in central obesity of humans, it is the increase in visceral omental rather than abdominal subcutaneous adipose tissue that best correlates with measures of insulin resistance [
1] and cardiovascular disease [
2‐
4]. Furthermore, obesity is associated with a mild inflammatory response in omental adipose tissue [
5‐
7] and inflammation has been considered the link between diabetes and obesity [
8,
9]. The deleterious effects of obesity with regard to the development of hypertension and type 2 diabetes are primarily seen in extremely obese humans and corrected by weight loss surgery [
10‐
12]. Furthermore the reduction in morbidity due to weight loss surgery has been attributed to a reduction of inflammatory mediators [
12].
One model system for studying the inflammatory response is the in vitro incubation of explants of omental adipose tissue from extremely obese humans for 48 h. IL-8 is a chemokine/adipokine whose circulating level is elevated in obese humans [
13,
14]. More IL-8 is released by adipose tissue explants or adipocytes over 4 h incubation than any other adipokine [
15]. Fain et al. [
16] reported that in human adipose tissue there is a marked up-regulation of IL-6 or IL-8 mRNA as well as release of IL-6 and IL-8 over a 5 h incubation of explants. The up-regulation of IL-8 mRNA was seen within 3 h and about half of this increase was abolished by blocking the effects of endogenous TNFα and IL-1β [
16]. Up-regulation of IL-6 is also seen when freshly isolated rodent adipocytes are incubated in vitro and attributed to effects of collagenase digestion [
17]. However, most of the increase in IL-6 and IL-8 mRNA is seen in the cells, other than fat cells, present in human adipose tissue and seen to the same extent in cut pieces of tissue as in the fractions obtained by collagenase digestion [
16]. Because IL-8 is a chemokine that could play a major role in recruitment of monocytes into adipose tissue [
14] and because of the evidence that TNFα and IL-1β regulated its release by human fat [
16] we focused on these adipokines.
The present studies were designed to utilize fat cells and nonfat cells derived from omental adipose tissue as well as omental fat explants obtained from extremely obese women. The three major aims were to investigate [a] the influence of albumin on the inflammatory response in omental adipose tissue explants, [b] whether the up-regulation is in fat cells or the nonfat cells of omental fat and [c] whether co-incubation of nonfat cells with fat cells, both derived from omental adipose tissue, affected their inflammatory response.
Methods
Visceral omental adipose tissue was obtained from obese women undergoing laparoscopic gastric bypass with Roux-en-y gastroenterostomy surgery for the treatment of extreme obesity in a clinical practice setting. The average body mass index [BMI] of the women whose fat was used for these experiments was 46.0, the age was 43.4 and the blood glucose was 5.4 mM. Each experimental replication involved tissue from a separate individual. Approximately one-third were taking anti-hypertensive agents and another third drugs for diabetes, but were fairly well controlled since the mean plasma glucose was 5.4 mM. The study had the approval of the local IRB and all patients involved gave their informed consent.
The adipose tissue was transported to the laboratory within 15-30 minutes of its removal from the donor. The handling of tissue and cells was done under aseptic conditions. The tissue was cut with scissors into small pieces (5-10 mg) and incubated in buffer [3 ml/g of tissue] for approximately 2-5 min to reduce contamination of the tissue with blood cells and soluble factors. The tissue explants were then centrifuged for 30 sec at 400-× g to remove blood cells and pieces of tissue containing insufficient fat cells to float.
Fat and nonfat cells were isolated by incubating 1.0 g of cut adipose tissue in 2 ml of incubation medium containing 1.3 mg of collagenase in a rotary water bath shaker [100 rpm] for two hours. The collagenase preparation was isolated from Clostridium histolyticum (Type 1) and obtained from Worthington Biochemical Corporation of Lakewood, NJ (lot CLS1-4197-MOB3773-B, 219 U/mg). The collagenase digest was then separated from undigested tissue by filtration through 200 μm mesh fabric. Five ml of medium was then added back to the digestion tubes and used to wash the undigested matrix on the filter mesh. This wash solution was combined with the collagenase digest and stromovascular [SV] cells were separated from fat cells and medium by centrifugation in 15 ml tubes for 1 min at 400-× g. The SV cells and fat cells were each suspended in 5 ml of fresh buffer and centrifuged for 10 sec at 400-× g. This medium was removed. The undigested tissue retained on the nylon mesh and the SV cells were combined to obtain the nonfat cells. One gram of adipose tissue explants, the nonfat cell fractions or fat cells obtained by digestion of 1 g of tissue were incubated in a volume of 5 ml for the indicated times. The average diameter of the isolated omental fat cells was 107 microns.
The buffer ordinarily used for incubation of adipose tissue was Dulbecco's modified Eagle's medium/Ham's F12 (1:1, Sigma-Aldrich No. 2906) containing 17.5 mM of glucose, 121 mM of NaCl, 4 mM of KCl, 1 mM of CaCl2, 25 mM of HEPES, 22 mM of sodium bicarbonate, 10 mg/ml of defatted bovine serum albumin [unless otherwise stated], 90 μg/ml of penicillin G, 150 μg/ml of streptomycin sulfate and 55 μM of ascorbic acid. The pH of the buffer was adjusted to 7.4 and the buffer filtered through a 0.2 μm filter. IL-8 and TNFα release to the medium was determined using ELISA assays with Duoset reagents from R & D Systems of Minneapolis, MN. Defatted bovine serum albumin powder prepared by heat treatment of serum plus organic solvent precipitation (Bovuminar, containing <0.05 moles of fatty acid/mole of albumin) was obtained from Intergen (Purchase, NY). The low endotoxin bovine albumin was prepared by a similar procedure [#A2934] and obtained from Sigma-Aldrich of St. Louis, MO.
For studies involving mRNA isolation, the nonfat cells, fat cells or tissue were separated from the medium and RNA extracted by Polytron homogenization as described by Chomczynski and Sacchi [
18] using 5 ml of a monophasic solution of phenol and guanidine isothiocyanate [Trlzol reagent from Invitrogen of Carlsbad, CA]. The extracts were then spun at 12,000-× g for 10 minutes at 2 to 8°C to separate the fat from the extract.
The assay of mRNA involved real-time qPCR [
19,
20]. The cDNA was prepared using the Transcriptor First Strand cDNA synthesis Kit from Roche Diagnostics. The quantification of mRNA was accomplished using the Roche Lightcycler 480 Real-time RT-PCR system and their Universal Probe Library of short hydrolysis Locked Nucleic Acid [LNA] dual hybridization probes in combination with the primers suggested by their web-based assay design center
http://www.universalprobelibrary.com. Integrated DNA Technologies of Coralville, IA, synthesized the primers. In each assay 70 ng per tube of total RNA [determined by absorption at 260 nm in a spectrophotometer] was used and the ratio of the right to left primers was 1 for each assay. The data were obtained as crossing point values [Cp] obtained by the second derivative maximum procedure as described by Roche Applied Science technical notes LC10/2000 and 13/2001
http://www.roche-applied-science.com/sis/rtpcr/htc/index.jsp. The Cp values are comparable to crossing threshold [Ct] values as defined by ABI or quantification cycle [Cq]
http://www.rdml.org. Samples with higher copy number of cDNA have lower Cp values, while those with lower copy numbers have the reverse.
The data were normalized by either the use of cyclophilin mRNA as the recovery standard/calibrator/reference gene or total RNA concentration as recommended by Bustin [
21]. The Cp values for cyclophilin A were the same in the nonfat cells as in the fat cells derived from omental adipose tissue [Cp = 28.9 ± 0.3 as the mean ± sem with n of 41 for nonfat cells and 28.5 ± 0.4 for fat cells] while that in unincubated omental adipose tissue was 29.0 [
19]. However, over a 24 or 48 h incubation there were significant increases [2.1× at 48 h] in cyclophilin A, so for time course studies the absolute Cp values were used [
21]. In this case the ratios were calculated from the ΔCp between unincubated tissue and tissue incubated for a particular time. Relative quantification of the data was calculated using the comparative Cp method, which eliminates the need for standard curves. The arithmetic formula to calculate ratios from ΔCp is based on a log
2 scale [2
-ΔCp]. This method is identical to the Comparative C
T procedure described in the ABI PRISM 7700 Sequence Detection System user Bulletin #2 for quantitative RT-PCR. The calculation of ratios was done without an efficiency correction by assuming that the number of target molecules doubles with every PCR cycle. Caution should be used in comparison of the Cp values between different genes because of the relative efficiencies of the particular primers and probes used for each gene may be different.
A two-tailed Student t-test was used to determine whether differences were significant at a P-value of < 0.05. Statistical analysis of mRNA values was based on the ΔCp values before log2 transformation to ratios.
Discussion
In mice, given enough lipopolysaccharide to kill 40% of the mice by 24 h, increases in MCP-1, IL-6, nerve growth factor, TNFα and HIF-1α were seen in adipose tissue within 4 h [
24]. Furthermore, there was a marked increase in HIF1α protein accompanied by even greater changes in mRNA [
24]. It is unclear how endotoxin elevates HIF-1α in the fat of mice and this could be independent of hypoxia. We observed a similar rapid increase in HIF1α and NFκB
1 expression simply by incubating human adipose tissue explants in vitro.
While albumin enhanced the release of IL-8 and its gene expression, it did not affect the early increase in the inflammatory response as judged by increases in expression of HIF1α or NFκB1. Furthermore albumin effects were primarily due to factors other than endotoxin contamination, which is in agreement with the findings of Schlesinger et al [
22]. These investigators found that while 2% bovine albumin, but not 0.7%, significantly stimulated the release of IL-6, IL-8 and TNFα by freshly isolated human adipocytes. However, albumin had much greater effects on in vitro differentiated human adipocytes [
22]. Exactly what accounts for the effects of albumin is unclear but albumin is able to bind many non-polar molecules and can bind up to 7 moles of fatty acid per mole of albumin [
25]. The albumin we used was isolated by a heat-shock process in the presence of octanoic acid resulting is a low fatty acid content, less than 0.05 moles/mole, but it is unclear whether this small amount of fatty acid can account for the effects. Traditionally adipose tissue or fat cells are incubated in the presence of 1 to 4% albumin to bind fatty acids released during lipolysis [
23]. This is done because lipolysis by rat fat cells is inhibited in the absence of albumin to bind fatty acids released during lipolysis [
26]. Albumin has been shown to influence inducible nitric oxide synthase in macrophage and smooth muscle cells [
27] and induce an inflammatory response in proximal tubular cells [
28]. While what is responsible for the inflammatory effect of albumin remains to be established, it had no effect of the increases in HIF1α or NFκB1 expression suggesting that albumin effects are exerted at a step between their activation and that of enhanced IL-8, TNFα, and IL1-β gene expression.
Whether the inflammatory response seen when adipose tissue is incubated in vitro is due to relative hypoxia secondary to cutting the blood supply remains to be established. Trayhurn et al [
29] have emphasized the pervasive effects of hypoxia on the inflammatory response of adipose tissue in obesity. The present results are compatible with this hypothesis as an explanation for the inflammatory response seen when human omental fat explants are incubated in vitro. The effects of hypoxia in tissues appear to be mediated in part through HIF1α, which is a major transcription factor that responds to hypoxia [
6,
29]. While initial studies on the role of HIF1α suggested that activation was primarily translational control of its proteolytic degradation, more recently HIF1α gene activation has been shown to play a role [
30]. Hypoxia activates other transcription factors and one of them is NFκB
1, which is also what we observed in human adipose tissue. The gene expression of both HIF1α and NFκB1 was elevated after only a 20 minute incubation of adipose tissue but at that time other inflammatory response genes were also activated making it impossible to determine a causal relationship. The finding that HIF1α mRNA up-regulation was far greater in intact adipose tissue explants than in nonfat cells or isolated fat cells suggests that incubated tissue is a more hypoxic environment. However, we did not measure HIF1α protein whose altered rate of degradation in the presence of hypoxia is the primary regulator of the inflammatory response.
The 9-fold up-regulation of HIF-1α mRNA over a 48 h incubation of nonfat cells isolated from omental adipose tissue is comparable to what Gesta et al [
31] reported using explants of human subcutaneous adipose tissue. They suggested that this was due to the relative hypoxia of tissue explants and accounted for the increase in TNFα mRNA. For reasons that are unclear, they found a different time course for TNFα in that the maximal increase in TNFα mRNA was seen at 48 h while we previously reported an increase that was maximal at 4 h and declined over the next 44 h [
32].
The inflammatory response as measured by accumulation of IL-1β, TNFα and IL-8 mRNAs was seen in both fat and nonfat cells of human omental fat. However, the increases in the nonfat cells for IL-1β, TNFα and IL-8 obtained after 2 h isolation procedure were identical to those seen when intact tissue was incubated for the same amount of time. This indicates that collagenase digestion is not responsible for the up-regulation as initially suggested by Ruan et al [
17]. The expression of IL-1β, TNFα and IL-8 was rather less in fat cells than was seen in the nonfat cells in agreement with studies on the release of TNFα and IL-6 over a 4 h incubation where the nonfat cells accounted for over 90% of total release [
32].
The role of fat cells as primary triggers for the inflammatory response in nonfat cells of adipose tissue could not be determined during the first 2 hours of incubation because it took that long to separate fat cells from nonfat cells. However, we found that the subsequent incubation of the nonfat cells with the fat cells for 48 h had no effect on up-regulation in fat cells. If there is paracrine cross-talk between fat cells and nonfat cells, it clearly has little influence upon the inflammatory response with respect to IL-8 since it was seen to the same extent in isolated fat cells, isolated nonfat cells or intact tissue explants. But we cannot exclude the importance of paracrine interactions between fat cells and the nonfat cells of omental adipose tissue prior to the start of the incubation during the time required to isolated the fat cells by digestion with collagenase. Our data also do not exclude cross talk between factors released by macrophages and other cells in the nonfat cell fraction.
One finding of interest was that while 11β-HSD1, which is initially enriched in fat cells by 4-fold, is up-regulated over 48 h by 9-fold in the nonfat cells but not in the fat cells. 11β-HSD1 is thought to be involved in the conversion of cortisone to cortisol and elevated levels of cortisol are associated with hypertension and insulin resistance [
33]. Furthermore, 11β-HSD1 gene expression is enhanced in visceral obesity [
34], which could contribute to insulin resistance by enhancing local conversion of cortisone to cortisol [
33,
34].
It should be noted that all the data were obtained with samples of omental adipose tissue from extremely obese women and whether the findings are applicable to fat from men and/or non-obese women remains to be established. Furthermore, the protein levels may not correlate as well with gene expression levels as they did with IL-8 and TNFα.
There is a growing consensus that massive obesity is accompanied by an inflammatory response in adipose tissue [
5‐
12] and that this is primarily due to visceral obesity [
4]. This can be mimicked in vitro by incubating explants of human omental fat from severely obese women and results in a rapid inflammatory response that can be seen within 20 minutes with respect to gene expression of inflammatory response proteins such as HIF1α and NFκB as well as inflammatory adipokines such as TNFα and IL-1β, and IL-8. Enhanced release of IL-8 could be seen after a 40-minute lag period and the present results provide further support for the hypothesis that this primarily occurs in the nonfat cells. Exactly what it is about obesity that induces an inflammatory response in vivo is unclear but may well relate to relative hypoxia for the large fat cells. The initial trigger could be breakdown of large fat cells and/or enhanced release of factors such as fatty acids that recruit mononuclear cells into adipose tissue. IL-8 is a chemokine that could well be involved in monocyte recruitment and the accumulation of mononuclear cells in adipose tissue is enhanced in obesity [
35,
36]. It is probable that the majority of the release of adipokines by nonfat cells in human adipose tissue is due to macrophages and other mononuclear phagocytic cells. These adipokines could account for generalized inflammation secondary to the release of inflammatory factors into the circulation. These factors and/or enhanced release of fatty acids could be responsible for the development of hypertension and diabetes in obesity.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JF designed the experiments, analyzed the data and drafted the manuscript. PC carried out the laboratory studies and analysis of mRNA. DT and AM selected the donors, obtained the samples of fat and aided in the interpretation of the data. All authors read and approved the final manuscript.