Increasing evidence has established correlative and causative links between chronic inflammation and insulin resistance. To further explore the molecular pathways underlying obesity-induced insulin resistance in general, we have studied mouse models of genetic and high-fat diet–induced obesity/diabetes. We have found that many macrophage-specific or -enriched genes, exemplified by ADAM8, MIP-1α, MCP-1, MAC-1, F4/80, and CD68, are dramatically upregulated in WAT of obese mice. In WAT, these genes are predominantly expressed in the stromal-vascular fraction where macrophages reside. Additional histological data provide evidence for a significant increase of macrophage numbers. We have also demonstrated, through immunohistochemical methods, that macrophages, not granulocytes or lymphocytes, are the cells involved in adipose inflammation. Because we used multiple obesity models, our results suggest that macrophage inflammatory response in WAT is a general phenomenon associated with morbid fat-mass expansion.
Macrophage infiltration is the likely explanation for the increase of macrophage numbers, since it is well established that extravasations of leukocytes from circulation lead to accumulation of immune cells at the site of inflammation. The observed upregulation of chemotaxin genes, such as MIP-1α
, supports the conclusion. It was previously reported that stromal-vascular cells from obese mice express more TNF-α
mRNA than do those from lean controls (1
). We confirmed by ELISA that more TNF-α was released from an equal number of stromal-vascular cells from obese mice than from lean controls (data not shown). Since TNF-α is a potent activator of macrophages, it is reasonable to conclude that the macrophages accumulated in WAT in the obese state are functionally active.
Recently it was reported that preadipocytes could be converted to macrophages under certain conditions (26
). Our study indeed showed that TNF-α could trigger preadipocytes to increase expression of some inflammation genes, although not the three macrophage markers we have studied (Supplemental Figure , http://www.jci.org/cgi/content/full/112/12/1821/DC1). This finding provides some support for converted preadipocytes as a source for the macrophage activity in obese WAT. It is possible that a combination of conversion of local preadipocytes to macrophages and activation and recruitment of resident macrophages and circulating monocytes has contributed to the increased macrophage activity observed in obese WAT. Future experiments are necessary to determine the significance of each mechanism during the physiological increase of adiposity.
This obesity-induced macrophage accumulation in WAT appears to be highly correlated with, if not causative of, insulin resistance in various mouse models. This is most strikingly demonstrated by tracking of the expression level of the macrophage markers over time in WAT of mice on a diet of high fat content. While body weight increases steadily over time on high-fat diet, fasting blood insulin levels remained relatively unchanged until 16 weeks on high-fat diet, when hyperinsulinemia became apparent. The expression levels of the selected macrophage and inflammation genes were upregulated before the increase in circulating-insulin levels. These data indicate that macrophage activities occur after the increase of adiposity but before insulin resistance.
The dramatic upregulation of macrophage-related genes is mostly restricted to WAT; the mRNA expression of these genes was barely detectable and essentially unchanged in muscle and liver of the obese mice. In addition, there was little change of expression of these genes in lung and spleen in the obese state, further demonstrating that the observed inflammation was adipose-specific. In the progression study, after 26 weeks on high-fat diet, a significant upregulation of CD68 was observed in the liver, although the absolute expression level was much lower than that in fat. These data suggest that obesity-induced inflammation is initiated in WAT but may become systemic with the steady increase of adiposity or insulin resistance.
Several recent studies have shown that TZDs can attenuate macrophage activation in vitro and improve atherosclerosis in vivo, suggesting direct involvement of these agonists in anti-inflammatory activities, most likely mediated by PPARγ in macrophages (30
). Part of their inhibitory effect on macrophage activation may even be independent of PPARγ (34
). Here we show that rosiglitazone suppressed the increased expression of the inflammation genes in WAT of ob/ob
mice. We have also observed reduced macrophage numbers in ob/ob
WAT upon rosiglitazone treatment (data not shown). In this scenario, reduced inflammation is correlated with increased insulin sensitivity. Our results are consistent with the known anti-inflammation function of TZDs and suggest the possibility that TZDs improve insulin sensitivity at least partially by suppressing the inflammatory events in obese WAT.
Macrophage accumulation is likely a direct response to the abnormal fat metabolism caused by the increasing adiposity. The molecular signals that trigger the macrophage activity in obese WAT are not yet known, but several good candidates exist. Adipocytes are known to secrete hormones, cytokines, and FFAs, most of which have been shown to play some role in inflammation and systemic insulin resistance. Adiponectin, known to improve insulin sensitivity, is downregulated in obese/diabetic states (37
). Adiponectin can inhibit adhesion of macrophages to endothelial cells, an essential process in the pathogenesis of atherosclerosis (38
). The decrease in adiponectin therefore might have contributed to the increased macrophage activity in obese adipose tissue. Leptin, which increases in proportion to fat mass, promotes cholesterol ester synthesis in macrophages in a hyperglycemic environment, an important process in the formation of foam cells in atherosclerosis (25
). This would suggest that lack of leptin signaling in the ob/ob
models might have some protective effect against the inflammatory response. This is consistent with our observation that some inflammation genes are even more dramatically upregulated in DIO mice than in leptin-deficient mice, although more detailed studies are required to address this issue. Complement factor C3, the precursor of C3a, a potent activator and chemotaxin of macrophages (40
), could also play a role. Indeed C3 is highly expressed in adipocytes and has been related to obesity and insulin resistance previously (41
). Another interesting candidate is MCP-1, an essential signal for macrophage activation and recruitment (44
). In our own studies, increased MCP-1 expression in WAT both preceded (at 3 weeks on high-fat diet) and was concurrent with (by 16 weeks on high-fat diet) insulin resistance in obese mice, supporting a possible causative role of this chemokine in the observed macrophage infiltration.
Once macrophage activation and infiltration is initiated, it is reasonable to expect that this macrophage-mediated inflammatory response would lead to impaired insulin response in adipocytes. Macrophages, upon activation, secrete numerous cytokines and chemokines, such as TNF-α, IL-1, IL-6, and MCP-1, that are known to cause insulin resistance in adipocytes (2
). These cytokines and chemokines further activate macrophages to increase lymphokine production and secretion. These feedback loops are schematically summarized in Figure . As a consequence, insulin signaling in adipocytes could become increasingly impaired, eventually leading to massive adipocyte lipolysis, necrosis, and systemic insulin resistance.
Figure 7 Hypothetical model of chronic inflammation and adipocyte insulin resistance. When adiposity reaches a certain threshold, factors derived from adipocytes induce macrophage activation and infiltration. Activated macrophages secrete cytokines that can impair (more ...)
It remains to be shown how the inflammatory response initiated in WAT ultimately causes systemic insulin resistance. Increased lipolysis is one possible piece of the puzzle. The observed increase of lipolysis in obese WAT could result in the release of a large amount of FFAs, and an increased FFA level in the circulation has been shown to result in resistance to insulin signaling in skeletal muscle and liver (48
). Therefore, it is possible that FFAs are an important link between chronic adipose inflammation and systemic insulin resistance. Additional work will be required to explore the molecular mechanisms involved in the initial inflammation response and the subsequent insulin resistance in WAT, and how they might relate to defective response to insulin in other peripheral tissues.