To our knowledge, this is the first detailed analysis of a mouse model that directly supports the hypothesis that enabling a massive expansion of the subcutaneous adipose tissue mass potently counteracts the strong trends toward the development of insulin resistance associated with excess caloric intake. We have chosen to address this issue in the ob/ob mouse model. These mice displayed hyperphagia and early onset obesity, resulting in hyperglycemia, hyperinsulinemia, and dyslipidemia. We achieved a normalization of all metabolic parameters through a modest overexpression of adiponectin. We were employing a transgene under the control of the fat cell–specific aP2 promoter that we previously characterized in the background of WT mice. This transgene encodes a version of adiponectin that carries a deletion within the collagenous domain of the protein, leading to the formation of a subset of mixed WT/mutant complexes. The detailed mechanism of action of these mutant protein complexes is currently under study in our laboratory. The net effect is an improved efficacy of the secretory pathway with respect to adiponectin assembly and release, leading to a 2- to 3-fold elevation of steady state levels of WT adiponectin complexes in plasma.
We challenged these mice metabolically by breeding them into the ob/ob background. The leptin deficiency induced hyperphagia. This resulted in a gradual increase in adipose tissue mass. Under these conditions, this expansion of the fat mass was associated with a reduction of circulating adiponectin levels. However, in the transgenic mice, this obesity-induced downregulation did not occur, but instead, there was a constitutive elevation of both intracellular and plasma adiponectin levels despite a continued increase in fat mass. The end result of this chronic elevation of adiponectin in transgenic ob/ob mice was a very significant expansion of the fat mass.
Phenotypically, this resulted in improvements in all metabolic parameters examined as they relate to glucose and lipid metabolism. Concomitant with the metabolic improvements, there was a positive impact on the inflammatory profile. The underlying mechanisms for these improvements are complex, but are likely related to the general increase in the local activity of PPARγ in adipocytes. Ultimately, this increased PPARγ activity in adipocytes resulted in a redistribution of lipids from ectopic deposits in liver and muscle to the subcutaneous adipose depots. As a result, hepatic insulin sensitivity increased, translating into systemic improvements in insulin sensitivity and preservation of β cell mass.
This suggests a role for adiponectin as a “starvation signal” released by the adipocytes, providing a systemic indication that the average adipocyte size is small and that adipose tissue is in need of accumulating higher levels of triglyceride. We have previously examined seasonal variations of adiponectin levels in yellow-bellied marmots and found that adiponectin levels are highest during the summer and fall months, a time when the marmots massively increase their fat stores (17
). In addition, Froguel and colleagues recently identified a nucleotide polymorphism in the gene encoding adiponectin that is associated with severe forms of childhood obesity and, importantly, increased levels of adiponectin in the plasma of these children (18
Adiponectin is consistently upregulated in the lean state with further elevation in anorexic states (19
) and correspondingly downregulated in overweight and obese states (21
). This type of regulation is a fundamentally different from the regulatory mechanism in place for leptin. Leptin levels tend to be directly proportional to fat mass and do not display the inverse relationship that adiponectin displays with adipose tissue mass (23
). Generally, adiponectin and leptin are regulated in an opposite fashion under many different physiological states (24
). Hence, elevated adiponectin levels go hand in hand with low leptin levels and may send comparable systemic signals. As an example, we have previously reported that transgenic overexpression of adiponectin in WT mice (under normal conditions a reflection of limited systemic energy stores) causes infertility in females, a phenomenon that can also be triggered with low leptin levels (25
Much excitement has been created by the original observation that obesity is associated with increased infiltration of macrophages into the growing fat pads. Several different models have been proposed to mechanistically explain why macrophages infiltrate obese fat pads with higher frequency. Monocyte chemoattractant protein–1 (MCP-1) and its receptor C-C motif chemokine receptor-2 (CCR2) have recently been implicated in this process (26
). MCP-1 is produced at higher levels in obese fat pads and hence attracts a higher number of macrophages (28
). An alternative model has recently been proposed by Cinti and colleagues (29
). These authors suggest that adipocytes that reach a maximal size upon lipid loading spontaneously undergo necrosis, which is associated with increased infiltration of macrophages around the dying adipocyte. Our massively obese mice do not allow us to differentiate between these 2 models. We report a reduction in both average adipocyte size and local inflammation. However, our transgenic line supports strongly the hypothesis that macrophage infiltration is not a function of absolute adipose tissue mass but rather relates to the “quality” of the individual fat cell in the adipose pad. Transgenic overexpression of adiponectin led to a massive expansion of adipose tissue mass, yet the level of macrophage infiltration was quite minimal, likely due to the fact that these mice showed hyperplasia but not hypertrophy. We conclude that macrophage infiltration is clearly not a function of adipose tissue quantity alone but rather a reflection of the quality of the individual adipocyte.
Adiponectin overexpression also has an impact on lipid levels. This is of interest in the context of ob/ob
mice. In contrast with humans, HDL cholesterol levels are increased in genetic mouse models of obesity (ob/ob
) due to their decreased clearance rates (9
). The appearance of large apoE-rich HDL-1 particles in these animals suggested a normal lipidation process, thus implying that the decreased clearance can account for the increase in total HDL levels in ob/ob
mice compared with WT mice (31
). In the present study, adiponectin-overexpressing ob/ob
mice exhibited significantly lower total plasma cholesterol levels than the ob/ob
controls (data not shown). Gel filtration analysis of lipoprotein profiles show markedly decreased HDL-1 and HDL peaks. More importantly, the lipoprotein profiles of adiponectin-overexpressing ob/ob
mice closely resemble the profiles of WT mice, suggesting that adiponectin normalizes HDL concentration and HDL particle size in ob/ob
mice. The adiponectin-related effects on HDL levels are consistent with a general improvement seen for the lipid profiles in the context of rodent obesity. This is not a peculiarity of adiponectin overexpression in the ob/ob
background, since we found that adiponectin overexpression also further reduces HDL-1 and HDL peaks in WT mice (our unpublished observations). This finding constitutes a novel observation that is consistent with the increased activity of LPL in adipose tissue of adiponectin-overexpressing mice. This is likely to be responsible for increased triglyceride clearance from circulation after an oral lipid challenge. Correspondingly, we demonstrate that triglyceride concentrations were decreased in the VLDL fractions in the presence of excess adiponectin.
The differences in HDL size and plasma concentrations also suggest a role for adiponectin in HDL metabolism. Although the detailed molecular mechanisms for the decrease in HDL cholesterol of adiponectin-overexpressing mice are not known, it is likely that HDL turnover is accelerated in adiponectin-overexpressing mice. Factors known to affect HDL turnover are under current investigation. Important differences in lipoprotein metabolism between humans and mice have to be taken into account though. Epidemiological data suggest a strong correlation between HDL levels and adiponectin levels (33
). In light of this, the relationship of adiponectin overexpression to HDL-cholesterol levels and atherosclerosis in mice is paradoxical. While we see a marked reduction of HDL-cholesterol levels in this model of adipocyte-derived overexpression of adiponectin, hepatic overexpression of globular adiponectin reduces atherosclerosis and neointima formation after arterial injury (35
). These apparently contrasting findings may be due to the ability of adiponectin to facilitate reverse cholesterol transport. Previous studies with mice overexpressing scavenger receptor class B type 1 (SR-B1) in liver demonstrated marked increases in reverse cholesterol transport and significantly reduced HDL cholesterol levels (37
) but resulted in reduction of atherosclerosis (38
). Hence, this leads to a model of adiponectin action that promotes reverse cholesterol transport triggering beneficial effects on atherogenesis without raising HDL cholesterol levels. Furthermore, recent clinical studies also supported a strong correlation of plasma adiponectin levels with plasma LPL activity and an inverse correlation with hepatic lipase activity independent of insulin resistance or inflammation (39
). Combined, our findings strongly support an integral role of adiponectin in lipoprotein metabolism beyond its well-known role in insulin sensitization (6
To study the metabolic phenotype of the transgenic animals in more detail, we used calorimetric cages. Metabolic cage studies revealed that transgenic ob/ob
mice were not hyperphagic compared with their ob/ob
controls. In fact they ate much less when adjusted for total body weight. This is in contrast to what one would expect considering their obese phenotype and considering the fact that they lack leptin. The massive obesity may be unleashed because leptin and adiponectin exert antagonistic effects. Lack of one of these adipokines combined with overexpression of the other leads to the phenotype described here. We have previously shown that mice lacking functional leptin are hyperresponsive to an injection of recombinant adiponectin (41
). Even though food intake was similar between both groups, there are clear metabolic differences. Core temperature as well as activity levels were considerably lower in transgenic ob/ob
mice. The reduced activity levels could be explained in part by the massive obesity. Consistent with these observations, oxygen consumption was also decreased compared with controls. RER levels were similar overall. However, at the beginning of the dark period, RER levels were increased in transgenic mice, which may reflect increased carbohydrate utilization, which spared triglyceride oxidation and stimulated the accumulation of adipose tissue. To investigate whether the animals had problems switching from carbohydrate to fat utilization as a main source of energy, the animals underwent a longer-term and short-term metabolic challenge by fasting and by administration of central 2DG, respectively. In general, the transgenic animals showed an altered response to fasting, switching more rapidly to FFA-based metabolism. The transgenic mice lost considerably less weight upon food deprivation, an indication that they had difficulty dipping into their triglyceride stores during caloric deprivation. This was particularly apparent in female mice, which developed a more severe form of hypoglycemia than males. Consistent with that, mRNA levels encoding gluconeogenic enzymes such as PEPCK and G-6-P were reduced in the females, which displayed a significantly higher degree of adiponectin overexpression than the males (6
). Nevertheless, the reduced body core temperature and reduced motility enabled both males and females to sustain glucose levels allowing them to survive the fast, despite the strong trend toward a metabolically inappropriate preservation of fat mass during times of starvation.
An additional striking observation was the total lack of diurnal rhythmicity of several metabolic parameters in the transgenic animals, which was restored to the level of ob/ob
mice during an acute fast. These changes suggest that adiponectin interacts in an inhibitory fashion with circadian oscillators in multiple central and peripheral tissues involved in maintaining energy homeostasis (43
While factors such as the maintenance of body core temperature, oxygen consumption, and locomotor activity are centrally regulated and were clearly altered in the transgenic mice, other aspects of central regulation of glucose homeostasis were normalized. Recent data has demonstrated the powerful effects of central hypothalamic regulation on hepatic glucose homeostasis (44
). In the transgenic mice, these central regulatory mechanisms were restored to the level of WT controls. Central infusion of minute amounts of 2DG offered a powerful challenge, effectively mimicking acute peripheral hypoglycemia. When performed in WT animals, this triggered release of glucose from the liver, a process which was effectively restored in the transgenic animals, while ob/ob
animals failed to respond peripherally to this central starvation signal.
In summary, we have described a model of chronic overexpression of adiponectin that leads to a massive increase in subcutaneous fat mass and protection against diet-induced insulin resistance. We believe that this mouse model is an excellent preclinical model to study the effects of adipose tissue expansion in the subcutaneous region in patients as well. There is the well-appreciated fact that not all obese patients are insulin resistant. Even more dramatically, not all morbidly obese patients display insulin resistance, and an elevation of adiponectin levels is frequently associated with improved metabolic profiles (47
). Hepatic steatosis is often associated with systemic insulin resistance (5
). The mechanism of action of antidiabetic compounds such as PPARγ agonists heavily relies on the ability of these ligands to reduce hepatic lipid content associated with a concomitant increase in subcutaneous fat mass. In the context of PPARγ agonists, this effect critically depends on their ability to induce adiponectin (48
). Independently, Cooper and colleagues have demonstrated the potent antisteatotic effects of adiponectin in the liver (50
). Furthermore, our analysis here demonstrates the closely linked gene expression pattern in white adipocytes from adiponectin-overexpressing adipocytes with the transcriptional hallmarks seen for PPARγ agonist treatment in the same cells. This parallel between adiponectin overexpression and PPARγ agonist action warrants further study. There is a striking overlap between the 2 conditions. PPARγ activity is increased by adiponectin, and adiponectin increases PPARγ activity. This demonstrates for what we believe is the first time a remarkable feed-forward loop between adiponectin and PPARγ, which appears to be functional only when one or the other is constitutively stimulated (transgenic adiponectin overexpression or pharmacological PPARγ activation). Under more physiological conditions (e.g., intact ob/ob
mice), this loop appears to be interrupted by unknown mechanisms.
All of these observations strongly suggest that adiponectin is a central player antagonizing the metabolic axis of evil involving obesity, hepatic lipid deposition, and local inflammation, leading ultimately to systemic insulin resistance. Therefore, the inability to sustain elevated adiponectin levels during times of excess caloric intake leads to reduced lipid deposition in the subcutaneous region, causing ectopic fat accumulation in liver and associated insulin resistance. Type 2 diabetes could therefore be viewed as a failure to appropriately expand fat mass in the context of a positive energy balance (51