Adipokines as novel biomarkers and regulators of the metabolic syndrome


  • Yingfeng Deng,

    1. Touchstone Diabetes Center, Department of Internal Medicine, University of Texas Southwestern Medical Center, Dallas, Texas.
    Search for more papers by this author
  • Philipp E. Scherer

    Corresponding author
    1. Touchstone Diabetes Center, Department of Internal Medicine, University of Texas Southwestern Medical Center, Dallas, Texas.
    2. Department of Cell Biology, University of Texas Southwestern Medical Center, Dallas, Texas
      Address for correspondence: Philipp E. Scherer, Touchstone Diabetes Center, Department of Internal Medicine, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, TX 75390-8549.
    Search for more papers by this author


This article is corrected by:

  1. Errata: Erratum for Ann. N. Y. Acad. Sci. 1212: E1–19 Volume 1226, 50, Article first published online: 26 May 2011

Address for correspondence: Philipp E. Scherer, Touchstone Diabetes Center, Department of Internal Medicine, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, TX 75390-8549.


Over the past two decades our view of adipose tissue has undergone a dramatic change from an inert energy storage tissue to an active endocrine organ. Adipose tissue communicates with other central and peripheral organs by synthesis and secretion of a host of molecules that we generally refer to as adipokines. The levels of some adipokines correlate with specific metabolic states and have the potential to impact directly upon the metabolic homeostasis of the system. A dysregulation of adipokines has been implicated in obesity, type 2 diabetes, hypertension, cardiovascular disease, and an ever-growing larger list of pathological changes in a number of organs. Here, we review the recent progress regarding the synthesis, secretion, and physiological function of adipokines with perspectives on future directions and potential therapeutic goals.

Adipose tissue has been recognized as an active endocrine organ in addition to its role as the main storage depot for triglycerides.1 An increasing number of adipocyte-derived secretory factors (“adipokines”) are described in the literature,2,3 highlighting the central role of adipose tissue in regulating whole body energy homeostasis, not only by partitioning lipids into various depots, but also through adipokine-mediated modulation of a number of signaling cascades in target tissues. It is well-established that individuals that are obese and/or suffer from the metabolic syndrome display a characteristic imbalance of their adipokine profile. This altered adipokine profile leads to profound changes in insulin sensitivity and other biochemical alterations of metabolites, making an individual more prone to metabolic disorders. Through their autocrine, paracrine, and endocrine functions, adipokines influence a number of organs critical for energy homeostasis. The changes in each individual adipokine are the result of a coordinated change of specific transcriptional programs that affect entire groups of adipocyte gene products as well as posttranslational mechanisms that affect the release of specific proteins differentially.

Among these adipokines, adiponectin is one of the most potent molecules with respect to its insulin-sensitizing activity. However, unlike the vast majority of adipocyte-derived factors, the levels of adiponectin in circulation display an inverse correlation with adiposity. Given the established beneficial roles of adiponectin on whole body metabolism and its profound protective effects against many chronic diseases, a better understanding of the regulation of adiponectin secretion is very important. Here, we focus on the regulation of adiponectin secretion from the adipocyte as a paradigm of protein release from the secretory pathway of the adipocyte and the changes it undergoes in the context of obesity and other pathological settings. Beyond the mechanics of protein release, we will extend the discussion to other recent developments in the area of adipokines and their effects on metabolism.

The ever increasing list of adipokines: from molecular sensors, messengers to regulators of energy homeostasis

In light of its role beyond mere energy storage organ, adipose tissue has been the target of multiple studies focusing on the identification of secreted bioactive molecules. Numerous molecules have been identified with autocrine, paracrine, or endocrine functions and they are now generally referred to as adipokines.

A dysregulation of these adipokines is observed under conditions of both excessive adipose tissue as well as under conditions associated with a lack of adipose tissue, i.e. both under obese as well as under lipoatrophic states in which total fat mass exceeds or is insufficient for proper function.4,5

In contrast, overconsumption of nutrients and caloric restriction (CR) exert opposite effects on metabolic health and lifespan and have differential effects on the complement of secretory factors released from adipocytes. The normalization of the adipokine secretion profile associated with weight loss due to long-term CR correlates well with the normalization of metabolic parameters, consistent with the idea that adipokines play an important role as molecular messengers and regulators of whole body energy balance. The concerted change of adipokine expression is part of a coordinated but differential regulation of the release of each adipokine in response to altered metabolic conditions in each individual fat cell. What are the key adipokines that we should be focusing on? This is, at this point, still a difficult question to answer in light of the emerging functions of many of these factors. Although comprehensive lists of adipokines have been generated, we will discuss below only a subset of them and refer to other review articles in the field that offer more extensive overviews of the entire secretome of adipocytes.3,6,7 However, we have assembled some of the more important adipokines described to date in Table 1 with a brief annotation of postulated function(s). Generally, the majority of these adipocyte-derived factors falls into one of several major categories that include (I) factors directly affecting metabolism; (II) proinflammatory factors and acute phase reactants; (III) extracellular matrix components; and (IV) promitogenic and proangiogenic factors.

Table 1.  Adipokines3,8–36,203
AdipokineMetabolismProinflammatory factors and acute phase reactantsExtracellular matrix componentsPromitogenic and proangiogenic factors
adipocyte fatty acid binding protein (aP2)   
Apolipoprotein E   
Insulin-like growth factor 1 (IGF-1)  
Lipoprotein lipase   
Alpha 1 acid glycoprotein   
Macrophage migration inhibitory factor (MIF)   
Macrophage chemoattractant protein (MCP1)   
Serum amyloid A3   
Alpha 2 macroglobin   
Collagen I   
Collagen III   
Collagen IV   
Collagen VI   
Lysyl Oxidase   
Angiopoietin 1   
Angiopoietin 2   
Fibroblast growth factor (FGF)   
Hepatic growth factor (HGF)   
Nerve growth factor   
Stromal derived factor (SDF-1)   
Tissue factor   


Functioning as an insulin sensitizer, increasing circulating adiponectin bears great potential for therapeutic purposes. In the ob/ob mice as well as the type I diabetic NOD mice, administration of recombinant adiponectin even after the development of diabetes significantly ameliorated the hyperglycemia.8–10 Furthermore, adiponectin is critical for PPARγ agonists to develop their full antidiabetic potential, particularly after exposure to a high fat diet.11 As part of its beneficial roles, adiponectin is also generally considered to have antiinflammatory, antiapoptotic, and proangiogenic activities,12,13 with a detailed unifying mechanism of action still to be established.

Hypoadiponectinemia has been found in a variety of human metabolic and cardiovascular disease states including type 2 diabetes mellitus (T2DM), lipodystrophy, nonalcoholic hepatic steatosis, essential hypertension, and coronary artery disease even after body mass index (BMI) is matched. Genetic hypoadiponectinemia caused by a missense mutation has been reported. The patients carrying this mutation also exhibit a much higher propensity to develop the metabolic syndrome.14 As the decrease of adiponectin precedes the development of insulin resistance and myocardial infarction in humans, low levels of adiponectin are likely to be a causal component of those disorders. A study in Pima Indians showed that individuals with high levels of adiponectin were less likely to develop T2DM, suggesting high adiponectin concentration is a protective factor against development of T2DM.15 Similarly, reconstituting adiponectin levels back to normal with recombinant adiponectin in a mouse model of diabetes ameliorated the insulin resistance.10

Other adipokines

With the exception of adiponectin and adipsin (complement factor D), most other adipokines described to date show a positive correlation between their circulating levels and adipose tissue mass, i.e., their levels are increased in the obese state. Many of them act as inflammatory cytokines and are critical mediators of the adverse effects associated with excess adipose tissue. Notably, some of these inflammatory factors directly inhibit adiponectin production and release in an autocrine fashion beyond other negative effects, thereby exerting their negative impact at multiple levels. A few select examples of adipokines are briefly discussed below.


As leptin resistance usually develops with increased leptin levels, we should view obesity as a state of reduced leptin function. Leptin exerts the bulk of its metabolic effects centrally.16 In fact, restoring leptin receptor function in the brain in the background of a db/db mouse (i.e., a complete absence of leptin receptor function in the periphery) causes a normalization of the metabolic phenotype,17 strongly arguing for the importance of central leptin action. However, there are clear peripheral effects as well that include interactions with immune cells, proangiogenic cells as well as a direct involvement in mammary tumor growth.18


Resistin is an exciting molecule and the founding protein of an entirely novel family of polypeptides that share a common higher order structure.19 To date, we do not understand the detailed functions of resistin or any of the other resistin-like molecules (RELMs). We appreciate that resistin can cause hepatic insulin resistance and that it may, along with its closely related homologs, interact with immune cells as well.20–22 As we still do not know the identity of the resistin receptor, we will have to await the further characterization of this signaling pathway to gain a better understanding of the function of this interesting factor.


RBP4 has been implicated in insulin resistance recently.23 It is secreted from both adipose tissue and the liver, and more prominently expressed in visceral fat depots compared to subcutaneous depots.24,25 Type II diabetic individuals have elevated levels of RBP4 in plasma along with elevated levels of transthyretin, a molecule that stabilizes RBP4 and extends its half-life.26 A number of studies have highlighted interesting correlations between RBP4 levels and plasma parameters in the context of the metabolic syndrome.23 Additional clinical studies will have to determine how useful the measurements of plasma levels of RBP4 will be as an indicator of general metabolic dysfunction.

TNFα and IL-6

These are factors that are upregulated in adipocytes undergoing proinflammatory stimulation. This can rank from high-level stimulation by bacterial lipopolysaccharide to subclinical inflammatory stimuli as frequently observed in the obese state. Although TNFα acts predominantly locally and cannot be measured at elevated levels in circulation under metabolically challenged conditions, IL-6 is released effectively from adipose tissue. In the case of visceral adipocytes, IL-6 is released into the portal vein where it is shunted directly into the liver.27 A significant gradient of IL-6 can be measured across the splanchnic bed,28 and IL-6 induces C-reactive protein production in hepatocytes.29


Visfatin is also known as pre-B cell colony-enhancing factor (PBEF) or Nicotinamide phosphoribosyltransferase (NAMPT) since it is the limiting enzyme in nicotinamide adenine dinucleotide (NAD) biosynthesis.30 Visfatin is expressd in leukocytes, adipocytes, muscle cells, and hepatocytes. In adipose tissue, however, it may be primarily the product of infiltrating macrophages as opposed to adipocytes. The research on visfatin remains relatively elusive. Although visfatin may frequently be upregulated in the obese state, its impact on insulin sensitivity and the underlying mechanisms are not clear at this point.


Omentin is more prominently expressed in omental fat depots.31 Omentin is found at lower levels in patients with glucose intolerance and diabetes. Although recombinant omentin enhances insulin-stimulated glucose uptake in adipose tissue, the molecular mechanism by which it achieves this beneficial effect remains to be worked out.

Apelin and VEGF

Apelin plays an important role in the regulation of blood pressure, may have an effect on several immune cells, and has proangiogenic properties.32 As such, it has similar functions to VEGF, another important proangiogenic factor. In light of the rapid expansion that adipose tissue can undergo in the content of caloric excess, along with constant remodeling even at steady state, proangiogenic factors play a major role in the preservation of appropriate nutrient and oxygen supply within the tissue. Generally, levels of these factors tend to be lower in the obese state, and the inability to appropriately upregulate these factors in response to the local hypoxia prevailing in adipose tissue is a significant contributor to local adipose tissue dysfunction.33

In summary, the growing list of adipokines is a reflection of the inherent heterogeneity of adipose tissue with respect to resident cell types. In addition to adipocytes, adipose tissue also contains preadipocytes, adipose tissue macrophages (ATM), additional immune cells, as well as fibroblasts and vascular constituents. The vast majority of adipose tissue-derived factors characterized to date are secreted from adipocytes or ATMs.34 Given the cellular source of these factors, it is not surprising that many of the adipokines exert a proinflammatory impact on the microenvironment. Therefore, obesity-associated chronic subclinical inflammation is, for the most part, due to direct release of these inflammatory factors from adipose tissue.35 In addition to the multiple cell types within a given fat pad, there is also heterogeneity among the different adipose depots due to differential anatomic distribution.36

Therapeutic potential of adipokines

Despite their potent physiological effects, there have been only a limited number of applications of adipokines as protein therapeutics. This is, at least in part, due to the relatively recent discovery of many of these factors, but is also a reflection of the complex physiological effects that many of these adipokines exert. Leptin is most advanced in this area with early beneficial applications in lipodystrophic patients.5 A combination of amylin treatment with leptin was shown to have synergistic effects on weight loss.37 More recently, intriguing preclinical data suggests that pharmacological leptin doses may exert beneficial effects for type 1 diabetics.38 Although there are widespread efforts to use other adipocyte-derived factors as protein therapeutics, there are no reports to date indicating that any other adipokine has moved into clinical trials.

Adipokines and adiposity

Secreted from adipose tissue, the adipokine profile is intimately linked to various parameters of adiposity (total body fat, percentage body fat, and fat distribution) that are not necessarily directly reflected in other parameters, such as BMI or waist to hip ratio. Generally, adipokine levels are positively correlated with fat mass (with very few exceptions, such as adipisin and adiponectin).

Genetic factors determining fat regional distribution and adipokine levels

The most obvious genetic factor determining adiposity is gender. There is no evidence for significant site- and sex-related differences in early development, although a striking difference in total fat mass and regional differences in adipose tissue distribution develop in childhood. Adult males have approximately 1.5 times the lean body mass of females, and females have on average twice as much body fat as males. Females generally have a characteristic gynecoid body fat distribution, with adipose tissue prominently developing in the subcutaneous depots around the hips and thighs; males, in contrast, have an android body fat distribution, with fat distributed prominently in the abdominal area.39 This difference is largely regulated by endocrine factors, with critical roles played by sex hormones.40 Striking sexual dimorphisms have been reported for multiple adipokines, including adiponectin and leptin in both rodents and humans.41 As adults, premenopausal females have about three times the circulating leptin concentration of males.42 Adiponectin levels are lower in males compared to females.43,44 This sexual dimorphism develops during puberty39,41 and studies in both human and rodents suggest that the inhibition by circulating androgens is one reason to explain the lower levels of leptin and adiponectin in males.45 Lower levels of adiponectin have been linked to a higher incidence of cardiovascular disease in males due to a diminished impact of the cardioprotective and antiatherogenic effects of this molecule in the context of hypoadiponectinaemia.46–48

Another underlying mechanism for the sexually dimorphic pattern of adipokine secretion is differential fat distribution. Adipokines and free fatty acids (FFAs) released from visceral fat depots have more potent effects on the liver compared to subcutaneous fat and this may explain some metabolic abnormalities in subjects with upper-body obesity,49,50 primarily due to the fact that the visceral fat depots drain their secretory components directly into the portal vein with immediate first pass effects on the liver. On top of the differential effects on other organs due to the anatomic location,51 there is a regional difference in protein expression profile by adipose tissue,52 which is reflected in the site-specific variations in adipokine production. Leptin mRNA levels as well as the rate of leptin release are higher in subcutaneous than in visceral adipose tissue.53,54 Additionally, only subcutaneous leptin production is correlated with circulating leptin levels.55 Similar results were also found for adiponectin as human adiponectin gene expression is lower in visceral compared to subcutaneous fat.56 Although the secretion of adiponectin from cultured human omental adipocytes was higher than from subcutaneous adipocytes, it was also found that the secretion of adiponectin from omental adipocytes showed a strong negative correlation with body weight, whereas secretion from the subcutaneous cells was unrelated to body weight.57 These differences in adiponectin production are, however, difficult to gauge, because neither mRNA nor intracellular protein levels are a reflection of the rate of protein release from the cell.

Regional expression differences have also been reported for other adipokines. Angiotensinogen has been reported to be higher in omental compared to visceral fat.58,59 Adipsin is also higher in omental compared to subcutaneous adipose tissue,58 and IL-6 is higher in visceral compared to subcutaneous fat.27 However, recent studies showed there was no regional variation in gene expression of TNFα in visceral and subcutaneous fat pads.54,58,60

The underlying mechanism for differential gene expression remains, however, still poorly understood. We appreciate that a differential induction of lipoprotein lipase upon PPARγ agonist exposure can lead to increased flux of lipids to subcutaneous tissues at the expense of visceral depots.61 Although it is clear that PPARγ and its coactivators and co-repressors are the main driving force for the differential gene expression, it is not clear how the differential distribution and activity of these transcriptional complexes are brought about and how they respond to different metabolic cues. These transcriptional changes in different fad pads cause a differential flux of metabolites to the different depots, and these altered fluxes are ultimately responsible for the specific structural changes observed under pathophysiological conditions as well as upon pharmacological intervention.

Visceral fat accumulation alters the circulating adipokine profile and is causally linked to metabolic disease and cancer

Obesity is a major risk factor for comorbidities associated with metabolic syndrome, including T2DM, stroke, hypertension, and cardiovascular disease.62 The increased risk of developing cancer in the context of obesity has also found widespread acceptance for various cancer types. In particular, prostate, postmenopausal breast, liver, kidney, colon, ovarian, and endometrial cancers show strong epidemiological connections to obesity.63–70 The underlying mechanistic connection to cancer is not well-established, but is likely to involve the interaction of local adipocyte-derived factors with tumor cells.71 In addition, general adipose tissue dysfunction in obesity results in increased ectopic fat distribution in several cell types that, in turn, may cause increased rates of ROS production, mitochondrial dysfunction and an elevated mutagenesis rate.

Visceral fat accumulation, rather than whole body adiposity, has been implicated in the development of diabetes, lipid disorders, hypertension, and astherosclerosis.72 Clinical studies suggest a strong association of visceral fat accumulation and liver steatosis in morbidly obese individuals.73 Notably, visceral fat accretion is also a “normal” phenomenon associated with aging in humans.74 In contrast, CR potently protects against obesity, T2DM, hypertension, atherosclerosis, and other aging-associated metabolic diseases.75 Whereas insulin resistance usually accompanies obesity and is the basis for many associated comorbidities,76 CR increases insulin sensitivity, which is the potential underlying mechanism for beneficial effects of CR.77

Using a mouse model of diet-induced obesity and T2DM, the impact of visceral fat removal on serum adipokine levels was assessed.78 The removal of visceral fat dramatically improved the impaired insulin signaling and normalized serum adipokine levels. In clinical studies, weight loss resulting from long-term CR restored the adipokine secretion profile and normalized the metabolic disorders associated with obesity. Kahn and colleagues found subcutaneous adipose tissue can correct metabolic phenotypes when transplanted to a visceral area, and this effect cannot be seen when visceral fat was used for transplantation.79

Consistent with the reduced levels of anti-inflammatory adipokines in visceral obesity, the circulating proinflammatory cytokine and acute phase reactant levels are higher in obese individuals and lower in patients on a CR regimen.80–82 Recent observations have implicated the increase of inflammatory factors in higher levels of adipocyte death in visceral fat pads. Interestingly, this is not seen in subcutaneous fat pads.83 The crown-like structures (CLS) frequently found in adipose tissue represent collections of cells comprising a dead adipocyte that is surrounded by a number of adipose tissue macrophages.84 The true functional role of CLSs remains not well-understood. However, the observations that visceral fat is richer in CLS density than subcutaneous fat,85–87 CLS prevalence is increased as much as 30-fold in obese mice and humans,84 and CLSs are associated with increased levels of circulating inflammatory adipokines88 suggest an increased rate of adipocyte cell death in visceral fat in the context of obesity. Apoptosis and necrosis of cells, including cell death of adipocytes, is invariably associated with tissue remodeling, which in turn again contributes to inflammation and local insulin resistance.86,89,90

An imbalance of adipokine levels leads to a further exacerbation of the situation. Inflammatory adipokines such as TNFα, IL-6, and IL-1β suppress the transcription of adiponectin in both subcutaneous and visceral fat tissues.27,91 The increased production of those inflammatory adipokines due to visceral fat accumulation exerts a further negative effect on the production of adiponectin.92

Adipokines regulate adiposity and metabolic homeostasis

Adipogenesis is initiated in utero and progresses until birth. The developmental stages between humans and rodents are distinct. In mice, expression of adipocyte markers can be seen as early as day 15 (with a total of 21 days of gestation time).93 In humans, the differentiation of fat tissue, as judged by the appearance of lipid laden cells, can be observed during the second trimester of gestation (between weeks 14 and 24 of gestation). Accumulation of adipocytes can be seen primarily around the head and neck, with a subsequent progression to the trunk. At the beginning of the third trimester, at about 28 weeks, adipocytes can be observed in most of the areas that give rise to the main fat pads in the adult.94 In both rodents and humans, the adipocyte development is extremely sensitive to nutrient availability in utero. A lack of nutrients during early development limits fetal fat formation, whereas bone growth and thus fetal size are less affected.95 Poor fetal nutrient exposure has been linked to subsequent development of insulin resistance and diabetes.96–98 The mechanisms underlying this “fetal programming”99,100 have not yet been established, but epigenetic changes are clearly one driving force for the long-lasting effects of suboptimal fetal nutrient availability. This also triggers an altered secretion profile of adipokines from the fetal or neonatal adipose tissue, with lower leptin, normal or lower adiponectin, and higher fetal/neonatal visfatin levels implicated as a result of suboptimal intrauterine conditions.101 The altered adipokine profile, on the other hand, may regulate fetal/neonatal fat development. Importantly, during the developmental stages that precede adipogenesis in utero, the developing fetus may be particularly prone to the lipotoxic effects that excessive maternal FFAs exert, since the fetus lacks that ability to properly buffer and neutralize these cytotoxic lipids.

Although our understanding of in vivo adipogenesis during development and in the adult stage is limited, isolated cell lines that can be differentiated in culture facilitate the study of adipocyte differentiation. The classical 3T3-L1 preadipocyte cell line was initially isolated by Green et al., which was the basis for later studies.102 3T3-L1 preadipocytes differentiate in vitro accompanying a striking change to the morphological and functional characteristics of fat cells in vivo. In the cultured preadipocytes, virtually all the adipokines start their transcription after the differentiation is initiated. Consequently, a number of defined factors including insulin, growth hormone, glucocorticoids, sex steroids, thyroid hormones, and indomethacin, which can accelerate adipocyte differentiation in vitro, also have a profound effect on adipokine secretion. Meanwhile, interferon, retinoic acid, IL1, and TNFα, which inhibit adipocyte differentiation, are also found to interfere with adiponectin secretion.103 The 3T3-L1 cell line has been invaluable as a tool to elucidate the basic transcriptional steps toward full adipogenic differentiation as well as some basic cellular processes, such as translocation of GLUT4 glucose transporters to the plasma membrane in response to insulin. Unquestionably though, the tissue culture system has major limitations. Under normal culture conditions, it does not allow one to study the complex phenomena associated with the three-dimensional growth of these cells. They lack sympathetic innervation, and a host of basic biochemical processes occur differentially in vitro versus in vivo. Although 3T3-L1 adipocytes accumulate lipid droplets predominantly through endogenous lipogenesis, primary adipocytes rely to a much smaller extent on lipogenesis. Furthermore, conventional culture conditions that use very high glucose levels introduce a host of artifactual changes of these cells during differentiation.104

Using 3T3-L1 preadipocytes, a number of interesting observations were reported about the role of adiponectin in adipogenesis. Overexpression of adiponectin using lentiviral approach in 3T3-L1 cell line led to enhanced adipocyte differentiation and lipid accumulation.105 This effect has been suggested to occur through an autocrine mechanism, mediated by secreted adiponectin as the conditioned medium from adiponectin overexpressing cells was capable of enhancing lipid accumulation in control cells. In contrast, a temporal relationship for the induction of adiponectin and the accumulation of lipid droplets during differentiation has been reported in 3T3-L1 preadipocytes.106 The negative correlation of intracellular adiponectin and lipid droplet formation is consistent with a model that suggests that cells containing small lipid droplets express higher levels of adiponectin. Obesity is characterized by an increase in the size of fat cells with excess lipid droplets, or a combination of both increase in cell size and cell number.107 In this respect, the paradoxical correlation between adiponectin and body fat mass essentially reflects an imbalance of large and small adipocytes, with obesity reflecting a bias toward generally larger fat cells with reduced adiponectin production.

The role of adiponectin in regulating adiposity is also examined in in vivo studies. Transgenic mice with increased circulating adiponectin levels have been achieved through an aP2 promoter-driven collagen domain truncated adiponectin, which enhances the secretion of endogenous adiponectin.108 These mice show an increase in circulating adiponectin that falls within the physiological range. Consistent with the positive role of adiponectin in regulating glucose and lipid metabolism, these mice display increased insulin sensitivity and enhanced metabolic flexibility of adipose tissue.109 The mice show a relatively normal weight in a wildtype background with lower levels of visceral adipose tissue and higher levels of subcutaneous and brown adipose tissue. In addition, the overexpression of collagen domain truncated adiponectin in ob/ob background led to striking obesity beyond that seen in the ob/ob background alone.110 The transgenic mice had much greater fat mass, particularly in the subcutaneous depots, but less relative visceral obesity than ob/ob mice and less triglyceride deposits in muscle and liver.

Regulation of adiponectin secretion

Unique features of adiponectin secretion

Adiponectin was first discovered in the 1990s, around the same time that leptin was first identified. After its original name “adipocyte complement-related protein of 30kD” (Acrp30),111 a number of additional groups used other names, such as AdipoQ,112 GBP28,113 and apM1,114 before its current name, adiponectin,115 became widely accepted. Adiponectin is produced predominantly by adipocytes. Adiponectin is an abundant, adipokine reaching 3–30 μg/mL in circulation.41 Its plasma concentrations are fairly stable with limited diurnal variability.116 Adiponectin clearance in rodents is unexpectedly rapid with a half-life of ∼45–75 minutes.117 The plasma abundance, in addition to a relatively short half-life of circulating protein, reflects an extremely high level expression of adiponectin in adipocytes. Of note is the inverse correlation with adiposity, especially visceral obesity.

Another unique feature of adiponectin is the fact that it is released in multiple forms from adipocytes. There are three major complexes: homotrimer, low molecular weight complexes consisting of two covalently linked trimers (LMW), and high molecular weight complexes built from six trimers (HMW).118,119 Impaired adiponectin multimerization leads to defects in adiponectin secretion.120 During weight-loss from long-term CR, total circulating adiponectin increased with the HMW form increased more than the other forms, whereas the decrease of plasma adiponectin in obese patients reduces the levels of HMW complex more than the trimer and LMW adiponectin. The different adiponectin complexes display distinct biochemical characteristics and exert non-overlapping biological functions.121,122 Interestingly, the HMW complex has the most potent insulin-sensitizing activity of all the complexes.123 Clinical studies concur that the relative ratio of the HMW form of adiponectin, rather than the total adiponectin level itself, correlates more significantly with key features of metabolic health.124–126 Biochemical analysis of purified complexes and in vivo studies suggest that different forms of adiponectin do not interconvert after secretion.121 Taken together, the multimerization of adiponectin is a critical step in regulation of adiponectin secretion as well as its function in insulin sensitization.

Beyond a general enhancement of adipogenesis,105 adiponectin has been suggested to play a positive role in regulating the distribution of subcutaneous and visceral fat.110 Although there is no data at this point to directly support this hypothesis, we suspect that this may be linked to differential intracellular accumulation of adiponectin in the different fat pads. Adiponectin accumulates at very high levels in the endoplasmic reticulum. In this process, adiponectin may exert an “intracrine” role, i.e., it may exert its effects intracellularly, eventually leading to differential activation of PPARγ in the different depots. This will, however, have to await further experimental support. In addition, it is very likely that adiponectin exerts important central effects that may trigger peripheral responses through activation of the sympathetic nervous system. Qi et al. have shown that central adiponectin administration can stimulate energy expenditure.127 They also observed that the adiponectin effects in the absence of functional leptin are more long-lasting and more profound, suggesting that there is an antagonistic relationship between these two hormones under some circumstances. Peripheral adiponectin, particularly its trimeric form, can be found in cerebrospinal fluid;128 however, we cannot formally exclude expression of adiponectin in specific neuronal subpopulations.

Regulation of adiponectin production

Consistent with the high demand for adiponectin in circulation, a substantial amount of adiponectin protein and mRNA can be detected within the adipocyte. In fact, adiponectin transcripts are among the most abundantly expressed mRNAs in adipocytes.114 The adiponectin promoter harbors binding sites for several transcription factors including PPARγ,129 C/EBPα,130 and SREBP-1c,131 all of which can upregulate adiponectin transcription. Recent studies using adiponectin promoter show highly specific expression of transgene in rodents.132 Meanwhile, several repressors have been identified that assert a negative effect on adiponectin transcription either directly through DNA-binding competition with those activators or indirectly through downregulation of the concentration or activity of the activators. Synthesized as a single polypeptide of 30 kDa, different posttranslational modifications including glycosylation and hydroxylation are added to adiponectin and those modifications have been implicated in the control of adiponectin oligomerization and secretion.133 The regulation of adiponectin transcription and posttranscriptional modifications have been comprehensively reviewed recently.134 Among the posttranscriptional steps, multimerization is one of the key regulatory steps of adiponectin release. A series of chaperones are involved in the maturation of adiponectin to ensure correct formation of the tertiary and quaternary structure of the adiponectin complex.135 ERp44 and disulfide-bond A oxidoreductase-like protein (DsbA-L) are two of these chaperones found to interact with adiponectin to assists its multimerization. ERp44, an ER/Golgi resident chaperone, binds adiponectin through Cys-39. This interaction retains adiponectin intracellularly before further oxidative folding with the assistance of Ero1-Lα and protein disulfide bond isomerase (PDI).136 DsbA-L, a mammalian homolog of the primary oxidase DsbA within the E. coli periplasm, interacts with adiponectin to enhance the formation of HMW form and its release from adipocytes.137 However, an additional cofactor might be required in this process, as incubation with DsbA-L alone was insufficient to promote adiponectin multimerization in vitro.

As a secretory molecule, subcellular trafficking of adiponectin has been extensively studied ever since it was first discovered. GGA-coated vesicle-dependent trafficking of adiponectin secretion from the trans Golgi network (TGN) has been reported in 3T3-L1 adipocytes.138 Caveolin-1 containing vesicular structures may also be involved in trafficking and secretion.139 Microscopic studies have shown some ambiguous results about the subcellular localization of adiponectin in 3T3-L1 adipocytes. An early study reported that adiponectin has a relatively equal distribution throughout the endoplasmic reticulum (ER) with less perinuclear accumulation.140 However, another study later showed a predominant perinuclear localization of adiponectin that colocalizes with several TGN markers.138 The discrepancy may arise from the different approaches employed. The TGN staining pattern was mostly detected with a transient transfection of adiponectin, which may cause the TGN accumulation due to overexpression of adiponectin. Although the TGN pattern was also reported with endogenous adiponectin, this may change at different differentiation stages of cells, sampled. The synthesis and secretion of adiponectin starts between day 2 and 4 of the in vitro differentiation protocol, but a dynamic change in the secretion occurs during the subsequent days of maturation. Adiponectin secretion is shown to diminish from day 6 to day 10.141 In fact, when the population heterogeneity during adipogenesis of 3T3-L1 preadipocytes was taken into account with high magnitude microscopy imaging, several phenotypically distinct subpopulations of cells were observed as judged by differential expression of the adipogenesis markers PPARγ, lipid droplet formation and adiponectin levels.106 Adiponectin levels were visibly heterogeneous in individual differentiating cells, and an unexpected negative correlation between adiponectin and lipid droplet levels was also observed. This is consistent with the negative correlation that has long been seen between the plasma adiponectin and body fat mass.142 Importantly, a temporal sequence of the phenotypic changes of the identified subpopulations has been proposed, in which the accumulation of intracellular adiponectin occurs early and is then reduced in the completely differentiated cells that contain more lipid droplets. These studies conclude that the reduced secretion of adiponectin may be coupled to increased lipid droplet accumulation.

Nutritional regulation of adiponectin secretion

Overnutrition, insulin resistance, and adiponectin secretion.

Insulin has been shown to stimulate adiponectin secretion in 3T3-L1 adipocytes in early studies,111 which is consistent with the in vivo observations that adiponectin secretion is reduced under conditions of insulin resistance. However, the situation may be more complex. During hyperinsulinemic clamp studies, circulating adiponectin was decreased significantly at the end of the study.143 A mouse model lacking insulin receptors in adipocytes showed an elevation in plasma adiponectin,144 suggesting that loss of insulin signaling enhances adiponectin secretion. Clinical studies demonstrate that patients with antibodies against insulin receptors showed a significant increase in serum adiponectin.145 In addition, hyperinsulinemia selectively downregulates the high-molecular weight form of adiponectin.146 This is consistent with the existence of a vicious cycle between insulin and adiponectin in T2DM patients, in which insulin levels increase due to the development of insulin resistance, and the higher levels of insulin suppress adiponectin secretion, which further deteriorates insulin sensitivity.

However, adiponectin serum levels are not affected by an acute glucose challenge in humans with a concomitant increase in insulin.39 Adiponectin levels remain unaffected in the postprandial stage,147 after a high fat load,148 or by acute elevations of FFAs.149 Some of these results are in contrast to other reports regarding the chronic and acute influence of insulin in vivo and in vitro systems. In subjects who have developed insulin resistance, many additional changes such as inflammation, ER stress, and mitochondrial dysfunction prevail,104,150 which in turn may exert secondary effects on adiponectin expression unrelated to an increase in insulin.

During the past decade, it became clear that inflammation is a key feature of obesity and T2DM.151 In contrast to the traditional types of inflammatory responses that exert important functions in the context of injuries and infections, the inflammation associated with chronic nutrient excess and metabolic surplus has different features and is often not beneficial at first sight.152 Chronic inflammation is characterized by the infiltration of adipose tissue by macrophages that trigger increased acute-phase reactants and inflammatory cytokines.151 The inflammatory adipokine TNFα has a potent role in suppressing adiponectin production in vivo and in vitro.91 Recently, additional inflammatory adipokines such as IL-6 and IL-1β have also been shown to have a negative impact on adiponectin secretion.153 Thus, the increased amount of local presence of inflammatory cytokines in circulation that behave as adiponectin suppressors becomes one causal factor for the decrease of adiponectin in obesity and other hypoadiponectinemia syndromes.

An additional underlying cause for inflammation in the obese state is ER stress.152 ER stress is increased in adipose tissue in response to both dietary and genetic obesity.154,155In vitro studies confirm that the induction of ER stress is sufficient to cause insulin resistance, and this is mediated through an intracellular crosstalk between inflammatory signaling pathways and insulin signaling pathways.156–158 Adiponectin is a secretory molecule that requires a thiol-mediated retention in the ER for the formation of the HMW complexes.136 ER stress causes a redox imbalance within the lumen,159 which in turn has potent effects on adipocyte function in general and adiponectin secretion specifically.160 An additional link between nutrient surplus and ER stress is the hexosamine biosynthesis pathway (HBP), which is an effective sensor for glucose availability. Activation of HBP has been associated with insulin resistance.161 Recently, the activation of HBP has been found to trigger the upregulation of ER proteins including Bip, calnexin, and careticulin, which are classical markers of ER stress.162 The rate-limiting enzyme in the HBP is Glutamine:fructose-6-phosphate amidotransferase (GFAT). As expected, the ap2-GFAT transgenic mouse, which has an increased activity of HBP in adipocytes, shows reduced circulating level of adiponectin.163 A direct effect of ER stress on adiponectin secretion has been recently demonstrated in 3T3-L1 adipocytes.164 Consistent with the role of DsbA-L in adiponectin maturation, this study also shows that DsbA-L can enhance adiponectin release in the context of ER stress.

Mitochondrial dysfunction and resultant defective oxidative phosphorylation are other causal factors for development of insulin resistance. On the one hand, reduced β-oxidation and the resulting lower oxidative metabolism lead to the accumulation of intrahepatic and intramyocellular lipids that can activate serine kinases, leading to the inhibition of insulin signaling.165–167 On the other hand, dysfunctional oxidative phosphorylation may result in increased production of reactive oxygen species (ROS), which further impairs insulin signaling.168,169 Additionally, oxidative stress generated from mitochondrial defects has been linked to the metabolic syndrome.170

Mitochondrial function has also been reported to play an essential role in adiponectin synthesis.171 Inhibitors of mitochondrial electron transport components selectively reduce adiponectin secretion but do not affect other adipokines release such as resistin from cultured adipocytes.171 Several in vivo studies focus on the correlation between impaired mitochondrial function and reduced adiponectin secretion in ob/ob and db/db mouse models.172 However, the direct mechanistic link between mitochondrial function and adiponectin release remains to be elucidated. In a recent study, intermediate metabolites of the TCA cycle have been proposed to affect adiponectin release.173 The inability to finish oxidative phosphorylation in mitochondria leads to accumulation of fumarate and NADH in the mitochondrion. These metabolites subsequently cause an increase in succination of adiponectin at cystine39, which prevents adiponectin from forming HMW complexes and thus its secretion. When the intracellular adiponectin from db/db and wild type mice were compared, the succinated adiponectin was only detected in adipose tissues from db/db mice. These results suggest that adiponectin succination may relate to the reduced adiponectin secretion in obesity and T2DM. Interestingly, succinyl-CoA transferase was reduced significantly in the fat-specific insulin receptor knock-out (FIRKO) mouse which has been shown to have increased circulating adiponectin levels.144

The profound effects of both the ER and mitochondria in regulating adiponectin secretion have been further substantiated by the use of thiazolidinediones (TZDs), a PPARγ agonist class of compounds widely used for T2DM treatment. These agonists have potent effects leading to an increase in circulating adiponectin, predominantly on HMW,174 which has been implied as a potential mechanism of insulin sensitization.175 Obesity reduces the cellular levels of PPARγ, which may downregulate adiponectin transcription.134 However, activation of PPARγ increases plasma adiponectin mostly through a posttranscriptional mechanism whereas the mRNA levels are only minimally affected.176 The precise mechanism by which TZDs mediates the increased release of adiponectin is not yet clear, but several possibilities exist. First, TZDs, through upregulation of several ER chaperons that include ERp44, Ero1-Lα, and DsbA-L, can alleviate ER stress-induced adiponectin downregulation.135,137,164 Second, TZDs improve mitochondrial mass and function, which improves cellular ROS levels and thereby enhances adiponectin secretion.177,178 Recent data from Spiegelman et al. suggest that Cdk5-mediated phosphorylation of PPARγ downregulates adiponectin transcription and, more importantly, reduces adiponectin secretion from adipocytes.179

CR, SIRT1, and adiponectin secretion

CR can prevent and reverse the harmful consequences of obesity, T2DM, hypertension, and other aging-associated metabolic alterations and diseases. SIRT1, a NAD+ dependent deacetylase, plays a critical role in CR as the effect of CR is lost in SIRT1 knock-out mice. SIRT1 is an important regulator of energy metabolism and becomes a potential target affected in the context of the metabolic syndrome.180 SIRT1 mediates the normal response to diet through pleiotrophic effects in different cell types, including hepatocytes, adipocytes, myocytes, and β cells.181–184 In adipocytes, activation of SIRT1 promotes fat mobilization through repression of PPARγ.185 Not surprisingly, CR can also increase circulating adiponectin in rodents and human,108,186 consistent with the general rule that adiponectin inversely correlates with adiposity. However, when the potential role of SIRT1 in adiponectin secretion was explored, in vitro studies have suggested that SIRT1 negatively regulates adiponectin secretion in response to the nutrient levels through Ero1-Lα, an oxidoreductase located in the ER and required for adiponectin HMW complex release.141 In a cellular system, SIRT1 activation lead to a downregulation of Ero1-Lα, which decreased secretion of adiponectin. The inhibition of Ero1-Lα by SIRT1 occurred at the transcription level through the competition of DNA binding by SIRT1 and PPARγ. These findings establish a direct link between nutrient status and adiponectin secretion. Intriguingly, SIRT1 is upregulated during CR in white adipose tissue, liver, muscle, kidney and brain.185,187 CR has been associated with a normalization of the adipokine profile, including an increase in adiponectin levels.188,189 An open question is whether the SIRT1-Ero1-Lα mediated regulation of adiponectin secretion also applies to an in vivo situation, and, if so, how much this SIRT1-Ero1-Lα axis contributes to the change of adiponectin secretion observed in CR and obesity.

In fact, in vivo studies support a positive role of SIRT1 on adiponectin secretion, potentially mediated through enhancing mitochondrial function. In liver, SIRT1 stimulates the gluconeogenic pathway in concert with PGC-1α, but does not influence the expression of mitochondrial genes regulated by PGC-1α.182 Interestingly, CR promotes mitochondrial biogenesis by inducing the expression of eNOS in white adipose tissue (WAT), brown adipose tissue (BAT), liver, heart, and brain.187 In eNOS-deficient mice, SIRT1 is reduced in WAT. Very recently, eNOS was shown to be required for adiponectin synthesis in adipocytes.190 Notably, other studies also suggest SIRT1 can downregulate the expression of several proinflammatory adipokines, since the SIRT1 activator resveratrol exerts antiinflammatory effects.191–193 Given the inhibitory effect of the proinflammatory adipokines on adiponectin expression, resveratrol is expected to increase the secretion of adiponectin through reducing inhibitory proinflammatory adipokines. In fact, several studies have revealed that activation of SIRT1 by resveratrol significantly inhibits the production of the proinflammatory adipokines TNFα, MCP1, IL6, PAI-1, and increases adiponectin production in 3T3-L1 adipocytes.194,195 The antiinflammatory effect of resveratrol on adipokine expression and secretion is also observed in human adipose tissue explants, in which resveratrol treatment reversed the IL1β-stimulated decrease of adiponectin mRNA levels.196

In summary, the role of SIRT1 in adiponectin secretion still needs to be better defined. Although genetic studies have found an association of SIRT1 gene variations with visceral obesity in man,197 and CR leads to a reduced fat mass,185,198,199 SIRT1 expression and activity in the fat tissues in obesity remain elusive. How does a complete loss of SIRT1 or mere insufficiency affect adiponectin and other adipokines in vivo? Is SIRT1 required for CR mediated plasma adiponectin increase? These questions need to be addressed to fully understand the regulatory role of SIRT1 in adiponectin release under physiological and pathophysiological conditions. On the other hand, given its high production rate, does adiponectin release regulate SIRT1 activity, since the high level production of adiponectin may alter NAD+/NADH ratio mediated through the oxidative protein folding in the ER? This may be one of the reasons why a paradoxical correlation of adiponectin circulating levels and some metabolic diseases has been observed. For example, increased levels of adiponectin have been found in patients with type 1 diabetes200 and patients with anti-insulin receptor antibodies in circulation.146 Strikingly, in one clinical study, the levels of serum adiponectin showed a positive correlation with the progression of type 1 diabetes in the first three months after diagnosis.201 The other report showed that the higher levels of circulating adiponectin in elder people are associated with a higher risk of cardiovascular disease.202 The hyperadiponectinemia presumably suggests that there is some defect of adipocytes in regulating normal rate of adiponectin secretion, such as a defect in SIRT1 action, which may affect the release of the adipokine and alter the other aspects of adipocyte function.

Conclusions and perspectives

The discovery of adipokines has reshaped the view on adipose tissue from an inert lipid storage depot to an active endocrine organ. Theories postulated decades ago that adipocytes could sense the relative level of fat accumulation through a feedback loop employing molecules released from them. By now, a host of adipokines have been identified from adipose tissue, which potentially fulfill the role of biomarkers and biosensors. On the other hand, the production and secretion of adipokines itself might play a role in sensing the nutrients, coordinating cellular function, and modulating physiological processes within the adipocyte. Many adipokines have been shown to regulate multiple aspects of the body energy homeostasis (appetite, thermogenesis, as well as energy expenditure) in a coordinated way to maintain a stable degree of adiposity and activity. However, insights related to the release of these adipokines and the connection to adipose tissue remodeling during obesity are still rudimentary at best. The elucidation of the molecular mechanisms by which adipokines are produced and secreted from adipocytes will be the next big challenge toward a more comprehensive understanding of the role of adipose tissue physiology in whole body energy homeostasis.


This work was supported by NIH grants R01-DK55758, RC1-DK086629, P01-DK088761, and R01-CA112023 (PES). YD is supported by a postdoctoral fellowship from the ADA (7–08-MN-53). We thank Dr. Zhao Wang for helpful discussions.

Conflicts of interest

The authors declare no conflicts of interest.