‘Metabolically healthy obesity’: Origins and implications

3.2. Adipocyte stress and fibrosis

Lean and obese humans are born with a limited number of adipocytes, thus obese individuals have been thought to expand WAT through hypertrophy, not hyperplasia (Spalding et al., 2008). Adipocytes are individually ‘wrapped’ in a supporting sheath of extracellular matrix (ECM), in particular collagens. Remodeling of the ECM, and cycles of collagen breakdown/ re-deposition in particular, are essential for adipocyte and adipose tissue expansion (Chun et al., 2006). However, in the obese state, excessive, dysregulated deposition of collagens and other ECM components (i.e., fibrosis) eventually constrains adipocyte expansion, thereby promoting adipocyte stress, inflammatory/stress kinase activation and resulting AT and systemic metabolic dysfunction (Henegar et al., 2008; Halberg et al., 2009; Khan et al., 2009; Divoux et al., 2010). Collagen VI is specifically implicated in the pathogenesis of obesity-associated AT fibrosis and metabolic dysfunction, as collagen VI deposition is increased in WAT of obese, insulin resistant humans (Spencer et al., 2010), and the absence of this ECM component in obese (knockout) mice permits greater adipocyte hypertrophy while normalizing key metabolic parameters (Khan et al., 2009; see Section 5). Collagen VI deposition and AT fibrosis in human WAT is coincident with the presence of ‘alternatively activated’ (CD150⁺) adipose tissue macrophages (ATMs) that are known to promote ECM remodeling and wound healing (Spencer et al., 2010; see also Shaul et al., 2010).

4.1. MONW

The phenotype of the ‘metabolically obese but normal weight’ (MONW) individual (Fig. 1), first identified by Ruderman and colleagues (Ruderman et al., 1981; Ruderman et al., 1998) and characterized by hyperinsulinemia, hyperglycemia, IR, impaired glucose tolerance, hypercholesterolemia and hypertriglyceridemia, but normal adipocyte volume and BMI, suggests that there must be genes and signal transduction pathways that ordinarily couple obesity to IR, and that these are deficient in certain individuals. These characteristics in a lean individual mark a departure from common human patterns, in which metabolic disease is a consequence of weight gain. Yet, these phenotypes are very prevalent. For example, PCOS has been diagnosed in up to 10% of one study of women of reproductive age (Goodarzi et al., 2011) and MONW in 12.7% of normal-weight subjects in a Korean study (Lee, 2009). Elevated risk for CVD is seen among both MONW (Ruderman et al., 1982) and PCOS individuals (Goodarzi et al., 2011), as well as elevated risk for hypertension, T2D and other metabolic complications. Thus, metabolic dysfunction and CVD risk come in many sizes and shapes (Fig. 1).

Open in a separate window

Fig 1

A classification model for obese and metabolic phenotypes. A ‘two-by-two’ map of the effects of obesity on health, with two off-diagonal entries.

4.2. Mouse models of the MONW individual

Recent study of Goto-Kakizaki rats (Xue et al., 2011) has shown that the Major Histocompatibility Complex (MHC) genes RT1-Ba, Bb and Db1, which are homologous to human HLA-DQA1, HLA-DQ beta 1 chain-like and HLA-DRB1, respectively, are differentially expressed in a phenotype associated with IR and increased inflammation, but a defect in pre-adipocyte differentiation and lack of obesity, which taken together resembles MONW individuals. The field would benefit from additional models of this phenotype.

5.2. Reduced adipocyte stress

In humans, WAT fibrosis as measured by collagen VI expression is positively correlated with IR and inflammatory markers, such as the number of ATMs (Spencer et al., 2010). The relationship between stress/fibrosis and unhealthy WAT (Pasarica et al., 2009) supports a hypothesis that alleles of genes that encode different forms of collagen or enzymes that modify collagen, such as lysyl oxidase (Huang et al., 2010), correlate with the ability of WAT to expand and remodel in obesity while avoiding stress and remaining metabolically healthy. Adipocyte size per se in both omental and subcutaneous depots is also strongly negatively correlated with metabolic health (Klöting et al., 2010; Spencer et al., 2010), with smaller adipocytes and preserved insulin-sensitive glucose transport characteristic of MHO individuals. A hypothesis follows that some MHO humans will show increased adipogenesis, based on functional allelic variants of PPARγ, PGC-1α, PRDM16 or other components of the adipogenic transcriptional program that expand subcutaneous WAT. Not much clinical evidence has yet been marshaled to support this idea, although Patti et al. (2003) noted in a study of Mexican–American subjects that expression of PGC-1α and PPARγ-directed transcriptional networks are decreased in pre-diabetic and T2D obese subjects. Their metabolic dysfunction, including reduced oxidative metabolism and attenuated mitochondrial electron transport, is consistent with defective PGC-1α and PPARγ function, although primary adipogenesis was not studied in this cohort. Well matched MHO and IR obese human subjects should be evaluated for these hypothesized variants.

6.1. Adiponectin transgenic mouse

Adiponectin is an anti-inflammatory, insulin-sensitizing adipokine expressed exclusively by adipocytes that exerts beneficial effects on lipid and glucose homeostasis via multiple mechanisms, most notably via ceramidase activity and AMPK/ SIRT1-dependent activation of PGC-1α (Iwabu et al., 2010; Holland et al., 2011). Elevated (~2-fold) circulating adiponectin is a hallmark and predictor of the MHO phenotype in humans (Klöting et al., 2010). Adiponectin transgenic mice on an ob/ob background (AdTg) have 2- to 3-fold higher levels of circulating adiponectin than ob/ob controls (Kim et al., 2007). These AdTg mice become extraordinarily obese on a normal ‘chow’ diet, weighing on average 40 g more than ob/ob mice. This extreme obesity largely reflects preferential hyperplastic expansion of subcutaneous AT depots, although visceral depots and adipocytes are also reduced in size to wild-type (ob/+) levels. Remarkably, despite morbid obesity, AdTg mice remain insulin sensitive and glucose tolerant, have fewer AT macrophages and inflammatory markers, less liver steatosis and improved circulating insulin, glucose and triacylglycerol. Thus, adiponectin overexpression results in a hyper-obese mouse with predominantly subcutaneous fat, smaller adipocytes, attenuated inflammation and metabolic protection – i.e., all hallmark features of human MHO.

Significant effort is currently directed toward the development of novel therapeutics that elevate production and secretion of the metabolically effective, high molecular weight (HMW) form of adiponectin. Some success has been reported with natriuretic peptides (Tsukamoto et al., 2009) and biflavonoids (Liu et al., 2007). More recently overexpression of the disulfide bond A oxidoreductase-like protein in fat (fDsbA-L) was reported to increase levels of total and HMW adiponectin and to confer resistance to obesity-associated IR through these effects on adiponectin expression (Liu et al., 2012).

6.2. Brd2 hypomorph

The Brd2 gene is located near the TNF-α locus in the class II MHC region of human chromosome 6. Recent work has shown that a leaky knockout of this gene in mice produces a hypomorphic phenotype characterized by extreme obesity coincident with elevated adiponectin (nearing levels of male AdTg mice) and protection from IR (Wang et al., 2009). These mice show hepatosteatosis and hyperinsulinemia, but hypoglycemia and better clamped glucose infusion rates than wild type (Wang et al., 2012; Jornayvaz et al., forthcoming). Brd2 ordinarily co-represses PPARγ-dependent transcription, thus reduced Brd2 in the animals resembles TZD treatment and adipogenesis is potentiated (Denis et al., 2010). Reduced expression of Brd2 also creates a low-inflammatory environment (Belkina and Denis, 2010, 2012; Belkina et al., 2010; Belkina et al., forthcoming) that protects against cytokines that would ordinarily cause IR in adipocytes. Endotoxin-stimulated production of pro-inflammatory cytokines such as TNF-α, IL-1β and IL-6 from bone marrow-derived macrophages is reduced by about half, compared to wild type controls, establishing a ‘protective deficiency’. In MHO humans, it is not yet clear if certain alleles in inflammatory loci create such mild hypo-responsiveness to inflammation signal transduction to confer protection. Consideration of the association among MHC alleles and complex human diseases with an inflammatory component provokes the hypothesis that certain MHO individuals are protected from adipose tissue inflammation and metabolic complications because of specific alleles in their MHC. Several genes in the MHC are likely candidates, including LTA, which encodes lymphotoxin-α and is important for the inflammatory component of coronary artery disease (Ozaki et al., 2002); TNF, BRD2 and HLA-DRB1. These alleles should be evaluated for correlation with the MHO phenotype.

6.3. COL6 KO mice

As discussed above, recent studies suggest that excessive deposition of collagen(s) or other ECM components (fibrosis) promotes inflammatory and metabolic pathology in individuals with hypertrophic obesity. In a seminal study, collagen VI (COL6) KO mice crossed onto the genetically obese ob/ob background developed dramatically-hypertrophic adipocytes and greater AT mass in both the gonadal and mesenteric depots, but had lower fasting glucose, enhanced glucose tolerance and reduced circulating triacylglycerol following lipid challenge as compared with control ob/ob mice (Khan et al., 2009). This metabolic protection was associated with reductions in ER stress markers, stress kinase (JNK, ERK) activation and adipocyte death, as well as with lower systemic IL-6 and TNF-α following LPS challenge. Thus, reducing AT collagen promotes the development of metabolically ‘protective’ adipocyte hypertrophy and an MHO phenotype. A plausible interpretation is that the (hyper)obese COL6 KO mouse uncouples hypertrophy per se from obesity complications and suggests that therapeutic approaches that reduce adipocyte stress can maintain metabolically healthy fat storage in enlarged adipocytes.

6.4. TBP-2/Txnip/VDUP1 KO mice

Thioredoxin binding protein (TBP)-2 is a member of the α-arrestin protein family with demonstrated roles in the regulation of cell fate, immune responses and energy metabolism. TBP-2 is a negative regulator of adipogenesis, in part through its inhibitory actions on PPARγ expression and activity (Chutkow et al., 2010; Yoshihara et al., 2010; Chutkow and Lee, 2011; see also Chen et al., 2008). Consistent with these observations, TBP-2 KO mice fed either a normal or HFD diet or TBP-2 KO mice on an ob/ob background consumed more energy and gained significantly more weight (up to 100%) and adiposity (up to 50%) than similarly-fed WT or ob/ob mice (Chutkow et al., 2010; Yoshihara et al., 2010). Notably, greater adipose mass reflected adipocyte hyperplasia in multiple AT depots (Chutkow et al., 2010). Despite greater adiposity, both models of obesity in TBP-2 KO mice were more glucose tolerant and insulin sensitive than obese controls, reflecting augmented glucose transport in AT and skeletal muscle and improved glucose-stimulated insulin secretion (GSIS). Thus, as in AdTg mice (see above) adipose accretion by adipocyte hyperplasia is associated with an MHO phenotype in obese TBP-2 KO mice. HBC-19 mice that express a naturally-occurring TBP-2 truncation mutation also become more obese than WT mice but retain glucose tolerance and insulin responsiveness (Chen et al., 2008). Moreover, HcB-19 mice crossed with the ob/ob mice become even more obese than ob/ob mice, but are protected against peripheral insulin resistance, β-cell apoptosis and subsequent T2DM. The protection from β-cell apoptosis despite morbid obesity in mice lacking functional TBP-2 was demonstrated in β cell-specific TBP-2 KO mice to reflect enhanced Akt/Bcl-xL signaling and associated mitochondrial protection (Chen et al., 2008).

6.5. Tumor-progression locus 2 (TPL2) KO mice

TPL2 (MAP3K8) is a serine/threonine kinase that is activated by TNFα, TLR4-signaling and inflammatory mediators that activate the NFκB and MAPK pathways (Vougioukalaki et al., 2011). Thus, TPL2 is uniquely situated to integrate inflammatory signaling pathways that promote obesity-associated inflammation and IR. Perfield et al. (2011) placed TPL2 KO mice on a HFD for 16 weeks and reported increases in body weight, adipose depot weights and gonadal adipocyte size comparable to those observed in HFD-fed WT mice. However, HFD-fed KO mice had decreased fasting glucose and insulin, and improved glucose dynamics in euglycemic/hyperinsulinemic clamp studies. This protected metabolic phenotype was associated with reduced MAPK (ERK, JNK) activation in peripheral tissues, as well as with reduced frequency of CLS and attenuated inflammatory gene expression in the epididymal fat depot. Thus, the absence of TPL2-dependent inflammatory signaling confers an MHO phenotype of reduced AT inflammation and enhanced insulin sensitivity despite hypertrophic obesity.

GVD is the 2011–2012 Chair of Basic Science Section of The Obesity Society and gratefully acknowledges the Society for its strong support and promotion of scientific interactions and collaborations. Drs. C. Apovian, B. Corkey, S. Fried, and B. Nikolajczyk are particularly acknowledged for valuable discussion related to this work, which is supported by the National Institutes of Health (R56 DK090455), and the Boston University Clinical and Translational Science Institute (UL1-TR000157).