Architecture and the extracellular matrix: the still unappreciated components of the adipose tissue


  • A. Divoux,

    Corresponding author
    1. INSERM, U872, Nutriomique (Team 7)
    2. Université Pierre et Marie Curie – Paris6, Centre de Recherche des Cordeliers, UMR S 872
    Search for more papers by this author
  • K. Clément

    1. INSERM, U872, Nutriomique (Team 7)
    2. Université Pierre et Marie Curie – Paris6, Centre de Recherche des Cordeliers, UMR S 872
    3. Assistance Publique-Hôpitaux de Paris, APHP, Pitié Salpetrière Hospital, Departement d'Endocrinologie et Nutrition, Paris, France
    Search for more papers by this author

A Divoux, INSERM, U872, Nutriomique (Team 7), F-75006 Paris, France. E-mail:


Fibrosis is usually characterized by the modification of both the amount and composition of a wide panel of extracellular matrix (ECM) proteins. In the liver, pancreas, kidney and lung the accumulation of fibrosis disrupts cellular processes and appears detrimental for organ function. This review highlights the available evidence supporting an important ECM remodelling in adipose tissue (AT) and, in particular, during the development of obesity. The modifications and occurrence of new adipose ECM components leads to an abnormal accumulation of fibrosis in this tissue. This phenomenon was well described in rodent models and evidence is beginning to emerge in humans; however, the origin and potential impact of these depots in AT biology are unclear. Two animal models with disruptions in ECM components (secreted proteins acidic in nature rich in cysteine null mice and ob/ob collagen VI null mice) suggest that fibrosis limits adipocyte hypertrophy and may cause the metabolic disorders associated with obesity. Over-expression of Hypoxia-inducible factor 1 leading to an increase in collagen expression suggests a role for hypoxia in fibrosis development. We conclude this review with possible hypotheses regarding the cellular and molecular contributors of fibrosis initiation.


Repair of damaged tissues is a fundamental biological process allowing the ordered replacement of dead or injured cells in response to inflammation. The repair process begins with a regenerative phase, in which injured cells are replaced by cells of the same type. If the damage persists, this initial phase is followed by a phase of fibrosis accumulation in which connective tissue replaces normal parenchymal tissue. Although this process is initially beneficial, the healing process can become pathogenic leading to considerable tissue remodelling and the formation of permanent scar tissue. Ultimately, this may contribute to organ failure and death. Adipose tissue (AT) undergoes a strong structural remodelling when fat mass increases, as seen with obesity. In this review, we focus on a relatively neglected phenomenon characterizing AT in obesity, i.e. modification of the AT extracellular matrix (ECM) and its potential consequence(s) on AT biology.

Multiple properties of extracellular matrix in tissue architecture and function

The ECM is composed of various elements ensuring not only tissue architecture, but also having a pivotal role in the biological functions of different organs. Components of the ECM include structural proteins (collagens) and various classes of adhesion proteins, such as fibronectin, laminin, elastins and proteoglycans (e.g. perlecan and decorin). A variety of collagen forms exist but type I, III and VI collagens are the isotypes most often associated with an organ's fibrotic depots. Type IV collagen and laminin form the basal lamina of epithelia.

The ECM actively orchestrates the keys steps in the programme of wound healing and regeneration (1). Its components vary quantitatively and qualitatively in different tissues providing structural support and anchorage for cells, segregating tissues from one another and regulating intercellular communication. ECM is also extremely important for the function of almost any cell type. The ECM components provide a structural lattice to which cells may adhere, facilitating their organization in the tissue. Integrins embedded in the plasma membrane ensure the communication between the matrix components and the intracellular environment. Any changes in the ECM trigger a signalling pathway through integrins to intracellular microtubules cytoskeleton and actin filaments. The transduction of this signal can trigger a cascade of chemical events in all adjacent cells which allows cohesion. Integrins also have a further role in the regulation of cell movement during development, cell proliferation, differentiation, migration, apoptosis and gene induction.

Proteoglycans constitute a reserve of molecules implicated in tissue biology, such as growth factors and matrix metalloproteinases (MMPs). The MMPs degrade structural components of the ECM and regulate tissue architecture (2). Furthermore, MMP-dependent proteolysis can create space for cells to migrate and can produce specific substrate-cleavage fragments with biological activity. E.g. MMP-2 and MMP-9 can release TGF-β from an inactive extracellular complex; the activated form of TGF-β alters cell migration and regulates ECM protein expression (in particular inducing the expression of collagens and fibronectin, and suppressing the expression and activity of MMP-2) (3).

Extracellular matrix and development of white adipose tissue

In AT, the ECM is crucial for maintaining the structural integrity of adipocytes and plays a pivotal role in adipogenesis and whole tissue formation. The first reported observations of ECM structure in AT came from Napolitano et al. in 1963. Using electron microscopy, the authors examined white adipose tissue (WAT) development in the rat and observed adipose cells embraced by a network of collagen fibres (4). In 1984, Cinti et al. reported an ECM rich in collagen fibrils surrounding adipose cells and capillaries during the development of fat organs in young rats (5). Several years later, immunohistochemistry studies highlighted the presence of collagen IV, laminin, fibronectin and heparan sulphate proteoglycan surrounding human adipocytes (6). Interestingly, the level of fibronectin expression is low in differentiated subcutaneous adipocytes; this finding is in line with in vitro studies showing a significant decrease of fibronectin synthesis during adipocyte differentiation (7).

The importance of the ECM in preadipocyte differentiation was further substantiated by cellular studies performed mostly with rodent models. Preadipocyte cell lines showed that during differentiation there was an accretion on the cell surface of type I–VI collagens, laminin, and fibronectin in comparison to undifferentiated cells (8). Modifications in the network structure of type V and VI collagens are observed during differentiation. More specifically, the fine and spiny fibrils present at the beginning of differentiation disappeared and were replaced by a rough and thick network. During the differentiation of mouse 3T3-F442A cells, the presence of fibronectin was found to strongly interfere with the cytoskeleton and morphological changes necessary for the expression and regulation of genes encoding lipogenic proteins (such as glycerophosphate dehydrogenase and fatty acid synthase) (9). A pivotal role for fibronectin in adipogenesis was also confirmed in humans via cell models studies. Specifically, a significant decrease in fibronectin gene and protein expression accompanied in vitro human preadipocytes differentiation (10).

In conclusion, an ECM rich in collagen fibrils and fibronectin was observed in AT and appears essential for regulating both tissue architecture, and preadipocyte differentiation. While these findings provide important preliminary indices, more research is necessary in order to appreciate the exact composition and role of the ECM in AT physiology and in the different fat depots. In pathological conditions such as obesity, variations in ECM components appear and may have consequences on cell and tissue biology.

Alterations of specific extracellular matrix components in obese white adipose tissue

Matricellular proteins include thrombospondins (THBS) 1 and 2, secreted proteins acidic in nature rich in cysteine (SPARCs), osteopontin and the cyr61, CTGF, Nov (CCN) family of proteins which includes Connective tissue growth factor (CTGF). These proteins modulate cell function by interacting with cell-surface receptors, proteases and hormones, as well as with structural matrix proteins such as collagens (11). During obesity, their level of expression is modified in WAT.

The THBS1 is a regulator of transforming growth factor β (TGF-β) activity, a well-known profibrotic factor. It is defined as an adipokine being highly expressed in obese patients and its expression was found to decrease during human preadipocyte differentiation (12). In mouse AT, THBS1's expression was showed to correlate with inflammatory mediators like plasminogen activator inhibitor-1 (PAI-1) and chemokine Monocyte chemotactic protein (MCP)-1 (13). The protein SPARC is also enriched in obese human WAT (14). This mediator of collagen deposition, which promotes fibrosis, is known to modulate interactions between cells and the ECM. The SPARC was shown to inhibit in vitro mouse adipocyte differentiation (15). More specifically, SPARC inhibits the expression of key adipogenic transcription factors and adipocyte-specific genes, such as CCAAT enhancer binding protein (CEBP)-α, CEBP-β and peroxisome proliferator-activated receptor γ2 PPARγ2(15). Osteopontin is a matricellular protein that plays an important role in tumor cell invasion due to its capacity to regulate the activity of at least two ECM degrading proteins: matrix metalloproteinase-2 (MMP-2) and urokinase plasminogen activator (16). Three Osteopontin (OPN) isoforms (a, b and c) were detected in human AT but OPN-a and OPN-b were predominantly expressed in the AT of obese subjects (17). Mice exposed to a high-fat diet exhibit increased plasma osteopontin levels with elevated expression in AT (18), suggesting a relationship between this ECM component and inflammation in obese AT. The CTGF, discovered more than a decade ago as a protein secreted by human endothelial cells, is induced by TGF-β and is considered a downstream mediator of the effects of TGF-β on fibroblasts. In contrast with the situation in normal fibroblasts, CTGF is constitutively expressed in dermal fibrotic lesions, such as in scleroderma and renal, pancreatic, lung and liver fibrosis. In WAT, CTGF was observed in mouse epididymal adipocytes (19). CTGF gene expression was found in 3T3L1 preadipocytes and increased during differentiation with TGF-β1 continual exposure; TGF-β1 expression increasing in AT of both animal and human models of obesity (20).

The ECM proteoglycan and associated proteins may also be altered during obesity. Decorin is a small leucine-rich proteoglycan required for the correct folding of ECM components such as collagens. Bolton K et al. showed that decorin is highly expressed in obese AT (21). The authors suggested its implication in a new ECM structure formed when AT expands. Another ECM molecule described in AT is tenascin C. It is involved in central nervous system development and was identified in human inflammatory preadipocytes (22). In obese diabetic mice (db/db on high-fat diet) tenascin C gene expression increased in parallel with the expressions of procollagen I, III, V and VI (23).

In this context, matrix metalloproteinases, MMPs, have been considered as important but complex player in AT remodelling and fat mass increase (2). These proteases are constitutively expressed at very low levels, but are rapidly induced during development of obesity. Rodent studies show conflicting results. Alexander et al. observed that inhibition of MMPs promotes 3T3L1 adipocyte differentiation (24) whereas Croissandeau et al. blocked adipogenesis by the inhibition of MMP activities (25). Bouloumiéet al. found the same result using another mice cells line (3T3F442A) (26) and Chavey et al. using primary rat preadipocytes (27). An administration of a broad-spectrum MMP inhibitor impairs AT growth, suggesting a functional role of MMPs on AT development in rodents (28). Nevertheless, only minor effects of partial MMP inhibition were observed on AT development and cellularity (29) while the deficiency of MMP-2 impairs fat mass expansion and is associated with adipocyte hypotrophy (30). In marked contrast, MMP-3 and MMP-11 deficient mice consuming a high-fat diet developed more AT than their wild type counterparts (31)(32). Cysteine proteases (i.e. members of the cathepsin family), expressed in AT (33) might also contribute to AT remodelling (34) since mice deficient for cathepsins L and K also show resist to high-fat diet-induced weight gain.

Taken together, these studies emphasize the relevance, but complexity, of different components of the ECM, including a variety of proteases, not only in weight modification, but also in the development of pathological characteristics of obese AT, such as inflammation. While the list of proteins that contribute to ECM remodelling in AT has grown in recent years, further research along this axis will undoubtedly lead to the discovery of other key players.

Global remodelling of extracellular matrix in obese white adipose tissue; evidence of fibrous depots

Prolonged positive energy balance induces a myriad of changes in AT, including adipocyte hypertrophy, new adipocyte formation, accumulation of inflammatory cells, and neovascularization (35). These changes lead to functional alterations in AT and modify the secretion of adipose produced molecules. To accommodate the changes, ECM remodelling occurs by degradation of the existing ECM and the production of new ECM components. This phenomenon was recently described in mice. Feeding C57BL/6 mice a high-fat diet for 20 weeks in order to induce obesity, Strissel et al. highlighted an intense remodelling of AT during the progression to an obese state, associated with collagen deposition and an increase in inflammatory markers (36). The AT secreting matrix-cellular proteins and proteases appear to play an important role in the complex processes occurring during the dynamic of weight gain.

During human obesity, ECM components previously mentioned, are modified. Global analyses have been undertaken in which the transcriptomic signature of human subcutaneous AT in lean and obese subjects was explored. The bioinformatic analysis of microarray studies further demonstrated the strong relationship between body mass index, inflammatory processes, and ECM components, including the structural components described above (37). This pattern was observed more clearly when AT samples were obtained using a surgical biopsy method, which better maintained tissue structure (38). Drastic and rapid weight loss induced by bariatric surgery lead to a significant change in the expression of more than 200 genes encoding ECM structural components and a myriad of proteases; suggesting an intense ECM remodelling is not only associated with weight gain, but also with weight loss (39). Moreover, an important mobilization of genes implicated in inflammatory process was observed (39).

Mirroring observations made in other organs, such as the liver, in which fibrosis is the consequence of major alterations in both the amount and composition of ECM, AT may be a site of fibrous depots. Based on this hypothesis and on the evidence of major expression changes in ECM genes in obese AT, our team measured the abundance of collagens depots in lean and obese subjects. In 2008, Henegar et al. showed fibrous areas grouped in bundles in human obese subcutaneous AT for the first time (Fig. 1). The quantification of red picrosirius staining, a marker for collagen, showed a higher amount of fibrosis in obese subcutaneous WAT compared to lean subjects Moreover, the percentage of subcutaneous WAT fibrosis correlated with the score of fibro-inflammation in ten morbidly obese subjects used for this study (37). Divoux et al. also recently identified a significant amount of pericellular fibrosis (i.e. collagens surrounding adipocytes) being more important in obese WAT omental and subcutaneous depots compared to lean depots (Divoux et al. work in revision) (67). More recently, Pasarica et al. showed the presence of collagen VI in subcutaneous human AT and that its expression increased with BMI and fat mass (40). Importantly, col6a3 is principally stained around adipocyte in human AT slides (Divoux et al. work in revision). Obese subjects with high col6a3 expression have increased visceral AT mass and inflammation, suggesting the importance of fibrosis in AT modifications characterizing obesity (40).

Figure 1.

Identification of fibrosis in white adipose tissue (WAT). Slides of lean (a) and obese (b) subjects were stained with red picrosirius. Arrows show accumulation of fibrosis at the proximity of adipocytes.

Consequence of modulating extracellular matrix in obese adipose tissue biology: recent rodent models

While fibrosis depots have been identified in obese AT, the origin and potential pathological impact of these depots in AT biology are unclear. The relationships between ECM components, fat cell size, metabolism, and inflammation have been proposed following investigations with genetically modified mouse models.

The protein SPARC may elicit changes in cell shape, inhibit cell-cycle progression, and influence the synthesis of ECM components (41). The expression of SPARC is low in adult tissues but its synthesis increases when intense tissue remodelling occurs. The disruption of SPARC in dermal ECM showed a drastic reduction in collagen content (42). Furthermore, SPARC-null mice appeared resistant to the weight gain induced with a high-fat diet; a phenotype unexpected in a condition where diminished ECM amount may enable fat mass expansion. Nevertheless, sections of skin and epididymal fat pads from SPARC-null mice revealed an increase in the subdermal adipose layer. An increase in the size of epididymal fat pads in SPARC-null mice was also found in comparison with those of wild-type animals. DNA quantification of fat pads and isolated adipocytes showed that SPARC-null pads contained a lower number of hypertrophic adipocytes. This animal model raises the possibility that a loose ECM structure favours adipocyte hypertrophia.

Khan T et al. recently showed a strong up-regulation of collagens in the epididymal fat pads of db/db mice (43), a well-known genetic model of obesity and metabolic dysregulation. In the fat pads of wild type mice, collagen 1a1, 4a1 and 6a1, were abundant; however, in contrast, db/db mice express only collagen 6a1 in high abundance. Collagen VI-null ob/ob mice (generated by disrupting the collagen6a1 isoform) were produced to determine the impact of collagen VI on the obese AT and metabolic phenotypes. ColVI−/−ob/ob mice showed ameliorations in glucose and lipid metabolism. The increased number of membrane caveolae, where the insulin receptors are localized, and the increased AKT activation upon insulin stimulation both supported the improvement in insulin signalling. Metabolic improvements in ColVI−/−ob/ob mice were associated with a decrease in inflammatory markers, a reduction of necrotic cell death, and a decrease of endoplasmic reticulum stress (i.e. as reflected by a reduced Xbp1 expression) in AT. The authors observed that ColVI−/−ob/ob mice had a remarkably loose and disorganized AT; concomitant with an increase in adipocyte size. Lumican and decorin are implicated in the epithelial-mesenchyme transition required for fibrotic processes. Lumican stimulates the pro-fibrotic factor TGF-β, while decorin has an inhibitory effect. In ColVI−/−ob/ob mice, lumican mRNA was strongly down-regulated and decorin mRNA up-regulated. This was associated with a decrease in TGF-β expression. The phenotype of ColVI−/−ob/ob mice is complex when one considers that in humans an increase in adipocyte size generally associates with increased inflammation and metabolic disturbances (44); however, an attractive hypothesis has been proposed to explain this phenomenon. When adipocytes reach their maximum size limit, the relationships between adipocyte and ECM are modified. The β1-integrin/extracellular signal-regulated kinase (ERK) signalling pathway is activated and promotes cell death, a situation that contributes to local inflammation. The lack of collagen VI may allow adipocytes to increase their size without any constraints by the ECM; thereby preventing the activation of the death signalling pathway. Thus, the absence of a limitation on adipocyte expansion may favour lipid storage over ectopic lipid accumulation. Consequently, this diminishes adipocyte necrosis resulting in a decrease in macrophage accumulation and, subsequently, in the reduction of the inflammatory microenvironment. This hypothesis would still support the improvement in insulin sensitivity. The molecular mechanism underlying this hypothesis is summarized in Fig. 2.

Figure 2.

Assumptions about implication of fibrosis on metabolic disorders observed during obesity. During obesity, adipocyte hypertrophy is limited by white adipose tissue (WAT) extracellular matrix (ECM) components (left side). A direct consequence of this constraint is ectopic fat deposition in other organ (1). Furthermore, interactions between ECM and hypertrophic adipocytes promote activation of C-jun N-terminal kinase (JNK) pathways, resulting in cells apoptosis. Adipocyte death leads finally to monocyte infiltration and moreover metabolic disorders (2). Activation of JNK pathways could also promote adipocyte invaginated calveolae increase, unable to stabilize the insulin receptor at the plasmic membrane and altering consequently insulin signalling (3). Loss of collagenVI (coll6−/−ob/ob mice) or collagenI (SPARC−/− mice) (right side) leads to an uncontrolled adipocyte expansion. In consequence adipocytes store more lipids, avoiding abnormal lipid accumulation in other tissue (4). The ECM modifications prohibit also transmission of apoptotic signal and then the followed macrophages accumulation (5). Thus deficient in collVI or collI could improve metabolic alterations described in obesity. MAPK, Mitogen-activated protein kinase; SPARC, secreted proteins acidic in nature rich in cysteine.

Proposed hypothesis of cellular and molecular contributors to human white adipose tissue fibrosis

Common mechanisms were reported in organs affected in different fibrotic diseases (45). The formation of fibrosis associates a deregulated wound healing capacity with successive steps including the release of inflammatory mediators, an antifibrinolytic-coagulation cascade, leukocyte recruitment, and the activation of myofibroblasts producing MMPs. This phase is followed by the production of ECM components by myofibroblasts and vessel formation by endothelial cells. Persistent inflammation leads to chronic activation of myofibroblasts and an accumulation of abnormal ECM components. The interactions of lymphoid cells sustain the production of growth factors, proteolytic enzymes, and fibrogenic cytokines. To date, no study has clearly demonstrated the presence of myofibroblasts in AT.

The preadipocyte, a fibroblast-like cell, may play a critical role in the formation of AT fibrosis. A recent study demonstrated the profound modification of the human preadipocyte phenotype when placed in an inflammatory environment. Stimulated by human macrophage secretions, preadipocytes were shown to increase the synthesis of many fibrotic molecules (such as collagens and fibronectin), which in turn promoted preadipocyte migration and proliferation (22).

The molecular mechanisms underlying abnormal ECM accumulation in AT are not yet well defined. Nevertheless, this previous in vitro study suggested that some inflammatory mediators secreted by AT cells, especially macrophages, are candidates to promote the synthesis of fibrous depots. Indeed, inflammatory mediators secreted by AT cells, especially macrophages, are candidates to promote the synthesis of fibrous depots. E.g. IL-6, a cytokine that has been largely explored in obese AT, stimulates the production of collagen by dermal fibroblasts (46). IL-10, another cytokine shown to be secreted by AT macrophages, was implicated in pancreatic fibrosis processing via TGF-β modulation (47). More specifically, IL-10-null mice have higher intra-pancreatic collagen content than controls; which coincides with higher plasma levels and intra-pancreatic transcription of TGF-β. Chemokines are also induced during obesity (48). Chemokines cooperate with profibrotic cytokines by recruiting effector cells to sites of injury (e.g. myofibroblasts and macrophages). Seki et al. demonstrated the role of MIP1α, MIP1β, and CCL5 and their receptors CCR1 and CCR5 in hepatic fibrosis (49). CCR1 and CCR5 deficient mice showed a reduction in hepatic fibrosis parallel to a reduction in monocyte accumulation. CCR5 exerts a profibrogenic effect on resident liver cells, whereas CCR1 acts via the bone marrow-derived cell population. Smith et al. showed that the expressions of MCP1 (a CCR2 ligand) and MIP1α were increased in a model of pulmonary fibrosis and that fibrosis was attenuated by an anti-MIP1α treatment (50). Moreover, CCR2-null mice were protected from fibrosis in pulmonary fibrosis models (51). Blocking CCR1 also reduced cell accumulation and fibrosis in a renal fibrosis model (52). These studies demonstrate that the interruption of specific chemokine signalling pathways may impact the development of fibrosis diseases. Gene expression of CC chemokines involved in monocyte chemotaxis, such as CCL2 and CCL5, and their receptors was shown to be higher in the AT of obese subjects (53)(54), and was found to participate in monocyte/macrophage adhesion and transmigration (55). The specific roles of these chemokine on AT fibrous depots have not yet been delineated. Membrane type 1 metalloprotease (MT1-MMP), a transmembrane metalloprotease directly degrades ECM components (such as collagen I) and indirectly via the activation of pro-MMP-2. The MT1-MMP deficient mice have a diminished adipocyte size and fat mass. The expression of genes regulating lipid metabolism was also decreased in fat pads from deficient mice (56). While the absence of this metalloprotease in vivo evokes default in preadipocyte differentiation concomitantly to a failure in type I collagen degradation, preadipocytes isolated from MT1-MMP null mice differentiated normally. This apparent discrepancy was solved by an elegant in vitro study in which the culture conditions were modified (57). Preadipocyte differentiation was analysed in a fibrillar gel of type I collagen, which mimicks the dense collagen fibrils observed in vivo. Conversely to a very organized network of actin displayed by MT1-MMP+/+ preadipocytes, MT1-MMP−/− preadipocytes had an abnormal morphology lacking stress fibres. Levels of CREB and its phosphorylated form, an early factor of adipogenesis, were dramatically reduced in MT1-MMP−/− preadipocytes. MT1-MMP−/− cells restored a normal phenotype when type I collagen gel concentration was reduced. MT1-MMP retroviral vector expression in null-preadipocytes reversed the defective development of these cells in 3D culture. Thus, MT1-MMP−/− preadipocytes lost the ability to degrade type I collagen in 3D conditions and these cells, entangled in a dense collagen network, do not differentiate properly. This study revealed MT1-MMP as an important modulator of adipocyte morphology and lipid metabolism.

The involvement of hypoxia via HIF-1α activation in AT fibrosis development was suggested in a recent report (58). HIF-1α, the hypoxia-sensing transcriptional factor, is rapidly degraded by an oxygen-dependent mechanism. Its accumulation reflects a local hypoxia and stimulates the expression of a high number of genes. Transgenic mice, constitutively expressing active HIF-1α in AT, were used to address the influence of local hypoxia on the development of fibrosis. In transgenic mice, HIF-1α was highest in subcutaneous AT (scAT) and coincides with an increase in adipocyte size. When fed a high-fat diet, the transgenic mice exhibited deterioration in glucose metabolism. When crossed on the ob/ob background, HIF-1α transgenic animals showed liver steatosis and scAT inflammation, as evaluated by macrophage accumulation. These mice developed AT fibrosis and had an increase in the expression of several ECM genes, such as col1a1, col3a1, col4a1, col6a1, col18a1, elastin and lumican. Lysyl oxidase (LOX), a HIF-1α target gene and cross-linker of collagen fibres, was also induced. The treatment of transgenic HIF-1α mice with a LOX inhibitor improved glucose tolerance, reduced AT macrophages, and reduced collagen streaks. These observations suggest a link between LOX, fibrillar collagen deposition, glucose metabolism, and macrophage accumulation. A time-course study of high-fat diet revealed that, in the early stages, adipocyte hypertrophy coincided with an increase in HIF-1α, LOX, col1a1 and col3a1 expression. Macrophage infiltration and inflammation markers were detected at later stages. Thus, HIF-1α-driven fibrosis appears to be an initial event that precedes AT inflammation. In contrast with the previously described observations in collVI−/−ob/ob model, where the loss of a fibrotic component led to an improvement in metabolic and inflammatory profiles, the over-expression of HIF-1α by promoting LOX, col1a1, and col3a1 production, triggers AT macrophage accumulation and insulin resistance. In human obesity, AT hypoxia appears to be one hallmark of expanded AT, as shown by gene expression studies (59) and the direct measurement of AT hypoxia (60).

Other factors than hypoxia may promote AT fibrosis. E.g. the renin-angiotensin system and the fibrinolytic (plasminogen/plasmin) system, known to exert a role in AT biology and development (2), could play a critical role in the development of fibrosis, as observed in other organs, such as the liver (61)(62)(63). Indeed the main physiological inhibitor of the fibrinolytic system, PAI-1, is expressed in murine and human ATs. Tissue-type plasminogen activator, the main target of PAI-1, impairs AT development and the use of synthetic low inhibitors of PAI-1 may have the potential to reduce obesity (64).

However, their links with AT inflammation and fibrosis development have to be further unravelled.

Conclusion and perspectives

Integrity of the ECM is essential for the proper development of AT in a physiological context. In pathologic conditions, more knowledge is nevertheless necessary to understand the contribution of ECM modifications and the consequence of fibrous depots. Rodent models showed that disruption of the matrix causes metabolic changes that coincide with inflammatory modifications. Observations done in different mouse models, summarized in Table 1, suggest a link between ECM components modification, inflammation and adipocyte biology. To definitively establish this link and understand the chronology of events more studies are necessary. It is important to decipher which mediators are involved in this complex process before considering the utility of antifibrotic molecules targeted to AT fibrosis. Recent study performed in HIF-1α knock-out mice, suggests hypoxia as a major factor initiating ECM remodelling. ECM molecules and remodelling events play also a key role in regulating lymphangiogenesis (65). Fibrosis accumulation was shown to inhibit lymphatic regeneration in a experimental mouse model (66). ECM modifications in WAT have also an impact on lymph drainage from the tissue, causing a decrease of the tissue clearance and an increased of local oxygen demand. While not very well studied, obesity is a well-known risk factor of lymphoedema apparition, it would be interesting to investigate the role of scWAT remodelling and fibrosis associated in the onset of lymphoedema.

Table 1.  Consequences of different ECM component modification in mice
 SPARC−/−ob/ob Coll6−/−MT1-MMP−/−Over-expression of HIF1α
  1. ↓, decrease; ↑, increase; ↔, equal; AT, adipose tissue; ECM, extracellular matrix; HIF1, Hypoxia-inducible factor 1; MCP1-SAA3, Monocyte chemotactic protein-1-Serum amyloid A3; MT1-MMP, Membrane type 1 metalloprotease; PPARγ, peroxisome proliferator-activated receptor gamma; SPARC, secreted proteins acidic in nature rich in cysteine.

ECM components    
 Collagen↑ Coll1α1, 3α1, 4α1, 6α1, 8α1
 ECM protein ↓ Lumican & elastin
↑ Decorin
 ↑ Elastin & lumican
Adiposity markers    
 Food intake   
 Weight, fat pad size  
 Serum leptin  
 Adipocyte size 
Metabolism parameters    
 Glucose tolerance 
 Insulin sensitivity ↑ Caveoles number and Akt activityNormal lipogenic response 
 Lipid metabolismPositive clearance test↓ Metabolism lipidic gene expression↑ Circulated triglycerides
Inflammatory markers    
 AT molecules ↓ MCP1-SAA3
 Macrophages accumulation  

Finally, the consequence of qualitative changes in nutritional condition in ECM remodelling of AT are poorly known. Modifications in AT mRNA expression of genes involved in ECM degradation (MMP-12, Cathepsine K, L, S) and in ECM components (procollagen I, III, VI) were induced specifically by a high-fat diet rich in satured fatty acids in db/db mice (23). The changes were reversed with an n-3 PUFA rich diet. Specific effect of diet nutrient types on AT remodelling is an important field of investigation and will be challenging in humans.

Conflict of Interest Statement

No conflict of interest was declared.


The authors wish to thank supports by the Commission of the European Communities (Collaborative Project ADAPT, contract number HEALTH-F2-2008-1100), Hepadip consortium (, contract LSHM-CT-2005-018734), and the « Programme Hospitalier de Recherche Clinique », Assistance Publique-Hôpitaux de Paris (AOR 02076) and the ‘Contrat de Recherche Clinique (CRIC)’. This work was supported from Fondation pour la Recherche Medicale (to A. D. and K. C.). The authors wish to thank Dr David Mutch for critical reading of the manuscript.