Chemotactic proteins in obesity and type 2 diabetes

Chemotactic proteins, particularly those of the chemokine family, have been shown to be related in vivo to the metabolic syndrome, obesity and type 2 diabetes (Table 1) and to be adipokines secreted from adipocytes or other cell types residing in adipose tissue. Chemokines are small proteins that attract various immune cells such as monocytes, neutrophils, T lymphocytes, basophils or eosinophils (each chemokine activating one or more target cell types)⁽Reference Rollins⁸⁾. Chemokines are characterized by the presence of four highly-conserved cysteine residues. CXC chemokines have two amino-terminal cysteine residues separated by only one amino acid. In CC chemokines, the other main subfamily of chemokines, the amino-terminal cysteine residues are adjacent⁽Reference Rollins⁸⁾. In addition, other chemoattractant proteins such as chemerin, which has been shown to attract macrophages and dendritic cells but is not structurally related to any chemokine family, comprise the adipokines and factors shown to be involved in obesity and obesity-related pathologies⁽Reference Bozaoglu, Bolton and McMillan⁹⁾.

Table 1. Clinical data showing the association between chemotactic cytokines and obesity and type 2 diabetes

MCP, monocyte chemoattractant protein; CCL, chemokine CC motif ligand; CXCL, chemokine CXC motif ligand; IP-10, 10 kDa interferon γ-induced protein.

Monocyte chemoattractant protein (MCP)-1 is a chemokine and a member of the small inducible cytokine family that plays a role in the recruitment of monocytes and T lymphocytes to sites of injury and infection⁽Reference Baggiolini¹⁰⁾. Its main receptor is the chemokine CC motif receptor (CCR) 2. Plasma MCP-1 levels are markedly higher in obese patients⁽Reference Christiansen, Richelsen and Bruun^11, Reference Kim, Park and Kawada¹²⁾ and patients with diabetes⁽Reference Herder, Baumert and Thorand¹³⁾, and in relation to these pathologies MCP-1 is one of the most studied chemokines. In obese patients different depots of adipose tissue such as visceral, subcutaneous and epicardial adipose tissue show increased expression of MCP-1⁽Reference Malavazos, Ermetici and Coman^14, Reference Huber, Kiefer and Zeyda¹⁵⁾. Clinical data provide good evidence for a relationship between serum MCP-1 levels and insulin resistance, as well as type 2 diabetes. Several studies have demonstrated that patients with type 2 diabetes display elevated MCP-1 levels⁽Reference Herder, Baumert and Thorand^13, Reference Piemonti, Calori and Mercalli^16, Reference Simeoni, Hoffmann and Winkelmann¹⁷⁾. High MCP-1 levels have been shown to contribute to diabetes risk independently of previously-described clinical, metabolic and immunological risk factors⁽Reference Herder, Baumert and Thorand¹³⁾. Conversely, diabetes treatments such as exercise⁽Reference Troseid, Lappegard and Claudi¹⁸⁾, pioglitazone⁽Reference Di Gregorio, Yao-Borengasser and Rasouli¹⁹⁾ and weight loss⁽Reference Schernthaner, Kopp and Kriwanek²⁰⁾, all of which improve insulin sensitivity in obese patients, reduce MCP-1 plasma concentrations. Expression of MCP-1 has been found to be higher in visceral adipose tissue than in subcutaneous tissue and is closely related to the number of resident macrophages⁽Reference Bruun, Lihn and Pedersen²¹⁾. Conversely, obese patients that lose weight after bariatric surgery show decreased levels of MCP-1⁽Reference Schernthaner, Kopp and Kriwanek²⁰⁾, probably in parallel with lower macrophage infiltration in adipose tissue⁽Reference Cancello, Henegar and Viguerie²²⁾.

Fewer data are available for other MCP such as MCP-2, -3 and -4 but it is clear that these adipokines are elevated in obese patients⁽Reference Huber, Kiefer and Zeyda^15, Reference Huber, Kiefer and Zeyda^23–Reference Jiao, Chen and Shah²⁵⁾. Measurement of these factors in adipose tissue has shown a marked increase in expression together with an increased expression of the corresponding receptors in obese patients⁽Reference Huber, Kiefer and Zeyda¹⁵⁾.

Other CC chemokines such as RANTES (or chemokine CC motif ligand 5) and eotaxin (or chemokine CC motif ligand 11) are also elevated in the serum of obese patients as compared with lean controls⁽Reference Hashimoto, Wada and Hida^23, Reference Herder, Haastert and Muller-Scholze^26, Reference Vasudevan, Wu and Xydakis²⁷⁾. Eotaxin is overexpressed in visceral adipose tissue of obese patients as compared with lean controls and subcutaneous fat. While RANTES has also been found to be associated with type 2 diabetes in a large German study cohort, eotaxin is not associated with insulin resistance⁽Reference Herder, Haastert and Muller-Scholze²⁶⁾.

IL-8 and 10 kDa interferon γ-induced protein (or chemokine CXC motif ligand (CXCL) 10) are CXC chemokines. IL-8 is secreted from adipose tissue and its plasma levels are increased in obesity⁽Reference Bruun, Pedersen and Richelsen^28–Reference Rotter, Nagaev and Smith³⁰⁾. However, a correlation between higher levels of IL-8 in obesity and increased insulin resistance has not yet been fully established, as the association between IL-8 and diabetes⁽Reference Herder, Baumert and Thorand¹³⁾ is attenuated by multivariable adjustment for BMI and other metabolic and immunological risk factors. Another study has demonstrated that IL-8 expression is markedly increased in human fat cells from individuals who are insulin-resistant⁽Reference Rotter, Nagaev and Smith³⁰⁾. Serum levels of 10 kDa interferon γ-induced protein are increased in obese patients but are not associated with insulin resistance⁽Reference Herder, Haastert and Muller-Scholze^26, Reference Herder, Schneitler and Rathmann³¹⁾. CXCL5 has very recently been revealed to be a new adipokine that is present in markedly increased levels in obese subjects as compared with lean controls⁽Reference Chavey, Lazennec and Lagarrigue³²⁾. The same study has also shown that the serum concentration of this chemokine decreases in obese subjects after weight reduction.

Recently, the rapidly-growing adipokine family has expanded to include chemerin, a secretory chemoattractant protein. Initially discovered in body fluids associated with inflammatory processes⁽Reference Wittamer, Franssen and Vulcano³³⁾, chemerin and its receptor chemokine-like receptor 1 (CMKLR1) (or ChemR23) are also highly expressed in adipose tissue⁽Reference Bozaoglu, Bolton and McMillan^9, Reference Goralski, McCarthy and Hanniman³⁴⁾. In vivo data have shown that chemerin is elevated in adipose tissue of diabetic Psammomys obesus (sand rat; an animal model of obesity and type 2 diabetes) compared with controls⁽Reference Bozaoglu, Bolton and McMillan⁹⁾. However, there is no difference in chemerin levels between patients with diabetes and control patients despite a correlation between chemerin levels and BMI, blood TAG and blood pressure⁽Reference Bozaoglu, Bolton and McMillan⁹⁾.

Chemotactic adipokines: data from animal models and cell culture

Many chemokines have been shown to possess biological activity beyond the recruitment of immune cells, which also applies to adipose tissue-derived chemokines such as MCP-1, for which insulin resistance-inducing capacity is postulated⁽Reference Sartipy and Loskutoff^35, Reference Sell, Dietze-Schroeder and Kaiser³⁶⁾ (Table 2). MCP-1 is secreted from adipocytes in rodents⁽Reference Sartipy and Loskutoff^35, Reference Gerhardt, Romero and Cancello³⁷⁾ and human subjects⁽Reference Christiansen, Richelsen and Bruun^11, Reference Dietze-Schroeder, Sell and Uhlig³⁸⁾. Large adipocytes release higher levels of MCP-1 together with other pro-inflammatory cytokines⁽Reference Skurk, Alberti-Huber and Herder³⁹⁾. It appears, however, that adipocytes only partly contribute to the MCP-1 output from adipose tissue⁽Reference Fain and Madan⁴⁰⁾. In vitro, MCP-1 expression and secretion is highly regulated in adipocytes, i.e. increased by insulin, TNFα, growth hormone and IL-6⁽Reference Fasshauer, Klein and Kralisch⁴¹⁾, all of which are increased in obese patients. Conversely, treatment of 3T3-L1 adipocytes with MCP-1 impairs glucose uptake, indicating that this cytokine may contribute to the pathogenesis of insulin resistance⁽Reference Sartipy and Loskutoff³⁵⁾. MCP-1 does not, however, cause insulin resistance by acting only in an autocrine or paracrine manner. In primary human skeletal muscle cells it has been shown that even hypophysiological levels of MCP-1 induce robust insulin resistance⁽Reference Sell, Dietze-Schroeder and Kaiser³⁶⁾.

Table 2. In vitro evidence that chemotactic cytokines are adipokines with a possible role in insulin resistance

MCP, monocyte chemoattractant protein; CCL, chemokine CC motif ligand; MIP-1, macrophage inflammatory protein 1; GRO-α, growth-regulated oncogene α; SV, stroma vascular; CXCL, chemokine CXC motif ligand; IP-10, 10 kDa interferon γ-induced protein.

The use of mouse models has revealed that specific overexpression of MCP-1 in adipose tissue alone can mimic the effects of diet-induced obesity such as insulin resistance, macrophage infiltration into adipose tissue and liver steatosis, which occurs in the absence of any increase in body weight⁽Reference Kanda, Tateya and Tamori⁴²⁾. The same study has also shown that in contrast to MCP-1 overexpression, MCP-1 deficiency in diet-induced obese mice or inhibition of MCP-1 expression in db/db mice ameliorates insulin resistance and reduces the number of macrophages in adipose tissue⁽Reference Kanda, Tateya and Tamori⁴²⁾. On the other hand, conflicting data from another group suggest that MCP-1 deficiency does not reduce obesity-induced inflammation in adipose tissue⁽Reference Inouye, Shi and Howard⁴³⁾. Another study using mice with adipose tissue overexpression of MCP-1 has demonstrated that MCP-1 can reduce insulin sensitivity in an endocrine manner in skeletal muscle⁽Reference Kamei, Tobe and Suzuki⁴⁴⁾.

Thus, the role of MCP-1 in adipose tissue inflammation is not fully understood, which is also the case for its receptor CCR2. One study with CCR2-knock-out mice has demonstrated that disruption of MCP-1 signalling does not prevent obesity induced by a high-fat diet⁽Reference Chen, Mumick and Zhang⁴⁵⁾. Another study, however, has found that when CCR2 is lacking the efficiency of diet-induced obesity is decreased concomitantly with reduced macrophage number and an ameliorated inflammatory profile together with reduced insulin resistance⁽Reference Weisberg, Hunter and Huber⁴⁶⁾. Furthermore, pharmacological inhibition of CCR2 has been shown to improve glucose homeostasis and inflammatory markers both dependently and independently of adipose tissue⁽Reference Tamura, Sugimoto and Murayama^47, Reference Yang, IglayReger and Kadouh⁴⁸⁾.

The release of the chemokines MCP-1, macrophage inflammatory protein 1α and β, growth-regulated oncogene α and IL-8 is inhibited by adiponectin⁽Reference Dietze-Schroeder, Sell and Uhlig³⁸⁾. Adiponectin is a prominent adipokine that is decreased in obesity and that positively influences insulin sensitivity⁽Reference Lihn, Pedersen and Richelsen⁴⁹⁾. Accordingly, low plasma adiponectin levels observed in obesity are good indicators of insulin resistance and the development of diabetes⁽Reference Tschritter, Fritsche and Thamer⁵⁰⁾. It has been demonstrated that adiponectin acts as an autocrine regulator of adipocyte secretion and by decreasing the release of adipokines simultaneously prevents insulin resistance in myocytes undergoing co-culture with adipocytes⁽Reference Dietze-Schroeder, Sell and Uhlig³⁸⁾. In addition, several chemokines, including IL-8, macrophage inflammatory protein 1β and MCP-1, induce insulin resistance in skeletal muscle cells⁽Reference Sell, Dietze-Schroeder and Kaiser³⁶⁾ and thereby may represent a link between obesity and type 2 diabetes. In the case of eotaxin there are no data to suggest that it is regulated by adiponectin, but one clinical study has demonstrated a link between high eotaxin levels and hypoadiponectinaemia⁽Reference Herder, Hauner and Haastert⁵¹⁾. Eotaxin is also released from adipose tissue but stroma vascular cells appear to be the major source of this chemokine⁽Reference Vasudevan, Wu and Xydakis²⁷⁾.

CXCL5, a very recent addition to the adipokines⁽Reference Chavey, Lazennec and Lagarrigue³²⁾, is mainly secreted by the macrophage fraction of adipose tissue. Like MCP-1 this chemokine induces insulin resistance in muscle, pointing to a link between adipose tissue inflammation and insulin resistance in peripheral tissues. In addition, blocking CXCL5 signalling in insulin-resistant mice using either an anti-CXCL5 antibody or an antagonist for the corresponding receptor, chemokine CXC motif receptor 2, improves insulin sensitivity without changing body weight or food intake. Also, chemokine CXC motif receptor 2-knock-out mice display enhanced insulin responsiveness when compared with wild-type mice. It should be mentioned that CXCL5 has only been studied by one group so far, so these results need verification by other studies. Furthermore, in light of the varying phenotypes of CCR2-knock-out mice it is difficult to discuss the role of CXCL5 and its receptor chemokine CXC motif receptor 2 definitively at this point.

Chemerin and CMKLR1 are necessary for adipogenesis, as viral knockdown of expression of both proteins completely inhibits this process⁽Reference Goralski, McCarthy and Hanniman³⁴⁾. Chemerin mRNA expression increases with adipogenesis⁽Reference Bozaoglu, Bolton and McMillan^9, Reference Goralski, McCarthy and Hanniman^34, Reference Roh, Song and Choi⁵²⁾. In human adipocytes a comparison of chemerin and CMKLR1 mRNA expression before and after differentiation shows a more pronounced increase in CMKLR1 than in chemerin⁽Reference Goralski, McCarthy and Hanniman³⁴⁾. Human adipocytes also release measurable amounts of chemerin, the secretion of which is up regulated by TNFα (H Sell and J Eckel, unpublished results). In adipose tissue chemerin can also be found in the stroma vascular fraction, suggesting a contribution of various adipose tissue cell types to chemerin production. It has been demonstrated that macrophages express CMKLR1 and are chemerin responsive⁽Reference Zabel, Ohyama and Zuniga⁵³⁾. A comparison of different animal models of obesity and diabetes reveals that chemerin expression is not increased in adipose tissue of genetically-obese mice⁽Reference Goralski, McCarthy and Hanniman³⁴⁾, is lower in db/db mice⁽Reference Takahashi, Takahashi and Takahashi⁵⁴⁾ but is higher in obese insulin-resistant P. obesus ⁽Reference Bozaoglu, Bolton and McMillan⁹⁾. A single study in human subjects has reported a correlation between blood chemerin levels and BMI that is independent of glucose tolerance⁽Reference Bozaoglu, Bolton and McMillan⁹⁾. However, it is difficult to speculate on the overall contribution of adipocyte-derived chemerin to serum levels of this chemokine. Concentrations and the origin of chemerin in the liver, lung and other chemerin-producing organs have to be taken into account. Surprisingly, chemerin itself increases glucose uptake in 3T3-adipocytes⁽Reference Takahashi, Takahashi and Takahashi⁵⁴⁾, although another study has reported the opposite effect on adipocytes⁽Reference Kralisch, Weise and Sommer⁵⁵⁾ and it has been demonstrated that chemerin induces insulin resistance in skeletal muscle cells (H Sell and J Eckel, unpublished results). Chemerin expression in adipocytes is up regulated by IL-1β⁽Reference Kralisch, Weise and Sommer⁵⁵⁾. Thus, chemerin may exert different effects by its endocrine and paracrine or autocrine actions.

The current knowledge of chemerin is complicated because the actions of this protein involve targets other than chemerin and its receptor CMKLR1. New receptors have been identified as well as peptides derived from chemerin that have been shown to have completely different modes of action. Chemerin is synthesized as prochemerin, which has a low affinity to CMKLR1⁽Reference Wittamer, Franssen and Vulcano³³⁾. Prochemerin is converted rapidly to a CMKLR1 agonist by proteolytic cleavage of a carboxy-terminal peptide involving serine proteases of the coagulation and inflammation cascades⁽Reference Wittamer, Franssen and Vulcano³³⁾. Carboxy-terminal peptides derived from chemerin by cysteine protease cleavage bind to CMKLR1 with much higher affinity than chemerin itself and exert potent anti-inflammatory effects on activated macrophages⁽Reference Wittamer, Gregoire and Robberecht^56, Reference Cash, Hart and Russ⁵⁷⁾. This divergent effect of chemerin and chemerin-derived peptides can be explained by binding to receptors other than CMKLR1, which have been identified recently. Chemerin binds to two G-protein-coupled receptors, GPR1 and CCR-like 2⁽Reference Cash, Hart and Russ^57, Reference Zabel, Nakae and Zuniga⁵⁸⁾. More specifically, chemerin binds with its carboxy-terminal domain to CMKLR1, directly activating cells; however, chemerin can also bind to CCR-like 2 with its amino-terminal domain and present the carboxy-terminal domain to CMKLR1 on neighbouring cells. In contrast, chemerin-derived peptides only binding to CMKLR1 inhibit an inflammatory response, a process that is comparable with that for other chemokines such as MCP-1 or RANTES⁽Reference Zhang and Rollins^59, Reference Proudfoot, Power and Hoogewerf⁶⁰⁾. The role of the novel chemerin receptors and chemerin-derived peptides in the context of obesity and type 2 diabetes is not known.

Mechanisms of adipose tissue inflammation with a potential role for chemoattractants

Obesity is associated with a state of chronic inflammation in adipose tissue. In addition to increased release of pro-inflammatory markers, macrophage infiltration has recently been shown to be characteristic of expanding adipose tissue⁽Reference Weisberg, McCann and Desai⁶¹⁾. However, obesity is not associated with increased macrophage numbers in muscle or liver. It has been proposed that the main source of pro-inflammatory adipokines is in fact macrophages, although other cells in adipose tissue such as adipocytes, preadipocytes and vascular cells contribute to adipose tissue secretion⁽Reference Fain⁶²⁾. Clinical studies have provided evidence for a good correlation between BMI and macrophage infiltration into adipose tissue, particularly in relation to the visceral fat depot⁽Reference Zeyda, Farmer and Todoric⁶³⁾. Paracrine and endocrine signals as well as adipocyte hypertrophy and hyperplasia might contribute to macrophage infiltration into adipose tissue. In adipose tissue of obese patients crown-like structures of macrophages surrounding apoptotic adipocytes have been found⁽Reference Murano, Barbatelli and Parisani⁶⁴⁾. The expression of several chemotactic cytokines is increased in the obese state concomitantly with increased expression of chemokine receptors such as CCR2 in newly-recruited macrophages, making it possible that ligands for this receptor contribute to macrophage attraction and activation⁽Reference Lumeng, Deyoung and Bodzin⁶⁵⁾. Characterization of adipose tissue-resident macrophages has shown that the latter express surface markers for alternatively activated macrophages (M2) that are able to secrete anti-inflammatory cytokines in addition to pro-inflammatory cytokines, a process that may be necessary for the uptake of large, apoptotic or necrotic adipocytes⁽Reference Lumeng, Deyoung and Bodzin⁶⁵⁾. Weight reduction in human subjects is accompanied by the occurrence of more M2-like macrophages in adipose tissue⁽Reference Clement, Viguerie and Poitou⁶⁶⁾ while diet-induced obesity is characterized by switching the macrophage phenotype towards classical inflammatory M1 status⁽Reference Lumeng, Bodzin and Saltiel⁶⁷⁾. However, it must be emphasized that the mechanisms of macrophage recruitment to adipose tissue in obesity are not yet understood.

The study of hypoxia in adipose tissue in the context of obesity is timely, as some very enlightening studies have put this theory in a physiological context in recent years⁽Reference Wood, Pérez de Heredia and Wang⁶⁸⁾. Hypoxia has been observed in both physiological and pathological situations. In relation to adipose tissue, it has been demonstrated in mice that oxygenation is comparable with general tissue oxygenation in lean animals, while their obese littermates are characterized by an approximately 60% lower O₂ pressure in fat⁽Reference Ye, Gao and Yin⁶⁹⁾. In adipose tissue of mice hypoxia underlies the increased production of adipokines and the development of obesity and the metabolic syndrome⁽Reference Hosogai, Fukuhara and Oshima⁷⁰⁾. Furthermore, it has been demonstrated in human subjects that hypoxia occurs in the obese state⁽Reference Cancello, Henegar and Viguerie²²⁾. Mechanistically, hypoxia leads to activation of the transcription factor hypoxia inducible factor 1α, which has a key role in the adaptive response to decreased O₂ availability in tissues. Hypoxia inducible factor 1α increases the transcription of various genes that affect, for example, cell proliferation, angiogenesis, glucose metabolism and the extracellular matrix⁽Reference Semenza⁷¹⁾. Hypoxia studies in isolated adipocytes have shown that hypoxia causes various changes in protein expression and secretory behaviour in this cell type. Hypoxia in isolated adipocytes leads to the same dysregulation of secretory function as that observed in expanded adipose tissue, including increased release of IL-6, leptin and vascular endothelial growth factor⁽Reference Wang, Wood and Trayhurn⁷²⁾. In contrast, the release of adiponectin is decreased in hypoxia, possibly through activation of endoplasmic reticulum stress⁽Reference Hosogai, Fukuhara and Oshima⁷⁰⁾. The release of RANTES is increased by hypoxia⁽Reference Skurk, Mack and Kempf⁷³⁾, while MCP-1 secretion is slightly decreased ⁽Reference Wang, Wood and Trayhurn⁷²⁾. The regulation of other chemotactic proteins by hypoxia is not yet known.

Another mechanism of adipose tissue inflammation associated with hypoxia currently under investigation is endoplasmic reticulum stress. There are several explanations of why endoplasmic reticulum stress occurs particularly in fat in obesity, including increased protein synthesis as a result of increased energy availability or even glucose deprivation as a result of insulin resistance in adipose tissue⁽Reference Gregor and Hotamisligil⁷⁴⁾. Hypoxia has also been proposed to be a cause of endoplasmic reticulum stress⁽Reference Hosogai, Fukuhara and Oshima⁷⁰⁾. Furthermore, hypoxia and endoplasmic reticulum stress might be closely related, as signalling pathways for both forms of stress merge in common pathways such as activation of mammalian target of rapamycin or c-Jun N-terminal kinase⁽Reference Hosogai, Fukuhara and Oshima⁷⁰⁾.