WAT as an endocrine organ

WAT has long been thought to be a mere storage tissue, only involved in FA release depending on the metabolic needs. But WAT is also a real endocrine organ able to receive hormonal signalling coming from the whole body and to react by secreting hormones named adipokines. These hormones have an impact on many target tissues, such as the liver or skeletal muscles and participate in the general control of energy balance. Some of these adipokines, like leptin or adiponectin activate neural circuits in the hypothalamus, mainly aimed at regulating EE and lipid catabolism [114]. Moreover, under stress, WAT secretes pro- or anti-inflammatory cytokines regulated by the WAT mass and the physiological state of the organism [115].

Table 1: Adipokines secreted by WAT and their impact on the tissue

name expression Main functions Leptin adipocytes Reflects fat mass

Satiety hormone, direct action on the hypothalamus

Stimulates lipolysis, inhibits lipogenesis, increases FA oxidation Adiponectin adipocytes Increases insulin sensitivity

Increases FA oxidation Anti-inflammatory action Il-6 adipocytes Pro-inflammatory cytokine

Inhibits insulin and leptin pathways TNFα macrophages Pro-inflammatory cytokine

Induces insulin resistance Increases lipolysis in adipocytes Expression increased in obesity Pref-1 preadipocytes Inhibits adipogenesis

Its overexpression in WAT impairs insulin sensitivity [116]

36 TGFβ adipocytes/

macrophages

Growth factor

Anti-inflammatory cytokine

Induces proliferation, differentiation and apoptosis [117]

Expression increased in obesity MCP1 adipocytes/

macrophages

Anti-inflammatory chemokine

Recruits macrophages on inflammatory sites Increases lipolysis and leptin secretion 2.6 Lipid homeostasis

One of the main functions of WAT is to store excess energy as TG, which can be degraded into FAs to be used by other tissues in response to metabolic needs, such as during food restriction periods. Synthesis of TG from FAs is named lipogenesis. Adipocytes are able to store large amounts of TG in their lipid droplet surrounded by a protein named perilipin, without causing any lipotoxicity to the cell [118].

2.6.1 Lipogenesis

TG stored in adipocytes are synthetized from fatty acids (FA) and glycerol, after being transformed, respectively into acyl-CoA and glycerol-3-phosphate (G3P). Most FAs come from circulating plasma lipids, whereas G3P has two main origins: glycolysis and glyceroneogenesis.

Plasma circulating FA levels are mainly non esterified FA (NEFA) linked to albumin, or TG incorporated within lipoproteins, mainly very low density lipoproteins (VLDL) or chylomicrons.

TG are first hydrolysed by lipoprotein lipase (LPL), an enzyme linked to capillaries of WAT and skeletal muscles, in order to release FAs [119]. Its expression and activity is increased in WAT during the postprandial period, mainly under diet enriched in carbohydrates, due to a

37 stimulatory effect of insulin. In contrast, LPL expression and activity are decreased during a starvation period or in response to a high fat diet (HFD) [120].

Long chain fatty acids (LCFA) need specific transporters to cross the plasma membrane [121]. White adipocytes express different FA transporters, such as "cluster of differentiation 36" (CD36), "fatty acid transport protein" (FATP) and "fatty acid binding protein" (FABP). CD36 is responsible for the majority of FA uptake [122]. Insulin improves this transport by stimulating the expression of these transporters, as well as their presence at the plasma membrane level [123].

FA are soluble in the cytosol only. To avoid cytotoxic effects, they are linked to a cytosolic protein, the FABP, which transports FAs to the action site of acyl-CoA synthase.

Human white adipocytes express two type of FABP: "adipocyte protein2" (AP2) (product of the FABP4 gene) and the "keratinocyte lipid-binding protein" (KLBP). AP2 is exclusively expressed in adipocytes and is the main molecule involved in FA transport, whereas KLPB is also expressed in macrophages [124]. The first step in FA metabolism after being linked to FABP is the activation of LCFA to LCFA-CoA through the acyl-CoA synthase. LCFA-CoA are then oxidized or orientated to the synthesis of more complex lipids, such as TG. Oxidation of LCFA-CoA occurs in the mitochondria after their transport inside this organelle by carnitine-palmitoyl transferase I (CPT1).

Another source of FA is through de novo lipogenesis, which is the synthesis of new FA molecules from non-lipid substrates, mainly carbohydrates in mammals. Expression and activity of enzymes involved in lipogenesis and glycolysis are closely related in the liver and WAT, which are lipogenic tissues. De novo lipogenesis is less active in humans than in rodents and it contributes to a small amount of TG production in adipocytes [125, 126].

38 Synthesis of G3P is needed for the esterification of FA into TG. G3P is produced from glucose (via glycolysis) or from glyceroneogenesis [127]. The limiting step of this latter process is the cytosolic form of phosphoenol pyruvate carboxykinase (PEPCK). Relative contribution of glycolysis and glyceroneogenesis to produce G3P depends on nutritional and pharmacological factors. Global availability of G3P controls FA esterification coming from de novo lipogenesis or circulating lipids, as well as partial re-esterification of FA released by hydrolysis of TG.

Figure 6: Lipogenesis pathways in white adipocytes. These pathways enable energy storage as lipids. FA are transported as lipoproteins (chylomicrons and VLDL), FA and glucose are transformed into acylCoA and G3P, respectively, which are both needed for triacylglycerol synthesis (TAG on the Figure). DAG: diacylglycerol, MAG: monoacyglycerol, LPA:

lysophosphatidic acid, PA: phosphatidic acid.

39 2.6.2 Lipolysis

In WAT, TG are hydrolysed when energy needs are not completely covered by food intake. TG in adipocytes are successively hydrolysed into diacylglycerols (DAG) and monoacylglycerols (MAG) to ultimately produce three molecules of FA and one molecule of glycerol. Glycerol produced by lipolysis (TG degradation into FAs and glycerol) is released into the bloodstream to be transformed to G3P by glycerol kinase in the liver or to be used by other tissues. This glycerol release depends in part on aquaporine, a canal protein in the plasma membrane. FAs produced by lipolysis are also released in the circulation or re-esterified directly as TG, inside the adipocytes. The intracellular FA cycling depends on G3P availability and on the re-esterification of FAs thanks to the triglyceride synthase. In the basal state, this cycling is negligible, but it increases during starvation periods, stress [128] or in pathological situations, such as hyperthyroidism [129]. The extent of FA re-esterification plays a role in the regulation of plasma FA levels.

Hormone sensitive lipase (HSL) plays a key role in hydrolysis of TG. Lipolysis is activated by hormones or mediators activating the adenylate cyclase system (glucagon, adrenaline, noradrenaline), inducing an intracellular increase in cyclic adenosine mono phosphate (cAMP) that phosphorylates protein kinase A, which will then phosphorylate HSL and perilipin.

Inhibition of HSL expression in mice induces an accumulation of DAG instead of TG. HSL hydrolyses DAG and then a MAG lipase releases the last FA and one glycerol. MAG lipase is the limiting enzyme in the release of glycerol and free fatty acids (FAs).

40 Adipose triglyceride lipase (ATGL) is another lipase present in WAT. Inhibition of both ATGL and HSL induces an inhibition of 90% of TG hydrolysis [130]. In fact, TG hydrolysis depends mainly on ATGL activity and less on HSL activity.

Figure 7: Lipolysis pathways in white adipocytes. TG (TAG on the Figure) are degraded between meals when body needs energy. The FAs produced will be then be used by the mitochondria (β oxidation) to produce ATP or they will be exported to the tissues which are able to use FAs (mainly skeletal muscles, heart, liver, adipose tissue).

3) BAT

BAT is characteristic of mammals, more precisely of hibernating animals because this tissue is responsible for heat production by consuming lipids [131]. The association between BAT and hibernation was described for the first time by Gessner in 1551 in marmots.

41 In contrast to white adipose tissue, brown adipose tissue (BAT), which is darker due to the presence of cytochrome c in the mitochondria and a higher degree of vascularization, is specialised in adaptive thermogenesis [132]. Even if the role of BAT has been well studied in rodents and human new-borns, its persistence and importance in adult humans are currently actively studied and its functions remain to be precisely determined [133-135].

3.1 Brown adipocyte

BAT is composed of brown adipocytes mainly. In contrast to white adipocytes that contain a unique and big lipid droplet, brown adipocytes are multilocular, have a central nucleus and a large cytoplasm that contains a lot of mitochondria. This tissue is highly vascularized and innervated.

Figure 8: Brown adipocyte. Histology of brown adipose tissue stained with hematoxylin eosin (HE). As schematized on the right panel, the cell has a central nucleus, a large cytoplasm full of active mitochondria and lot a small lipid droplets.

Brown adipocytes have a common origin with skeletal muscles, which is not the case for white adipocytes [136]. Thus, it was demonstrated by cell tracing that both brown adipocytes and myocytes express Myf5, strongly suggesting that they share a common precursor [137].

42 3.2 UCP1 in mitochondria

In eukaryotes, mitochondria are responsible for oxidative metabolism. It contains enzymes that catalyse the citric acid cycle and FA oxidation, ultimately producing ATP as an energy source for the cell. This organelle has a smooth outer and an inner membrane. Porine is a constituent of the outer membrane, which allows free circulation of proteins with a molecular weight lower than 10kDa. The inner membrane contains a lot of invaginations named cristae. The number of cristae (increasing the membrane surface) reflects the respiratory activity of the cell. The proteins that achieve the electron transport, linked with oxidative phosphorylation, lie within the inner membrane. This membrane contains 75% of proteins and is richer in this constituent than the outer membrane. It is only permeable to O2, CO2 and water. It also contains many proteins involved in the transport of metabolites, such as ATP, ADP, pyruvate, Ca²⁺ and phosphate. Internal compartment is constituted of a gel (50%

of water) named matrix, containing high concentrations of soluble enzymes involved in oxidative metabolism (such as citric acid cycle enzymes). Overall, the various systems present in mitochondrial membranes allow for the maintenance of ionic gradients through the membranes.

Major implication of the mitochondria in energy metabolism is due to the occurrence of many metabolic pathways within this organelle, as well as its ability to convert energy into ATP. Pr Peter D. Mitchell received a Nobel Prize in 1978 for his chimio-osmotic theory [138]. According to this theory, electron transfer coming from NADH or FADH2 to oxygen, via different complexes of the respiratory chain, is associated with protons’ expulsion from the mitochondrial matrix into the intermembrane space. This proton flux creates an

43 electrochemical gradient on either side of the inner membrane of the mitochondria.

Dissipation of this gradient by ATP synthase in the presence of ADP allows for ATP synthesis.

Thus, ATP production depends, via the protons’ gradient, on the function of the respiratory chain. It is named coupled respiration [139].

Figure 9: Mitochondrial coupled respiration. Taken from: Mitochondrial uncoupling proteins in the CNS: in support of function and survival. Horvath et al, Nature Reviews Neuroscience 6, 829-840, November 2005. Glucose or FA oxidation produces NADH and FADH2, releasing electrons that will be used by the electron transport chain or complexes one to four. This releases protons into the intermembrane space. Proton’s reuptake by ATP synthase produce ATP from ADP.

This mechanism implies that the inner membrane is impermeable to protons. It is the case for the majority of cells, but one exception exists in animals, in the mitochondria of brown adipocytes responsible for thermogenesis. This process is possible thanks to a high permeability for protons through the inner membrane. Indeed, it allows dissipating the electrochemical gradient independently from ATP synthase. The consequence is that the

44 respiratory chain and ATP production are dissociated and energy is released as heat. The whole process is called uncoupled respiration. A few years after its initial description by Nicholls [138], a protein, which was a member of the mitochondrial anionic transporters, was discovered and its gene cloned [140]. As this protein, inserted in the inner membrane, acted as a proton channel and was responsible for an increased inner membrane permeability to protons, it was called the “uncoupling protein 1” or UCP1. Thus, UCP1 decreases ATP production by using protons from the respiratory chain [141].

Figure 10: Uncoupled respiration. Adapted from: Mitochondrial uncoupling proteins in the CNS: in support of function and survival. Horvath et al, Nature Reviews Neuroscience 6, 829-840, November 2005. UCP1 dissipates the protons’ gradient resulting from the electron transport chain to generate heat, decreasing ATP production by ATP synthase.

Importance of UCP1 in the metabolic regulation was highlighted by several transgenic models in mice. Surprisingly, Ucp1 KO mice have the same phenotype as wild-type mice at 22°C. At this temperature, mice develop other pathways to induce thermogenesis, such as shivering thermogenesis, which probably explains why they can maintain a normal body weight gain and fat content. However, at thermoneutrality (30°C), Ucp1 KO mice become

UCP1

45 obese, probably due to the fact that BAT, via UCP1, is the unique way to dissipate energy excess when animals do not need to maintain their body temperature [142]. It was also shown that under a HF diet, Ucp1 KO mice become more obese than wild-type mice, demonstrating that UCP1 activity is indispensable for diet-induced thermogenesis [143].

3.3 Beige adipocyte

Occurrence of brown-like adipocytes in rodents in WAT was shown to contribute to non-shivering thermogenesis [144]. These recruitable adipocytes were named BRITE for

“BRown to whITE” or beige adipocytes. Their morphology is very close to that of brown adipocytes. They have a central nucleus and various multilocular lipid droplets. However, in contrast to brown adipocytes, they are not coming from a Myf5+ cell lineage [137]. Moreover, differences between mice strains were observed in the development of UCP1-positive cells in WAT, whereas no differences were observed in BAT, suggesting a different genetic regulation of beige and brown adipocytes [145]. Interestingly, it is now possible to distinguish mature brown, beige and white adipocytes, depending on cell surface markers [146, 147]. Lhx8 (LIM homeobox protein 8), Zic1 (zinc finger protein of the cerebellum 1) for brown adipocytes, Cd137 (tumour necrosis factor receptor superfamily, member 9), Tmem26 (transmembrane protein 26) or Tbx1 (T-Box 1) for beige adipocytes and Tcf21 (transcription factor 21) or TLE3 (transducin-like enhancer of split 3) for white cells are examples of markers, which are specific of one tissue [148]. Finally, if brown adipocytes have a thermogenic activity under standard conditions, beige cells need a stimulation to be activated and to produce heat [149, 150].

46 Figure 11: Brown, beige and white adipocytes repartition in mice and humans. Taken from Adipose tissue browning and metabolic health. Alexander Bartelt et al, Nat Rev Endocrinol.

2014 Jan;10(1):24-36. In mice, beige adipocytes are localised in the perigonadal and in the inguinal regions, whereas in humans, they are disseminated in the neck and the supraclavicular areas. Specific markers exist for each depot.

3.4 In humans

BAT is present in human new-borns, as initially described by Shinkishi Hatai in 1902. Its presence was then demonstrated during the first decade of life, but it was thought that this tissue was completely disappearing after the third decade in adults [151]. Interestingly, a combination of PET and computed tomography (CT) with the glucose analogue ¹⁸ F-fluorodeoxyglucose (18F-FDG) is often used to detect tumours. In this context, BAT was commonly observed but interpreted as an artefact [152]. This tissue was rediscovered when,

47 for the first time, this technique was used for BAT detection and activity in vivo in adult humans [153]. Indeed, in 2009, three studies simultaneously demonstrated the presence of BAT and its metabolic activity in adults [154-156]. In children, BAT is mainly present in the interscapular area, as in rodents, but some depots are also present along the aorta and in the retroperitoneal area [157, 158].

The proof of an activated BAT in humans was obtained in a study using ¹¹C-acetate PET imaging. In fact, it is well known from studies in rodents that the main fuel of BAT is FAs and not glucose [159]. With this tracer, higher oxidative metabolism and FA uptake were observed in BAT of all subjects compared to WAT and skeletal muscles [159]. These results were recently confirmed by exposing the subjects to cold [160].

It was initially reported that BAT found in supraclavicular area in humans is more related to beige than to classical BAT cells, as they highly express beige markers, such as CD137, TMEM26, and TBX1 [146]. However, more recently, it was demonstrated that fat depots in the neck region are heterogeneous and that some parts are composed of classical BAT adipocytes, because they can produce heat without stimulation [161]. Moreover, this tissue increases its uncoupled respiration under noradrenaline or beta-3 adrenergic stimulation, indicating the presence of beige adpocytes [162].

48 Figure 12: Precise brown or beige adipose tissue localisation in humans performed by PET histology. Taken from: Brown adipose tissue in adult humans: a metabolic renaissance, Lee et al, The Endocrine Society, June 2013, 34(3):413–438.

4) UCP1 Regulation and Function

4.1 Transcriptional regulation in brown and beige adipocytes

The Central nervous system (CNS) can stimulate thermogenesis via skeletal muscles, the heart or BAT. Skeletal muscles and the heart produce heat indirectly as a result of contraction. BAT is the only tissue producing non-shivering thermogenesis as a main function [157].

Chemoreceptors in the skin relay signals to the hypothalamus to activate the SNS through beta3 adrenoreceptors (β3R) [163, 164]. Macrophages in this tissue produce catecholamine

49 under cold exposure, reinforcing activation of this pathway [165]. As BAT is highly innervated and rich in β3R, it is highly sensitive to a β3 adrenergic stimulation, allowing for a strong and fast response to stimuli [166]. In brown adipocytes, β3 adrenergic receptors, activated by catecholamines induce a signal transduction cascade, ultimately resulting in increased Ucp1 expression. Thus, noradrenaline induces a release of cAMP, activating protein kinase A (PKA) and leading to the activation of mitogen-activated protein kinase (MAPKs) [167, 168]. PKA phosphorylation activates different pathways, such as the cAMP-response element binding protein (CREB), inducing expression of peroxisome proliferator-activated receptor gamma, coactivator 1 alpha (PGC1α) and peroxisome proliferator-activated receptor gamma (PPARγ), notably, two transcriptional factors promoting UCP1 expression [169]. In fact, PGC1α is a transcriptional coactivator interacting with PPARγ in brown and beige adipocytes. This transcriptional factor is a key regulator of UCP1, mitochondrial biogenesis and oxidative metabolism, thereby of the whole thermogenic program [169]. Paradoxically, PGC1α is not required for UCP1 expression and activation in BAT. Thus, PGC1α KO mice have a normal UCP1 expression in BAT [170]. This phenomenon may be due to other PGC1β family members that would compensate the role of PGC1α, therefore allowing for a normal UCP1 expression.

However, PGC1α remains essential for UCP1 activation by β3 agonist treatment in brown and beige adipocytes [171-173].

PR domain zinc finger protein 16 (PRDM16), a transcriptional regulator, induces the program of brown adipocyte differentiation by increasing UCP1 expression in classical brown adipocytes through an activation of PGC1α or β. This cofactor is highly enriched in brown adipocytes compared to white adipocytes and is essential for UCP1 expression [174]. In inguinal subcutaneous WAT, its expression induces an increase in UCP1, and its expression is required for WAT beiging under β3 agonist treatment [175]. In WAT, an ectopic expression of

50 PRDM16 or COX-2, a downstream effector of β-adrenergic signalling, increases the number of beige cells, inducing a protection against diet-induced obesity [175, 176].

Figure 13: Transcriptional regulation of UCP1. Taken from: Brown and beige fat: development, function and therapeutic potential. Matthew Harms & Patrick Seale

Other pathways exist to activate thermogenesis in BAT but they are less efficient. As an example, phospholipase A2 (PLA2), present in the inner membrane of the mitochondria, generates long chain fatty acids (LCFA) that directly bind and activate UCP1 [177]. CIDEA, a

51 protein highly expressed in BAT, induces a suppression of UCP1 activity. Thus, CIDEA KO mice are lean and resistant to diet-induced obesity and diabetes [178].

51 protein highly expressed in BAT, induces a suppression of UCP1 activity. Thus, CIDEA KO mice are lean and resistant to diet-induced obesity and diabetes [178].