Vitamin D: a brief overview
Vitamin D (VD; calciferol) is a hormone mainly described for its role as a regulator of phosphate and calcium homeostasis( Reference Holick 1 ). It can be obtained through animal (VD3, cholecalciferol) or plant (VD2, ergocalciferol) food sources. Only a few foodstuffs contain significant amounts of VD, the main sources being fish liver oils, fatty fish (sardines, herring and mackerel) and egg yolk( Reference Holick, Binkley and Bischoff-Ferrari 2 , Reference Schmid and Walther 3 ), but small quantities are also found in fortified milk, orange juice, bread and cereals. Alternatively, VD3 is produced endogenously in the skin after UVB irradiation from the precursor 7-dehydrocholesterol to give pre-VD3, which is further isomerised to VD3 before being released into the circulation( Reference Holick 4 ). Classical estimates have assigned a majority (70–90 %) of VD supply to dermal synthesis, but a recent paper revised this figure down to just 10–25 % of VD supply( Reference Heaney, Armas and French 5 ) and posited that dietary intake of 25-hydroxy-vitamin D (25(OH)D) is a significant contributor to total VD input.
Adipose tissue is a major storage site for vitamin D
Despite limited data, it is widely accepted that adipose tissue is a reservoir for VD in human subjects and rats( Reference Mawer, Backhouse and Holman 6 – Reference Marcotorchino, Tourniaire and Landrier 10 ). Interestingly, visceral fat was found to contain 20 % more VD than subcutaneous fat( Reference Beckman, Earthman and Thomas 11 ). Heaney et al.( Reference Heaney, Horst and Cullen 12 ) calculated that 65 % of total VD in the body is in the form of D3, for which adipose tissue and skeletal muscle appear to be the main body stores (accounting for 73 and 16 %, respectively). Regarding 25(OH)D, 34 % of it is found in adipose tissue, 30 % in serum and 20 % in skeletal muscle. However, 25(OH)D was recently detected in human subcutaneous adipose tissue( Reference Piccolo, Dolnikowski and Seyoum 13 ). Fat tissue may thus contain about 60 % of total body VD, but the amount of VD present in fat tissue varies strongly between individuals and is not correlated to serum 25(OH)D levels( Reference Pramyothin, Biancuzzo and Lu 8 ), whereas 25(OH)D adipose tissue content seems to be correlated to 25(OH)D plasma levels( Reference Piccolo, Dolnikowski and Seyoum 13 ). Other factors such as VD status and amount of intake can also influence adipose tissue storage. Indeed, it was calculated that for low VD intakes and when serum 25(OH)D concentration is below 88 nm/l, almost all VD is converted to 25(OH)D in the liver with very little deposited in tissues( Reference Heaney, Armas and Shary 14 ), indicating that low 25(OH)D status is associated with limited tissue stores as well.
A recent pilot study using time-of-flight secondary ion mass spectrometry confirmed that both VD and 25(OH)D were present in human adipose tissue, and also reported for the first time that 1,25-dihydroxy-vitamin D (1,25(OH)2D) was also detectable in fat tissue( Reference Malmberg, Karlsson and Svensson 15 ). This pilot study also suggested that all these molecules were located in adipocyte lipid droplets and that VD and 25(OH)D concentrations in adipose tissue were lower in obese than lean subjects, although this last observation was only generated with a very limited number of samples( Reference Malmberg, Karlsson and Svensson 15 ).
Parallel between systemic vitamin D metabolism and adipose tissue metabolism
Uptake of vitamin D
VD from food is partially absorbed in the distal part of the small intestine, in emulsion with bile salts( Reference Borel, Caillaud and Cano 16 ). Its intestinal absorption occurs not only by passive diffusion but also via at least two cholesterol carriers( Reference Reboul, Goncalves and Comera 17 , Reference Reboul 18 ). VD and its metabolites are majorly transported in the plasma bound to vitamin D-binding protein (DBP), a globulin produced in the liver( Reference Daiger, Schanfield and Cavalli-Sforza 19 ), but also to albumin, LDL and chylomicrons for dietary VD( Reference Haddad, Jennings and Aw 20 ). Plasma DBP is in large excess compared with VD and metabolites, thus leaving only a very limited amount of circulating unbound VD and metabolites( Reference Speeckaert, Huang and Delanghe 21 ). Interestingly, only the unbound part of VD is considered biologically active and able to diffuse in any target cells( Reference Speeckaert, Huang and Delanghe 21 ), thus leading to the free-hormone hypothesis( Reference Mendel 22 ). Indeed, despite having low plasma levels of the different forms of VD compared with wild-type animals, Dbp-null mice do not show any signs of disrupted calcium homeostasis, suggesting that free VD levels can cover the needs of physiological functions as long as diet is VD-sufficient( Reference Safadi, Thornton and Magiera 23 ). Subsequent work showed that kidney content of 1,25(OH)2D was not different from that of wild-type animals( Reference Zella, Shevde and Hollis 24 ). Taken together, these data suggest that tissues may uptake VD and metabolites from the free pool through a DBP-independent mechanism.
The molecular mechanisms involved in VD uptake/secretion by adipose tissue have not yet been investigated but may well involve the megalin/cubulin pathway (described later) as suggested by Abboud et al.( Reference Abboud, Gordon-Thomson and Hoy 25 ) and/or cholesterol transporters as described in the intestine( Reference Reboul, Goncalves and Comera 17 ) or for other lipophilic micronutrients in adipose tissue( Reference Moussa, Gouranton and Gleize 26 ).
25-hydroxylation
Whatever its origin (endogenous or exogenous), calciferol is taken to the liver via the circulation where the VD 25-hydroxylase enzyme catalyses the synthesis of 25(OH)D. 25(OH)D is the major circulating form of VD and its serum concentration is classically used as a marker of VD status. 25(OH)D has a relatively long half-life (15 d), and mean plasma 25(OH)D concentration varies between 20 and 50 ng/ml (50–125 nm/l)( Reference Jones, Schoenmakers and Bluck 27 ). Several enzymes can accomplish this first hydroxylation of 25(OH)D, but CYP2R1 seems to be the key one( Reference Cheng, Motola and Mangelsdorf 28 , Reference Schuster 29 ). Interestingly, Cyp2r1−/− mice only display a 50 % reduction in serum 25(OH)D compared with wild-type or heterozygous animals, suggesting that other enzymes help maintain circulating 25(OH)D levels and/or compensate for CYP2R1 dysfunction( Reference Zhu and DeLuca 30 ).
In human subjects, other P450 cytochromes such as CYP3A4( Reference Gupta, Hollis and Patel 31 ), CYP2J2( Reference Aiba, Yamasaki and Shinki 32 ) and CYP27A1( Reference Guo, Strugnell and Back 33 ) display 25-hydroxylase activity towards VD molecules, but less efficiently (i.e. with a high K M relative to physiological substrate concentration)( Reference Zhu and DeLuca 30 ). CYP2J3( Reference Yamasaki, Izumi and Ide 34 ), CYP2D25 and CYP2C11 also show VD 25-hydroxylase activity but are only expressed in pigs and male rats, respectively( Reference Postlind, Axen and Bergman 35 , Reference Rahmaniyan, Patrick and Bell 36 ).
25-Hydroxylation seems to be functional in adipose tissue, as Zoico et al.( Reference Zoico, Franceschetti and Chirumbolo 37 ) recently reported that 25(OH)D release in the cell culture medium increased after 24 h incubation of 3T3-L1 adipocytes with VD. This production of 25(OH)D could be due to the presence of Cyp27A1, which is up-regulated by VD treatment. Interestingly, human adipose tissue biopsies have confirmed the expression of CYP27A1, CYP2R1 and CYP2J2( Reference Wamberg, Christiansen and Paulsen 38 ), suggesting that human adipose tissue and adipocytes are able to convert VD to 25(OH)D.
1α-hydroxylation
In renal proximal tubule cells, urinary loss of DBP–25(OH)D complexes is prevented by uptake via the membrane receptors megalin (also known as low-density lipoprotein receptor-related protein 2) and cubilin( Reference Nykjaer, Dragun and Walther 39 , Reference Dusso, Brown and Slatopolsky 40 ). After internalisation into vesicles, DBP is degraded into lysosomes and 25(OH)D is handled by intracellular DBP. Intracellular DBP have been identified in VD-resistant new-world primates (four isoforms have been reported so far), are related to the human heat-shock protein 70 family, and are thought to mediate 25(OH)D interactions with intracellular proteins( Reference Gacad, Chen and Arbelle 41 ). An additional binding protein termed cytosolic DBP has also been isolated from human intestinal cells( Reference Teegarden, Nickel and Shi 42 ). 25(OH)D is then either secreted into circulation or directed towards mitochondrial 1α-hydroxylase CYP27B1 to be metabolised into 1,25(OH)2D, the active form of VD. CYP27B1 is the key enzyme of 1α-hydroxylation and its activity is regulated by parathyroid hormone, fibroblast growth factor 23, calcium and phosphorus and self-regulated by 1,25(OH)2D via a negative-feedback mechanism( Reference Holick 1 ). 1,25(OH)2D has a very short half-life (about 4 h) and is 1000 times less concentrated than 25(OH)D in the plasma.
The ability of adipocytes to convert 25(OH)D into 1,25(OH)2D was initially demonstrated in 3T3-L1 cells via the activation of a gene reporter system and through the identification of radiolabelled 1,25(OH)2D derived from radiolabelled 25(OH)D( Reference Li, Byrne and Chang 43 ). The production of 1,25(OH)2D from 25(OH)D was then confirmed by Ching et al.( Reference Ching, Kashinkunti and Niehaus 44 ) and Nimitphong et al.( Reference Nimitphong, Holick and Fried 45 ). CYP27B1 expression has also been detected in murine adipocytes( Reference Li, Byrne and Chang 43 ) and in human adipose tissue biopsies( Reference Wamberg, Christiansen and Paulsen 38 ).
24-hydroxylation
Finally, vitamin D 24-hydroxylase (CYP24A1) is in charge of inactivating 1,25(OH)2D. This inactivation is self-regulated, since 1,25(OH)2D induces the expression of CYP24A1 that converts 25(OH)D and 1,25(OH)2D into less-active metabolites (e.g. 24,25(OH)2D and 1,24,25(OH)3D), which are further catabolised into inactive calcitroic acid( Reference Dusso, Brown and Slatopolsky 40 ).
In adipose tissue, CYP24A1 expression has been detected in murine and human adipocytes( Reference Li, Byrne and Chang 43 , Reference Nimitphong, Holick and Fried 45 ). In addition, the mRNA levels of CYP24A1 are strongly induced by 1,25(OH)2D incubation( Reference Li, Byrne and Chang 43 , Reference Nimitphong, Holick and Fried 45 ). CYP24A1 expression has also been confirmed in human adipose tissue biopsies( Reference Wamberg, Christiansen and Paulsen 38 ).
Vitamin D signalling
Even if few VD receptor (VDR)-independent effects of 1,25(OH)2D have been documented( Reference Lips 46 ), most biological activities of VD are mediated by the VDR, a member of the nuclear receptor superfamily that is the only nuclear receptor that binds 1,25(OH)2D with high affinity( Reference Carlberg and Seuter 47 , Reference Bikle, Gee and Pillai 48 ). VDR expression has been demonstrated in almost all human tissues( Reference Bouillon, Carmeliet and Verlinden 49 ), which means that all cells are potential targets of 1,25(OH)2D action. The VDR–1,25(OH)2D complex is associated with the retinoid X receptor( Reference Issa, Leong and Eisman 50 ), and the retinoid X receptor–VDR–1,25(OH)2D complex binds to the DNA of sites called VD response elements in the promoter region of genes whose expression is either activated or repressed( Reference Carlberg and Seuter 47 ). There are more than 1000 genes that are directly or indirectly regulated by 1,25(OH)2D and involved in various physiological processes such as cell proliferation, differentiation, apoptosis and angiogenesis( Reference Plum and DeLuca 51 ).
The presence of VDR in adipose tissue was first reported in the early 1990s( Reference Kamei, Kawada and Kazuki 52 ) and has since been widely confirmed. Interestingly, it was recently found that VDR expression is increased in obese compared with lean subjects( Reference Wamberg, Christiansen and Paulsen 38 , Reference Clemente-Postigo, Munoz-Garach and Serrano 53 ), but the physiological relevance of this up-regulation has not yet been elucidated.
Another VD-dependent signalling pathway has been described that involves ERp57 (also known as GRP58 or 1,25D3-MARRS), a protein disulfide isomerase involved in stress response( Reference Turano, Gaucci and Grillo 54 ) that mediates rapid cellular responses (i.e. within seconds or minutes) to 1,25(OH)2D stimulation( Reference Nemere, Safford and Rohe 55 , Reference Nemere and Hintze 56 ). ERp57–1,25(OH)2D complexes are internalised, which opens the possibility that ERp57 might also participate in 1,25(OH)2D intracellular trafficking, especially since ERp57 has been found to participate in nuclear complexes, including heat-shock protein 70, one of the human intracellular DBP. However, it is not yet known whether this signalling pathway is active in other tissues, particularly adipose tissue.
Taken together, these data demonstrate that on top of being a major storage site for VD, adipose tissue also expresses enzymes involved in VD metabolism and signalling, which points to the hypothesis that adipose tissue could be a target tissue that is also able to synthesise 25(OH)D and 1,25(OH)2D that could be locally active via paracrine, autocrine or even intracrine processes( Reference Lisse, Adams and Hewison 57 ). The regulation of this local metabolism has never been studied but certainly warrants future investigation. Nevertheless, there is increasing evidence of the impact of VD and its active metabolites on adipose tissue, notably in terms of adipogenesis control, adipokine expression and a host of other metabolic regulations.
In vitro and in vivo effects of vitamin D on adipose tissue and adipocyte biology
Regulation of adipogenesis
Many studies have examined the role of 1,25(OH)2D in the proliferation and differentiation of murine 3T3-L1 pre-adipocytes( Reference Kamei, Kawada and Kazuki 52 , Reference Ishida, Taniguchi and Baba 58 – Reference Mutt, Hypponen and Saarnio 61 ). Low 1,25(OH)2D concentrations were associated with an inhibition of adipogenesis and a reduction of TAG accumulation in 3T3-L1 cells, even if the opposite effects, i.e. induction of adipogenesis, have also been depicted in this cellular model( Reference Vu, Ong and Clemens 62 ). The mechanism governing these effects implies 1,25(OH)2D-mediated down-regulation( Reference Kong and Li 59 , Reference Blumberg, Tzameli and Astapova 60 ) of C/EBPα and PPARγ, the two master regulators of adipogenesis( Reference Farmer 63 ). Other mechanisms such as antagonisation of PPARγ activity and stabilisation of the VDR protein are also components of this complex regulation( Reference Kong and Li 59 ). Similar results, i.e. inhibited differentiation under 1,25(OH)2D, have also been reported in brown adipocytes( Reference Ricciardi, Bae and Esposito 64 ).
However, these data have been recently challenged in human adipocytes where 1,25(OH)2D enhanced adipocyte differentiation and lipid accumulation( Reference Nimitphong, Holick and Fried 45 ). Interestingly, a similar activation of adipogenesis was found in primary mouse pre-adipocytes (albeit at a more advanced stage of differentiation compared with 3T3-L1 cells, suggesting that stage of differentiation is a key factor in the nature of the effect of 1,25(OH)2D on adipogenesis). Similarly, 1,25(OH)2D also increased adipogenesis in human adult stem cells derived from adipose tissue, as revealed by lipid accumulation and expression of adipogenic markers( Reference Narvaez, Simmons and Brunton 65 ).
To summarise, the effects of 1,25(OH)2D and VDR on adipogenesis are not fully consistent: VDR appears to act as a promoter of adipogenesis, but 1,25(OH)2D has less clear effects. It is currently difficult to firmly conclude in favour of an anti- or pro-adipogenic effect, and further in vivo studies are required to clarify this point.
Regulation of gene expression linked to energy metabolism
VD and particularly 1,25(OH)2D may also influence adipose tissue and systemic biology. Indeed, 1,25(OH)2D directly up-regulated leptin expression and secretion, independently of fat mass modifications( Reference Kong, Chen and Zhu 66 ). 1,25(OH)2D was found to promote glucose uptake by adipocytes( Reference Marcotorchino, Gouranton and Romier 67 ), which could be related to the induction of GLUT4 protein expression and translocation observed in 3T3-L1 adipocytes( Reference Manna and Jain 68 ). We showed that VD supplementation in mice led to an increase of fatty oxidation (especially in brown adipose tissue) that could be responsible for a high-fat diet-induced limitation of body weight gain in VD-supplemented mice( Reference Marcotorchino, Tourniaire and Astier 69 ). Note that similar weight gain limitations in response to VD supplementation have been reported in mice( Reference Sergeev and Song 70 ) while VD-deficient old rats showed fat mass gain( Reference Domingues-Faria, Chanet and Salles 71 ). Likewise, a global dietary vitamin restriction was associated with an increase of adiposity and a disruption of glucose homeostasis( Reference Amara, Marcotorchino and Tourniaire 72 ) in mice.
Taken together, these data suggest that VD regulates energy expenditure, notably via its impact on adipose tissue biology.
Regulation of inflammation
Initial studies performed on human and mouse adipocytes (3T3-L1) found that 1,25(OH)2D up-regulates several inflammatory cytokines and down-regulates anti-inflammatory cytokines in both cell types( Reference Sun and Zemel 73 , Reference Sun and Zemel 74 ). These results have since been challenged by several groups( Reference Marcotorchino, Gouranton and Romier 67 , Reference Lorente-Cebrian, Eriksson and Dunlop 75 – Reference Mutt, Karhu and Lehtonen 77 ) who consistently report anti-inflammatory effects of 1,25(OH)2D whatever the model studied, which fits better with the well-described anti-inflammatory effect of VD in many other cell types( Reference Guillot, Semerano and Saidenberg-Kermanac'h 78 ). Indeed, it was shown that 1,25(OH)2D significantly decreased the release of IL-8, monocyte chemoattractant protein (MCP)-1 and IL-6 by human preadipocytes( Reference Gao, Trayhurn and Bing 76 ) and MCP-1 by human adipocytes( Reference Lorente-Cebrian, Eriksson and Dunlop 75 ). These anti-inflammatory effects were associated with inhibition of the NF-κB signalling pathway( Reference Mutt, Karhu and Lehtonen 77 ). We also demonstrated the anti-inflammatory properties of 1,25(OH)2D in murine and human adipocytes( Reference Marcotorchino, Gouranton and Romier 67 ). In these various models, 1,25(OH)2D was able to decrease the expression of inflammatory markers such as IL-6, IL-1β and MCP-1 in both basal and TNFα-stimulated conditions( Reference Marcotorchino, Gouranton and Romier 67 ). Similarly, 1,25(OH)2D reduced the expression of IL-6, IL-8 and MCP-1 in human adipose tissue biopsies submitted to IL-1β stimulation in vitro ( Reference Wamberg, Cullberg and Rejnmark 79 ). The molecular mechanisms have been investigated, and VDR and NF-κB signalling pathways and p38 mitogen-activated protein kinases were shown to be involved in 3T3-L1 adipocytes( Reference Marcotorchino, Gouranton and Romier 67 ). Interestingly, Zoico et al. recently demonstrated that VD, similarly to 1,25(OH)2D, was able to blunt the lipopolysaccharide-mediated pro-inflammatory effect in human adipocytes( Reference Zoico, Franceschetti and Chirumbolo 37 ). Using a microarray approach, we recently demonstrated that 1,25(OH)2D was able to down-regulate a large set of chemokines( Reference Karkeni, Marcotorchino and Tourniaire 80 ) induced by inflammatory stimulus( Reference Tourniaire, Romier-Crouzet and Lee 81 ) in human and murine adipocytes. This effect was accompanied by a reduction of macrophage migration mediated by an adipocyte-conditioned medium( Reference Karkeni, Marcotorchino and Tourniaire 80 ).
These anti-inflammatory properties of 1,25(OH)2D have also been established in several types of immune cells found in the adipose tissue, including lymphocytes and macrophages( Reference Guillot, Semerano and Saidenberg-Kermanac'h 78 ). Furthermore, 1,25(OH)2D reduced macrophage-induced inflammatory response in human adipocytes via inhibition of NF-κB and mitogen-activated protein kinase activation together with monocyte migration mediated by the adipocyte-conditioned medium( Reference Ding, Wilding and Bing 82 ). These data suggest that VD not only blunts adipocyte response to inflammatory stimulus, but also interferes with macrophage/adipocyte cross-talk, a key element of the propagation of metabolic inflammation in obesity( Reference Suganami, Nishida and Ogawa 83 ).
The anti-inflammatory effect of VD in adipose tissue has also been observed in vivo. It was reported that dietary treatment with 1,25(OH)2D-reduced IL6 protein content in the epididymal adipose tissue of obese mice( Reference Lira, Rosa and Cunha 84 ), whereas feeding with a VD-deficient high-fat diet was found to increase Il-6 expression in rat adipose tissue( Reference Roth, Elfers and Figlewicz 85 ). In addition, we recently demonstrated that VD supplementation of high-fat diet reduced the expression of proinflammatory cytokines and chemokines and inhibited macrophage infiltration in the adipose tissue of obese mice( Reference Karkeni, Marcotorchino and Tourniaire 80 ). Similar effects of VD supplementation were found in an acute inflammation model (intraperitoneal injection of LPS) where no modification of body weight was measured( Reference Karkeni, Marcotorchino and Tourniaire 80 ), which strongly suggests that the decreased inflammatory status of adipose tissue observed in obese mice is not only a consequence of reduced fat mass( Reference Marcotorchino, Tourniaire and Astier 69 ) but is also driven by an anti-inflammatory effect of VD per se.
To summarise, it has been demonstrated in vitro and in vivo that VD has a limiting effect on adipose tissue inflammation, acting on both inflammatory status in preadipocytes and adipocytes and on leucocyte infiltration.
Energy metabolism in transgenic mice models impacting vitamin D metabolism
Several transgenic mice models generated over the last decade have been used to gain insight into the role of VD metabolism in body weight management and adipose tissue metabolism. Studies using transgenic mouse models have shown that Vdr−/− and Cyp27b1−/− mice (which are unable to synthesise 1,25(OH)2D) are resistant to diet-induced obesity( Reference Narvaez, Matthews and Broun 86 , Reference Wong, Szeto and Zhang 87 ). This phenotype was linked to the co-induction of fatty acid β-oxidation and uncoupling proteins in adipose tissue leading to increased energy expenditure in these mice. Conversely, overexpression of human VDR in mouse adipose tissue induced an obese phenotype characterised by increased weight and fat mass due to decreased energy expenditure, reduced fatty acid β-oxidation and lipolysis( Reference Wong, Kong and Zhang 88 ).
Taken together, these data strongly suggest that VDR or 1,25(OH)2D has an impact on overall energy metabolism by acting on adipose tissue biology, but with a number of caveats. First of all, the use of global knockout models makes it difficult to attribute the observed phenotype to a specific tissue. In addition, these mice were fed a rescue diet containing large amounts of calcium and lactose, and calcium is strongly suspected to regulate energy homeostasis and well known to reduce intestinal lipid absorption( Reference Soares, Murhadi and Kurpad 89 ). Moreover, in wild-type mice, high-calcium rescue diet( Reference Narvaez, Matthews and Broun 86 ) has been shown to reduce 1,25(OH)2D to extremely low levels similar to Cyp27b1−/− mice( Reference Rowling, Gliniak and Welsh 90 ), making it unlikely that the observed effect on body weight is attributable to 1,25(OH)2D.
In all these animal models (Vdr−/−, Cyp27b1−/− and human VDR overexpression), phenotype appears tightly linked to uncoupling protein-1 modulation( Reference Narvaez, Matthews and Broun 86 – Reference Wong, Kong and Zhang 88 ). However, we now know that unliganded VDR can down-regulate uncoupling protein-1( Reference Malloy and Feldman 91 ), which means the VDR knockout could trigger an increase in uncoupling protein-1 leading to obesity resistance independently of plasma or adipose tissue levels of 1,25(OH)2D or other VD metabolites. In this case, the resistance to diet-induced obesity observed in Vdr−/− mice would be only VDR-dependent and not mediated by its ligand( Reference Wong, Szeto and Zhang 87 ). This would also be the case of human VDR overexpression where a down-regulation of uncoupling protein-1 is associated with weight gain( Reference Wong, Kong and Zhang 88 ). In the case of Cyp27b1−/−, the decrease in 1,25(OH)2D plasma levels likely results in a decrease of VDR, since VDR is able to induce its own expression( Reference Sun and Zemel 92 ), leading to the observed phenotype( Reference Narvaez, Matthews and Broun 86 ). The phenotype of these mice models could stem from other mechanisms too, e.g. modification of the bile acid pool, as recently evoked( Reference Bouillon, Carmeliet and Lieben 93 ).
Further research is clearly needed to provide an explanation (if any) that could reconcile data from transgenic mice and nutritional approaches on the impact of VD on energy metabolism regulation. An important aspect to investigate would be the concentrations of VD and its metabolites in adipose tissue in the different mice models (notably between lean and obese animals) and the regulation of VD metabolism in adipocytes, which remains largely unknown.
Effects of vitamin D on obesity and associated disorders in human studies
Numerous cross-sectional studies have reported a correlation between VD deficiency and obesity. Indeed, serum 25(OH)D is consistently lower in obese than lean individuals( Reference Gallagher, Yalamanchili and Smith 94 ). A recent meta-analysis found that prevalence of VD deficiency was 35 % higher in obese and 25 % higher in overweight compared with lean subjects( Reference Pereira-Santos, Costa and Assis 95 ). In addition, 25(OH)D plasma levels are inversely correlated to all the parameters of obesity, including BMI, fat mass and waist circumference( Reference Garcia, Long and Rosado 96 , Reference Cheng, Massaro and Fox 97 ), and increased dietary intake of VD, resulting in higher 25(OH)D plasma levels, is associated with a lower visceral adiposity( Reference Caron-Jobin, Morisset and Tremblay 98 ).
The fact that adipose tissue is the main storage site for VD and/or its metabolites in the body has prompted the hypothesis that VD and/or its metabolites gets sequestered in the excess fat mass in obese persons( Reference Wortsman, Matsuoka and Chen 99 ). However, the physiological mechanisms underlying this hypothesis have not been brought forward. Nevertheless, as pointed out in a recent study from Drincic et al.( Reference Drincic, Armas and Van Diest 100 ), it might just be that in individuals with a higher body mass, 25(OH)D is simply diluted in a higher volume, so they would require greater VD input than lean individuals to achieve a sufficient 25(OH)D status. Decreased plasma 25(OH)D levels could also result from a modification in VD metabolism occurring during obesity development. Indeed, modifications in the expression of genes encoding key enzymes of VD metabolism have been reported in the adipose tissue of obese people( Reference Wamberg, Christiansen and Paulsen 38 ).
The relationship between obesity and 1,25(OH)2D is less clear. Recent studies have reported an inverse relationship between 1,25(OH)2D and BMI and fat mass( Reference Parikh, Edelman and Uwaifo 101 , Reference Konradsen, Ag and Lindberg 102 ) while an older study found a direct relationship( Reference Bell 103 ). The origin of these inconsistent results is unclear but could stem from methodological bias in calcitriol measurement or else indicate that serum 1,25(OH)2D displays an inter-individual variability that is not adiposity-related. Also, no data are available on 1,25(OH)2D concentration in adipose tissue, which could be the critical effector in relation to obesity.
Several recent prospective studies have reported that low 25(OH)D plasma levels were associated with higher prevalence of obesity in adults( Reference Mai, Chen and Camargo 104 , Reference Gonzalez-Molero, Rojo-Martinez and Morcillo 105 ), children( Reference Gilbert-Diamond, Baylin and Mora-Plazas 106 ) and elderly women( Reference LeBlanc, Rizzo and Pedula 107 ). Low 25(OH)D was also associated with higher 5-year waist circumference( Reference Gagnon, Lu and Magliano 108 ). Low VD intake has also been considered as predictor of obesity( Reference Kamycheva, Joakimsen and Jorde 109 ). Mechanistic explanations to these prospective observations are scarce. Indeed, the impact of VD on the regulation of energy metabolism in human subjects is still unclear. Baseline 25(OH)D was positively correlated to diet-induced thermogenesis( Reference Teegarden, White and Lyle 110 ), which could at least partly explain why low 25(OH)D concentrations chronically modify energy balance. However, Boon et al. reported no effect of VD supplementation on energy expenditure and fat metabolism, but it should be noted that their supplementation regime was for 1 week only( Reference Boon, Hul and Sicard 111 ).
Intervention studies have been designed to study the causality between low plasma 25(OH)D levels and obesity. Except for Salehpour et al.( Reference Salehpour, Hosseinpanah and Shidfar 112 ) who reported that VD supplementation decreased body fat mass in healthy overweight and obese women, most randomised clinical trials have failed to demonstrate any benefit of VD supplementation in terms of weight loss( Reference Zittermann, Frisch and Berthold 113 – Reference Wamberg, Kampmann and Stodkilde-Jorgensen 115 ). These data were recently meta-analysed, and the lack of major effect of VD supplementation on weight loss was confirmed( Reference Pathak, Soares and Calton 116 ). However, a well-designed randomised controlled trial combining a weight loss programme together with placebo or VD supplementation highlighted that VD supplementation was associated with a significantly higher reduction of BMI and waist circumference in subjects with 25(OH)D levels raised to 80 nm/l( Reference Pathak, Soares and Calton 117 ), suggesting that beneficial effects of the supplementation only kick in with elevated plasma 25(OH)D levels.
A recent study using Mendelian randomisation of several thousands of volunteers of different age, gender and geographical location demonstrated that an increase in BMI could cause a decrease in 25(OH)D status, whereas VD insufficiency would at most result in only very minor effects on obesity( Reference Vimaleswaran, Berry and Lu 118 ). Note that the study only focused on genes related to 25(OH)D status (VDBP, DHCR7, CYP2R1 and CYP24A1) and chosen on the basis of a genome-wide association study( Reference Wang, Zhang and Richards 119 ) that explained 1–4 % of the variation in 25(OH)D concentrations, and so it cannot be ruled out that the use of polymorphisms present in other VD-related genes (such as VDR or CYP27B1) as instrumental variables of the Mendelian randomisation may lead to divergent results. Another large-scale study including Chinese women failed to show an association between obesity and genetic variants in genes in the pathway of VD metabolism( Reference Dorjgochoo, Shi and Gao 120 ), whereas several other studies found associations, notably between VDR polymorphisms and adiposity( Reference Ochs-Balcom, Chennamaneni and Millen 121 ). The origin of these discrepancies has not been established but could stem from ethnic specificities, since it is well established that ethnicity is a significant determinant of plasma 25(OH)D( Reference Sulistyoningrum, Green and Lear 122 ).
Taken together, these data suggest that VD may limit weight gain in human subjects even if it has no clear effect on weight reduction in obese or overweight people. However, clinical trials are needed to provide conclusive proof and to define the mechanisms governing this potential preventive effect of VD against obesity development. The question of the impact of VD supplementation and polymorphisms present in genes coding for key enzymes of VD metabolism remains unclear, and will require further investigations. Equally useful would be studies on the impact of VD supplementation on adipose tissue biology in well-designed randomised clinical trials that should follow several criteria, chiefly low baseline 25(OH)D, use of doses necessary to raise 25(OH)D concentrations up to 75–80 nm/l, and genotyping subjects.
Conclusions
Recent data from different research groups are converging to highlight the impacts of VD on adipose tissue/adipocyte biology. One of the best-documented effects is the ability of VD to limit the expression for inflammatory markers in adipose tissue and adipocytes. However, several key points urgently warrant further investigation, chiefly VD metabolism and its regulation in adipose tissue, which needs to be clarified, especially in the context of physiopathological disorders such as obesity. The last 5 years have seen some very interesting data in transgenic mice and rodents subjected to VD supplementation or restriction, but still without convergent findings. It is urgent to identify the origin of these discrepancies, where a key factor could be the quantification of VD metabolites in adipose tissue. Randomised clinical trials have been performed, but again the results remain contrasted, leaving persistent uncertainty over a beneficial role of VD supplementation on weight management. However, results from a recent intervention study suggest that a minimum level of plasma 25(OH)D has to be reached in order to elicit a beneficial effect from supplementation, since only subjects that became replete showed improvements in several parameters( Reference Pathak, Soares and Calton 116 ). This observation is consistent with successful investigations in mice where plasma 25(OH)D levels were inflated. This same study also demonstrates that it is important to stratify the data according to 25(OH)D status in order to explore and interpret differential physiological responses at the end of the supplementation period, an approach that should be used in future clinical studies. Recent prospective studies have presented low plasma 25(OH)D levels as a predictor of body weight gain, suggesting that VD may limit the prevalence of obesity. If these observations are confirmed in dedicated well-designed clinical studies, it could pave the way to the use of VD in preventive nutrition to limit the development of obesity and associated disorders, notably by reducing the inflammatory status.
To conclude, even if preclinical studies have provided strong support for beneficial impacts of VD supplementation, well-designed clinical studies are urgently needed to demonstrate real valuable utility for limiting obesity in human subjects.
Acknowledgement
The authors thank all the JFL group members.
Financial support
This work has been funded by INRA, AMU and INSERM.
Conflict of Interest
None.
Authorship
All authors participated in writing up this review.
