White adipose tissue is the major tissue for storage of excessive energy in the form of triglycerides, releasing fatty acids into the circulation in times of fasting and starvation. In addition to being an energy reserve for fatty acids, adipose tissue is also an active endocrine organ secreting many different substances, including adipokines with paracrine or endocrine function [10,11,12]. It has become increasingly clear that adipose tissue is an essential regulator of whole-body energy homeostasis [13]. In light of increased rates of obesity in countries worldwide, and the consequent increase in associated morbidity, understanding adipose tissue development and physiology has become of increasing importance. Below we highlight the role of retinoid metabolism in this context.

Retinoids in adipose tissue are derived from several sources. Retinol bound to retinol-binding protein in the circulation is one source used by adipose tissue. Furthermore, retinyl ester and β-carotene present in chylomicrons postprandially reach adipose tissue. Lipoprotein lipase present on endothelial cells facilitates the uptake of these postprandial retinoids into adipose tissue [14]. The presence of β-carotene monoxygenase type 1 (Bcmo1) in adipose tissue enables the conversion of β-carotene to retinaldehyde [15,16].

It has been estimated that adipose tissue stores 10–20% of total retinoids in the body as retinyl esters [17]. Under conditions of dietary vitamin A deficiency, the adipose retinol/retinyl ester stores are readily mobilized as evidenced by decreasing levels [18,19,20].

Mice deficient in lecithin:retinol acyltransfrase (LRAT), the enzyme that catalyzes the esterification of retinol, have very low hepatic retinyl ester stores, but adipose retinyl ester levels in these mice is higher than in wild-type mice. This indicates that adipose tissue can store retinol as retinyl esters when hepatic retinol storage is defective. These data also indicate that LRAT is the dominant enzyme for hepatic retinol esterification but in adipose tissue a different enzyme, and a yet unidentified acyl-CoA:retinol transferase (ARAT), may be important for retinol esterification [21,22]. Diacylglycerol acyltransferase type 1 (DGAT1) has been proposed to be this ARAT [23,24] however, in mice lacking both DGAT1 and LRAT, adipose retinyl ester levels were not significantly different from those in control mice [25], suggesting the possible existence of an additional enzyme involved in esterifying retinol.

In addition to esterification, retinol is converted to retinaldehyde and retinoic acid in adipose tissue. Both, aldehyde dehydrogenase (ADH1), catalyzing the first step in retinol oxidation and retinaldehyde dehydrogenase (Raldh1), responsible for retinoic acid synthesis, have been identified in adipose tissue [26]. The role of the latter as it may relate to adipocyte differentiation is discussed below [26].

Intracellularly, protein binding may be an important mechanism to direct retinol towards enzymes that metabolize retinol to retinaldehyde and then to retinoic acid or retinyl esters; this would protect retinol from nonspecific oxidation/isomerization [27,28,29]. These roles have been attributed to intracellular proteins that specifically bind retinol; they are known as cellular retinol-binding proteins (CRBP) [27,29,30]. Thus far, three types, CRBP-I, CRBP-II and CRBP-III, have been described in murine tissues [27,31,32]. Together with cellular retinoic acid-binding proteins (CRABP) and fatty acid-binding proteins (FABP), they belong to the family of intracellular lipid-binding proteins (iLBP) [30]. It is currently thought that the primary role of CRBP-I and CRBP-III is to facilitate the esterification of retinol to retinyl ester catalyzed by lecithin:retinol acyltransferase (LRAT) [31,33]. The lack of CRBP-I in a mouse model leads to a 50% reduction of vitamin A storage in the liver and higher hepatic turnover of vitamin A in mice [33,34]. Similarly, we have shown that the absence of CRBP-III leads to decreased retinol esterification during lactation resulting in decreased milk retinyl ester levels in mice [31].

CRBP-I and CRBP-III have high protein homology but they differ in their affinity for retinol and in their tissue and cellular expression patterns [31,32,35]. CRBP-I binding affinity for retinol is almost two orders of magnitude greater than that of CRBP-III. Both binding proteins are expressed in adipose tissue, with CRBP-I expression restricted to preadipocytes and CRBP-III expression localized to endothelial cells and mature adipocytes [31,32,35]. This localization suggests that they may have different functions and this possibility is discussed below in more detail.

In addition to CRBP, other members of the intracellular retinoid-binding protein family, cellular retinoic acid-binding proteins (CRABP), which bind retinoic acid with high affinity, are also expressed in adipose tissue and are involved in delivering RA into the nucleus. Of particular interest is CRABP-II as it has recently been shown to be involved in controlling adipogenesis (see below).

While the association between dietary vitamin A intake and obesity in humans is difficult to assess, several studies employing rodent models have examined this relationship with various outcomes [36,37,38,39,40,41]. Supplementation of obese rats with high levels of dietary vitamin A (129 mg of vitamin A/kg) led to significant weight loss compared to a control group receiving lower levels of dietary vitamin A (2.6 mg of vitamin A/kg) [36]. A similar supplementation regimen in lean mice had no effect [37]. In contrast, short-term supplementation with retinoic acid resulted in weight loss, decreased adiposity and increased insulin sensitivity in several lean and/or obese mouse models [37,38,39,40,41]. Similarly, a diet deficient in vitamin A (no detectable vitamin A levels) led to an increase in adiposity in mice [40].

It has long been appreciated that all-trans-retinoic acid can inhibit adipocyte differentiation by binding to and activating retinoic acid receptor (RAR) [63,64,65]. All-trans-retinoic acid inhibits adipogenesis early in the differentiation process by blocking the transcriptional activity of C/EBPβ [66]. Suppressed C/EBPβ activity lowers the expression of PPARγ, thus decreasing adipogenesis. In addition, retinoic acid may affect the cell cycle, keeping cells in a proliferative state rather than allowing growth arrest, which is essential for adipogenesis [67]. Retinoic acid no longer has an inhibitory effect on adipocyte differentiation if added to the cell culture after early differentiation is completed [16,68].

Retinoic acid applied to mature adipocytes appears to induce lipolysis and influence the production of several adipokines that are secreted into the tissue culture media. However, β-carotene and the subsequent intracellular generation of retinoic acid decreased adipocyte differentiation after early differentiation providing mechanistic data for the increased adiposity observed in mice lacking Bcmo1 [15]. The addition of β-carotene to 3T3-L1 cells during the mid stage of differentiation (day 5) results in a repression of adipocyte differentiation and loss of lipids [16]. Inhibition of the activity of Bcmo1 abolishes this effect. Interestingly, the gene for Bcmo1, which is a PPARγ target gene, is highly expressed in mature adipocytes, allowing for β-carotene to be converted to retinoic acid. The addition of β-carotene, however, leads to repression of PPARγ expression in the cell model, through a mechanism has not yet been identified.

This differential effect of retinoic action on adipocyte differentiation, i.e., that retinoic acid has an inhibitory role at early but not at later stages of the differentiation process may be explained by the fact that retinoic acid can be transported to the nucleus via two binding proteins, namely CRABP-II or FABP5 [38]. CRABP-II will deliver retinoic acid for binding to RAR while FABP5 will deliver retinoic acid to PPARβ/δ. Thus, in cells that express high levels of CRABP-II, retinoic acid will activate RAR and, conversely, in cells high in FABP5 will activate PPARβ/δ and their respective downstream pathways [38]. Recently, Berry et al. [62] provided data indicating that CRABP-II is highly expressed during early adipocyte differentiation but its expression is repressed during later stages. The lack of CRABP-II after early differentiation may thus explain that retinoic acid only inhibits differentiation early but not in later stages of differentiation.

Similar to retinoic acid, retinaldehyde is also capable of inhibiting adipogenesis in vitro and in vivo. As shown above (section 3.2.2), in the absence of Raldh1, increased intracellular retinaldehyde levels are accompanied by the development of decreased adiposity [26]. In contrast to the effect of retinoic acid on adipogenesis, the effect of retinaldehyde on adipogenesis might be due to the repression of PPARγ activity [26] pointing to retinaldehyde as a transcriptional regulator. Retinaldehyde appears to be able to bind both PPAR and RXR resulting in a repression of receptor activation and thus offering an explanation for retinaldehyde mediated inhibition of adipocyte differentiation.

We have recently shown that CRBPs are important participants in adipogenesis. Both CRBP-I and CRBP-III are expressed in adipose tissue with CRBP-I present in preadipocytes and CRBP-III in adipocytes. We have identified CRBP-III to be a PPARγ target gene and thus examined whether it participates in adipogenesis [47]. Using the 3T3-L1 cell model we showed that CRBP-III is expressed during mid- and late-stage differentiation [47]. To test the role of CBRP-III in vivo we used our CRBP-III knockout mouse model. When fed a high-fat diet, mice lacking CRBP-III developed decreased adiposity. Based on our in vitro data, it appears that lack of CRBP-III, PPARγ is markedly downregulated indicating that CRBP-III is important in maintaining adipogenesis (unpublished). In contrast, our data indicate that CRBP-I is an important inhibitor of adipogenesis [46]. Lack of CRBP-I in vivo led to increased adiposity. Similarly, in vitro silencing of CRBP-I led to enhanced adipocyte differentiation in 3T3-L1 preadipocyte cells and mouse embryonic fibroblasts (MEF). In addition, although CRBP-I is specifically expressed in preadipocytes it does not appear to regulate known inhibitory genes of adipogenesis. Our data point to a role of CRBP-I affecting PPARγ function. Further examination of molecular markers revealed that the expression and activity of PPARγ but not of other transcriptional regulators important during the different stages of differentiation were specifically increased in the absence of CRBP-I in vitro and in vivo. Interestingly, retinoid homeostasis was not altered in cells lacking CRBP-I. This finding is very similar to data presented on the enzyme retinol saturase (RetSat), where the function of RetSat in adipocytes was not readily explained by its enzymatic role of generating 13,14-dihydroretinol from retinol [69]. Taken together CRBP-I and CRBP-III appear to play important opposing roles in adipogenesis with CRBP-I repressing and CRBP-III maintaining adipogenesis.