In this study, we report the effects of ATRA on differentiation and UCP1 expression in various mouse and human adipocytes. We find that high concentrations of ATRA inhibit mouse and human adipogenesis, whereas lower concentrations enhance UCP1 expression in mouse, but not in human, adipocytes. In addition, we show that the effects of ATRA are mediated by RARs and not by PPARδ or other ATRA-activated nuclear receptors. Moreover, the enhanced expression of UCP1 in response to ATRA is independent of PGC-1α.
Adipocyte and adipose tissue function are impacted by ATRA [47, 48]. Expression of UCP1 is reduced in BAT of mice fed vitamin A-depleted feed [17, 49] and exogenous ATRA enhances expression of UCP1 in both WAT and BAT of mice and rats [16, 17, 28, 32, 49, 50]. Expression of UCP1 is induced by ATRA in primary brown adipocytes from mice and rats as well as in mouse brown adipocyte cell lines [24, 26, 28, 29, 51–53]. Moreover, UCP1 expression is strongly induced in MEF-derived white adipocytes . Contrary, ATRA has been reported not to induce expression of UCP1 in mature 3T3-L1 adipocytes and mouse primary white adipocytes [16, 30, 54]. Exposure to ATRA leads to activation of p38 mitogen-activated protein kinase (MAPK), an activation that is required for full induction of UCP1 expression by ATRA [29, 31].
ATRA has been reported to activate three nuclear receptors besides RARs, namely PPARδ, COUP-TFII and TR4. The induction of UCP1 observed in the mouse cells applied in this study is unlikely to be mediated by COUP-TFII and TR4, as the app. EC50 of ATRA are 20 μM  and 24 μM , respectively, which is 20 to 240 times higher than the concentrations inducing UCP1. Although the EC50 of ATRA for PPARδ is much lower than for COUP-TFII and TR4 (app. 200 nM) , PPARδ is not mediating the effects of ATRA either. Firstly, the RAR agonist TTNPB mimics the effects of ATRA (see Figure 3), but does not bind to PPARδ . Secondly, a potent PPARδ agonist does not enhance expression of UCP1 (see Figure 5). Thirdly, a RAR antagonist attenuates the effects of ATRA (see Figure 3). In this study, we have not addressed the potential involvement of nongenomic effects of ATRA, e.g. activation of p38 MAPK and the cell surface receptor responsible for retinol uptake called stimulated by retinoic acid gene 6 .
At intermediate concentrations of ATRA, we consistently observe an induction of UCP1 expression in mouse adipocytes. This does not only occur in the mouse cells shown here, but was also observed with WT-1 brown adipocytes [56, 57] and 3T3-F442A white adipocytes (data not shown). Thus, our results demonstrate that ATRA can cause an induction of UCP1 expression in white adipocyte cell models of mouse origin. It is tempting to speculate that exposure to ATRA will cause white preadipocytes and mature adipocytes to transdifferentiate into brown-like adipocytes in vitro. However, in order to confirm if a transdifferentiation event has taken place in our study, a more detailed gene expression analysis is required combined with a characterization of mitochondrial function.
Using three cell models of human origin, SGBS and hMADS cells as well as primary subcutaneous adipocytes from two different donors, we failed to detect an induction of UCP1 expression by ATRA (see Figures 7 and 8). hMADS cells have been proposed to represent brown or brown-like adipocytes, the latter due to the induction of UCP1 expression in response to prolonged culture in the presence of rosiglitazone [58, 59] or upon treatment with atrial natriuretic peptide . Despite being considered white fat cells, SGBS and primary subcutaneous human adipocytes have the ability to induce expression of UCP1 in response to genetic manipulation [61, 62]. Thus, the lack of effect of ATRA in the human adipocyte models studied here cannot be explained by an inherent inability to induce expression of UCP1. Consistently, to our knowledge, an induction of the endogenous human UCP1 gene by ATRA has never been reported. Nevertheless, we cannot rule out that the lack of response in our study is due to the experimental setup or the human cell models used. In particular, it remains to be shown if primary human brown adipocytes respond to ATRA by increasing UCP1 expression. However, as we consistently observe enhanced expression of UCP1 by intermediate concentrations of ATRA in mouse fat cells, we find this difference between mouse and human adipocytes noteworthy.