FABP4 Controls Fat Mass Expandability (Adipocyte Size and Number) through Inhibition of CD36/SR-B2 Signalling

Adipose tissue hypertrophy during obesity plays pleiotropic effects on health. Adipose tissue expandability depends on adipocyte size and number. In mature adipocytes, lipid accumulation as triglycerides into droplets is imbalanced by lipid uptake and lipolysis. In previous studies, we showed that adipogenesis induced by oleic acid is signed by size increase and reduction of FAT/CD36 (SR-B2) activity. The present study aims to decipher the mechanisms involved in fat mass regulation by fatty acid/FAT-CD36 signalling. Human adipose stem cells, 3T3-L1, and its 3T3-MBX subclone cell lines were used in 2D cell cultures or co-cultures to monitor in real-time experiments proliferation, differentiation, lipolysis, and/or lipid uptake and activation of FAT/CD36 signalling pathways regulated by oleic acid, during adipogenesis and/or regulation of adipocyte size. Both FABP4 uptake and its induction by fatty acid-mediated FAT/CD36-PPARG gene transcription induce accumulation of intracellular FABP4, which in turn reduces FAT/CD36, and consequently exerts a negative feedback loop on FAT/CD36 signalling in both adipocytes and their progenitors. Both adipocyte size and recruitment of new adipocytes are under the control of FABP4 stores. This study suggests that FABP4 controls fat mass homeostasis.

Adipogenesis is induced and regulated by lipogenesis, including de novo lipid synthesis and lipid uptake. In 2D cell cultures in presence of high glucose concentration (4.5 g/L), an improved intracellular lipid accumulation during adipogenesis is obtained in response to oleic acid. Adipogenesis promotes gene transcription of adipogenic markers FAT/CD36, FABP4, and CIDEC [12]. FAT/CD36 is characterized by a high adipogenic potential required for long-chain fatty acid uptake as well as lipolysis [16]. Fatty acid interaction with

FAT/CD36 Is Involved in Lipid Storage but Reduces Lipogenesis
The role of FAT/CD36 in the regulation of adipocyte size was investigated on rat adipose tissue explants. In a previous study, we showed using AdipoRed probe incorporation that oleic acid uptake induced size increase with better efficiency in high (HG, 4.5 g/L) versus low glucose media (LG, 1 g/L) [23]. In the present study, rat explants were treated for 48 h in either LG or HG media and/or optimized concentration of FAT/CD36 inhibitor AP5258 ( Figure 1A). Triglyceride (TG) contents were then assessed using the fluorescent probe AdipoRed incorporation in 40 min. We found that in HG media AdipoRed incorporation was higher in HG than in LG media, suggesting that fatty acid de novo synthesis from glucose was detected in these experiments. AP5258 inhibited this effect, suggesting that FAT/CD36 plays a major role in the regulation of TG synthesis in adipose tissue. An example of rat adipose tissue explant labeled with AdipoRed (red) and Hoechst 33258 (blue) is presented, the scale bar represents 1 mm. Triglyceride contents were quantified using AdipoRed (AR) uptake after 40 min, fluorescence intensity was normalized at the time the marker was added (T0). Inhibition of FAT/CD36 induced a reduction of lipid content increase observed in HG media by comparison to LG media. Data are presented as mean fold changes AR after 40 min to T0 +/− SEM (n = 8), different letters represent significant differences (ANOVA test, p < 0.05). (B) Analysis of droplet size regulation 3 days after treatment with oleic acid (10 μM) in HG culture media in 3T3-L1 adipocytes. In presence of either lipolysis inhibitor ATGListatin (1 μM), AP5258 (1 μM), or cell-permeable FABP4 inhibitor (FABP4i 20 μM), lipid droplet sizes were increased. PPARA or PPARG (inhibited respectively by GW6471 and GW9662, 10 μM each) did not affect both cell lines. Insulin (0.05 U/mL) was used as a control of the inhibitory effect on droplet size increase. Results were obtained in independent experiments and normalized to corresponding control media (n = 8 biological replicates), and significant Student t-test p-values (p < 0.05) are indicated.
Whether fatty acid/FAT/CD36 signalling pathways were involved in the regulation of cell size was monitored in 2D cell cultures ( Figure 1B). The 3T3-L1 adipocytes were treated with a low dose of oleic acid in HG culture media in presence of several inhibitors at optimized concentrations over 3 days. Fatty acid uptake is a fast event and this protocol was previously optimized in order to avoid cell detachment or toxicity, given the time for synthesis of the proteins required for lipid accumulation and droplet size increase. This means also that in such experiments, important cell reorganization could be observed, such as basal lipolysis and fatty acid exchange between adipocytes [23]. As previously observed, insulin reduced the effect of OA on droplet size increase. Inhibition of PPARG (by GW9662) or PPARA (by GW6471) did not have any significant effect, suggesting that their regulation by FAT/CD36 was not involved in these conditions. However, inhibition of FAT/CD36 (by AP5258), together with those of ATGL (by ATGListatin) and intracellular FABP4 (by FABP4i), were related to an increase of droplet size in 3T3-L1 cells. This result shows that ATGL-dependent lypolysis, FAT/CD36, and FABP4 play a major role in the regulation of cell size.

Transcriptional Regulations by Fatty Acid in Absence of FAT/CD36: Emerging Role of FABP4
The regulation of adipocyte size proceeds through the transcriptional induction of genes involved in fatty acid storage. We previously found that oleic acid is a major activator of FAT/CD36-PPARG signalling pathway [12,23]. Whether FAT/CD36 inactivation regulates gene transcription was analyzed using comparative analyzes of gene datasets. The gene dataset regulated by fatty acids was compared to the gene datasets regulated by FAT/CD36 knock-out. The method consists in comparing significant enrichment of genes (genome as reference) [12,23]. Although the datasets are obtained from experiments performed in different conditions and on different kinds of samples, previous studies have shown that significant enrichments in gene datasets allow the identi- LG HG AR / AR Whether fatty acid/FAT/CD36 signalling pathways were involved in the regulation of cell size was monitored in 2D cell cultures ( Figure 1B). The 3T3-L1 adipocytes were treated with a low dose of oleic acid in HG culture media in presence of several inhibitors at optimized concentrations over 3 days. Fatty acid uptake is a fast event and this protocol was previously optimized in order to avoid cell detachment or toxicity, given the time for synthesis of the proteins required for lipid accumulation and droplet size increase. This means also that in such experiments, important cell reorganization could be observed, such as basal lipolysis and fatty acid exchange between adipocytes [23]. As previously observed, insulin reduced the effect of OA on droplet size increase. Inhibition of PPARG (by GW9662) or PPARA (by GW6471) did not have any significant effect, suggesting that their regulation by FAT/CD36 was not involved in these conditions. However, inhibition of FAT/CD36 (by AP5258), together with those of ATGL (by ATGListatin) and intracellular FABP4 (by FABP4i), were related to an increase of droplet size in 3T3-L1 cells. This result shows that ATGL-dependent lypolysis, FAT/CD36, and FABP4 play a major role in the regulation of cell size.

Transcriptional Regulations by Fatty Acid in Absence of FAT/CD36: Emerging Role of FABP4
The regulation of adipocyte size proceeds through the transcriptional induction of genes involved in fatty acid storage. We previously found that oleic acid is a major activator of FAT/CD36-PPARG signalling pathway [12,23]. Whether FAT/CD36 inactivation regulates gene transcription was analyzed using comparative analyzes of gene datasets. The gene dataset regulated by fatty acids was compared to the gene datasets regulated by FAT/CD36 knock-out. The method consists in comparing significant enrichment of genes (genome as reference) [12,23]. Although the datasets are obtained from experiments performed in different conditions and on different kinds of samples, previous studies have shown that significant enrichments in gene datasets allow the identification of signalling pathway crosstalks. A pairwise comparison of gene datasets regulated by either fatty acids or in FAT/CD36 knock-out mice was performed. Both gene datasets were compared to human gene datasets regulated by a series of either extracellular signals, intracellular pathways, or transcriptional regulators. In another way, their relevance to adipose phenotype, i.e., to signalling pathways, is significantly over-represented in either adipose stem cells (ASCs), in vitro differentiated adipocytes (dA), or adipose tissue (AT) datasets has been analyzed (Suppl Table S1). Only signalling pathways significantly over-represented in both fatty acid and FAT/CD36 knock-out gene datasets are presented in Figure 2. fication of signalling pathway crosstalks. A pairwise comparison of gene datasets regulated by either fatty acids or in FAT/CD36 knock-out mice was performed. Both gene datasets were compared to human gene datasets regulated by a series of either extracellular signals, intracellular pathways, or transcriptional regulators. In another way, their relevance to adipose phenotype, i.e., to signalling pathways, is significantly over-represented in either adipose stem cells (ASCs), in vitro differentiated adipocytes (dA), or adipose tissue (AT) datasets has been analyzed (Suppl Table S1). Only signalling pathways significantly over-represented in both fatty acid and FAT/CD36 knock-out gene datasets are presented in Figure 2. Eight extracellular signalling pathways are commonly regulatable by FAT/CD36 inhibition and free fatty acid treatment, including glucose, insulin, and several pro-inflammatory signals. Sixteen intracellular signalling pathways are commonly involved in fatty acid signalling without FAT/CD36, they are related to oxidative stress, such as Glucose oxydase, Janus kinase 2 (JAK2), and NADPH oxidase 1 (NOX1). Eight transcriptional regulators, including intracellular FABP4 (FABP4i), were also found in this analysis. Importantly, PPARs did not emerge from this study as expected since they are activable by fatty acids through FAT/CD36 signalling. Intracellular FABP4 is involved in fatty acid-induced signalling in absence of FAT/CD36, in pathways over-represented in either ASCs, in in vitro differentiated adipocytes (dA) as well as in adipose tissue (AT). These data suggest that FABP4 plays a major role in adipose cells with low FAT/CD36 activity. FABP4 was also identified as an external signal with significant transcriptional activities commonly modulated in FAT/CD36 knock-out cells but without a significant link with fatty acid activity (Suppl Table S1A). Moreover, both extracellular and intracellular activities of FABP4 present a significant similarity in gene transcription activity (Suppl Table S1D). The signalling pathways commonly enriched in extracellular and in- Eight extracellular signalling pathways are commonly regulatable by FAT/CD36 inhibition and free fatty acid treatment, including glucose, insulin, and several proinflammatory signals. Sixteen intracellular signalling pathways are commonly involved in fatty acid signalling without FAT/CD36, they are related to oxidative stress, such as Glucose oxydase, Janus kinase 2 (JAK2), and NADPH oxidase 1 (NOX1). Eight transcriptional regulators, including intracellular FABP4 (FABP4i), were also found in this analysis. Importantly, PPARs did not emerge from this study as expected since they are activable by fatty acids through FAT/CD36 signalling. Intracellular FABP4 is involved in fatty acid-induced signalling in absence of FAT/CD36, in pathways over-represented in either ASCs, in in vitro differentiated adipocytes (dA) as well as in adipose tissue (AT). These data suggest that FABP4 plays a major role in adipose cells with low FAT/CD36 activity. FABP4 was also identified as an external signal with significant transcriptional activities commonly modulated in FAT/CD36 knock-out cells but without a significant link with fatty acid activity (Suppl Table S1A). Moreover, both extracellular and intracellular activities of FABP4 present a significant similarity in gene transcription activity (Suppl Table S1D). The signalling pathways commonly enriched in extracellular and intracellular FABP4, fatty acids and CD36 knock-out datasets are reported in Figure 2. They include MAPK1/ErK1/2 and JAK2 pathways ( Figure 2, central panel). Commonly with fatty acids, FABP4 may also regulate genes regulated by transcription factors CEBPG and Z, LXR, and the USF. These results suggest that FABP4 regulates gene transcription through several signalling pathways independently of FAT/CD36. Thus the role of FABP4 in the regulation of fatty acid signalling and fat mass was further investigated.

Intracellular FABP4 Regulates Oleic Acid-Induced PPARG Activity in Adipocytes
At the intracellular level, FABP4 plays a pivotal role in the regulation of lipid storage, through its implication in the processes of lipolysis, lipid uptake, and PPARG-mediated gene transcription. FABP4 intracellular accumulation could result from (1) de novo synthesis after fatty acid induction of its gene transcription through FAT/CD36-PPARG signalling and (2) inhibition of lipolysis. The role of intracellular FABP4 in the regulation of size was then assessed in 3T3-MBX adipocytes ( Figure 3). In fully differentiated 3T3-MBX adipocytes, in the presence of OA, FABP4 inhibition hindered the OA-induced increase of droplet size ( Figure 4B-D). This result suggests again that intracellular FABP4 is required at least during fatty acid-induced adipogenesis. In the next step, the role of FABP4 in fatty acid mediated FAT/CD36 signalling was further explored. The recombinant FABP4 protein (FABP4r) and the cell-permeable in-hibitor (FABP4i) were used to study its extracellular and intracellular activities, respectively. Activators of PPARG (rosiglitazone) and AMPK (AICAR) were used as controls for FAT/CD36 signalling related to gene transcription and basal lipolysis, respectively. The regulation of FAT/CD36 activation allowing fatty acid signalling was assessed using a specific APC-coupled antibody directed against the extracellular part of FAT/CD36. In 3T3-L1 adipocytes, the extracellular distribution of FAT/CD36 (related to its activable form) was inversely related to lipid droplet size ( Figure 4A), suggesting that extracellular FAT/CD36 is downregulated when adipocyte size increases.
The 3T3-L1 adipocytes were treated with oleic acid in order to increase lipid content and maturation. The presence of FAT/CD36 inhibitor promoted lipid droplet size increase without affecting the frequency of "mature" adipocytes (i.e., droplets up to 50 µm diameter). This suggests that FAT/CD36 was no longer involved in the increase of size after maturation ( Figure 4B). Interestingly, inhibition of intracellular FABP4 induced an increase of extracellular FAT/CD36 without influencing lipid droplet size but by reducing the frequency of "mature" adipocytes ( Figure 4B). These results suggest that FABP4 is involved in the regulation of the activable form of FAT/CD36 during the pro-adipogenic process and is involved in the regulation of adipocyte size. Taken together, these results support the hypothesis that concomitant reduction of FAT/CD36 active form and intracellular accumulation of FABP4 is required to obtain mature adipocytes.
In order to mimic fatty acid induction of adipocyte maturation, 3T3-MBX adipocytes were treated over 3 days with a low dose of oleic acid (5 µM) in the first step. In the second step, the adipocytes were treated over 48 h with either recombinant FABP4r, cellpermeable inhibitor of FABP4 (FABP4i), PPARG activator (rosiglitazone), or AMPK activator (AICAR) ( Figure 4E). As expected, FABP4, FAT/CD36, and CIDEC, another adipogenic marker used as control, were induced by PPARG activation. Both FABP4r addition and inhibition of intracellular FABP4 downregulated FAT/CD36 and CIDEC. In the third step, their lipid storage capacity was assessed by treatment with oleic acid 10 µM over 24 h ( Figure 4F). AMPK activation potentiated the effect of oleic acid and promoted droplet size increase, suggesting involvement of its protective effect against basal lipolysis. Both PPARG activation and inhibition of intracellular FABP4 reduced lipid storage capacity. Intracellular FABP4 was required to increase droplet size, since it is involved in CIDEC induction and/or maintenance, although FABP4 reduced FAT/CD36 activity. FABP4 and PPARG activities were thus inversely regulated. The extracellular activity of FABP4 was further explored.

Extracellular FABP4 Inhibits the Recruitment of Adipogenic Precursors by Oleic Acid-Induced FAT/CD36 Signalling
FABP4 is secreted by adipocytes during lipolysis. Human adipocytes were treated with oleic acid over 48 h in order to increase droplet size ( Figure 5A). Fatty acid uptake is a fast event. Forty-eight hours after treatment with OA, an accumulation of FABP4 in culture media was detected. This observation suggests that basal lipolysis occurred in high glucose culture media.
In a previous study, we found that basal lipolysis can be detected in high glucose culture media of highly differentiated adipocytes with heterogeneous sizes [23]. We used fully differentiated 3T3-MBX adipocytes obtained by pre-treatment with OA over 3 days ( Figure 5B,C) in order to study in co-cultures how basal lipolysis and FABP4 release can modulate the recruitment of new adipocytes using 3T3-L1 fibroblasts. The extracellular FABP4 activity was assessed in 3T3-L1 fibroblasts co-cultured with fully differentiated 3T3-MBX adipocytes ( Figure 5D). After 3 days of co-culture, in 3T3L-1 cells the lipid content was higher in high (4.5 g/L) versus low (1 g/L) glucose culture media, due to the uptake of fatty acids released by lipolytic 3T3-MBX adipocytes. However, at the transcriptional level, FAT/CD36, CIDEC, and G0S2 were reduced in high versus low glucose co-culture media, suggesting a reduction of the adipogenic process ( Figure 5E). The role of lipolysis products i.e., fatty acids (represented by oleic acid) and FABP4, were tested on human ASCs in real-time experiments (Figure 6). ASCs exposed to oleic acid during the proliferative phase increased their capacity to differentiate in a dosedependent manner, the non-toxic dose was defined for a Kd 0.5 µM ( Figure 6A). Oleic acid induction of adipogenesis promoted the increase of droplet size according to time ( Figure 6B). Recombinant FABP4 did not modulate proliferation, differentiation ( Figure 6C), or OA uptake ( Figure 6D). Similar results were observed in adipocytes (Suppl Figure S1). However, OA complexation to FABP4 counteracted the differentiation process induced by OA complexed to albumin ( Figure 6E), without significant effect on lipid accumulation ( Figure 6F) but by altering lipid droplet formation ( Figure 6G). Whether extracellular FABP4 stabilization of extracellular addressing of FAT/CD36 correlates with its ability to reduce FAT/CD36 activity was further explored. Fluorescently labeled FABP4 interacts with high affinity to 3T3-MBX cell membranes, with apparent co-localization with polymerized beta-actin cytoskeleton detected with phalloïdin ( Figure 7A,B). FABP4 restored extracellular FAT/CD36 addressing in high glucose media in short-term experiments but reduced extracellular FAT/CD36 addressing after long-term exposure ( Figure 7C), without affecting the lipogenic activity of oleic acid nor the protective effect of insulin (Suppl Figure S1). This result suggests that FABP4 promotes FAT/CD36 activation but that its accumulation reduces FAT/CD36 expression. The results are consistent with the results obtained on the bio-informatic gene dataset analyses showing common transcriptional activities of both extracellular and intracellular FABP4 also commonly regulated by fatty acids and in FAT/CD36 knock-out ( Figure 2). Among them, MAPK1/2 and JAK2 were identified. Thirteen transcription regulators datasets are commonly over-represented in both extracellular and intracellular FABP4 activities, 11 of them are also involved in JAK2 signalling, and among them, 6 are also involved in LKB1/AMPK1 signalling (Suppl Figure S1D).
Taken together these results suggest that FABP4 could be internalized to promote the signalling pathways regulated by intracellular accumulation of FABP4 in adipocytes. Since it was not significantly detected in adipogenic precursors, FABP4 transcriptional activity in presence of oleic acid was analyzed on 3T3-L1 fibroblasts and compared to PPARG activation with rosiglitazone ( Figure 8A). CD36, CIDEC, G0S2, and FABP4 are commonly induced by OA and PPARG, they are inhibited by OA-FABP4 as well as by OA when FAT/CD36 is inhibited. Both FABP4r and FAT/CD36 inhibition were linked to the reduction of active FAT/CD36 form ( Figure 8B). Taken together these results suggest that FABP4 antagonizes OA-FAT/CD36 signalling pathways in both adipocytes and their precursors. The results are consistent with the results obtained on the bio-informatic gene dataset analyses showing common transcriptional activities of both extracellular and intracellular FABP4 also commonly regulated by fatty acids and in FAT/CD36 knock-out ( Figure 2). Among them, MAPK1/2 and JAK2 were identified. Thirteen transcription regulators datasets are commonly over-represented in both extracellular and intracellular FABP4 activities, 11 of them are also involved in JAK2 signalling, and among them, 6 are also involved in LKB1/AMPK1 signalling (Suppl Table S1D).
Taken together these results suggest that FABP4 could be internalized to promote the signalling pathways regulated by intracellular accumulation of FABP4 in adipocytes. Since it was not significantly detected in adipogenic precursors, FABP4 transcriptional activity in presence of oleic acid was analyzed on 3T3-L1 fibroblasts and compared to PPARG activation with rosiglitazone ( Figure 8A). CD36, CIDEC, G0S2, and FABP4 are commonly induced by OA and PPARG, they are inhibited by OA-FABP4 as well as by OA when FAT/CD36 is inhibited. Both FABP4r and FAT/CD36 inhibition were linked to the reduction of active FAT/CD36 form ( Figure 8B). Taken together these results suggest that FABP4 antagonizes OA-FAT/CD36 signalling pathways in both adipocytes and their precursors.

Discussion
In obesity fat mass expandability is a result of multifactorial processes including dysregulation of adipocyte size (hypertrophy) and recruitment of new adipocytes from adipose stem cells (hyperplasia). In normal adipose tissue, in non obese individuals, adipocyte cell size distribution is highly homogeneous i.e., in a range of 50 to approx. 100 µm [5,[35][36][37], suggesting the existence of a limiting factor to the size increase. In the present study, we identified the pivotal role of FABP4 in the regulation of this homeostasis. The fatty acid-induced activity of FAT/CD36 was highly regulated by FABP4. FABP4 controls adipocyte size as well as the recruitment of new adipocytes.
Using 2D cell cultures we previously determined that lipid droplets modulate cell adhesion force, which can be monitored in real-time experiments using an xCelligence sensor. Fatty acid uptake is a fast event. Pre-treatment of adipocyte precursors by oleic acid promotes induction of adipogenesis ( Figure 6A), OA increases the rate of differentiation, and adherent mature adipocytes could be obtained 3 days after uptake of oleic acid (10 µM in HG glucose, 4.5 g/L, culture media) [12,23]. Both induction of adipogenesis and droplet size increase are accompanied by OA/PPARG-induced expression of CIDEC and G0S2, involved in droplet formation, FAT/CD36 and FABP4 adipogenic markers ( Figures 4A and 8A).
FAT/CD36 expression is induced by PPARs activation during adipogenesis [16]. It is highly induced in both adipocytes and their precursors through FAT/CD36-PPARG signalling ( Figures 4E and 8A) [12,23]. FAT/CD36 interacts with FFA when their extracellular concentration is low but is not required for high lipid uptake [38]. However, we found that FAT/CD36 is required to retain the capacity of lipogenesis induced by high glucose in adipose tissue ( Figure 1A). FAT/CD36 level of expression together with its trafficking are determinants of lipolysis [6,39]. OA treatment activates FAT/CD36 in adipocyte precursors ( Figure 8B) although this active form is inversely correlated with droplet size increase ( Figure 4A). These observations correlate with the increase of adipocyte size observed in PPARG knock-out mice [40], its low level of expression in mature primary human adipocytes [22], and the inhibitory effect of rosiglitazone on induction of size increase ( Figure 4F). PPARG and CEBPA over-expression in obese AT [41] are probably a result of increased recruitment of new adipocytes (hyperplasia) rather than of adipocyte size increase since PPARG downregulation is correlated with fat mass reduction and adipocyte hypertrophy [40]. Furthermore, PPARG overexpression and activation in mature adipocytes do not influence adipose cell hypertrophy [42]. Therefore reduction of PPARG activity during the process of lipid uptake allows an increase in size.
During adipogenesis, FABP4 expression is induced by PPARG, and its intracellular activity is required for adipogenesis (Figures 3 and 4F) and lipolysis ( Figure 1B). During lipolysis, fatty acids are complexed to FABP4, which represents approximately 1% of all soluble proteins in adipose tissue [43,44]. We found that potential extracellular FABP4 uptake could occur in adipogenic precursors (Figure 7). Intracellular interaction of FABP4 with cytokeratin, a cytosquelettal associated protein [45], and exosome-like export of FABP4 from adipocytes also support the hypothesis of FABP4 uptake by ASCs [46]. FABP4 is required for both fatty acid uptake and release through lipolysis but also in fatty acid signalling to regulate PPARG transcriptional activity [34,43]. Our study suggests that the increase of FABP4 stores due to uptake in adipose progenitors ( Figures 6E-G, 7 and 8) or in adipocytes ( Figure 4E,F) were correlated with the reduction of the active form of FAT/CD36. FABP4 abundance is inversely correlated with that of PPARG in human adipose tissues and FABP4 induces proteasomal degradation of PPARG [34]. FAT/CD36 activation by FFA activates AMPK, which in turn inhibits lipolysis through the phosphorylation of ATGL [29].
Thus the relative availability of FABP4 on either fatty acid uptake, lipolysis, and modulation of PPARG gene transcription may result from the regulation of its interactome, including HSL, JAK2, and FAT/CD36 [34,47]. FABP4 was found to regulate gene transcription through interaction with JAK2 signalling (Figure 2). FABP4 signalling through JAK2 may induce SOCS3 gene transcription, resulting in the inhibition of AMPK, increased lipolysis, and reduction of PPARG transcriptional activity [48]. FABP4 retention in the absence of lipolysis and accumulation due to PPARG-induced gene transcription [49] could explain the consecutive increase of PPARG inhibition after long-term transcriptional activation of FAT/CD36. Finally, the reduction of FAT/CD36 levels of expression ( Figure 4A) increases basal lipolysis as found in FAT/CD36 deficient adipocytes [30] and after its inhibition in adipose tissue explants ( Figure 1A).
Fatty acid-FAT/CD36 signalling through PPARG for gene transcription and AMPK for lipolysis finely regulates the size of adipocytes. Our results suggest that FABP4 trafficking is a central regulator of the imbalance between reduction versus an increase in size (Figure 9). On the basis of comparative analyses of homogeneous adipocyte maturation (3T3-MBX cell line) versus heterogeneous differentiation allowing to measure the impact of basal lipolysis (3T3-L1 cell lines and human primary cells), we propose that AMPK and PPARG activations are required during early adipogenesis in order to respectively reduce basal lipolysis and increase lipid uptake and storage capacities through transcriptional regulations. The increase of adipocyte size due to dietary fatty acid uptake is associated with an accumulation of intracellular FABP4 which in turn reduces PPARG activity and AMPK activation leading to basal lipolysis. Finally, free fatty acid release and FABP4 secretion improve lipid uptake, and adipocyte size increase in adipocytes with lower size. This hypothesis suggests that FABP4 plays a major role in the regulation of adipocyte size, in order to avoid hypertrophy. The tight regulation of adipocyte size in adipose tissue may explain the stable bimodal distribution of adipocytes according to the cell size as "small" adipocytes (50-70 µm) and "large" adipocytes (70-120 µm) [5,18,35,37]. Figure 9. Putative mechanism of Fatty acid binding protein 4 (FABP4) role in the regulation of fat mass. Fat mass regulation proceeds through pro-adipogenic processes allowing lipid storage (red pathways) requiring fatty acid-induced PPARG gene transcription, lipid uptake, and inhibition of lipolysis. Fatty acid interaction with FAT/CD36 controls both AMPK anti-lipolytic activity and PPARG mediated gene transcription, including those of FAT/CD36 and FABP4. FAT/CD36 increases during adipogenesis together with FABP4. Increased accumulation of FABP4 in absence of lipolysis in mature adipocytes is a limiting factor inducing reduction of FAT/CD36 through PPARG inhibition (negative feedback), which in turn alters its antilipolytic activity. Finally, FABP4 exerts a limitation to a fatty acid-induced increase of adipocyte size through the restoration of basal lipolysis (green pathways), then promotes small or immature adipocytes to reach the optimal size but inhibits the recruitment of new adipocytes. The model suggests that under normal conditions, lipid storage is then optimized to obtain a homogenous population of adipocytes (in a range of 50-100 µm) allowing to monitor the metabolic response to diet uptake versus energy expenditure.
Finally, induction of adipogenesis in human ASCs by oleic acid was inhibited when the fatty acid was complexed to FABP4 ( Figure 5) and inversely regulates the transcription of PPARG-induced adipogenic markers ( Figure 8A). Our results are in accordance with in vitro as well as in vivo studies showing that FABP4 interferes with differentiation, downregulates adipogenic markers such as adiponectin and leptin, increases lipolysis in adipocytes, and inhibits differentiation of ASCs at the intracellular level [50,51].
FABP4 might enter into cells through endocytosis and then induce intracellular signalling, in addition to intracellular FABP4 accumulation by synthesis, in adipocyte progenitors and mature adipocytes. Thus, FABP4 seems to constitute a central regulator of fat mass expandability.
At the time of treatment, an oleic acid stock solution (400 µM complexed to lipid-free bovine serum albumin BSA 10%) was diluted in cell culture media with either 1% BSA (mouse cell cultures) or human lipid-free albumin (hSA) for human cell cultures then incubated at 42 • C over 2 h. Cell cultures were treated with 1 volume of 2× inhibitor media with 1% BSA and then 1 volume of 2× oleic acid solution in order to avoid fatty acid toxicity. AICAR, ATGListatin, BSA, human free fatty acid albumin, dexamethasone, human recombinant FABP4, FABP4 inhibitor, GW6742, GW9662, IBMX, oleic acid, rosiglitazone, and human albumin were purchased by Sigma Aldrich, insulin Actrapid by Novo Nordisk A/S, Danemark), and AP5258 by Clinigenetics (Lyon, France)

Size Determination, Triglyceride, and FAT/CD36 Content Analyses
The methods were previously described in detail [12,23]. Mouse adipose cell lines 3T3-L1 and sub-clone 3T3-MBX were used to analyze either heterogeneous (allowing detection of basal lipolysis) or homogeneous (90-100% differentiation efficiency) cell cultures, respectively. Briefly, cell cultures were analyzed in real-time experiments. XCelligence sensor (ACEA, Agilent Technologies, Les Ulis, France) detects modifications of adipocyte adhesion force related to TG accumulation (reduction of cell index) or release (an increase of cell index). Realtime cell imaging was performed on Cytation 3 plateform (Biotek Instruments, Winooski, VT, USA) [12,23]. Image and fluorescence data were retrieved using the software Gen5 2.08 (Biotek Instruments).
Analyses of size distribution were performed using a Multisizer (Beckman Coulter, Villepinte, France) with adipocytes fixed in formalin 3% after cell detachment by trypsin 0.05% (Sigma Aldrich, Saint-Quentin-Fallavier, France). Triglyceride (TG) contents were quantified in either cell cultures or adipose tissue explants using a cell permeable AdipoRed fluorescent marker (Lonza, Ozyme, Montigny Le Bretonneux, France) on 96-wells plate cultures in real-time experiments performed at the end of treatments.
Effects of treatments on FAT/CD36 expression, droplet number, droplet size, and cell number, were analyzed by automated quantification of fluorescence after fixation in paraformaldehyde 3 % before labeling with a specific anti-mouse APC-coupled antibody directed against the extracellular domain of FAT/CD36 (2.5 µg/mL, Texas Red filter, Ex 650/Em 661, Biolegend, Ozyme) and/or AdipoRed for TG (FITC filter, Excitation 485/Emission 528) without permeabilization, then Hoechst 33258 in cells permeabilized with 0.1% triton (Sigma Aldrich, Dapi filter, Ex. 355/Em. 465) as previously described [23]. In experiments without detection of FAT/CD36 cells were fixed with formalin 3%. Lipid droplets and nuclei counts were determined for each wavelength according to the controls on images obtained at objective ×4 in 6 to 12 biological replicates.
Analyses of droplet volume and number per cell (i.e., droplet number normalized to Hoechst counts) were performed on ×4 images. TG content (volume) per cell is the result of the mean number of droplets per cell × mean droplet volume. In each experiment, imaging parameters (i.e., led intensity, camera gain, threshold) were optimized on controls and applied to each sample. In cell cultures, mature adipocytes were considered for diameters up to 50 µm and were analyzed according to mean droplet volume and frequency as previously described [23]. Extracellular FAT/CD36 was quantified by APC intensity detection (normalized to nuclei counts on ×4 images). Realtime AdipoRed uptake by explants was performed by fluorescence intensity measurement after 40 min normalized to the time of AdipoRed treatment on Cytation 3 plateform at 37 • C with 5% CO 2 .

Gene Transcription Analyzes
Gene transcription analyses were performed by qRT-PCR after total RNA extraction with Trizol (Sigma Aldrich), reverse transcription using Superscript II on 500 µg total mRNA, and quantified with SYBRGreen kit (Roche Diagnosis, Meylan, France) as delta Ct (cycle threshold) as previously described [1,2]. Results were normalized to hypoxanthine guanine phosphoribosyl transferase (HPRT) standard gene quantification and performed in three to four biological replicates. The list of primers has been previously published [12,23].

Bioinformatic Gene Dataset Analyses
The method used for human gene dataset analyses was previously described [12,52,53]. Briefly, human gene datasets were retrieved from either Gene Expression Omnibus (GEO) datasets or published experiments leading to the identification of lists of human genes regulatable according to adipose phenotype (adipose stem cells ASCs, differentiating adipocytes dA, isolated adipocytes or tissues, AT), extracellular signals, intracellular signalling pathways or transcription regulators. Signalling pathways enriched in fatty acids and insulin human gene datasets are reported in Berger & Géloën 2022 [23], those of glucose have been previously published [53]. Mouse FAT/CD36 knock-out gene datasets were retrieved from Sabaouni et al. [54], and that of intracellular FABP4 from Hurley et al., 2012 [55].

Statistics
Experimental data were analyzed on representative experiments on three (12-wells plates) to 6-10 biological replicates (E96-wells plates). In experiments using inhibitors, which were performed in independent experiments, fold changes to corresponding controls were prefered to absolute values, and statistically significant differences were considered for Student t-test p-values p < 0.05. Multiple conditions were analyzed with StatView 4.5 software through an ANOVA test followed by Fisher's protected least significance difference, post hoc test. ANOVA significantly different values p < 0.05 are reported as letters. In bio-informatic data analyses, significant enrichments were considered for Z-test confidence levels > 95% using Epitools software (https://epitools.ausvet.com.au, accessed 3 February 2020, Ausvet Europe).

Conclusions
Regulation of adipocyte size requires (1) increased fatty acid storage through lipid uptake and (2) high PPARG transcriptional activity for the synthesis of major pro-adipogenic markers FAT/CD36 and FABP4. A finely tuned management of lipid uptake, storage, and release has been found to be orchestrated by the regulation of PPARG transcriptional activity and AMPK mediated antilipolytic activity through a finely tuned regulation of FAT/CD36 signalling by FABP4 (Figure 9). Although FABP4 is required for adipogenesis, its accumulation by either retention in non-lipolytic adipocytes or uptake in adipogenic precursors may induce a negative control of respectively, size increase resulting in homogenization of adipocyte size, and fatty acid-induced adipogenesis involved in the recruitment of new adipocytes. This hypothesis is in accordance with the observation that FABP4 lack is associated with increased body weight, as well as an increase in body fat without changes in glucose concentration or lipid homeostasis [56]. Thus, FABP4 is a major controller of fatty acid-induced FAT/CD36 mediated regulation of adipose tissue homeostasis by FAT/CD36 signalling.