Bisphenol S Induces Adipogenesis in Primary Human Preadipocytes From Female Donors

Abstract

Human exposure to bisphenol A has been associated with negative health outcomes in humans and its use is now regulated in a number of countries. Bisphenol S (BPS) is increasingly used as a replacement for bisphenol A; however, its effects on cellular metabolism and potential role as an endocrine disruptor have not been fully characterized. In the current study, we evaluated the effect of BPS on adipogenesis in primary human preadipocytes. The effect of BPS on the differentiation of human preadipocytes was determined after treatment with BPS at concentrations ranging from 0.1nM to 25μM by quantifying lipid accumulation and mRNA and protein levels of key adipogenic markers. Treatment of preadipocytes with 25μM BPS induced lipid accumulation and increased the mRNA and protein levels of several adipogenic markers including lipoprotein lipase and adipocyte protein 2 (aP2). Cotreatment of cells with the estrogen receptor antagonist ICI-182,780 significantly inhibited BPS-induced lipid accumulation and affected aP2 but not lipoprotein lipase protein levels. Cotreatment of cells with the glucocorticoid receptor antagonist RU486 had no effect on BPS-induced lipid accumulation or protein levels. Furthermore, reporter gene assays using a synthetic promoter containing peroxisome proliferator-activated receptor-γ (PPARG)-response elements and a PPARG-responsive human aP2 promoter region showed that BPS was able to activate PPARG. To our knowledge, this study is the first to show that BPS induces lipid accumulation and differentiation of primary human preadipocytes, and this effect may be mediated through a PPARG pathway.

Bisphenol A (BPA) has been used in the manufacture of polycarbonate plastic and other consumer products for decades. Human exposure to BPA is ubiquitous and has been measured in humans from around the world (1, 2). Exposure to environmental chemicals such as BPA has been suspected to play a role in the development of a number of human disorders including cardiovascular disease and obesity (3–5). We and others have previously shown that BPA and its metabolite BPA-glucuronide (BPA-G) can induce lipid accumulation and adipogenesis in murine 3T3L1 and primary human preadipocyte cell models (6–9).

Bisphenol S (BPS) is now commonly used as a substitute for BPA, due to regulatory restrictions, and has been found in a number of consumer products including thermal receipt papers, polycarbonate plastics, canned foods/soft drinks, epoxy resins and plastic baby bottles (10–14). BPS is a close analog of BPA in which the functional phenol groups are linked by a sulfonyl (SO₂) group instead of the dimethylmethylene (C(CH₃)₂) group. BPS has also been detected in the environment in sediment and indoor dust samples collected from North America and China (11, 15). Several studies have now measured BPS in human urine (15, 16) with 1 study reporting BPS in 81% of urine samples tested with a mean concentration of 0.654 ng/mL (0.003μM) (15). Because of structural similarities with BPA, there is a risk that BPS may have similar negative effects on endocrine systems (17). Of particular concern is the fact that BPS has been reported to be much less biodegradable than BPA in seawater (18). Several reports have examined the biological effects of BPS in animal models. Zebrafish exposed to BPS showed skewed sex ratios, decreased egg and sperm counts, increased estradiol concentrations and decreased body weight, thyroxine and testosterone levels in males (19, 20). Much like BPA, BPS has also been reported to have estrogenic activity in vitro (17, 21, 22).

We have previously reported the effect of BPA and its metabolite BPA-G increased both lipid accumulation and expression of differentiation markers in human and murine preadipocytes (6, 9). It is well documented in human and rodent cell models, that adipocyte differentiation is controlled by the timely expression of key transcriptions factors, namely, peroxisome proliferator-activated receptor-γ (PPARG) and CCAAT-enhancer-binding protein-α (CEBPA), resulting in adipogenesis and lipid accumulation (23). The differentiation process is also affected by hormonal cues from glucocorticoids and insulin (24). In particular, the glucocorticoid receptor (GR) has been linked to BPA-induced adipogenesis (25). To date, only 1 report has showed that very low concentrations of BPS can induce lipid accumulation and glucose uptake in the murine 3T3L1 preadipocyte cell line (26).

In the current study, the effect of BPS on the differentiation of primary human preadipocytes was examined. To our knowledge, we show for the first time that BPS can induce lipid accumulation in human preadipocytes and increase expression of several key adipogenic markers at both the mRNA and protein levels. In order to determine a potential mechanism of action of BPS-induced adipogenesis, reporter gene assays in the presence of the known adipogenic activators PPARG and GR were performed. The data show that BPS can activate PPARG using a PPARG-response element (PPRE) reporter plasmid but could not activate GR using a glucocorticoid-response element (GRE) reporter plasmid. Interestingly, BPS could also increase human adipocyte protein 2 (aP2) promoter activity only in the presence of both PPARG and GR. The data presented here suggest that further studies are needed to evaluate the extent of the biological effects caused by BPS on adipogenesis and human health.

Materials and Methods

Adipocyte differentiation

Primary human sc preadipocytes from hip thigh or abdomen depots (Zenbio, Inc) from female donors with body mass indexes less than 25, age 25–57, and who gave informed consent were differentiated in Preadipocyte Medium (ZenBio) containing 33μM biotin, 17μM pantothenate (Sigma-Aldrich), and 100nM insulin (Roche Diagnostics) for 14 days as well as 500μM 3-isobutyl-1-methylxanthine (Sigma-Aldrich) from day 0 to 4 and 5μM troglitazone (Sigma-Aldrich) from day 2 to 4. The indicated concentrations of BPS or 1μM dexamethasone (DEX), as a positive control, were also included in the differentiation media and replenished every 2 days (both Sigma-Aldrich). Ethics approval for the use of human cells was obtained by the Health Canada Research Ethics Board. For the estrogen receptor (ER) and GR antagonist studies, 1μM ICI-182,780 (ICI) or 1μM RU486 (both Sigma-Aldrich) was also added as indicated.

Lipid staining and quantification

Primary human preadipocytes were differentiated as described above for 14 days with the indicated treatments. Cells were then fixed with 10% formalin and stained with Nile red and 4′,6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI) (both 1 μg/mL) as previously described (27). Nile red (staining for lipid droplets) was imaged at 530 nm and DAPI (staining for cell nuclei) at 405 nm using fluorescence imaging on an Olympus IX71 microscope. Cells were imaged at ×100 magnification. Nile red fluorescence was quantified at 485/528 nm (excitation/emission) and normalized to DAPI staining measured at 360/460 nm using a Synergy 2 Microplate Reader (BioTek Instruments, Inc).

Determination of mRNA expression levels by quantitative real-time-PCR

Total RNA were extracted from differentiating cells treated as described using the RNeasy kit and genomic DNA was eliminated using the RNase-Free DNase kit (both from QIAGEN). RNAs (1 μg) were reverse transcribed into cDNA using the iScript Advanced cDNA Synthesis kit (Bio-Rad). Sample cDNA was amplified and quantified in a CFX96-PCR Detection System using the iQSYBR SsoFast EvaGreen Supermix (Bio-Rad). The primer pairs for each gene target were: CEBPA forward-TGGACAAGAACAGCAACGAG and reverse-CCATGGCCTTGACCAAGGAG; lipoprotein lipase (LPL) forward-CAGGATGTGGCCCGGTTTAT and reverse-CGGGGCTTCTGCATACTCAA; sterol regulatory element binding transcription factor 1 SREBF1 forward-GAGCCATGGATTGCACTTTCG and reverse-AAGAGAGGAGCTCAATGTGGC; PPARG forward-TCCGAGGGCCAAGGCTTCAT and reverse-GCAAACCTGGGCGGTCTCCA; aP2 forward-CATCAGTGTGAATGGGGATG and reverse-GTGGAAGTGACGCCTTTCAT; perilipin PLIN forward-CCCAGGAGTGACAGGAATTGT and reverse-AGGCCTTTGTTGACTGCCAT; and β-actin (ACTB) forward-GACTTCGAGCAAGAGATGGC and reverse-CCAGACAGCACTGTGTTGGC. Primer efficiencies were more than or equal to 90% and specificity was confirmed by sequence blast and melting curve analysis. All target gene transcripts were normalized to ACTB expression (which was not affected by BPS treatment) and fold inductions were calculated using time-matched control (ethanol)-treated samples and the comparative cycle threshold (ΔΔC_T) method was used for data analysis (28).

Western blot analysis

Cells were lysed in radioimmunoprecipitation assay buffer containing protease inhibitors (Roche Diagnostics). Primary antibodies for aP2, LPL (both R&D Systems), and ACTB (13E5) (both Cell Signaling Technology) were used with appropriate horseradish peroxidase-labeled secondary antibodies (Table 1). Blots were developed using Clarity Western ECL Substrate (Bio-Rad). Western blottings were visualized using a ChemiDoc Imager and quantified using Image Lab software (Bio-Rad). Protein levels were normalized to ACTB.

Table 1

Antibody Table

Peptide/Protein Target	Antigen Sequence (if Known)	Name of Antibody	Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody	Species Raised in; Monoclonal or Polyclonal	Dilution Used
aP2	Escherichia coli-derived recombinant human FABP4/A-FABP Cys2-Ala132 accession number P15090	Human FABP4/A-FABP antibody	R&D Systems, AF3150	Goat; polyclonal	1 μg/mL
LPL	E. coli-derived recombinant human Lipoprotein Lipase/LPL accession number P06858	Human/mouse LPL/LPL antibody	R&D Systems, AF7197	Goat; polyclonal	1 μg/mL
ACTB	Synthetic peptide corresponding residues near the amino terminus of human ACTB	13E5	Cell Signaling Technology, 4970	Rabbit; monoclonal	1:1000

Reporter gene assays

COS-7 cells (ATCC) were seeded in phenol red-free DMEM (Wisent) containing 5% dextran-coated charcoal stripped serum (Sigma-Aldrich). Twenty-four hours after plating, the cells were transfected with plasmid DNA using FuGENE HD (Roche Diagnostics) according to the manufacturer's instructions and dosed 6 or 24 hours after transfection. For the estrogen-response element (ERE) transcriptional assay, COS-7 cells were transfected with 125 ng of 3X-ERE-luciferase reporter plasmid (ERE-luc), 25-ng pVP16-ERα (ERα expression plasmid) (both from Addgene) and 10-ng pRL-CMV Renilla (Promega) as an internal control. For the GR transcription assays, cells were transfected as above with 125 ng of 3X-GRE-luc (containing GR-response elements), 25-ng pTL2-GR (a GR expression plasmid), and 10 ng of pRL-CMV as an internal control. For the PPARG transcriptional assays, cells were transfected with 25 ng of pcDNA human (h)PPARG1 (a PPARG expression plasmid) (29) and 3X-PPRE-luc (30) (both kind gifts from Dr Jae Bum Kim, Seoul National University, Seoul, Korea, and Dr Didier Junquéro, Centre de Recherche Pierre Fabre, Castres, France, respectively). Six hours after transfection, cells were treated with vehicle control, and the indicated concentrations of BPS, 1μM DEX, or 5μM PPARG agonist rosiglitazone (ROSI). For the transcriptional assays using the endogenous 5403-bp human aP2 promoter (29), COS-7 cells were transfected with either 25-ng pcDNA hPPARG1 alone, pTLrGR alone, or cotransfection of both in the presence of 125 ng of human aP2 promoter luciferase (haP2-luc) and 10 ng of pRL-CMV as an internal control. Six hours after transfection, cells were treated with either 5μM ROSI, 25μM BPS, 1μM DEX or both BPS and DEX. For all transcriptional assays described above, 24 hours after treatment, cells were lysed in Passive Lysis buffer (Promega), and luciferase activity was quantified with the Dual Luciferase Assay kit (Promega) using a Glomax96 Luminometer (Promega). Luciferase activity was normalized to the internal control and to vehicle control.

Statistical analyses

Statistical significance was analyzed by one-way ANOVA with Holm-Sidak post hoc test analysis using SigmaPlot 12.5. A paired Student's t test analysis was conducted between treated and untreated samples on the mRNA data generated in Figure 2 below (25μM BPS and 0.01μM BPS). A repeated measures ANOVA was also performed for the 25μM BPS treatment for the mRNA expression data using GraphPad Prism Software.

Figure 2

Effect of BPS treatment on mRNA expression of adipogenic genes. Differentiation and treatment of preadipocytes with ethanol (MIT control), 0.01μM or 25μM BPS (A–F) or 1μM DEX (G–L) was performed as described. Total RNAs extracted from MIT controls (n = 8), 25μM BPS (n = 8), or 0.01μM BPS (n = 4) were used for qRT-PCR analysis of the adipogenic genes normalized to ACTB gene expression. Values are expressed as mean fold change relative to day 2 control ± SEM. *, P < .05 relative to control calculated by a paired Student's t test analysis (0.01μM BPS) and a repeated measures ANOVA (25μM BPS).

Results

Effects of BPS on lipid accumulation in human preadipocytes

The effect of BPS on lipid accumulation was assessed by Nile red lipid staining and quantification in human preadipocytes treated with 0.0001μM–25μM BPS for 14 days. We observed a nonmonotonic response to BPS, where lipid accumulation was increased by almost 2-fold at concentrations between 0.1nM and 1nM, although not statistically significant, decreased at 10μM and increased further at 25μM BPS. In fact, 25μM BPS induced a statistically significant 4.2-fold increase in lipid accumulation, whereas treatment with DEX (positive control) resulted in a 15.3-fold increase (Figure 1, A and B). No cell death was observed under any of the treatment conditions (data not shown).

Figure 1

BPS-treated human preadipocytes and lipid accumulation. Human preadipocytes were treated with ethanol (control), increasing concentrations of BPS or 1μM DEX as a positive control, and lipid accumulation was visualized using Nile red staining (A) and quantification (B) on day 14 of differentiation. Values are expressed as mean fold change relative to control ± SEM from at least 6 donors performed in triplicate wells. *, P < .001 relative to control calculated by one-way ANOVA with Holm-Sidak post hoc test analysis.

Effects of BPS on mRNA expression levels of adipogenic markers

To further evaluate BPS-induced preadipocyte differentiation, mRNA levels of key adipogenic factors were evaluated by quantitative real-time PCR in response to low (0.01μM) and high (25μM) concentrations of BPS. We have evaluated the expression levels of the key adipogenic transcription factors PPARG, CEBPA, and SREBF1 as well as the markers of mature adipocytes LPL, PLIN, and aP2. Treatment of the cells with 10nM BPS and 25μM BPS both resulted in a statistically significant increase of 1.6-fold in PPARG expression at day 4 (Figure 2A). CEBPA expression was up-regulated by 25μM BPS at day 2 and day 4 by approximately 2-fold, reaching statistically significance when compared with MIT (3-isobutyl-1-methylxanthine, insulin, troglitazone) control at day 4 (Figure 2B). The same trend was observed for SREBF1 where a significant increase of 2-fold in mRNA expression was obtained with the high dose treatment of BPS at day 4 (Figure 2C). Although similar trends were observed with 10nM BPS, the treatment did not reach significance for CEBPA and SREBF1 expression. Notably, treatment with the positive control DEX resulted only in an increase of 2.7- and 2-fold for PPARG at day 2 and day 4, respectively, and an increase of 3-fold in CEBPA expression at day 4. LPL mRNA levels were significantly up-regulated by 25μM BPS treatment when compared with MIT controls at days 6, 8, and 10 after the initiation of differentiation by 2-, 4-, and 3-fold, respectively (Figure 2D). No change in LPL expression was observed for the 10nM BPS treatment (Figure 2D) when compared with the MIT controls. aP2 expression was significantly up-regulated by 25μM BPS as compared with the day-matched MIT controls by 10-fold at day 2 and 2-fold at day 4 (Figure 2E). The high fold increase at day 2 is a result of very low expression of aP2 in the MIT control at this time point. PLIN was up-regulated by BPS treatment by 3- and 2-fold at day 2 and day 4 at the high dose (Figure 2F). Treatment of cells with DEX as a positive control showed, as expected, a significant increase in the mRNA expression levels of all adipogenic markers examined albeit at different times throughout differentiation (Figure 2, G–L). Taken together, the data show that the observed BPS-induced lipid accumulation also resulted in changes in the mRNA expression levels of key adipogenic genes during differentiation.

Effects of BPS on protein levels of adipogenic markers

The effects of increasing concentrations of BPS treatment on the protein levels of the adipogenic markers LPL, aP2, and PLIN were also evaluated by Western blot analysis on day 14 at the end of differentiation. The data show that the protein levels of LPL and aP2 were all significantly increased after treatment with 25μM BPS by 8.8- and 4.5-fold, respectively, relative to control (Figure 3, A and B). Protein levels of aP2 and LPL were also significantly increased by 2.5- and 5.3-fold after treatment with 10μM BPS. Treatment of cells with DEX resulted in significant increases of 6.2- and 11.9-fold in the protein levels of aP2 and LPL, respectively, relative to control (Figure 3, C and D). These results clearly show that the effect of BPS on mRNA expression can also be seen at the protein level during differentiation.

Figure 3

Effect of BPS treatment on protein levels of adipogenic markers. Human preadipocytes were differentiated as described with the indicated concentrations of BPS for 14 days, and protein levels of the adipogenic markers LPL and aP2 were assessed by Western blotting (A) and densitometry (B). Human preadipocytes were also differentiated as described with 1μM DEX as a positive control, and protein levels were assessed by Western blotting (C) and densitometry (D). Images are representative of results from at least 6 different human donors. ACTB was used as the protein loading control. Densitometry values are expressed as means of 4–8 donors ± SEM. *, P < .05 relative to control calculated by one-way ANOVA with Holm-Sidak post hoc test analysis.

The effects of ER and GR antagonists on BPS-induced differentiation

Both BPA and BPS have previously been shown to induce the transcriptional activity of ERα, whereas the GR has been implicated in BPA-induced adipogenesis (21, 25). Therefore, the potential roles of both the ERα and GR in BPS-induced adipocyte differentiation were investigated using selective antagonists for both receptors. Human preadipocytes were treated with 25μM BPS in the presence of either 1μM ICI (ER antagonist) or 1μM RU486 (GR antagonist) and lipid accumulation was determined as described. The data show that ICI or RU486 treatment alone does not induce significant lipid accumulation, whereas 25μM BPS induces a significant increase in lipid accumulation relative to control (Figure 4, A–C). Cotreatment of BPS with ICI significantly inhibited BPS-induced lipid accumulation by 37.5%, whereas cotreatment with RU486 appeared to cause a small but not statistically significant reduction. Treatment of cells with the GR agonist DEX resulted in significant lipid accumulation which was decreased, as expected, after RU486 cotreatment. Treatment of cells with estradiol had no effect on lipid accumulation (data not shown) and as previously reported (6, 9). Moreover, ERα and ERβ mRNA expression levels were undetectable in the human preadipocytes under these conditions (data not shown). The effect of ICI and RU486 on BPS-induced adipocyte marker protein levels was also examined. Cotreatment of cells with BPS and ICI caused a small but significant decrease in aP2 but not LPL expression levels (Figure 5, A and B). RU486 cotreatment led to no significant changes in BPS-dependent protein levels (Figure 5, C and D). However, cotreatment of RU486 with DEX caused a significant decrease in aP2 expression levels but not LPL. The data show that although ICI inhibited BPS-induced lipid accumulation, only some protein levels were affected.

Figure 4

The effects of ER and GR antagonists on BPS-induced lipid accumulation. Preadipocytes were treated with ethanol (control), 25μM BPS, or 1μM DEX in the presence and absence of 1μM ICI or RU486 as indicated and lipid accumulation was visualized (A) and quantified (B and C) by Nile red staining. Values are expressed as mean ± SEM for at least 3 donors. *, P < .05 relative to control calculated by one-way ANOVA with Holm-Sidak post hoc test analysis.

Figure 5

The effects of ER and GR antagonists on BPS-induced protein levels of adipogenic markers. Preadipocytes were differentiated as described, and protein levels of the adipogenic markers LPL and aP2 were assessed by Western blotting (A and C) and densitometry (B and D) analysis at day 14 of differentiation after treatment with ethanol (control), 25μM BPS, or 1μM DEX in the presence or absence of 1μM ICI or RU486 as indicated. ACTB was used as the protein loading control. Values are expressed as mean ± SEM for at least 3 donors. Statistical significance was assessed relative to control by one-way ANOVA with Holm-Sidak post hoc test analysis.

The effects of BPS treatment on transcriptional activity

We first confirmed that BPS has estrogenic activity, as previously described (21, 22), using a transcription reporter assay. We show that BPS can activate an ERE-luciferase reporter in an ERα-dependent manner at concentrations as low as 0.1μM (Figure 6A). In order to determine other possible mechanisms by which BPS treatment induces adipogenesis of primary human preadipocytes, we investigated whether BPS can activate PPARG or GR, both transcription factors involved in adipogenesis (25, 31). COS-7 cells were transfected with expression vectors containing PPARG and/or GR, and the transcriptional activity of the receptors was assessed using luciferase reporter plasmids containing either a 3X-PPRE or a 3X-GRE, respectively. As indicated, BPS did not activate GRE-luciferase activity (Figure 6B); however, treatment with 25μM or 50μM BPS was able to induce the transcriptional activity of PPARG on a PPRE-luciferase reporter by 1.5- and 1.6-fold, respectively (Figure 6C). The data suggest that BPS can activate PPARG directly and promote its transcriptional activity. The ability of BPS to activate PPARG was further evaluated using a human aP2 promoter luciferase reporter plasmid (haP2-luc) that contains the genomic region 5.4 kb upstream from the transcriptional start site of aP2 (Figure 6, D–F). Using the PPARG agonist ROSI, we show that it was able to activate PPARG transcription on the endogenous PPREs in the haP2-luc plasmid; although cotransfection of GR with PPARG enhances the ROSI-dependent effects on the promoter, whereas GR transfection alone had no effect (Figure 6D). In contrast, in the presence of PPARG alone or GR alone, BPS did not increase aP2 luciferase activity; however, in the presence of both PPARG and GR, BPS was able to significantly induce aP2 luciferase activity by 1.6-fold (Figure 6E). Moreover, cotreatment with BPS and DEX did not enhance these effects, whereas treatment with DEX alone did not induce aP2 luciferase activity (Figure 6F). This suggests that GR functions as a coactivator at the human aP2 promoter and cannot induce aP2 activity on its own in the promoter region that was studied. Our results suggest that PPARG and GR may both be required for BPS-induced adipogenic effects and up-regulation of the aP2 gene.

Figure 6

The effect of BPS on ER, GR and PPAR activity. A, COS-7 cells were transfected for 24 hours with an ERE-luciferase plasmid as well as ERα expression plasmid and then treated with increasing concentrations of BPS for an additional 24 hours. Luciferase activity was then assessed as described. B, COS-7 cells were transfected with a GRE-luciferase plasmid and pTL2-rGR expression plasmid for 6 hours and then treated with increasing concentrations of BPS for an additional 24 hours. Luciferase activity was then assessed as described. C, COS-7 cells were transfected with a PPRE-luciferase plasmid and pcDNA hPPARG1 expression plasmid for 6 hours and then treated with increasing concentrations of BPS for an additional 24 hours. Luciferase activity was then assessed as described. D–F, COS-7 cells were transfected with pcDNA hPPARG1 and/or pTL2-rGR in the presence of haP2-luciferase reporter plasmid (haP2-luc). Cells were transfected for 6 hours and then dosed with DMSO (untreated control), 1μM ROSI, 25μM BPS, or 1μM DEX as indicated, and luciferase activity was then assessed as described. For all panels, values are expressed as mean ± SEM for 3 separate experiments; *, P < .05 relative to untreated/DMSO control; #, P < .05 relative to pcDNA hPPARG1 ROSI-treated cells calculated by one-way ANOVA with Holm-Sidak post hoc test analysis. G, Schematic of the human aP2 promoter with putative GRE and PPRE sites highlighted. Potential transcription factor binding sites were determined using Genomatix.

Discussion

Increasing evidence suggests that BPS, the predominant replacement chemical for BPA, can now be detected in humans and may have unwanted health effects (17, 32, 33). The current study is the first to report that BPS can induce adipocyte differentiation in primary human preadipocytes and that it can activate the adipogenic transcription factor PPARG. Our data suggest that, similar to BPA, BPS increases adipogenesis in human preadipocytes and that further study is required to better understand potential hazards of widespread BPS exposure. The few reports available now indicate that BPS can affect endocrine function as demonstrated by studies showing decreased testosterone, androstenedione and cortisol levels in ex vivo and in vitro models (17, 33). To our knowledge, only 1 study has recently shown that BPS can also induce lipid accumulation and glucose uptake in the murine 3T3L1 cell line at low concentrations (26).

Adipogenesis involves the timely expression and activity of a series of transcription factors such as PPARG, GR, CEBPA, and SREBF1, leading to differentiation of preadipocytes into mature adipocytes (23, 25, 34). We have previously shown that BPA and its metabolite BPA-G can both induce lipid accumulation and adipocyte differentiation (6, 8, 9). Here, we show that BPS induces lipid accumulation as well as the mRNA expression of several adipogenic markers such as aP2, SREBF1, LPL, and PLIN. The lowest concentrations of BPS used in this study are within the physiological range (15); however, we mainly observed statistically significant effects at the higher concentrations of 10μM and 25μM BPS. Interestingly, the fold change increases in BPS-induced lipid accumulation and adipogenic mRNA expression levels appear to be higher than the fold changes observed previously for BPA and BPA-G (6, 9, 35) and (manuscript in preparation). This may indicate that BPS may be more adipogenic than those 2 chemicals and/or works through a different mechanism of action. Interestingly, the master regulators of adipocyte differentiation, PPARG and CEBPA, were significantly up-regulated at the mRNA level after BPS treatment, by almost 2-fold. Further, even the low concentration of BPS (10nM) increased PPARG expression by 1.6-fold. This is consistent with other reports in human preadipocytes showing that PPARG expression levels increase roughly 2-fold during the course of adipogenesis (36). To our knowledge, only 1 study examined adipogenesis in the murine 3T3L1 preadipocyte cell line in response to BPS (26). Although BPS did induce lipid accumulation in 3T3L1 cells in that report, it did not induce expression of Pparg, Srebf1 or aP2 mRNA, whereas expression of those same genes was increased by BPA. This suggests that BPS and BPA may be activating different sets of proadipogenic genes to induce adipocyte differentiation in murine preadipocytes. The discrepancy in the target genes between that study and the current one is likely due to the species-related differences (murine vs human), and the use of the GR agonist DEX in the BPS containing differentiation cocktail used in that report.

Both BPA and BPS have been shown to have estrogenic activity and BPA has been reported to activate GR (25, 37). Consistent with these findings we also show that BPS increased ERα-mediated ERE-luciferase activity (37), and that the ER-antagonist ICI could inhibit BPS-induced lipid accumulation, and aP2 protein, but not LPL. This indicates that the effect of ICI may be gene-specific. In addition, the markers of mature adipocytes such as aP2 and LPL do not have a direct function in lipid droplet formation in the cell. In fact, there is an entire family of proteins collectively named PLIN, adipophilin and tail interacting protein of 47-kDa (PAT) that control lipid droplet formation (38). If ICI specifically inhibits the expression of 1 or more of these proteins, it could affect lipid accumulation but not expression of the other adipogenic markers evaluated in this study. In contrast, the GR antagonist RU486 had no effect on either BPS-induced adipogenesis and BPS did not activate GR transcription. This suggests that similar to our results with BPA or BPA-G, BPS does not directly activate GR to induce adipogenesis.

Because GR did not seem to be involved in BPS-induced adipogenesis, we also examined the role of PPARG, the master regulator of adipocyte differentiation (31). We show for the first time that BPS can activate PPARG using a PPRE-luciferase reporter assay. Moreover, using the endogenous human aP2 promoter which contains putative binding sites for both PPARG and GR, we found that BPS can induce aP2 promoter activity in a luciferase reporter assay in the presence of both PPARG and GR, but not in the presence of only 1 of those transcription factors, suggesting that GR has a coactivator function in this context. Interestingly, these effects were independent of DEX indicating that a ligand bound GR is not required for the effects mediated by BPS on the human aP2 promoter. This is also consistent with the lack of an inhibitory effect by RU486 in this study as the presence of this inhibitor would not impact the function of unliganded GR. Moreover, DEX alone was unable to activate the human aP2 promoter, suggesting some cooperation between PPARG and GR on this promoter. In fact, cross talk between monomeric GR and other nuclear receptors and participation of GR in protein complexes in the absence of DNA binding have all been previously described (39, 40). GR has been also been shown to synergize with the PPARG-related family member PPARA where both GR and PPARA were shown to cooccupy specific chromatin sites (41). Whether GR and PPARG are found at the same DNA binding sites on the aP2 promoter and the role of GR and PPARG in BPS-mediated adipocyte differentiation remains to be fully characterized. Although the presence of PPARG is absolutely required for adipogenesis (42), the mRNA levels of PPARG are not known to be increased by much more than 2-fold even by DEX when compared with controls (36), consistent with our results with BPS. PPARG activity is much more affected by the presence of its fatty acid ligands (43), binding partners, such as retinoid X receptor (44) and posttranslational modifications such as phosphorylation and acetylation which are important factors determining the extent of PPARG activation and its effect on adipocyte differentiation (45).

Despite the very close structural similarities, it is now clear that BPS has similar but also distinct effects on cellular metabolism compared with BPA. Although both BPA and BPS have estrogenic properties, it has been shown that BPS does not have GR activating properties and displays less antiandrogenic activity in vitro, compared with BPA (17, 46, 47). Androgen levels have been linked to inhibition of adipogenesis (48) and decreases in abdominal fat (49); therefore, different effects by BPA or BPS on the androgen receptor could potentially result in differences in fat accumulation in vivo. In addition to differences in hormonal effects, BPA and BPS are also converted to gluco- and sulfo-conjugated forms in vivo with some small differences in the conjugating enzymes involved, which can result in distinct activities (50). There are also additional limitations to this study, we have used a limited number of subjects, varying in age and most of the effects reached statistical significance only at concentrations higher than human exposure. However, our data and the current literature suggest that although BPA and BPS have some similarities, there are also differences in BPS-mediated effects warranting further study.

In summary, we show for the first time that BPS can induce lipid accumulation and differentiation in primary human preadipocytes and that this effect may be mediated via direct activation of the nuclear receptor PPARG. We also show that at the human aP2 promoter, PPARG activation by BPS is enhanced by GR. This study suggests that BPS may not be a harmless substitute for BPA and more thorough toxicological and epidemiological investigations of BPS effects on human health are warranted.

Acknowledgments

We thank Dr Michael Wade and Dr Emily Tung for reviewing this manuscript and Adele Desjardins-Boudreau for technical assistance.

This work was supported by the Health Canada Chemical Management Plan (E.A.) and the Natural Sciences and Engineering Research Council of Canada Visiting Scientist Fellowships (J.G.B. and S.A.).

Disclosure Summary: The authors have nothing to disclose.

For News & Views see page 1321

Abbreviations

 
• ACTB

  β-actin

 
• aP2

  adipocyte protein 2

 
• BPA

  bisphenol A

 
• BPA-G

  BPA-glucuronide

 
• BPS

  bisphenol S

 
• CEBPA

  CCAAT-enhancer-binding protein-α

 
• DAPI

  4′,6-Diamidino-2-Phenylindole, Dihydrochloride

 
• DEX

  dexamethasone

 
• ER

  estrogen receptor

 
• ERE

  estrogen-response element

 
• GR

  glucocorticoid receptor

 
• GRE

  glucocorticoid-response element

 
• h

  human

 
• ICI

  ICI-182,780

 
• LPL

  lipoprotein lipase

 
• MIT

  3-isobutyl-1-methylxanthine, insulin, troglitazone

 
• PLIN

  perilipin

 
• PPARG

  peroxisome proliferator-activated receptor-γ

 
• PPRE

  PPARG-response element

 
• ROSI

  rosiglitazone

 
• SREBF1

  sterol regulatory element binding transcription factor 1.