Introduction

Obesity remains a major public health concern (Cao 2014; Capurso and Capurso 2012) and its prevalence is predicted to rapidly increase in the near future (Cao 2014; WHO 2018a; WHO 2018b). Obesity is associated with a state of chronic, low-grade inflammation that triggers metabolic diseases including type 2 diabetes mellitus (Hameed et al. 2015). The etiology of obesity and its associated diseases has been attributed to various factors, such as genetics, diet and lifestyle and certain environmental obesogens that disrupt the functions of the lipid-storing adipocytes (Acconcia et al. 2015; Brehm and D’Alessio 2014; Sheikh et al. 2017).

Environmental obesogens are a category of chemical endocrine disruptors that have detrimental effects on the endocrine control of metabolism. They are widely used in the manufacture of industrial and consumer products, mainly as plasticizers (Acconcia et al. 2015). In adipose tissue, obesogens can act on adipocyte precursors and differentiated adipocytes directly, or they can disturb the hormonal control of appetite and metabolism, leading overall to an increase in the number of adipocytes and facilitating fat storage by existing adipose cells (Grun and Blumberg 2009a; Grun and Blumberg 2009b; Schug et al. 2011). Obesogens also depress the utilization of fatty acids and lipoproteins, causing dyslipidemia and fatty liver disease (Grun and Blumberg 2009a; Grun and Blumberg 2009b). Drinking water, bottled mineral water and beverages and diet (seafood, cereals, food wrapped or stored in polycarbonate- or polyvinyl chloride-based materials) are the major sources of intake, with daily exposure in the micromolar concentration range (Chen et al. 2009; Goran et al. 2013; Grun and Blumberg 2009a; Grun and Blumberg 2009b, Welshons et al. 2006). Accordingly, obesogen chemicals can be routinely detected in human serum, tissues and urine at nanomolar concentrations, indicating existing or past exposure (Goran et al. 2013; Grun and Blumberg 2009b; Wang et al. 2013). Animal studies have shown that such quantities cause hyperlipidemia, fat accumulation, liver steatosis, impair oxidative phosphorylation and fatty acid utilization and can even evoke heritable changes in body adiposity and metabolism (Grun and Blumberg 2009b; Naville et al. 2013). These alarming findings highlight the strong negative impact of obesogens on metabolism and their likely contribution to the prevalence of obesity. However, little is known about the molecular mechanisms underlying the metabolic effects of obesogens.

Bisphenol A (BPA; 2,2-bis(4-hydroxyphenyl)propane) is one of the most widely used obesogen chemicals and can be found in food storage containers and the protective internal lining of cans used for food and beverages, in thermal printer paper, electronic equipment, medical devices and dental sealants (Flint et al. 2012; Konieczna et al. 2015; Repossi et al. 2016). The metabolic effects of BPA on adipocytes have been previously studied, with results showing that BPA contributes to obesity-associated diseases (Acconcia et al. 2015; Dong et al. 2018; Flint et al. 2012; Heindel 2011; Schug et al. 2011).

Obesity has an immune component and how adipocytes metabolize lipids is largely dictated by immune cells of the adipose tissue. Among these, adipose tissue macrophages (ATMs) are the most studied immune cells. ATMs can either support or impede insulin sensitivity of the adipocytes, cause adipose tissue fibrosis or initiate burning-off of stored lipids in the form of heat (Ivanov et al. 2018). How ATMs affect adipose tissue function is largely dependent on their number. Obesity is associated with an increase in ATM number (Amano et al. 2014; Haase et al. 2014; Lumeng et al. 2007), which favors the development of inflammation and, ultimately, an insulin-resistant state (Boutens and Stienstra 2016). How the ATM number is controlled is still not fully understood (Röszer 2018), although recent reports have shown that ATM self-renewal is a key element in maintaining a metabolically beneficial ATM pool (Amano et al. 2014; Braune et al. 2017; Haase et al. 2014; Röszer 2018; Waqas et al. 2017).

Because BPA accumulates in the adipose tissue (Geens et al. 2012), in the present study, we addressed whether BPA can affect macrophage self-renewal. We demonstrate that BPA increases macrophage self-renewal and this effect is associated with activation of the mitogen-activated ERK pathway and an increase in the levels of liver X receptor alpha (LXRα), with both changes regulating macrophage self-renewal. ERK is a member of the MAPK superfamily and plays a vital role in cell cycle progression by regulating cell cycle entry (Li et al. 2016). LXRα is a nuclear receptor, with a role in the control of macrophage immune functions (A-González and Castrillo 2011) and cell cycle (Pascual-García et al. 2011).

Results

Bisphenol A increases macrophage self-renewal

To assess the possible effects of BPA on macrophage function, we treated murine J774A.1 macrophages in vitro with BPA (Fig. 1a), using concentrations that mimic human environmental exposure (Ribeiro et al. 2017; Vandenberg et al. 2007, Welshons et al. 2006). We assessed viability, phagocytic activity and activation state; all of which are key determinants of the physiological functions of macrophages (Chang 2009). Treatment of J774A.1 macrophages with 1 μM BPA for 1 or 18 h and with 100 nM BPA for 18 h, resulted in overt cell death, as shown by a decrease in mitochondrial activity in cells assessed with the PrestoBlue cytotoxicity assay (Fig. S1). However, treatment with lower concentrations of BPA (1 or 10 nM) for the same time period did not impair metabolic activity (Fig. S1). Likewise, phagocytic activity was not affected by 1 and 10 nM BPA for 18 h (Fig. S2), corroborating the finding that macrophage viability and maturation are not affected by a low BPA concentration.

Fig. 1
figure 1

BPA increases macrophage self-renewal. a Chemical structure of BPA. b–f Gene transcription in response to 18 h treatment of J774A.1 macrophages with 1 and 10 nM BPA. g Cell number of J774A.1 macrophages after treatment with 1 and 10 nM BPA. h Quantification of BrdU incorporation in J774A.1 macrophages after 18 h stimulation with BPA. i Quantification of Ki67+ macrophages after treatment with 1 and 10 nM BPA. j ATM cell number after treatment with 10 nM of BPA. k BrdU+ ATMs after 18 h treatment. l Quantification of Ki67+ ATMs after 18 h stimulation with 10 nM BPA. m–o Gene transcription in response to 18 h treatment of ATMs with 10 nM BPA. ATMs were pooled from 3 to 5 mice and treated in triplicate. The unprocessed raw flow cytometric data are accessible in Flow Repository (https://flowrepository.org/) under the repository IDs FR-FCM-ZYP6 and FR-FCM-ZYP7. RNA expression of the indicated genes was normalized to Actb. qPCR and FACS values are expressed as mean fold change relative to vehicle ± SEM (n = 5–6). Statistical analysis was performed using unpaired Student’s test with Welch’s correction. Significance is indicated by *p < 0.05, **p < 0.01 and ***p < 0.001. BPA bisphenol A, BrdU 5-bromo-2′-deoxyuridine, ATMs adipose tissue macrophages

Full size image

Macrophage activation is key for macrophage function, by ensuring pathogen elimination and immunological self-tolerance and also proper metabolic performance (Ginhoux et al. 2015; Röszer 2015). To test whether BPA could affect macrophage activation, we measured the mRNA levels of hallmark genes of macrophage activation after treatment of J774A.1 cells with BPA for 18 h, finding that 10 nM BPA modestly increased Arg1 and Il6 transcription without affecting the transcription of Il4rα, Tnfa and Nos2 (Fig. 1b–f). These findings suggest that BPA does not substantially activate macrophages.

Finally, we tested the effect of BPA on macrophage self-renewal, a recently recognized trait that ensures the proper development of macrophage niches in their corresponding tissues and hence has impact on healthy organ development and function (Röszer 2018). We found that at concentrations of 1 and 10 nM, BPA increased J774A.1 macrophage number after 18 h of treatment (Fig. 1g). To test whether this was attributable to self-renewal, we examined the incorporation of 5-bromo-2′-deoxyuridine (BrdU) into the DNA during this time, finding that BrdU incorporation was higher in BPA-treated cells than in vehicle-treated cells (Fig. 1h). Taken together, these data suggest that low doses of BPA increase macrophage self-renewal but do not affect macrophage viability and maturation. Next, we questioned whether BPA treatment could induce cell cycle re-entry of quiescent macrophages. To do this, we used the nuclear proliferation marker Ki67, which is expressed in the active phases of the cell cycle but is absent in quiescent cells. We found that 10 nM BPA induced cell cycle re-entry of quiescent J774A.1 macrophages, measured as an increase in the number of Ki67+ macrophages (140.4% ± 9.51) (Fig. 1i).

The above findings suggest that BPA specifically affects macrophage self-renewal. Since BPA is a lipophilic substance and hence can be accumulated in adipose tissue (Geens et al. 2012), it is conceivable that ATMs are exposed to BPA. ATMs are capable of self-renewal (Braune et al. 2017; Waqas et al. 2017), hence we questioned whether BPA could affect ATM self-renewal. Similar to the findings in J774A.1 macrophages, we found that 18 h treatment with 10 nM BPA did not have any cytotoxic effects on ATMs (Fig. S3) but increased the number of ATMs (Fig. 1j). Importantly, self-renewal of ATMs was also increased by BPA, as shown by the increase in BrdU labeling in BPA-treated ATMs (Fig. 1k). Likewise, BPA increased the re-entry of quiescent ATMs into the cell cycle, measured as Ki67+ ATMs (Fig. 1l). Again, similar to the results in J774A.1 macrophages, the ATM activation state was not affected by BPA (Fig. 1m–o). BPA also increased the self-renewal of human monocytic THP-1 cells, macrophages differentiated from THP-1 cells in vitro (Figs. S4–14). Briefly, 10 nM BPA increased the number of monocytic THP-1 cells after 18 h treatment (Fig. S4), along with the increased amount of Ki67+ cells (Fig. S5) and cells in the G2/M-phase (Fig. S6). BPA increased the self-renewal of macrophages differentiated from THP-1 cells; however, this effect was apparent following 48 h treatment (Figs. S7–14). Taken together, these data suggest that BPA is able to trigger macrophage self-renewal.

Bisphenol A induces ERK phosphorylation in macrophages

Mechanisms that increase macrophage self-renewal are not well understood (Röszer 2018), but can involve the activation of mitogen-activated protein kinase (MAPK) signaling pathways including the extracellular signal-regulated kinase (ERK) cascade, which can be activated by growth factors to regulate, among other processes, cell cycle progression and survival (De Luca et al. 2012).

To examine whether BPA activated ERK signaling in macrophages, J774A.1 cells were treated with BPA and cell lysates were subjected to Western blotting. We found that 10 nM BPA induced the rapid phosphorylation of ERK, which was detectable 30 min after treatment (Fig. 2a–a”’) but not for 1 nM BPA (Fig. S15). ERK activation was not sustained by the 18-h-long BPA treatment (Fig. S16). Interestingly, the lower dose of BPA (1 nM) rather increased total ERK levels after 18 h (Fig. 2b–b”’). In accordance with these findings, BPA increased the transcription of the mitogenic genes Ccne and Ccnd1 (Fig. 2c, d); however, it had no effect on the transcription of the cell cycle inhibitor genes Cdkn1a, Cdkn1b and Trp53(Figs. S17–S19).

Fig. 2
figure 2

BPA upregulates signals involved in cell cycle progression. a Western blotting of phospho- (p)ERK/ERK in J774A.1 macrophages treated with 10 nM BPA for 30 min. a’–a”’ Quantification of pERK, ERK and pERK/ERK normalized to β-actin. b Western blotting of pERK/ERK in J774A.1 macrophages treated with 1 nM BPA for 18 h. b’–b”’ Quantification of pERK, ERK and pERK/ERK normalized to β-actin. c–f Gene expression levels in response to 18 h stimulation of J774A.1 macrophages with 1 or 10 nM BPA. g A representative of Western blotting of LXRα in J774A.1 macrophages. g’ Densitometry of LXRα normalized to β-actin, exposure time 5 s. qPCR and Western blot values are expressed as mean fold change relative to vehicle ± SEM (n = 3–6). Statistical analysis was performed using unpaired Student’s t test with Welch’s correction. Significance is indicated by *p < 0.05, **p < 0.01, and ***p < 0.001

Full size image

In addition to the ERK pathway, we searched for other possible mechanisms that could account for the BPA-induced macrophage self-renewal. In this regard, we have previously shown that macrophage self-renewal is associated with an increase in the transcription of N-Myc downstream-regulated gene 2 (Ndrg2) and a decrease in the expression of the transcription factor MAF bZIP transcription factor B (Mafb) (Menendez-Gutierrez et al. 2015; Waqas et al. 2017). We found that 1 nM but not 10 nM BPA increased the expression of Ndrg2 (Fig. S20), whereas the expression of Mafb was not affected at either concentrations (Fig. 2e).

LXR inhibits macrophage self-renewal

BPA, at least in part, increases cellular lipid storage through liver X receptor (LXR) signaling (Dahlman et al. 2006; Marmugi et al. 2012). LXR signaling can also inhibit macrophage self-renewal (Pascual-García et al. 2011; Röszer 2018), although its physiological impact and underlying mechanisms are largely unexplored. Thus, we asked whether BPA could alter LXR signaling and affect macrophage self-renewal. First, we measured the mRNA expression of Lxra and Lxrb, which encodes the two isotypes of LXR. Interestingly, we found that 10 nM BPA modestly increased LXRα expression both at the mRNA (Fig. 2f) and protein levels (Fig. 2g, g’), without affecting Lxrb mRNA levels (Fig. S21).

Next, we activated macrophage LXR signaling by treating J774.A1 cells with the synthetic and endogenous LXR ligands, GW3965 (1 μM) and 25-hydroxycholesterol (1 μg/ml), respectively. We found that 18 h treatment with either ligand induced the accumulation of macrophages in the G0/G1-phase and decreased the amount in the G2/M-phase (Fig. 3a, a’). Notably, this treatment was not cytotoxic for the cells (Fig. S22) but decreased self-renewal, as measured by the number of Ki67+ cells (Fig. 3b). The effects of both LXR ligands were similar in ATMs (Fig. 3c, d). Importantly, we verified that the LXR ligands activated LXR signaling by measuring the expression of the LXR target gene Abcg1 (Fig. 3e, f).

Fig. 3
figure 3

LXR ligand activation impedes macrophage self-renewal. a Cell cycle analysis of J774A.1 macrophages and a’ quantification of cell cycle distribution by propidium iodide (PI) staining. b Proportion of Ki67+ J774A.1 macrophages after treatment with 1 μM of GW3965 or 1 μg/ml of 25-hydroxycholesterol for 18 h. c Cell cycle analysis of in vitro cultured ATMs and c’ quantification of cell cycle distribution by PI staining. d Proportion of Ki67+ ATMs after treatment with 1 μM of GW3965 or 1 μg/ml of 25-OHC for 18 h. e and fAbcg1 transcription in J774A.1 macrophages and ATMs after 18 h stimulation. qPCR and FACS values are expressed as mean fold change relative to vehicle ± SEM (n = 3–6). The unprocessed raw flow cytometric data are accessible in Flow Repository under the repository ID FR-FCM-ZYP3 and FR-FCM-ZYPA. Statistical analysis was performed using unpaired Student’s t test with Welch’s correction, or with one-way or two-way ANOVA. Significance is indicated by *p < 0.05, **p < 0.01 and ***p < 0.001

Full size image

Our findings indicate that LXR activation decreases macrophage self-renewal, which is in line with other studies on the cell cycle-inhibitory effect of LXRs in various cell types (Hassan et al. 2015; Pascual-García et al. 2011; Warns et al. 2018). Turning this question on its head, we next assessed whether inhibiting LXR signaling with the inverse agonist SR9238 would activate macrophage self-renewal. We first tested a range of concentrations (40–500 nM) to determine the lowest effective dose that blocked LXR signaling without affecting cell viability. We found that 200 nM SR9238 effectively blocked LXR target gene transcription (Abcg1, Srebp1c and Fasn) (Fig. 4e–f’; Fig. S23) without apparent toxicity (Fig. S24). This concentration is in accord with the known IC50 of SR9238 for LXRα (200 nM) and LXRβ (40 nM) (Griffett et al. 2012). After treating J774A.1 macrophages with SR9238, we found an increased percentage of cells in the G2/M-phase (Fig. 4a, a’) accompanied by a prominent increase in the percentage of Ki67+ cells (Fig. 4b). Similar results were obtained with ATMs (Fig. 5c, d), with no changes in cell viability (Fig. S24).

Fig. 4
figure 4

Inhibition of LXR signaling increases macrophage self-renewal. a Cell cycle analysis of J774A.1 macrophages and a’ quantification of cell cycle distribution by PI staining. b Proportion of Ki67+ J774A.1 macrophages after treatment with 200 nM SR9238 for 48 h. c Cell cycle analysis of in vitro cultured ATMs and c’ quantification of cell cycle distribution by PI staining. d Proportion of Ki67+ ATMs after treatment with 200 nM SR9238 for 48 h. e–f’ Gene transcription in J774A.1 macrophages and ATMs after 48 h stimulation with 200 nM SR9238. ATMs were pooled from three to six mice and treated in triplicate. The experiment was repeated twice. qPCR and FACS values are expressed as mean fold change relative to vehicle ± SEM (n = 3–6). The unprocessed raw flow cytometric data are accessible in the flow repository under the repository ID FR-FCM-ZYP4 and FR-FCM-ZYP5. Statistical analysis was performed using unpaired Student’s t test with Welch’s correction. Significance is indicated by *p < 0.05, **p < 0.01 and ***p < 0.001

Full size image
Fig. 5
figure 5

Inhibition of in vivo LXR signaling by SR9238 increases macrophage number. a Body weight. b eWAT weight. c SVF cell number normalized to weight of eWAT. d Quantification of cell cycle distribution in the SVF by PI staining. e Percentage of CD45+/F4/80+ ATMs. f–f” Relative gene transcription in liver after LXR inhibition for 14 days. EWAT of two mice were pooled (n = 6). All data are presented as mean ± SEM. The unprocessed raw flow cytometric data are accessible in the flow repository under the repository ID FR-FCM-ZYP9. Statistical analysis was performed using unpaired Student’s t test with Welch’s correction or one-way ANOVA. Significance is indicated by *p < 0.05, **p < 0.01 and ***p < 0.001

Full size image

To assess whether these in vitro findings could be replicated in vivo, we treated male mice with 30 mg/kg SR9238 for 14 days to study the effect of systemic LXR blockade on macrophage self-renewal (Griffett et al. 2012; Griffett et al. 2015). We found that whereas SR9238 did not affect body weight (Fig. 5a), there was a trend for a reduction in the weight of epididymal white adipose tissue (eWAT) (Fig. 5b). Through the analysis of self-renewal, we observed an increase in the cell number in the ATM-rich stromal vascular fraction (SVF) normalized to eWAT weight in SR9238-treated mice (Fig. 5c). Also, cell cycle analysis of this fraction demonstrated a non-significant increase in cells in the G2/M-phase (Fig. 5d). Closer inspection of the SVF revealed that SR9238 significantly increased the number of CD45+/F4/80+ ATMs (Fig. 5e). Notably, SR9238 also increased the number of CD45+/F4/80+ Kupffer cells in the liver (Fig. S25), without altering liver weight or cell number (Figs. S26 and S27). Nevertheless, cell cycle analysis of non-parenchymal cells of the liver also demonstrated a non-significant increase in cells in the G2/M-phase (Fig. S28). We also assessed the expression of the LXR target genes Srebp1c, Abcg1 and Cd5l in the liver and found them to be downregulated (Fig. 5f–f”). Overall, these data suggest that the blockade of LXR signaling increases tissue-resident macrophage number.

Lxra expression distinguishes self-renewing from quiescent macrophages

Our results so far suggest that LXR signaling antagonizes BPA effects, hence limiting excessive self-renewal of macrophages. This raises the possibility that macrophages with reduced LXR levels are more prone to BPA-induced increase in self-renewal. We next examined whether Lxr transcript levels were different between self-renewing (Ki67+) and quiescent (Ki67−) ATMs. We found that Lxra expression was lower in Ki67+ ATMs than in Ki67− ATMs (Fig. 6a), whereas the levels of Lxrb were similar between the two groups (Fig. S29). To further confirm that self-renewing macrophages express low levels of Lxra, we treated macrophages with IL-4, which is known to induce macrophage self-renewal (Röszer 2018). As expected, macrophage self-renewal was increased by IL-4 treatment (Fig. 6c–c”) and this was accompanied by a decrease in both Lxra and Lxrb expressions (Fig. 6b, b’). By contrast, IFNγ, which is known to arrest macrophage cell cycle (Singh et al. 2011), decreased self-renewal (Fig. 6d) but increased Lxra and Lxrb transcription (Fig. 6e, e’).

Fig. 6
figure 6

Lxra expression is low in self-renewing macrophages. a Relative Lxra transcription in sorted Ki67− and Ki67+ ATMs. Actb Ct values were above 20 due to processing for cell sorting followed by RNA isolation. b and b’ Lxra and Lxrb transcription in J774A.1 macrophages treated for 18 h with 10 g/ml IL-4. c and c’ J774A.1 macrophages treated with vehicle or IL-4 for 18 h. c” Flow cytometric quantification of Ki67+ macrophages. d Flow cytometric quantification of BrdU+ macrophages upon 18 h stimulation with IFNγ. e Relative transcription of Lxra and Lxrb in J774A.1 macrophages after 4 h IFNγ stimulation. f and gCD68 and LXRA transcription in THP-1 cells upon differentiation with 100 nM PMA. h LXRα+ macrophages in Kupffer cells, ATMs and peritoneal macrophages. i Flow cytometric analysis of Ki67+ macrophages in Kupffer cells, ATMs and peritoneal macrophages. qPCR values are expressed as mean fold change relative to vehicle ± SEM. Statistical analysis was performed using unpaired Student’s t test with Welch’s correction, Mann-Whitney test or one-way ANOVA. All data are presented as mean ± SEM (n = 3–6). The unprocessed raw flow cytometric data are accessible in the flow repository under the repository ID FR-FCM-ZYP3 and FR-FCM-ZYPB. Significance is indicated by *p < 0.05, **p < 0.01 and ***p < 0.001. ATMs adipose tissue macrophages, PM peritoneal macrophages, PMA phorbol 12-myristate 13-acetate

Full size image

Given the above, we next evaluated whether a decline in self-renewal was associated with an increase in LXR expression in human cells. The human monocytic cell line THP-1 can be differentiated into quiescent THP-1 macrophages by chemical induction (Richter et al. 2016), allowing us to compare the levels of LXR expression in self-renewing and quiescent states in the same cell type. We measured LXRA and LXRB mRNA levels in proliferating monocytic- and quiescent macrophage-like THP-1 cells, finding that monocyte-macrophage differentiation and decreased self-renewal was associated with an increase in LXRA expression, which paralleled the expression of the macrophage marker CD68 (Fig. 6f, g). Moreover, LXRB transcription also increased after differentiation for 4 days (Fig. S30).

Finally, we questioned whether Lxra expression correlated with the degree of self-renewal ability of distinct tissue-resident macrophages in the mouse. We first measured the percentage of LXRα-expressing macrophages in distinct tissue-resident macrophage pools, finding that 78.35 ± 19.66% of Kupffer cells expressed LXRα, whereas 32.94 ± 7.7% of ATMs and only 18.96 ± 5.03% of resident peritoneal macrophages expressed the receptor (Fig. 6h). These differences in the percentage of LXRα levels were mirrored by the mRNA levels (Fig. S31). In turn, Kupffer cells had the lowest percentage of Ki67+ macrophages (7.7 ± 2.15%), whereas 17.22 ± 3.38% of ATMs and 27.48 ± 8.38% of peritoneal macrophages were Ki67+ (Fig. 6i). These findings suggest that the more quiescent tissue-resident macrophage pools contain more LXRα-expressing macrophages than those that are rich in self-renewing macrophages. These data suggest that an increased Lxra expression inhibits macrophage self-renewal.

Our overall findings show that low doses of BPA increase macrophage self-renewal through the ERK signaling pathway and also upregulate the expression of Lxra. Moreover, tissue-resident macrophages of various origins express different levels of Lxra, which mirrors their self-renewing capacities. Finally, we show that self-renewing macrophages in the mouse are characterized by their negligible expression of Lxra.

Discussion

Our study shows that BPA interferes with the control of macrophage number. We demonstrate that BPA, presumably through ERK/MAPK signaling, increases macrophage self-renewal without affecting macrophage maturation or activation. BPA also moderately increases LXRα expression in macrophages and this effect likely antagonizes BPA-induced cell cycle entry. This is a relevant, yet unexplored, effect of BPA that calls into question its safety in the context of tissue-macrophage homeostasis.

Regulation of tissue-resident macrophage homeostasis is highly relevant in health and disease and involves elimination and self-renewal of macrophages, which is crucial for the development and physiological function of tissues (Röszer 2018). ATM homeostasis in the steady-state is regulated by adipose type 1 innate lymphoid cells, which eliminate ATMs to prevent excessive macrophage expansion. In obesity, however, this regulation is disturbed (Boulenouar et al. 2017). Self-renewal is a newly discovered feature of macrophages (Röszer 2018) that is used to replenish local resident macrophage niches without the need for infiltration of monocyte-derived macrophages (Belhareth and Mège 2015; Jenkins et al. 2011). Intrinsic and extrinsic signals, such as M-CSF, IL-4, IL-6 and IFNγ and tissue-specific cues including neuropeptide FF, regulate macrophage niches by either promoting or repressing macrophage self-renewal (Belhareth and Mège 2015; Luo et al. 2018; Smith et al. 2013; Waqas et al. 2017; Yu et al. 2012). Dysregulation of macrophage self-renewal can lead to diseases that contribute to obesity in addition to atherosclerosis, cancer and neurodegenerative disease (Belhareth and Mège 2015). Nevertheless, we still lack a good understanding of the mechanisms that regulate macrophage self-renewal.

The environmental obesogen BPA, which is widely used in industrial, medical and consumer products, is a known endocrine disruptor (Flint et al. 2012; Huang et al. 2012; Repossi et al. 2016) and there is increasing evidence that vertebrates and invertebrates can be exposed to BPA under specific conditions. BPA has been shown to contribute to the development of obesity-induced diseases and also affects fertility and tissue development (Acconcia et al. 2015; Flint et al. 2012; Heindel 2011; Holtcamp 2012; Schug et al. 2011). Of particular concern is the persistence of BPA in the environment, despite its short half-life of 2.5–4 days (Flint et al. 2012). The concentration at which BPA exerts its effects varies between species, cell types and tissues (Ribeiro et al. 2017; Vandenberg et al. 2007), with increased bioaccumulation in lipophilic tissues, such as adipose tissue, over hydrophilic compartments (Geens et al. 2012). Particularly alarming are the findings that BPA can be detected in urine specimens from pregnant women (Lim et al. 2017; Ye et al. 2009). Furthermore, maternal BPA exposure has a detrimental effect on organ development (Fan et al. 2018) and, as a consequence of the early life exposure, BPA might affect children’s behavior and increase the prevalence of obesity (Braun et al. 2011; Ejaredar et al. 2017; Harley et al. 2013; Vafeiadi et al. 2016). Of note, macrophage self-renewal peaks in the neonatal period, allowing the development of tissue-resident macrophage niches (Ginhoux et al. 2015). Along this line, testicular macrophage expansion is negatively affected by BPA, which can impair sperm counts and male hormonal levels in adulthood (Peretz et al. 2014; Röszer 2018; Williams et al. 2014). The presence of lipophilic BPA in formula milk, baby food and plastic toys raises the possibility that self-renewing macrophages can be targets of BPA in infants (Andaluri et al. 2018; Cao et al. 2011; Doerge et al. 2010; Food and Administration 2009; Williams et al. 2014).

The obesogenic effects of BPA have been studied mainly in the context of endocrine signaling (Zhang et al. 2018) and its direct effects on adipocytes (Acconcia et al. 2015; Ariemma et al. 2016). Regarding macrophages specifically, BPA has been investigated with respect to modulation of inflammation (Byun et al. 2005; Liu et al. 2014; Rogers et al. 2013) but its potential effects on self-renewal have been overlooked. Resident ATMs govern adipose tissue metabolism (Luo and Liu 2016) but in the setting of obesity this regulation is disturbed, resulting in their activation towards a pro-inflammatory phenotype that is metabolically detrimental (Boutens and Stienstra 2016; Rogero and Calder 2018; Rosen and Spiegelman 2014). We demonstrate that BPA increases macrophage self-renewal. Notably, even 1 nM BPA was effective. The human plasma BPA level is within that range and has been shown to increase cell cycle progressions of LNCaP cells (Wetherill et al. 2006). Also, long-term BPA treatment increases the growth rate of prostate epithelial cells (Huang et al. 2017) and promotes carcinogenesis (Wang et al. 2017). Together, our results suggest that ERK activation by BPA may trigger macrophage self-renewal. Prolonged exposure to BPA, or bioaccumulation of BPA, may augment this effect. Accordingly, BPA might have a detrimental effect on macrophages during the development of obesity, which is associated with uncontrolled ATM self-renewal (Amano et al. 2014; Boutens and Stienstra 2016; Huang et al. 2012; Lumeng et al. 2007; Weisberg et al. 2003).

In the present study, we found that BPA increased LXRα expression. The importance of LXR signaling in the regulation of macrophage self-renewal became apparent when we inhibited LXR signaling. Whereas LXR ligand activation inhibited ATM self-renewal, its inhibition increased macrophage number. This latter finding is in line with recent data showing that the absence of LXR increases the number of F4/80+/CD11b+ macrophages in the liver (Endo-Umeda et al. 2018), consistent with its importance in the regulation of macrophage homeostasis (Menendez-Gutierrez et al. 2015; Pascual-García et al. 2011). Of note, the two LXR isoforms (LXRα and LXRβ) share a high degree of homology between their ligand-binding domains (Lund et al. 2006) but are differentially regulated in cell cycle control of various cell types. For instance, LXRα inhibits proliferation of LNCaP human prostate cancer cells (Fukuchi et al. 2004), whereas LXRβ inhibits cell cycle in human breast cancer cell lines, melanoma and pancreatic cancers (Candelaria et al. 2014; Hassan et al. 2015; Pencheva et al. 2014). Nevertheless, both isoforms can control cell cycle (Pascual-García et al. 2011; Vedin et al. 2013).

The mechanism of tissue-resident macrophage replenishment varies: they can either self-renew locally or they can be repopulated by monocytes (Murray and Wynn 2011; Okabe and Medzhitov 2015; Röszer 2018). For instance, peritoneal macrophages can maintain their population by local proliferation, whereas the self-renewing capacity of Kupffer cells is debatable (Davies et al. 2013; Röszer 2018; Schulz et al. 2012). In the steady-state, Kupffer cells maintain their homeostasis with low self-renewing capacity, while in the setting of inflammation and injury there is increased self-renewal (Davies et al. 2013; Zigmond et al. 2014). As a possible regulator of macrophage self-renewal, LXR signaling has been investigated in various cell types (Endo-Umeda et al. 2018; Hassan et al. 2015; Kim et al. 2010). Upon activation, LXRs regulate cholesterol and lipid metabolism, immune cell functions and also macrophage self-renewal (Beltowski 2011; Kim et al. 2010; Pascual-García et al. 2011; Spann and Glass 2013). LXR signaling is relevant for proper tissue development and gene regulation of macrophage survival and immunity (Joseph et al. 2004). For example, macrophage differentiation in the spleen marginal zone depends on LXR activation for the clearance of pathogens (Den Haan and Kraal 2012; Noelia et al. 2013). Notably, impaired LXR signaling leads to the accumulation of cholesterol and is a contributing factor for the development of atherosclerosis and autoimmune diseases (Ito et al. 2016; Pascual-García and Valledor 2012).

Our data suggest that the differences in LXR expression of distinct macrophages may correlate with the level of macrophage self-renewal. Future studies using lineage-tracing and gene expression analyses will be needed to evaluate whether LXR expression serves as an indicator for in situ tissue-resident macrophage self-renewal. In addition, LXR signaling and environmental obesogens may play a role in obesity-induced disease, providing the impetus to further examine the effect of BPA and other environmental endocrine disruptors on nuclear receptors and their contribution to the development of metabolic disease.

In summary, we identify herein a potential relationship between BPA and the homeostatic maintenance of macrophage number, which, at least in part, is a consequence of ERK and LXR signaling in macrophages.

Methods

Animals

We performed in vivo studies with C57BL/6 male mice, aged 8–13 weeks (Charles River Laboratories). Animal experimentation was performed with the approval of the National Research Institute for Radiobiology and Radiohygiene, Budapest. C57BL/6 male mice (aged 8–16 weeks) used for the isolation of ATMs were obtained from the Institute of Comparative Molecular Endocrinology, University of Ulm, as approved by the Tübingen Regional Government Office on Animal Welfare.

Isolation of adipose tissue stromal vascular fraction and in vitro culture of adipose tissue macrophages

The SVF was isolated from epididymal white adipose tissue (eWAT) depots of mice using collagenase digestion (Weisberg et al. 2003). Briefly, washed eWAT pads were minced and digested in 10 ml of digestion medium consisting of 7 ml of Hank’s balanced salt solution (H9269, Sigma-Aldrich), 3 ml of a 7.5% bovine serum albumin (BSA) solution and 20 mg of collagenase II (C6885, Sigma-Aldrich) for 1 h with agitation at 120 rpm. After digestion, adipocytes and lipids were separated from the SVF, which contains ATMs. The SVF was passed through a 100-μm strainer and centrifuged for 10 min at 1500 rpm at 4 °C. The pellet was resuspended in selection medium consisting of 2 mM EDTA and 0.5% BSA in 1 × phosphate-buffered saline (PBS), pH 7.4. The cells were then used for flow cytometry and/or in vitro culture.

For in vitro culturing, the SVF was cultured in Dulbeccos’s modified Eagle’s medium, supplemented with 10% of heat-inactivated FBS, 20% L-cell-conditioned medium and 1% penicillin/streptomycin for 4 h to allow seeding of ATMs. The medium was then carefully removed and the attached ATMs were treated with BPA (239658, Sigma), SR9238 (SML1510, Sigma), GW3965 (G6295, Sigma), or 25-hydroxycholesterol (H1015, Sigma), as indicated. Posttreatment cell viability was assessed with the resazurin-based PrestoBlue reagent (Thermo Fisher Scientific), according to the manufacturer´s instructions.

J774A.1 macrophages and THP-1 cells were cultured in RPMI-1640 (both cell lines were obtained from ATCC), supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin/streptomycin and were also treated as indicated. THP-1 cells were differentiated with 100 nM of phorbol 12-myristate 13-acetate (PMA) for 2 days, after which differentiated THP-1 macrophage-like cells were treated as described.

Preparative cell sorting and fluorescence-activated cell sorting analysis

The SVF was adjusted to a density of 106 cells/ml and resuspended in selection buffer consisting of 2 mM EDTA and 0.5% BSA in PBS, pH 7.2, for fluorescence-activated cell sorting (FACS) analysis. Cells were fixed and permeabilized with a 20-fold dilution of permeabilization buffer (00-8333-56, eBioscience) to ensure the labeling of the nuclear antigen Ki67. Cells were then incubated with the Fc receptor blocker (KT1632) (MA5-18012, Invitrogen) for 15 min, after which cells were stained with the respective antibodies. To identify ATMs, we used the following antibodies: phycoerythrin-conjugated F4/80 monoclonal antibody (BM8) (12-4801-82, eBioscience), allophycocyanin-conjugated CD11b monoclonal antibody (M1/70) (17-0112-81, eBioscience) and their respective isotype controls, rat phycoerythrin-conjugated IgG2a,κ (clone R35-95) (553930, BD Pharmingen) and rat allophycocyanin-conjugated IgG2b,κ (clone A95-1) (553991, BD Pharmingen). We also analyzed ATMs within the ATM-rich gate (Ampem et al. 2016). Other antibodies and their respective isotype controls used were as follows: phycoerythrin-cyanin 7-conjugated Ki67 monoclonal antibody (25-5698-80, eBioscience), rat phycoerythrin-cyanin 7-conjugated IgG2b,κ (clone eB149/10H5) (15380950, eBioscience), allophycocyanin-conjugated Ki67 monoclonal antibody (clone 20Raj1) (17-5699-41, eBioscience), phycoerythrin-cyanin 5.5-conjugated CD45 monoclonal antibody (30-F11) (35-0451-80, eBioscience) and rat phycoerythrin-cyanin 5.5-conjugated IgG2b,κ (Clone eB149/10H5) (35-4031-80, eBioscience). After 1 h incubation, cells were washed with selection buffer and used for analysis. The following unconjugated antibodies were also used: anti-LXRα antibody (ab176323, Abcam) and anti-BrdU (ab126556, Abcam). The unconjugated antibody for anti-LXR alpha was secondarily labeled with Alexa-488-conjugated goat anti-rabbit IgG (A11034, Invitrogen). We also used phycoerythrin-conjugated BrdU monoclonal antibody (BU20A) (12-5071-41, eBioscience) to label BrdU+ cells. Control cells were labeled with secondary antibody or isotype IgG.

For preparative cell sorting of proliferating (CD11b+/F4/80+/Ki67+) and non-proliferating (CD11b+/F4/80+/Ki67−) ATMs, we used an Aria cell sorter (Becton Dickinson). Cells were also incubated with an Fc receptor blocker and permeabilized. In some experiments, sorted cells were collected for RNA extraction. For cell cycle analysis, cells were washed twice with cold 1 × PBS and stained with Nicoletti buffer with propidium iodide staining solution (50 μg/ml) (00-6990-50; eBioscience). Analysis was performed on the BD LSR II cell cytometer (1000 events). For data acquisition and analysis, we used the BD FACSDiva 8.0.1 (BD Biosciences) and FlowJo (Tristar) software. The unprocessed raw flow cytometric data are accessible in Flow Repository, as indicated.

RNA extraction and quantitative PCR

Total RNA was isolated from macrophages with TRI Reagent (T9424, Sigma-Aldrich). For sorted ATMs, we used the RNeasy Micro Kit (Qiagen) for RNA isolation, to ensure an optimal RNA yield. Gene expression analysis was performed on the ViiA 7 quantitative PCR platform 7 (Thermo Fisher Scientific). Gene products were analyzed with SYBR Green chemistry (Applied Biosystems) and gene expression values were presented as the relative mRNA levels according to the ∆Ct and ∆∆Ct methods, with the Actb housekeeping gene as a reference. The sequence-specific primers used are listed in the Electronic supplementary material of this article (Supplemental Table 1).

Western blot analysis

Cells were lysed in RIPA buffer and 50 μg of protein was subjected to 10% SDS-polyacrylamide gel electrophoresis. Proteins were transferred onto 0.2-μm nitrocellulose membranes (1620112, Bio-Rad) and blocked with 5% skimmed milk in TBS containing 1% Tween-20 (TBS-T). Blots were incubated with primary antibodies at 4 °C overnight, washed in TBS-T and incubated with the appropriate conjugated secondary antibodies for 1 h. Blots were developed with the Pierce ECL Western blotting substrate (RPN2109, GE Healthcare). When appropriate, blots were stripped and re-probed with new primary antibodies. Gel quantification was performed with ImageJ software (NIH). The antibodies used are as follows: mouse monoclonal anti-β-actin antibody (A1978, Sigma), phospho-p44/42 MAPK (pErk1/2) (Thr202/Tyr204) antibody (9101, Cell Signaling Technology), p44/42 MAPK (Erk1/2) (Thr202/Tyr204) antibody (4695, Cell Signaling Technology) and anti-LXR alpha antibody [EPR6508(N)] (ab176323, Abcam). We used the following secondary antibodies for detection: goat anti-rabbit IgG (H + L) secondary antibody, HRP (65-6120, Invitrogen) and anti-mouse IgG HRP (P 0447, DAKO).

Statistics

All data represent the mean ± SEM. For statistical analysis, the unpaired Student’s t test with Welch’s correction, Mann-Whitney test, one-way analysis of variance (ANOVA), or two-way ANOVA with Dunnett’s post hoc test for multiple comparisons was applied, as appropriate. A p value less than 0.05 was considered significant. GraphPad Prism 6 (GraphPad Software) was used for statistical analysis and the illustration of results.