Culture medium Dulbecco's modified Eagle's medium/Nutrient Mixture F12 D-MEM/F-12 (cat. no. 12400-024), trypsin-EDTA (cat. no. 25200-072) were purchased from Gibco (Thermo Fisher Scientific, Inc.); phenol red-free D-MEM/F-12 (cat. no. 90-090-PBR) was obtained from Corning (Mediatech, Inc.); heat-inactivated fetal bovine serum (FBS) from Natocor; 3-isobutyl-1-methylxanthine (IBMX, cat. no. I7018), bovine serum albumin (BSA, cat. no. 10735086001) and dexamethasone (cat. no. D1756) were purchased from Sigma-Aldrich/Merck KGaA and rosiglitazone (cat. no. ab120762) was from Abcam.

Immortalized epithelial cell lines, derived from tumor (4T1, LM3 and MC4-L1) and non-tumor (NMuMG) mammary cells were used. Mouse metastatic mammary cancer cells 4T1 (ER⁻, PR⁻, HER2⁻) and NMuMG were obtained from the American Type Culture Collection (ATCC). LM3 is an immortalized mouse metastatic mammary cell line obtained from a primary tumor (ER⁻, PR⁻, HER2⁺) by Urtreger et al (22). MC4-L1 is an immortalized mouse metastatic mammary line obtained from a primary tumor (ER⁺, PR⁺) by Lanari et al (23). All cells were cultured in D-MEM/F-12 medium (cat. no. 12400-024; Gibco; Thermo Fisher Scientific, Inc.) containing 10% FBS (complete D-MEM/F-12 medium) in a humidified 5% CO₂ atmosphere at 37°C. Every 2–3 days, cell culture medium was removed and replaced by complete D-MEM/F-12 medium. All cells used were mycoplasma-free.

Murine 3T3-L1 preadipocyte cells (obtained from ATCC) were grown in complete D-MEM/F-12 medium in a humidified 5% CO₂ atmosphere at 37°C. Cells were used between passages 12-22; adipocyte differentiation was induced as previously described (24). Briefly, 3T3-L1 preadipocytes were seeded in 12-well plates and grown until confluence (day 0). Two days later, an adipogenic cocktail containing 0.5 mM IBMX, 0.1 µM dexamethasone and 0.1 µM rosiglitazone was added and cells were cultured further for 2 days. Subsequently, culture medium was replaced by complete D-MEM/F-12 containing 2 µM insulin until the end of the culture period, changing the medium every other day. After 9–10 days, 90–100% cells had differentiated into WAs, with increasing TG content and formation of LDs, which were visualized under light microscopy either directly or following Oil Red O staining. To obtain BAs, 3T3-L1 preadipocytes were differentiated in the presence of 1 µM rosiglitazone for the entire culture period, following the same protocol.

To prepare the conditioned media of mature WAs and BAs, serum-free D-MEM/F-12 medium with 1% BSA was added to 12-well plates containing WAs or BAs and cells were incubated for an additional 24 h at 37°C in 5% CO₂ (24). Subsequently, cells were removed by centrifugation, supernatants collected, aliquoted and immediately stored at −80°C. Control conditioned media (Ctrol-CM) were obtained by incubating empty 12-well culture plates with serum-free D-MEM/F-12 medium with 1% BSA for 24 h at 37°C in 5% CO₂. Cell-free CM were used undiluted in the experiments.

Preadipocyte 3T3-L1 cells were grown on coverslips and differentiated into WAs or BAs following the protocol described above. On day 10, coverslips were washed with phosphate-buffered saline (PBS) and adipocytes were fixed in 4% paraformaldehyde-sucrose (15 min), washed with PBS, permeabilized with 0.5% Triton X-100 in PBS (10 min) and blocked with high ionic strength buffer (20 mM Tris-HCl pH 8.0, 0.63 M NaCl, 1% BSA, 0.05% Tween-20 and 0.02% sodium azide) at room temperature (RT) for 1 h. Cells were incubated overnight (ON) at 4°C with polyclonal rabbit anti-UCP1 (dilution 1:500; cat. no. U6382; Sigma-Aldrich; Merck KGaA) or monoclonal mouse anti-GLUT4 (dilution 1:50; cat. no. sc-53566; Santa Cruz Biotechnology, Inc.), washed with PBS and blocked with 1% BSA in PBS (1 h at RT). Subsequently, coverslips were incubated with FITC (dilution 1:100; cat. no. F0382; Sigma-Aldrich; Merck KGaA) or Alexa Fluor 647 nm (dilution 1:100; Thermo Fisher Scientific, Inc.) conjugated secondary antibody for 1 h at RT. Nuclei were counterstained with Hoechst (cat. no. 33342; Thermo Fisher Scientific, Inc.) for 10 min. For the staining of LDs, coverslips were inverted onto 25 µl of Lipid-TOX (dilution 1:300; HCS Lipid-TOX™ Neutral lipid Stains, Invitrogen; Thermo Fisher Scientific, Inc.) following Hoechst staining, incubated 30 min at RT and mounted with Vectashield. Mitochondria were stained using MitoTracker™ Green FM (cat. no. M7514; Thermo Fisher Scientific, Inc.). Images were acquired with a Spinning Disk-TIRF-Olympus DSU-IX83 microscope (60× objective) and analyzed using ImageJ software (version 1.52p; NIH). A total of 200–300 cells were analyzed per condition and per experiment by two independent observers.

Total RNA was extracted using Tri Reagent (cat. no. TR118; Molecular Research Center) according to the manufacturer's instructions. RNA quantity and quality were assessed spectrophotometrically (NanoDrop 2000; Thermo Fisher Scientific, Inc.). Following reverse transcription with EasyScript Reverse Transcriptase (TransGen Biotech), 1 µl cDNA transcripts were amplified by real-time qPCR in a StepOne Real-Time PCR System in a 25 µl reaction mixture containing 12.5 µl SYBR-Green PCR Master Mix (both from Applied Biosystems; Thermo Fisher Scientific, Inc.) and the appropriate primer concentration. Primers used were: UCP1 forward, 5′-GTACCAAGCTGTGCGATGTC-3′ and reverse, 5′-ACATGATGACGTTCCAGGAC-3′; CIDEA forward, 5′-TCAAACCATGACCGAAGTAGC-3′ and reverse, 5′-TCCAGCACCAGCGTAACC-3′; GLUT4 forward, 5′-TCAACCAGCATCTTCGAGTC-3′ and reverse, 5′-AGATGGAGTGTCCGTCGTC-3′; β-actin forward, 5′-TGGCACCACACTTTCTACAAT-3′ and reverse, 5′-GGTACGACCAGAGGCATACA-3′. The amplification program used was as follows: Initial denaturation step at 95°C for 10 min, 40 denaturation cycles at 95°C for 15 sec and annealing and extension steps at 60°C for 1 min. The expression of selected genes was normalized to that of b-actin gene and the comparative ΔΔCq method (2^−ΔΔCq) (25) was used to calculate relative gene expression.

WB was performed to evaluate protein expression levels in cell lysates and conditioned media. Cells were lysed in lysis buffer (60 mM Tris-HCl, 1% SDS, pH 6.8). Sample protein content was determined by the Bradford method. Proteins (50–80 µg) were resolved in an SDS-PAGE 12% gel, and electrotransferred to polyvinylidene fluoride membranes (PVDF) (Bio-Rad Laboratories, Inc.). After blocking in 1% BSA for 1 h, the membranes were incubated with the different primary antibodies ON at 4°C as follows. Polyclonal rabbit anti-UCP1 (dilution 1:1,000; cat. no. U6382; Sigma-Aldrich; Merck KGaA), polyclonal rabbit anti-leptin (dilution 1:1,000; cat. no. ab117751; Abcam), polyclonal rabbit anti-caveolin 1 (dilution 1:2,000; cat. no. 610060; BD Transduction Laboratories), polyclonal goat anti-vimentin (dilution 1:50; cat. no. V4630; Sigma-Aldrich; Merck KGaA), polyclonal rabbit anti-perilipin 1 (dilution 1:2,000; cat. no. ab3526; Abcam), polyclonal rabbit anti-HSL (dilution 1:1,000; cat. no. ab45422; Abcam), monoclonal rabbit anti-FABP4 (dilution 1:1,000; cat. no. ab219595; Abcam) and monoclonal mouse anti-adiponectin (dilution 1:1,000; cat. no. sc-136131; Santa Cruz Biotechnology, Inc.) antibodies were used to measure protein expression in mature WAs and BAs on day 10 and in WA-CM and BA-CM. Polyclonal rabbit anti-MCT4 (dilution 1:500; cat. no. sc-50329 H90; Santa Cruz Biotechnology, Inc.) antibody was used to measure protein expression in adipocyte lysates. Vimentin, polyclonal rabbit anti-CD44 (dilution 1:2,000; cat. no. ab157107; Abcam), polyclonal rabbit anti-ObR (dilution 1:500; cat. no. sc-8325; Santa Cruz Biotechnology, Inc.), polyclonal mouse anti-MCT1 (dilution 1:500; cat. no. ab90582; Abcam), polyclonal goat anti-MMP-9 (1:500; cat. no. AF909; R&D Systems) and polyclonal rabbit anti-LDH (dilution 1:1,000; cat. no. ab47010; Abcam) antibodies were used to measure protein expression after treatment of non-tumor or tumor epithelial cells with CM. Following incubation in the primary antibody, membranes were washed and incubated with the corresponding HRP-conjugated secondary antibody: anti-rabbit (dilution 1:6,000; cat. no. A0545; Sigma-Aldrich; Merck KGaA), anti-goat (dilution 1:7,500; cat. no. A5420; Sigma-Aldrich; Merck KGaA) or anti-mouse (dilution 1:4,000; cat. no. sc-2005; Santa Cruz Biotechnology, Inc.). Antibody complexes were visualized by chemiluminescence [ECL solution containing luminol (cat. no. 123072; Sigma-Aldrich; Merck KGaA), p-coumaric acid (cat. no. C9008; Sigma-Aldrich; Merck KGaA) and H₂O₂]. Bands were quantified by densitometry using ImageJ software. Actin (dilution 1:1,000; cat. no. sc-1616; Santa Cruz Biotechnology, Inc.) levels were used as loading control.

Oil Red O staining and quantification of TG content were performed as previously described (26) using a 0.2% Oil Red O solution in 40% 2-propanol. After staining, cell images were captured in an inverted phase-contrast microscope (Olympus CKX-41; objective 20×) and the average number of LDs per cell recorded in WAs and BAs using ImageJ software. TG content was quantified by incubating culture plates in 100% 2-propanol for 10 min at RT on an orbital shaker to elute the Oil Red O dye. The eluted dye was transferred to a 96-well plate and the absorbance was measured at 510 nm using a multimodal microplate reader (Synergy HTX S1LFA S/N 18062619; BioTek Instruments). Assay was performed in quadruplicate.

To quantify extracellular lactate released by mature WAs and BAs on day 10, CM were obtained from adipocytes incubated with D-MEM/F-12, 1% BSA for 24 h. Supernatants were recovered, and cell debris was removed by centrifugation. Lactate was measured by a standard method involving the conversion of NAD⁺ to NADH and subsequent spectrophotometric determination of NADH at 340 nm. A commercial kit (Sigma-Aldrich; Merck KGaA) was used to carry out the determination as indicated by the manufacturer. Results are expressed as µg of lactate/100 µl.

The adhesion capacity of cell lines was analyzed following treatment with WA-CM or BA-CM. For this purpose, 1×10⁵ cells/well of non-tumor mammary epithelial cell lines (NMuMG) or tumor cells (LM3, 4T1 and MC4-L1) were seeded in 12-well plates in D-MEM/F-12 supplemented with 10% FBS. Twenty-four hours later, the cells were washed with PBS and incubated for another 24 h with WA-CM, BA-CM or Ctrol-CM. Cells were then harvested and re-seeded in duplicate or triplicate in a 96-well plate (5×10⁴ cells/well). After 1 h, medium containing the unattached cells was aspirated, adhered cells washed with PBS and, 3 h later, adhesion was assessed by the MTS assay measuring absorbance at 490 nm in a multimodal microplate reader (Synergy HTX S1LFA S/N 18062619; BioTek Instruments).

The effect of WA-CM or BA-CM on the motility of tumor and non-tumor mammary epithelial cell lines was evaluated by wound healing assays. NMuMG, LM3, 4T1 and MC4-L1 cells were grown on 96-well plates with supplemented D-MEM/F-12 until cells reached 100% confluence. Afterwards, cell monolayers were scratched with a 200-µl pipette tip, washed twice with PBS and WA-CM, BA-CM or Ctrol-CM were added to the wells. CM were obtained as previously described in a serum-free D-MEM/F-12 medium with 1% BSA and used undiluted. Images at time zero (0 h) were obtained to record the initial width of the wounds, and the recovery of the wounded area was evaluated at 6 and 12 h. Images of the wound closure were captured using an inverted microscope (Olympus CKX-41; objective 4×), quantitative analysis of wounds was performed by ImageJ software and the relative wound closure was calculated with respect to 0 h.

The effect of WAs or BAs on the motility of tumor and non-tumor mammary epithelial cells was analyzed by Transwell migration assay. NMuMG, LM3, 4T1 and MC4-L1 mammary cell lines (1.5×10⁴ cells/well) were placed into Transwells [8-µm pore membrane (top); Jet BIOFIL cat. no. TCS-013-024] containing serum-free D-MEM/F-12 1% BSA medium. The inserts were immediately transferred to 24-well plates with mature WAs or BAs at the bottom of the well. Given that LM3 and 4T1 epithelial cells have a high basal migration rate whereas NMuMG and MC4-L1 cell lines present a very low basal migration rate, Transwell migration assays were performed for 6, 12 and 24 h in all mouse mammary cells studied. After tumor and non-tumor mammary epithelial cells were allowed to migrate across the membrane, the inserts were removed, washed with PBS, and the mammary cells were fixed with cold methanol for 30 min, washed with PBS and stained with Hoechst (cat. no. 33342; Thermo Fisher Scientific, Inc.) 1 µg/ml for 10 min and non-migrating cells were removed with a wet cotton. Air-dried membranes were imaged at ×40 magnification under an inverted microscope (Olympus CK-41) and migrated cells were counted in 10 randomly chosen fields per membrane with ImageJ software.

Experiments were repeated at least twice with similar results. Data are presented as mean ±SEM. Student's t-test, the Mann-Whitney test or one-way analysis of variance (ANOVA) followed by Tukey's multiple comparison test was performed as needed. Differences were considered significant at P<0.05.

Indirect immunofluorescence was performed in 3T3-L1 adipose cells to assess the expression and localization of beige markers. As expected, the presence of 1 µM rosiglitazone (Rosi), a specific agonist of peroxisome proliferator-activated receptor γ (PPARγ), in the differentiation cocktail significantly increased the expression of UCP1 (Fig. 1A, bottom panels). UCP1 showed a punctate staining pattern like the one observed with Mitotracker (Fig. 1A vs. B, bottom panels). Interestingly, BAs displayed round-shaped mitochondria, suggesting sustained fission activity (Fig. 1B, white arrow head in bottom panels), while WAs showed a mix of tubular (white arrow in top panels, Fig. 1B) and partially fragmented mitochondria (white arrow head in top panels Fig. 1B). Fragmented mitochondria with round-shaped morphology were not found in undifferentiated 3T3-L1 preadipocytes treated with rosiglitazone (Fig. 1B, bottom panels, a fibroblast adjacent to an adipocyte occupies the bottom right in first panel and can be seen at higher magnification in third panel), excluding an effect of PPARγ activation on fission events.

There is evidence that brown/beige adipocytes express high levels of GLUT4, and consume glucose at significantly higher levels (20,27). An increase in GLUT4 fluorescence intensity was observed in BAs in comparison to WAs (Fig. 1C, bottom vs. top panels), and BAs showed an increase in the number of micro-LDs (Fig. 1D, +Rosi) but a decrease in TG accumulation compared to WAs (Fig. 1E +Rosi vs. -Rosi).

The increased fluorescence intensity of UCP1 and GLUT4 previously observed in BAs was confirmed by RT-qPCR (Fig. 2A). The expression of Ucp1, Cidea (another classic beige/brown marker), and Glut4 was significantly higher in 3T3-L1 adipose cells after PPARγ activation. Expression levels of UCP1, HSL, perilipin 1, adiponectin, leptin, FABP4, caveolin 1, vimentin and the metabolic markers MCT4 and LDH were evaluated in WAs and BAs (Fig. 2B). Perilipin 1, the main protein coat for LDs in mature adipocytes, and HSL are two proteins involved in the lipolytic process.

Perilipin 1 expression was slightly decreased in the BAs compared to the WAs; HSL levels were not affected by rosiglitazone treatment. Adiponectin and leptin are the main adipokines released by adipocytes; worthy of note, there was a higher leptin expression but significantly lower adiponectin expression in BAs when compared to the WAs (Fig. 2B). Caveolin-1 is an important protein, necessary for LD formation (28), and a marked decrease in caveolin 1 expression was observed in the BAs (Fig. 2B), in agreement with a decrease in lipid accumulation (Fig. 1E). The expression of FABP4, actively secreted and essential to modulate lipid fluxes, trafficking, signaling and metabolism in adipocytes, was increased in BAs, while a decrease in vimentin expression was observed in BAs when compared to WAs (Fig. 2B). The expression of MCT4, a mediator of lactate export, was increased in BAs relative to WAs (Fig. 2B); expression levels of LDH, enzyme catalyzing the conversion of pyruvate to lactate, were similar to those of MCT4 (data not shown).

To characterize soluble factors present in WA-CM and BA-CM, the expression levels of different proteins were determined by WB. According to Fig. 3A and B, the results showed significantly increased expression of leptin in BA-CM compared to WA-CM, while adiponectin and caveolin 1 expression was significantly decreased. Vimentin was present in both WA- and BA-CM, but the expression levels were not significantly different. Increased expression of FABP4 and UCP1 was found in BA-CM in comparison to WA-CM. Worthy of note, changes in expression levels of leptin, adiponectin, caveolin 1, UCP1 and FABP4 in CM (Fig. 3A and B) agree with those observed in the cellular extracts (Fig. 2B). Expression levels of HSL (Fig. 3A) and perilipin 1 (data not shown) were insignificant in both CM.

We investigated whether the upregulation of MCT4 and LDH upon differentiation into BAs translated into an increased lactate flux. Lactate efflux from WAs and BAs, evaluated in WA- and BA-CM using a colorimetric assay, showed that lactate transporter capacity was increased in 3T3-L1 differentiated into a beige phenotype (Fig. 3C).

To compare the presence of soluble factors in WA-CM and BA-CM, that could affect tumor cell adhesion, NMuMG, LM3, 4T1 and MC4-L1 epithelial cells were exposed to different CM (WA-, BA- or Ctrol-CM) and cell adhesion was determined as described in Materials and methods. WA-CM and BA-CM significantly decreased adhesion of LM3, 4T1 and MC4-L1 cells with respect to Ctrol-CM (Fig. 4). On the other hand, NMuMG cell adhesion was decreased significantly after incubation with BA-CM vs. both WA-CM and Ctrol-CM.

To evaluate the effect of WA-CM and BA-CM on the migratory capacity of tumor and non-tumor mammary epithelial cells, we performed wound healing assays. WA-CM and BA-CM significantly increased migration of LM3 and 4T1 (tumor) cells after incubating for 6 h, and of MC4-L1 (tumor) and NMuMG (non-tumor) after incubating for 12 h vs. the Ctrol-CM (Fig. 5). These results suggest that BA-CM increase the migration capacity of both tumor and non-tumor mammary epithelial cells in agreement with results found in the cell adhesion assays (Fig. 4).

To compare the chemotactic capacity of BAs and WAs on tumor and non-tumor epithelial cell migration, Transwell migration assays were performed as described in Materials and methods. BAs significantly increased cell migration of LM3 and MC4-L1 cells vs. the WAs, following incubation for 6 and 24 h, respectively (Fig. 6). BAs also increased 4T1 cell migration after incubation for 6 h though to a lesser extent, but there was no significant chemotactic effect of BAs or WAs on NMuMG migration at any time tested.

Taken together, these results suggest that crosstalk takes place between tumor cells and BAs and that this cell-cell communication may be necessary to modulate adipocyte action on epithelial cell migration, beyond the effect of soluble factors.

The effect of soluble factors released by WAs and BAs on the expression level of several tumor progression markers, was evaluated by WB in tumor and non-tumor epithelial cells incubated with WA-CM, BA-CM or Ctrol-CM for 24 h. ObR and MMP-9 expression in MC4-L1 cells was significantly higher when incubated with BA-CM compared to WA-CM or Ctrol-CM (Fig. 7, MC4-L1). CD44 expression showed an increasing trend in MC4-L1 cells incubated with BA-CM (Fig. 7, MC4-L1), but differences were not significant. By contrast, MMP-9 expression was significantly lower when NMuMG (non-tumor) cells were treated with BA-CM and WA-CM vs. Ctrol-CM and, interestingly, CD44 and ObR expression levels were significantly decreased in LM3 and 4T1, respectively, after treatment with WA-CM. Vimentin expression level was not affected by different CM treatments and, surprisingly, neither were MCT1 and LDH expression levels (Fig. 7).