Biomedicine & Pharmacotherapy 141 (2021) 111871 Contents lists available at ScienceDirect Biomedicine & Pharmacotherapy journal homepage: www.elsevier.com/locate/biopha Selective estrogen receptor modulators (SERMs) affect cholesterol homeostasis through the master regulators SREBP and LXR María E. Fernández-Suárez a, b, *, 1, Lidia Daimiel a, b, 2, Gemma Villa-Turégano a, María Vázquez Pavón a, 3, Rebeca Busto a, b, Joan C. Escolà-Gil c, d, e, Frances M. Platt f, Miguel A. Lasunción a, b, Javier Martínez-Botas a, b, Diego Gómez-Coronado a, b, * a Servicio de Bioquímica-Investigación, Hospital Universitario Ramón y Cajal, IRYCIS, Ctra. de Colmenar, km 9, 28034 Madrid, Spain b CIBER de Fisiopatología de la Obesidad y Nutrición (CIBEROBN), Instituto de Salud Carlos III, Spain c Institut de Recerca de l′ Hospital de la Santa Creu i Sant Pau, Institut d′ Investigacions Biomèdiques (IIB) Sant Pau, Barcelona, Spain d CIBER de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM), Instituto de Salud Carlos III, Spain e Departament de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, Spain f Department of Pharmacology, University of Oxford, Mansfield Road, Oxford OX1 3QT, United Kingdom A R T I C L E I N F O A B S T R A C T Keywords: Selective estrogen receptor modulators (SERMs) are nonsteroidal drugs that display an estrogen-agonist or es­ Tamoxifen trogen-antagonist effect depending on the tissue targeted. SERMs have attracted great clinical interest for the Raloxifene treatment of several pathologies, most notably breast cancer and osteoporosis. There is strong evidence that Toremifene SERMs secondarily affect cholesterol metabolism, although the mechanism has not been fully elucidated. In this SREBP study, we analysed the effect of the SERMs tamoxifen, raloxifene, and toremifene on the expression of lipid LXR Cholesterol metabolism genes by microarrays and quantitative PCR in different cell types, and ascertained the main mechanisms involved. The three SERMs increased the expression of sterol regulatory element-binding protein (SREBP) target genes, especially those targeted by SREBP-2. In consonance, SERMs increased SREBP-2 pro­ cessing. These effects were associated to the interference with intracellular LDL-derived cholesterol trafficking. When the cells were exposed to LDL, but not to cholesterol/methyl-cyclodextrin complexes, the SERM-induced increases in gene expression were synergistic with those induced by lovastatin. Furthermore, the SERMs reduced the stimulation of the transcriptional activity of the liver X receptor (LXR) by exogenous cholesterol. However, their impact on the expression of the LXR canonical target ABCA1 in the presence of LDL was cell-type depen­ dent. These actions of SERMs were independent of estrogen receptors. We conclude that, by inhibiting the intracellular trafficking of LDL-derived cholesterol, SERMs promote the activation of SREBP-2 and prevent the activation of LXR, two master regulators of cellular cholesterol metabolism. This study highlights the impact of SERMs on lipid homeostasis regulation beyond their actions as estrogen receptor modulators. 1. Introduction by two families of transcription factors, the sterol regulatory element- binding proteins (SREBP) and the liver X receptors (LXR), which In mammalian cells, cholesterol levels are tightly regulated, mainly govern the supply and removal, respectively, of cholesterol. The SREBP Abbreviations: BSA, bovine serum albumin; CholMCD, cholesterol-MCD complex; DHE, dehydroergosterol; DMSO, dimethyl sulfoxide; ER, estrogen receptor; E2, 17β-estradiol; FBS, fetal bovine serum; INSIG, insulin-induced gene proteins; LDL, low-density lipoprotein; LDLR, LDL receptor; LOV, lovastatin; LPDS, lipoprotein deficient serum; LXR, liver X receptor; LXRE, LXR response element; MCD, methyl-β-cyclodextrin; PBS, phosphate-buffered saline; PMA, phorbol 12-myristate 13- acetate; RAL, raloxifene; SERM, selective estrogen receptor modulator; RXR, retinoid X receptor; SCAP, SREBP cleavage-activating protein; SREBP, sterol regulatory element-binding protein; TAM, tamoxifen; TOR, toremifene. * Corresponding authors at: Servicio de Bioquímica-Investigación, Hospital Universitario Ramón y Cajal, IRYCIS, Ctra. de Colmenar, km 9, 28034 Madrid, Spain. E-mail addresses: maria.fernandez-suarez@pharm.ox.ac.uk (M.E. Fernández-Suárez), diego.gomez@hrc.es (D. Gómez-Coronado). 1 Present address: Department of Pharmacology, University of Oxford, Mansfield Road, Oxford OX1 3QT, United Kingdom. 2 Present address: IMDEA Food, Ctra. Cantoblanco, 8, 28049 Madrid, Spain. 3 Present address: Department of Physiology, National University of Singapore, Singapore 117597, Singapore. https://doi.org/10.1016/j.biopha.2021.111871 Received 26 March 2021; Received in revised form 10 June 2021; Accepted 28 June 2021 Available online 3 July 2021 0753-3322/© 2021 The Author(s). Published by Elsevier Masson SAS. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 family of basic-helix-loop helix transcription factors is composed of drugs [19,21]. However, the molecular mechanisms underlying the three members, SREBP-1a and SREBP-1c, encoded by the same gene, alteration of cellular cholesterol metabolism by SERMs are not fully SREBF1, but driven by different promoters and alternative splicing, and established. We hypothesized that the SERM-induced impairment of SREBP-2, encoded by a different gene, SREBF2 [1]. SREBPs reside in the intracellular cholesterol trafficking leads to reciprocal alterations in the membrane of the endoplasmic reticulum, bound to SREBP SREBP and LXR pathways. In the present work, our aim was to inves­ cleavage-activating protein (SCAP). SCAP contains a sterol sensing tigate the effect of SERMs and their combinations with LOV on the domain that upon sterol biding adopts a conformation that allows it to cholesterol-mediated regulation of the SREBP and LXR pathways. interact with insulin-induced gene proteins (INSIG). Under these con­ ditions, the whole complex is retained in the endoplasmic reticulum. 2. Materials and methods When cholesterol levels descend, lack of sterol binding to the sterol sensing domain produces a conformational change in SCAP, leading to 2.1. Materials the disruption of the interaction with INSIG. As a result, the SREBP-SCAP complex is released from the endoplasmic reticulum and Tamoxifen citrate and raloxifene hydrochloride were obtained from transported to the Golgi, where two proteases, S1P and S2P, sequentially Tocris Bioscience, and toremifene citrate was purchased from Sigma. cleave SREBPs yielding the amino-terminal fragment, which is the active Lovastatin and dimethyl sulfoxide (DMSO) were obtained from Merck transcription factor [2]. In the nucleus, active SREBPs bind to sterol Sharp & Dohme (MSD), 17β-estradiol (E2) and phorbol 12-myristate 13- response elements to promote the transcription of their target genes. acetate (PMA) from Sigma and T0901317 from Tocris Bioscience. SREBP-1c increases the expression of genes involved in fatty acid and Methyl-β-cyclodextrin was obtained from Trappsol (CTD, Inc) and 2- triglyceride synthesis, SREBP-2 stimulates the expression of genes hydroxypropyl-β-cyclodextrin from Fluka. Other chemicals used were of required for cholesterol biosynthesis and uptake, while SREBP-1a in­ analytical reagent grade quality. creases the expression of both cholesterogenic and lipogenic genes [1]. RPMI 1640 culture medium, DMEM (1 mg/ml glucose) with or The LXR family includes two members, LXRα and LXRβ, encoded by without phenol red were obtained from PAA Laboratories. RPMI 1640 different genes. They belong to the nuclear hormone receptor super­ without phenol red and fetal bovine serum (FBS) were obtained from family of ligand-activated transcription factors and form obligate het­ Gibco. Dextran/charcoal treated FBS, with reduced levels of estrogens, erodimers with the retinoid X receptors (RXRs) [3]. Natural ligands of was from HyClone. Antibiotics for supplementing the medium and non- LXR are some cholesterol-derived oxysterols, such as 22(R)-hydrox­ essential amino acids were obtained from Gibco, glutamine was from ycholesterol and 27-hydroxycholesterol, and 24(S),25-epox­ Sero-Med and sodium pyruvate from Sigma. Trypsin was obtained from ycholesterol, synthesized in a shunt pathway of cholesterol biosynthesis Sigma and phosphate-buffered saline (PBS) from Biomérieux. [4]. Upon cellular cholesterol overload, cholesterol-derived oxysterol levels increase in a coordinated fashion. These oxysterols bind to and 2.2. Low density lipoprotein (LDL) and lipoprotein deficient serum activate LXR, promoting its interaction with LXR response elements (LPDS) preparation (LXREs) in the regulatory regions of LXR target genes [5]. This results in the transcription of genes necessary for reverse cholesterol transport, Human LDL (1.019–1.063 kg/l) was isolated by vertical rotor density including cholesterol efflux from cells of peripheral tissues through gradient ultracentrifugation of plasma from hypercholesterolemic pa­ ATP-binding cassette (ABC) A1 and ABCG1, and cholesterol excretion tients, as previously described [22]. AcLDL was prepared from LDL by from the liver through ABCG5 and ABCG8 [5]. The coordinated, the addition of acetic anhydride [22]. LPDS was prepared from FBS or tissue-specific actions of the LXR pathway maintain not only systemic charcoal/dextran-treated FBS by ultracentrifugation at a density of 1.21 cholesterol homeostasis, but also regulate immune and inflammatory kg/l. responses [6]. Selective estrogen receptor modulators (SERMs) are a group of 2.3. Preparation of cholesterol-methyl-β-cyclodextrin complexes diverse, non-steroidal molecules characterised by their ability to bind (CholMCD) the estrogen receptors (ER), ERα and ERβ [7]. They act as ER agonists or antagonists depending on the cell type and tissue targeted, although The complex was prepared by the method previously described with estrogen receptor-independent effects have also been described [8–10]. minor modifications [23,24]. In brief, a solution of 14.7 mg/ml The SERMs tamoxifen (TAM) and toremifene (TOR), triphenylethylene cholesterol in 100% ethanol was added in small aliquots to a stirred derivatives differing only by the presence of a chlorine atom in the ethyl solution of 5% (w/v) methyl-β-cyclodextrin (MCD) on a heating block chain of TOR, are widely used to treat breast cancer. Raloxifene (RAL), (60 ◦ C) until a clear solution was achieved. The solution was lyophilized, another SERM, is a benzothiophene derivative indicated in Europe for and the dried complex was reconstituted in water to a sterol concen­ the treatment of osteoporosis in postmenopausal women [11,12]. tration of 3 mM. The sterol to cyclodextrin molar ratio was 1:10. We have previously demonstrated that the SERMs TAM, RAL, and TOR increase the expression and activity of the low-density lipoprotein 2.4. Cells and experimental design (LDL) receptor (LDLR) [13,14]. This effect is synergistic with that of lovastatin (LOV), a competitive inhibitor of the rate-limiting choles­ MOLT-4 (ATCC CRL-1582) and THP-1 cells (ATCC TIB-202) were terogenic enzyme 3-hydroxy-3-methylglutaryl coenzyme A reductase, in maintained in RPMI 1640 supplemented with 10% FBS, 2 mM gluta­ lymphoblastic MOLT-4 and hepatocytic HepG2 cell lines, but not in mine, 100 U/ml penicillin, 100 U/ml streptomycin and 10 μg/ml human primary lymphocytes [13,14]. Furthermore, we and others have gentamicin at 37 ◦ C in humidified atmosphere with 5% CO2. THP-1 shown that SERMs reduce cholesterol biosynthesis by inhibiting the monocytes were differentiated to macrophages with 50 ng/ml PMA activity of post-lanosterol enzymes [13,15,16]. These effects may ac­ for 48 h. HepG2 (ATCC HB-8065) and MCF-7 (ATCC HTB-22) cells were count for the low levels of circulating cholesterol described in patients maintained in DMEM supplemented with 10% FBS, 0.1 mM non- taking these drugs [15,17–19]. We have also found that SERMs block the essential amino acids, 1 mM sodium pyruvate, and the above­ egress of LDL-derived cholesterol from lysosomes, thus explaining the mentioned additives. synergistic effect with LOV on LDLR expression in cell lines [13,14]. For experiments, cells were pre-treated with the corresponding me­ Moreover, these drugs downregulate the expression of ABCA1 and dium supplemented with 10% LPDS for 24 h. Subsequently, TAM, RAL, ABCG1 [14,20], resulting in reduced cholesterol efflux from macro­ or TOR (10 µM final concentration), LOV (1 µM), T0901317 (1 µM), the phages [20]. The effects of SERMs described above are independent of indicated amounts of E2 or vehicle (DMSO, final concentration ERs [14,20] and may contribute to the anti-tumoral efficacy of these ≤0.044%, v/v) were added to the media as indicated, and 1 h later LDL 2 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 (120 μg/ml cholesterol) or the indicated amounts of CholMCD were were washed with Microarray Wash Buffers A, B, and C (ArrayIt). added to the treatments. 16 h after the addition of the drugs, E2 or vehicle, the cells were washed and processed to perform the corre­ 2.5.2. Microarray analysis sponding analysis. For ER-related experiments, DMEM without phenol Microarrays were scanned with the ScanArray Express (PerkinElmer) red and supplemented with 10% charcoal/dextran-treated LPDS was using the ScanArray Express 3.0 software. Background-corrected data used. were normalized by the locally weighted scatterplot smoothing (LOW­ ESS) method, based on the intensity of the Lucidea Universal ScoreCard 2.5. DNA microarrays control probes, and using an R script specifically developed for the Cholestchip. Normalised data were processed to eliminate the technical The focused cDNA expression microarray (Cholestchip™) developed controls, data with intensity below the scanner’s detection confidence by our group was used in these experiments [25]. This microarray, that limit, and inconsistent replicates. The arithmetic mean of the Cy5/Cy3 contained cDNA probes for 319 genes involved in lipid metabolism and coefficients for each probe was Log2-transformed (M value). M ≥ 0.85 cell cycle, was produced using a SpotArray 72 spotter (Perkin-Elmer) and M ≤ − 0.85 were considered significant changes. Next, data were with TeleChem Stealth SMP3 split pins. The full-length cDNA probes subjected to cluster analysis (hierarchical cluster and k-means), using were selected from the I.M.A.G.E. Consortium database and obtained Euclidean distances, with the MultiExperiment Viewer software (Mev from Open Biosystems. cDNAs were amplified using universal primers v3.1) [26]. Finally, we represented the changes in the cholesterol (M13 (− 21) universal forward 5′ -TGTAAAACGACGGCCAGT-3′ and biosynthesis pathway using the PathwayExplorer software [27]. M13/pUC reverse primer 5′ -CAGGAAACAGCTATGACC-3′ ). Quality of the PCR products was routinely checked on agarose gel and by 2.6. Real-time PCR assay sequencing prior to spotting. Each PCR sample was resuspended in Micro Spoting Solution (ArrayIt) and spotted three times in two different 500 ng of total RNA, extracted with TRI Reagent (Sigma), were locations on aminoxylan-treated glass slides (CapitalBio OPAmi­ reverse transcribed with random hexamers using the PrimeScript RT noSlide). Lucidea Universal ScoreCard probes (GE Healthcare Life Sci­ reagent kit (Takara). Real-time PCR amplification was performed on a ences) were spotted three times at the beginning and at the end of each LightCycler 480 II using the SYBR Green I Master kit (Roche). The subarray and used to validate the quality of the experiments. The thermocycler protocol consisted of a preincubation step at 95 ◦ C for 5 microarray contains a total of 2800 spots distributed in four subarrays. min, followed by 45 cycles at 95 ◦ C for 10 s, 60 ◦ C for 10 s, and 72 ◦ C for DNA was fixed to the slides by irradiation of 300 mJ of UV light (GS 10 s. To prove that only one PCR product was amplified, melting curves Gene Linker, BioRad) and subsequent incubation at 80 ◦ C for 2 h. After were obtained with a temperature ramp of 65 ◦ C to 97 ◦ C with changes each series of prints, SyberGold staining and hybridisations with sense of 0.11 ◦ C/s. Relative expression of each gene was determined by and antisense Cy5-labeled M13 were performed to analyse the quality of calculating the increase in CP (ΔCP), as previously described [28], the print. corrected for the efficiency of the assay for each pair of primers, and normalized by the housekeeping control RPLP0 (large ribosomal protein 2.5.1. DNA Labeling and microarray hybridisation gene P0). The sequence of primers used in this study can be found in Total RNA was extracted using Tri Reagent (Sigma) according to the Supplementary Table 1. manufacturer’s protocol. The mRNA was isolated from 100 µg of total RNA using oligo(dT) cellulose columns from the GenElute™ mRNA 2.7. LXRα luciferase reporter assay MiniPrep Purification Kit (Sigma) and following the manufacturer’s instructions. The mRNA was precipitated with 30:1 (v/v) 100% ethanol: Cells were transduced with a lentivirus (QIAGEN, Cignal Lenti Re­ 7.5 M sodium acetate, and glycogen at − 20◦ C for 16–20 h. After pre­ porter Assay, CLS-7041L) to stably express the LXRα response elements cipitation, the mRNA was washed with 75% ethanol and resuspended in upstream of a luciferase reporter gene, following the manufacturer’s DEPC water. mRNA integrity was verified on agarose gel. instructions. Transduced cells were selected performing a puromycin kill 1 µg of mRNA was labeled with Cy3-dUTP (control) or Cy5-dUTP curve. As an internal control for normalization, cells were transduced (test) (GE Healthcare Life Sciences) in a reverse transcription reaction with a constitutively expressing Renilla luciferase construct (QIAGEN, with 0.5 µg/µl oligo(dT) and 0.5 μg/µl random primers (Promega), 10 Cignal Lenti Reporter Assay, CLS-RHL). For the experiments, cells were mM dithiothreitol, 0.75 µl of dNTPs (10 mM dATP, 10 mM dCTP, 10 mM treated as described above and lysed for dual luciferase reporter assay dGTP and 2 mM dUTP; Applied Biosystems), 30 U RNAsin (Promega) (Promega), following the manufacturer’s instructions. Luminescence and 14 U of SuperScript II enzyme reverse transcriptase (Invitrogen) for intensity was measured using a Sirius dual-injector luminometer (Bert­ 2 h at 42 ◦ C. Lucidea™ Spike Mix Controls (Amersham-GE Healthcare hold). LXRα activity was calculated as firefly/Renilla luciferase activity Life Sciences) were added to the control and test samples prior to the ratios after subtracting the background (non-transduced cells). labeling reaction. Subsequently, the mRNA was degraded with 0.08 N NaOH (Merck) and 40 mM EDTA (Sigma) for 30 min at 65 ◦ C. The 2.8. Western blot analysis degradation was stopped with 0.17 M Tris-HCl, pH 7.5. Finally, the control and test samples were combined and purified using CyScrybeTM Cells were lysed in 20 mM Tris-HCl buffer, pH 8.0, containing 120 GFXTM columns (Amersham-GE Healthcare Life Sciences) following the mM KCl, 1 mM dithiothreitol, 1 mM Na2-EDTA, 2 mM EGTA, 0.1% manufacturer’s protocol. Samples were added with 0.4 µg/µl of poly(dA) Triton X-100, 0.5% Nonidet P-40, 1 mM benzamidine, 10 µg/ml anti­ (Sigma) and 0.08 µg/µl of Cot-1 DNA (Invitrogen), lyophilized, resus­ pain, 1 µg/ml leupeptin, 40 µg/ml aprotinin, 100 mM NaF, 20 mM so­ pended in 20 µl of hybridization buffer (25% formamide, 2.98% SSC, dium molybdate, 20 mM β-glycerophosphate, 2 mM sodium 0.20% SDS, and Denhardt’s 5X solution in water), and incubated at orthovanadate, 1 mM PMSF, and 1 μg/ml Caspase-3 inhibitor Ac- 55 ◦ C for 3 min. DMQD-CHO, for 20 min under stirring and then sonicated at 4 ◦ C. Microarray slides were blocked for 15 min at room temperature with Samples were passed 10 times through a 25-gauge needle and centri­ 1% bovine serum albumin (BSA) (Sigma) preheated to 50 ◦ C. Then, to fuged at 10,000 x g for 10 min. Protein concentration was measured in denature the probes, the slides were incubated in boiling water for 5 min the collected supernatant by the bicinchoninic acid method (BCA Pro­ and fixed for 2 min with 95% ethanol at 4 ◦ C. Labelled samples were tein Assay Kit, Pierce). Equal amounts of protein were subjected to SDS- hybridized on the microarrays using LifterSlip coverslips (Erie Scientific PAGE, transferred to a nitrocellulose membrane and probed with anti­ Company). Hybridization was carried out in a BioMixer II hybridizer bodies directed against SREBP-2 (ab30682, Abcam). Bound antibodies (CapitalBio) at 55 ◦ C for 16–20 h. After hybridization, the microarrays were visualized using secondary antibodies conjugated to IRDye 800CW 3 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 (LI-COR). Fluorescence was detected using an Odyssey 9120 scanner (LI- cells were fixed with 4% paraformaldehyde for 5 min and exposed to COR Biosciences). filipin (35 µg/ml in PBS) for 1 h. Coverslips were mounted on slides using ProLong Gold antifade reagent. Cells were examined on an Olympus BX51 fluorescence microscope. 2.9. Microscopy MOLT-4 cells were treated or not with the different SERMs. 2 h 2.10. Statistical analysis before finishing the treatments, 7 × 104 cells/cm2 were seeded on glass coverslips previously treated with poly-D-lysine and the incubation To evaluate the effects of SERMs and their interaction with other continued at 37 ◦ C to allow cell adhesion. For free cholesterol staining, drugs, the two-way repeated measures ANOVA analysis was used. Post Fig. 1. Microarray analysis of gene expression patterns in MOLT-4 cells treated with SERMs or their combinations with LOV. Cells were treated with 10 μM SERMs, 1 μM LOV or their combinations in the presence of 120 μg/ml LDL-cholesterol for 16 h. (a) Hierarchical tree cluster with Euclidean Distances. Only genes whose expression changed in at least one condition in the two independent experiments were included in the analysis. The heatmap contains representative data, and each column belongs to a treatment and each row to a gene. Red indicates overexpression, green repression and black no change. (b) Effect of TAM and LOV and their combination on the expression of genes of the cholesterol biosynthesis pathway in MOLT-4 cells. Each box represents one of the enzymes of the pathway. Within each box, the result on the left corresponds to LOV, on the centre to TAM and on the right to the combination of both. One experiment of two independent experiments is represented. The increases in intensity of red and green correspond to increase and decrease of expression, respectively. 4 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 hoc multiple comparisons were performed by Student-Newman-Keuls should be noted that we did not find differential effects between the test. The analyses were performed using the SigmaStat 2.0 software. p SERMs amongst the genes analysed. ≤ 0.05 was defined as statistically significant. We found 19 genes overexpressed in all the conditions (Table 1), 16 of which were targets of the transcription factors SREBPs. Among them, 3. Results the most represented were genes encoding enzymes of the cholesterol biosynthesis pathway, with 12 genes being upregulated. The effects of 3.1. Tamoxifen, raloxifene, and toremifene increase the expression of TAM on genes of the cholesterol biosynthesis pathway are represented in SREBP target genes Fig. 1b. The effects of RAL and TOR on the cholesterol biosynthesis pathway are represented in Supplementary Figs. 1 and 2, respectively. To determine which genes involved in lipid metabolism changed These results suggest that SREBPs, particularly SREBP-2, which is the their expression in response to treatment with SERMs, and to ascertain primary activator of cholesterol synthesis and LDL uptake genes, whether LOV could potentiate the effect of SERMs, we used the DNA mediate many of the effects of SERMs on lipid metabolism genes. To focused microarray Cholestchip™. The lymphoblastic cell line MOLT-4 confirm these results, we analysed, by real-time PCR, the expression of a was incubated in medium supplemented with LPDS for 24 h and set of representative SREBP-targeted genes: LDLR, HMGCR, SCD, FDFT1 treated with TAM, RAL, or TOR combined or not with LOV and in the and INSIG1. Real-time PCR results correlated well with those of the presence of LDL. After 16 h of treatment, gene expression was analysed. microarray. SERMs increased the expression of these genes (Fig. 2a-e). Of the 317 genes represented in the microarray, 119 genes (37.5%) were The treatment with LOV alone did not modify gene expression, but when expressed under all the conditions. Of them, 56 (47%) changed signifi­ LOV was combined with any of the SERMs, the effect surpassed that of cantly in at least one condition, with a cut-off M-value > 0.85 or < the SERM alone. Moreover, we analysed a set of non-SREBP target genes − 0.85 (Fig. 1a). We performed a cluster analysis using the KMC method, (DGAT2, encoding diacylglycerol O-acyltransferase 2; ACADM, encod­ with calculated K-means and Euclidean distances, and found four sets of ing medium-chain specific acyl-coenzyme A dehydrogenase; and genes with a similar response. The first cluster consisted of genes that HADHA, encoding hydroxyacyl-coenzyme A dehydrogenase/3-ketoacyl- increased their expression in response to SERMs and had a synergistic coenzyme A thiolase/enoyl-coenzyme A hydratase [trifunctional pro­ effect with LOV. The second cluster was similar to the previous one but tein], alpha subunit) whose expression was detectable but did not with an unclear enhanced expression when combined with LOV. The change according to the microarray analysis. As shown in Fig. 2f-h, the data of these two groups are represented in Table 1. The third cluster quantitative expression of these genes was not significantly altered by consisted of genes repressed upon SERMs treatment (Table 2). Finally, the different treatments. the fourth cluster contained genes that showed erratic behaviour. It We wanted to know if the effects of the SERMs observed in MOLT-4 Table 1 Genes that increased their expression by microarray analysis following treatment with SERMs or their combinations with LOV in MOLT-4 cells. Data are M values, M=log2 (fluorescence treated sample / fluorescence control sample), from two independent experiments. SREBP target genes are in bold letters. Gene LOV TAM TAM+LOV RAL RAL+LOV TOR TOR+LOV Name LSS 1.68 2.27 3.32 2.33 2.68 1.69 2.19 Lanosterol synthase 1.51 1.68 3.46 1.69 1.84 1.49 1.55 ACAT2 1.43 1.45 2.53 1.10 2.87 1.13 3.94 Acetyl-coenzyme A acetyltransferase 2 1.43 1.33 1.92 1.29 2.53 1.57 1.99 MVK 1.45 1.88 2.68 1.40 3.51 1.27 2.68 Mevalonate kinase 1.47 1.25 1.48 1.16 1.73 1.52 1.40 SC5DL 1.37 2.08 3.46 1.36 2.22 1.37 2.43 Lanosterol 5-desaturase 1.36 1.30 1.99 1.35 1.44 1.41 1.46 SREBF2 1.33 2.46 3.53 1.92 2.33 1.92 3.07 Sterol regulatory element binding transcription factor 2 1.16 1.57 1.73 1.35 1.53 1.60 1.47 CYP51 1.54 2.57 4.17 1.66 3.14 1.48 3.58 lanosterol 14-alpha demethylase 1.51 1.68 2.16 1.60 2.30 1.43 1.64 SR-1 1.46 1.92 2.81 2.00 2.95 1.33 3.63 C14 sterol reductase 1.62 1.57 2.03 1.41 2.19 1.32 1.36 VLDLR 1.57 1.80 1.59 1.74 1.78 1.45 2.50 Very low density lipoprotein receptor 1.55 1.62 1.72 1.29 1.48 1.58 1.28 UCP2 1.46 1.78 1.85 1.37 1.83 1.29 2.11 Uncoupling protein 2 1.37 1.17 1.45 1.09 1.39 1.39 1.32 LDLR 1.44 1.47 1.85 1.11 1.38 1.22 1.88 Low density lipoprotein receptor 1.38 1.41 1.89 1.23 1.29 1.23 1.12 HMGCR 1.44 2.33 4.44 1.52 4.38 1.47 5.28 3-hydroxy-3-methylglutaryl-coenzyme A reductase 1.39 1.77 2.73 1.55 3.05 1.42 1.72 SCD 1.72 2.81 4.29 2.07 4.35 2.04 5.10 stearoyl-coenzyme A desaturase 1.61 1.88 2.31 2.04 3.03 1.87 2.08 SLC2A1 1.35 2.66 5.46 1.80 4.20 1.83 5.94 Solute carrier family 2 1.56 1.77 2.71 1.73 3.12 1.97 2.30 DHCR24 1.38 2.83 5.35 2.20 4.50 2.17 6.19 24-dehydrocholesterol reductase 1.31 1.88 2.68 1.88 3.14 1.71 1.91 ERG25 1.46 2.79 5.10 2.14 5.98 1.58 6.73 Sterol C4 demethylase 1.69 1.68 2.77 1.87 3.39 1.59 2.03 DHCR7 1.43 2.99 5.98 2.14 5.50 1.88 6.15 7-dehydrocholesterol reductase 1.54 1.80 2.48 2.16 3.51 1.53 1.96 MVD 1.56 2.85 5.24 2.22 6.96 1.91 5.74 Mevalonate (diphospho) decarboxylase 1.52 1.74 2.57 2.01 3.51 1.40 1.59 FDFT1 1.38 2.91 5.31 2.13 4.44 2.28 5.70 Farnesyl-diphosphate farnesyltransferase 1 1.48 2.04 3.05 2.36 3.53 2.43 3.41 INSIG1 1.71 5.58 12.64 3.97 10.70 4.35 12.38 Insulin induced gene 1 1.78 2.87 4.23 3.58 5.28 2.41 3.46 5 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 Table 2 Genes that decreased their expression by microarray analysis following treatment with SERMs or their combinations with LOV in MOLT-4 cells. Data are M values, M=log2 (fluorescence treated sample / fluorescence control sample), from two independent experiments. Gene LOV TAM TAM+LOV RAL RAL+LOV TOR TOR+LOV Name PRKAG2 1.01 1.11 0.97 0.70 0.56 0.67 0.88 Protein kinase, AMP-activated, gamma 2 non-catalytic subunit 1.06 0.65 0.75 0.53 0.57 0.93 0.82 PPARGC1 1.20 0.80 0.61 0.51 0.37 0.59 0.66 Peroxisome proliferative activated receptor, gamma, coactivator 1, alpha 1.09 0.61 0.95 0.54 0.52 1.03 0.94 CDC2 1.21 0.78 0.68 0.51 0.55 0.65 0.77 Cell division cycle 2, G1 to S and G2 to M 1.04 0.67 0.73 0.61 0.64 0.89 0.78 CDK2 0.03 -0.36 -0.46 -1.07 -1.1 -0.53 -0.16 Cyclin-dependent kinase 2 0.18 -0.48 0.26 -0.68 -0.46 0.12 -0.18 cells were exclusive of this lymphoid cell type, or if they could be with the SREBP-mediated feed-back effect of cholesterol when cells are reproduced in a cell type important in lipid metabolism, such as hepa­ supplied with LDL-cholesterol, but not when supplied with non- tocytes. For this, we used the HepG2 human hepatocyte cell line. The lipoprotein cholesterol. cells were treated following the same experimental design as with The expression of the LXR target gene ABCA1 did not change upon MOLT-4 cells. Results of microarrays assays in HepG2 cells paralleled LDL addition in comparison to LPDS alone (Fig. 3d). The addition of the those obtained in MOLT-4 cells. From the genes expressed in all the different SERMs caused intense repression of ABCA1 expression relative conditions (Supplementary Fig. 3a), there was an enrichment of SREBPs to its control condition without SERM, independently of the combina­ targets amongst the stimulated genes (Supplementary Table 2), the tion with LOV. Contrary to the lack of response of the SREBP target majority of them belonging to the cholesterol biosynthesis pathway. The genes, in the presence of CholMCD, SERMs repressed the expression of effects of TAM, RAL, and TOR on this pathway are individually repre­ ABCA1, although to a lesser extent than with LDL (Fig. 3d). sented in Supplementary Figs. 3b, 4 and 5, respectively. Curiously, one To determine if these effects were exclusive to MOLT-4 cells we of the most upregulated genes was FABP1, a fatty acid transporter that, reproduced the previous experiment in HepG2 cells. In these cells, we although it has not been described as an SREBP target gene, had a studied the SCD gene, also a target for SREBP, instead of INSIG1, which similar behaviour to that of SREBP known targets. Repressed genes are was expressed at very low levels. The effects of SERMs on SREBP targets summarised in Supplementary Table 3. To corroborate the results ob­ (Supplementary Fig. 7a–c) were similar to those obtained in MOLT-4 tained by microarray analysis, we chose three representative upregu­ cells. Curiously, in HepG2 cells, SERMs or their combinations with lated genes, FABP1, MVD, and ERG25, and analysed their expression by LOV had no effect on the expression of ABCA1 in the presence of LDL nor real-time PCR (Supplementary Fig. 6). As with MOLT-4, real-time PCR in that of CholMCD (Supplementary Fig. 7d). results correlated with those of the microarray. Thus, it was confirmed To further study the role of cholesterol from different sources, we that FABP1 responded similarly to known targets of SREBP (Supple­ analysed a third cell line, human THP-1 monocytes differentiated to mentary Fig. 6a). macrophages (Supplementary Fig. 8). The effect of SERMs on SREBP target genes was analogous to those in MOLT-4 and HepG2 cells. Similar 3.2. SERMs modulation of SREBP and LXR target gene expression to HepG2, the expression of ABCA1 did not vary in response to SERMs depends on LDL-cholesterol uptake (Supplementary Fig. 8d). Since SERMs interfere with the lysosomal trafficking of cholesterol 3.3. LDL-cholesterol but not MCD-cholesterol accumulates within the cells [13,14,20], we wanted to explore whether SERMs could also affect the expression of lipid metabolism genes when cholesterol is administered In light of our results showing different effects of SERMs on gene to cells in a non-lipoprotein form. For this, we used a cholesterol and expression depending on the cholesterol source, we assessed the effects methyl-cyclodextrin complex (CholMCD), widely used to efficiently of SERMs and LOV on free cholesterol distribution when LDL or transfer cholesterol to cells independently of the endocytic pathway [23, CholMCD were supplied to MOLT-4 cells as before. Filipin staining 29–31]. We incubated the MOLT-4 cells with the different SERMs revealed that, in the presence of LDL, cells treated with SERMs and their combined or not with LOV in the presence of LDL or CholMCD. We combinations with LOV accumulated cytoplasmic vesicles containing analysed the expression of LDLR, HMGCR, and INSIG1, as representa­ free cholesterol in a distribution compatible with late endosomes/lyso­ tives of the SREBP-target genes, and ABCA1, as the main representative somes localisation, while LOV alone produced no effect (Fig. 4). When of the LXR-target genes. The amount of cholesterol added with CholMCD the cells were incubated with CholMCD alone the staining was similar to was adjusted so that the magnitude of the effect on the SREBP targets that of cells treated with LDL alone (Fig. 4). None of the SERMs or their was comparable to that produced by LDL. As references, we included combinations with LOV induced accumulation of free cholesterol in the two conditions without cholesterol in the medium, containing only LPDS presence of CholMCD. These results correlate with those of gene or LPDS plus LOV. expression and confirm that SERMs only significantly affect cholesterol As expected, the addition of LDL to the medium decreased the processed through the endosomal route. expression of the SREBP target genes LDLR, HMGCR, and INSIG1, compared to the condition with LPDS alone (Fig. 3a-c). The three SERMs 3.4. SERMs induce SREBP-2 processing and prevent LXR activation prevented the repression caused by the addition of LDL. LOV treatment did not change the expression of these genes versus LDL alone, sug­ Gene expression studies suggest that the main effects of SERMs on gesting the predominance of the repressive effect of LDL. However, the regulatory response to LDL may involve the SREBP-2 and LXR when LOV was combined with any of the SERMs, the expression of the pathways, both requiring the lysosomal export of LDL-derived choles­ three genes increased synergistically. When cells were supplied with terol for them to be inhibited and activated, respectively. Thus, we CholMCD, we observed a decrease in the expression of the SREBP target directly assessed the processing of SREBP-2 and the activation of LXR. genes relative to LPDS, similar to that in cells exposed to LDL. But, We selected HepG2 cells for this part of the study because of the rele­ contrary to what happened with LDL, the SERMs, in combination or not vance of hepatocytes to cholesterol metabolism. with LOV, did not prevent the repression caused by the addition of First, we evaluated the processing of SREBP-2 in cells treated with CholMCD (Fig. 3a-c). These results indicate that SERMs only interfere TAM in the presence of LDL for different incubation times. To perform 6 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 Fig. 2. Validation of microarray expression re­ sults of selected genes by real-time PCR in MOLT-4 cells. Cells were treated with 10 μM SERMs, 1 μM LOV or their combinations in the presence of 120 μg/ml LDL-cholesterol for 16 h. Data represent the mean ± SEM of three inde­ pendent experiments, expressed as the relative amount of mRNA compared to the control (C) condition in the absence of LOV (set to 1). * p < 0.05, ** p < 0.01, *** p < 0.001; •p < 0.05 compared to the same condition without LOV. these experiments, we used a combination of 0.1% hydroxypropyl- increase in the proportion of n-SREBP-2 was observed. At 16 h, the cyclodextrin (HPCD) and 1 μM LOV for 2 h, a condition that causes precursor form of SREBP-2 (p-SREBP-2) was nearly absent and the rapid cholesterol depletion and, consequently, stimulates SREBP-2 amount of n-SREBP-2 reached a maximum. Finally, at 19 h the effect processing and the appearance of the nuclear form of SREBP-2 (n- ceased. Based on these data, we focused on the 16 h treatment to extend SREBP-2). TAM produced an increase in the processing of SREBP-2, the study to RAL and TOR and their combination with LOV. As expected, although this effect was transitory (Fig. 5a). At 13 h a moderate given the presence of LDL in the medium, in the control condition almost 7 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 Fig. 3. Effect of SERMs and their combinations with LOV on the expression of selected genes in MOLT-4 cells. Cells were treated with 10 μM SERMs, 1 μM LOV or their combinations in the presence of 120 μg/ml LDL-cholesterol or 1.2 μg/ml cholesterol complexed with methyl-cyclodextrin (CholMCD). Alternatively, cells were incubated in the absence of cholesterol (LPDS) and SERMs as a reference. Data represents the mean ± SEM of three independent experiments, expressed as the relative amount of mRNA compared to the control (C) condition in the absence of LOV (set to 1). Results were analysed using the Two-Way Repeated Measures Anova test, and the Student-Newman-Keuls test post-hoc. * p < 0.05, ** p < 0.01, *** p < 0.001 compared to the same condition without any SERM (C); • p < 0.05 compared to the same condition without LOV. all SREBP-2 was unprocessed (Fig. 5b). The addition of LOV alone did macrophages with the SERMs in the presence of acetylated LDL (AcLDL), not increase the processing of SREBP-2 compared to the control condi­ which is actively taken up by macrophages. The capacity of macro­ tion. The addition of the SERMs increased the processing of SREBP-2, phages to activate LXR and increase cholesterol efflux is paramount to and the combination with the LOV produced a further increase. prevent cholesteryl ester deposition and the formation of foamy mac­ LXR activation was studied using a luciferase reporter gene con­ rophages like those that accumulate in atheroma plaques [32]. Virtually struction containing LXRE in its promoter. HepG2 cells were treated identical responses to the different experimental conditions were ob­ with TAM or remained untreated, in the presence of LDL, and LXR ac­ tained in THP-1 macrophages compared with those in HepG2 cells tivity was measured at different time points. From 8 h of treatment, (Supplementary Fig. 9a and b). Overall, SERMs efficiently suppress the control cells, with only LDL, showed increased LXR activity that peaked LDL-induced transcriptional activity of LXR. at 16 h (Fig. 5c). TAM prevented the activation of LXR by LDL. Next, we assessed the effects of TAM, RAL, and TOR and their combinations with 3.5. Estrogen receptor-independent effects of SERMs LOV on LXR activation at 16 h. The three SERMs reduced LXR activation versus the control condition, with LDL alone (Fig. 5d). LOV slightly Given that SERMs are ligands of ERs, we evaluated the role of these reduced LXR activity. The addition of LOV in combination with each receptors in the effects of SERMs. First, we measured the expression of SERM did not produce any additional effect. Moreover, we evaluated the the genes ESR1 and ESR2, coding for ERα and ERβ, respectively. effect of SERMs in the presence of LDL and the LXR activator T0901317. Expression of both ESR1 and ESR2 in HepG2 and THP-1 cells was un­ As expected, T0901317 increased LXR activity relative to the control, detectable (Ct > 40). In MOLT-4 cells, ESR1 mRNA was also undetect­ and only RAL could slightly oppose to this effect. Finally, when cells able and ESR2 mRNA was found in very small amounts (Ct = 30). These were incubated with CholMCD, LXR activity increased moderately. The findings suggest that ERs do not mediate the observed effects of SERMs. SERMs prevented this activation, but to a much lower extent than in the To confirm this, we compared the responses to SERMs and 17β-estradiol presence of LDL (Fig. 5d). (E2) in MOLT-4 cells to those in MCF-7 cells, a mammary gland ERα- Additionally, we assessed the effect of SERMs on LXR activation in positive cell line. The genes selected were LDLR, as a representative cholesterol-loaded macrophages. For this, we treated THP-1 SREBP-2 target, and BCL2, ESR1, and ESR2, as ERs targets. LDLR 8 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 Fig. 4. Effect of SERMs and their combinations with LOV on intracellular distribution of free cholesterol in MOLT-4 cells. Cells were treated with 10 μM SERMs, 1 μM LOV or their combinations in the presence of 120 μg/ml LDL-cholesterol or 1.2 μg/ml cholesterol complexed with methyl-cyclodextrin (CholMCD) for 16 h. Results correspond to a representative experiment of two independent experiments. Scale bar: 20 µm. expression did not change upon the addition of E2, but it did in response endosomal pathway. to SERMs in both cell lines (Fig. 6a). Consistently with the expression of Microarray assays indicated that SREBP-2 targeted gene expression ERα, BCL2 expression increased in MCF-7, but not in MOLT-4 cells in is increased by SERMs. In line with this, SREBP-2 processing is stimu­ response to E2 (Fig. 6b). The SERMs did not modify the expression of lated by SERMs, an effect that likely results from the blockade of the exit BCL2 in MOLT-4 nor in MCF-7 cells. In MCF-7 cells, ESR1 expression was of LDL-derived cholesterol from lysosomes, thereby preventing the inhibited only by E2 (Fig. 6c). The amount of ESR2 mRNA did not cholesterol-mediated inhibition of SREBP-2 processing. Moreover, the significantly change when MOLT-4 cells were treated with E2 or SERMs effect of SERMs on SERBP-2 processing and expression of its target genes compared to untreated cells (Fig. 6d). These results indicate that the is synergistic with that of LOV. This synergy can be the consequence of effects of SERMs on the SREBP and LXR pathways are independent of the the disruption of LDL-cholesterol trafficking by SERMs combined with ERs. the inhibition of cholesterol biosynthesis by LOV. Although SERMs reduce cholesterol biosynthesis, they are less potent than LOV [13] and, 4. Discussion hence, de novo synthesized cholesterol can still reach the endoplasmic reticulum and retain some SREBP-2. When only LOV is administered, it In this study, we demonstrate that the SERMs TAM, RAL and TOR does not produce any effect, since the repressive effect of influence the activity of the lipid metabolism master regulators SREBP LDL-cholesterol prevails over the effect of cholesterol biosynthesis in­ and LXR by an ER-independent pathway and in association with the hibition on SERBP-2 processing. Yet, when LOV is combined with any of interference with the trafficking of LDL-derived cholesterol via the the SERMs, the arrival of LDL cholesterol to the endoplasmic reticulum 9 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 Fig. 5. Effect of SERMs and their combinations with LOV on SERBP-2 and LXR activation in HepG2 cells. (a) Time-dependent effect of 10 μM TAM on SREBP-2 processing in the presence of 120 μg/ml LDL-cholesterol. As a positive control for SREBP-2 processing, cells were treated with 0.1% hydroxypropyl-cyclodextrin (HPCD) plus 1 μM lovastatin (LOV) for 2 h. (b) Effect of 10 μM SERMs, 1 μM LOV or their combinations on SREBP-2 processing at 16 h of treatment in the pres­ ence of 120 μg/ml LDL-cholesterol. (c) Time-dependent effect of 10 μM TAM on LXR transcriptional activity in the presence of 120 μg/ml LDL-cholesterol. Data represent the mean ± SEM of five independent experiments; au, arbitrary units. (d) Effect of 10 μM SERMs combined or not with 1 μM LOV or 1 μM T0901317 on LXR transcriptional activity at 16 h of treatment in the presence of 120 μg/ml LDL-cholesterol. The effect of SERMs in the presence of 30 μg/ml of cholesterol complexed with methyl-cyclodextrin (CholMCD) was also assessed. Results are expressed as log2 and relative to the activity in the condition with LDL alone, and represent the mean ± SEM of four independent experiments. Results were analysed using the Two-Way Repeated Measures Anova test, and the Student-Newman- Keuls test post-hoc. * p < 0.05, ** p < 0.01. Bars not sharing any letter are statistically different to the conditions only differing in the SERM treatment (p < 0.05); •p < 0.05 compared to the same SERM condition with only LDL (d). C, control (no SERM). as well as cholesterol biosynthesis are blocked, resulting in a synergistic ABCA1 expression was observed in MOLT-4 cells, but not in HepG2 cells effect on SREBP-2 processing and expression of target genes. or THP-1 macrophages, reflecting a cell-type-dependent regulation. The MCD complexes with the fluorescent cholesterol analog dehy­ reduced LXR activity observed in THP-1 macrophages treated with droergosterol (DHE) have previously been used for intracellular SERMs in the presence of AcLDL agrees with our previous report in cholesterol trafficking studies, and have shown that they deliver DHE which ABCA1 and ABCG1 protein expression were repressed in these directly to the plasma membrane and, from there, DHE is transported to conditions [20]. Here, in order to compare different cell types, we recycling endosomes or lipid droplets via non-vesicular transport, with assessed ABCA1 expression in the presence of native LDL. The differ­ no involvement of late endosomes or lysosomes [30,31]. SERMs did not ential effect on ABCA1 expression in THP-1 macrophages may lie in the prevent CholMCD-induced inhibition of SREBP-2 target gene expression. efficiency of uptake of each LDL type, much lower for native than for This correlates with the absence of free cholesterol deposits in cells chemically modified LDL [34]. In HepG2 cells, the lack of effect of treated in the same conditions. This result contrasts with the blockade of SERMs on ABCA1 expression may lie in the fact that ABCA1 in hepa­ cholesterol trafficking and the consequent accumulation of free choles­ tocytes is positively regulated by both SREBP-2 and LXR [35]. In this terol in late endosomes/lysosomes produced by SERMs in the presence case, the inactivation of LXR might be compensated by the activation of of LDL, previously described by our group [13,14,20] and confirmed in SREBP-2. Unexpectedly, in MOLT-4 cells, SERMs reduced the expression this study. of ABCA1 also when cholesterol was supplied complexed with MCD, The interference with intracellular cholesterol trafficking produced although to a much lesser extent than with LDL. A possible explanation by SERMs also reduced LXR activity. This may be a consequence of the may be the SREBP-induced expression of miR-33, encoded by an intron inability for LDL-derived cholesterol to leave late endosomes/lysosomes of SREBF2 and which targets ABCA1 mRNA [36]. As observed herein, and, thus, form oxysterols, which bind to LXR and activate the tran­ the SERMs also impaired LXR transcriptional activity when CholMCD scription of genes involved in the efflux of cholesterol, such as ABCA1 was added to cells, suggesting that these drugs are able to prevent LXR and ABCG1 [33]. Nonetheless, in the present study, the reduction in activation also through mechanisms independent of endolysosomal 10 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 Fig. 6. Effect of E2 and SERMs on the expression of LDLR, BCL2, ESR1 and ESR2 genes in MCF-7 and MOLT-4 cells. Cells were treated or not (C) with the indicated concentrations of SERMs or E2 for 16 h in the presence of 120 μg/ml LDL-cholesterol. Data represent the mean ± SEM of three and four independent experiments for MOLT-4 and MCF-7 cells, respectively, expressed as the relative amount of mRNA compared to the control (C) condition (set to 1). Results were analysed using the Repeated Measures Anova test, with Student-Newman-Keuls post-hoc. Bars not sharing any letter are statistically different (p < 0.05). cholesterol trafficking. This might relate to the inhibition of cholesterol In this work, we showed that ERs are not involved in the effect of biosynthesis [13,15,16] or the accumulation of 5,6-epoxycholesterols, SERMs on SERBP-2 and LXR-mediated gene expression. A line of evi­ of which the 5,6α diastereoisomer can modulate LXR activity and dence is the lack of expression of ER genes in HepG2 and THP-1 cells and display agonist or antagonist effects in a cell and gene the marginal expression of only ESR2 in MOLT-4 cells. Furthermore, in context-dependent manner [37,38]. MOLT-4 cells, E2 had no effect on the expression of SREPB-2 or ERs As stated above, the most numerous set of genes with increased target genes. These results contrast with those in MCF-7 cells in which expression in response to SERMs was related to cholesterol biosynthesis the expression of BCL2, an ERα target gene, increased and that of ESR1 and uptake, mainly targets of SREBP-2. However, these drugs also decreased, due to ERα-mediated negative feedback, by treatment with increased the expression of some SREBP-1c targets, such as SCD and E2, in agreement with other reports [45,46]. These results agree with INSIG1. SCD encodes stearoyl-coenzyme A desaturase, involved in our previous studies demonstrating that ERs do not mediate the lipogenesis [1]. This suggests that SERMs could stimulate lipogenesis in SERMs-induced inhibition of LDL-cholesterol trafficking and efflux in vivo and thus contribute to the increased incidence of hepatic steatosis macrophages [20] nor the increase in LDL uptake [14]. Although it has described for TAM and, to a lesser extent, for RAL and TOR [39,40]. This been described that TAM can stimulate LDLR expression by HepG2 cells mechanism could also be involved in the hypertriglyceridemia described through ERα, this was observed in ERα-transfected cells, whereas in in some patients treated with TAM [41]. Nonetheless, the poor repre­ non-transfected cells both TAM and E2 failed to activate transcription sentation of SREBP-1c target genes among the SERM-stimulated genes [47]. suggests that this transcription factor is not so markedly affected by the The effects of SERMs on intracellular cholesterol trafficking are lysosomal cholesterol trafficking blockade. This effect can be explained common to a set of molecules that, like TAM, RAL, and TOR, are cationic by the dual gene expression regulation of SREBP-1c by LXR and SREBP-1 amphiphilic drugs with lysosomotropic properties, that is, they accu­ itself [41]. In this sense, an increase in SREBP-1c expression, due to mulate in lysosomes [48]. The lipophilic characteristics in their struc­ reduced levels of cholesterol in the endoplasmic reticulum, could be tures allow them to passively diffuse through the lysosomal membrane compensated by a direct downregulation of LXR activity [42,43]. These when the molecule is in its basic non-ionized form, which is favoured by results are in line with the interdependence and crosstalk between a concentration gradient [49]. At lysosomal acidic pH, the base is pro­ SREBP-1, SREBP-2, and LXR in order to maintain lipid homeostasis [44]. tonated and cannot leave this compartment, a process called pH 11 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 partitioning or ion trapping [48]. The best known of this class of com­ [5] L.J. Millatt, V. Bocher, J.C. Fruchart, B. Staels, Liver X receptors and the control of cholesterol homeostasis: potential therapeutic targets for the treatment of pounds is U18666A [50], which, like SERMs, interferes with the endo­ atherosclerosis, Biochim. Biophys. Acta 1631 (2) (2003) 107–118. somal trafficking of cholesterol, producing a cellular phenotype that [6] B. Wang, P. Tontonoz, Liver X receptors in lipid signalling and membrane resembles that caused by Niemann-Pick disease type C. There are other homeostasis, Nat. Rev. Endocrinol. 14 (8) (2018) 452–463. compounds with similar chemical characteristics, such as certain anti­ [7] C.K. Osborne, H. Zhao, S.A. Fuqua, Selective estrogen receptor modulators: structure, function, and clinical use, J. Clin. Oncol. 18 (17) (2000) 3172–3186. psychotics, antidepressants and antifungals, that also cause cholesterol [8] U. Gundimeda, Z.H. Chen, R. Gopalakrishna, Tamoxifen modulates protein kinase accumulation in late endosome/lysosomes [51–54]. Some of these an­ C via oxidative stress in estrogen receptor-negative breast cancer cells, J. Biol. tipsychotics and antidepressants have been shown to stimulate SREBP Chem. 271 (23) (1996) 13504–13514. [9] M. Poirot, S. Silvente-Poirot, R.R. Weichselbaum, Cholesterol metabolism and activation and target gene expression [53,55–57]. resistance to tamoxifen, Curr. Opin. Pharmacol. 12 (6) (2012) 683–689. The ability of TAM, RAL, and TOR to interfere with endosomal [10] S. Silvente-Poirot, M. Poirot, Cholesterol and cancer, in the balance, Science 343 trafficking has been effectively used, in vitro, to inhibit Ebola virus entry (6178) (2014) 1445–1446. [11] V.C. Jordan, SERMs: meeting the promise of multifunctional medicines, J. Natl. and infection [51,58–60]. More studies are needed to explore the pos­ Cancer Inst. 99 (5) (2007) 350–356. sibility of using these SERMs in the treatment of different viral in­ [12] J.H. Pickar, T. MacNeil, K. Ohleth, SERMs: progress and future perspectives, fections. The endosome-related effects of SERMs can give to these widely Maturitas 67 (2) (2010) 129–138. [13] Y. Suarez, C. Fernandez, D. Gomez-Coronado, A.J. Ferruelo, A. Davalos, used drugs a new clinical purpose beyond the ER pathway. J. Martinez-Botas, M.A. Lasuncion, Synergistic upregulation of low-density lipoprotein receptor activity by tamoxifen and lovastatin, Cardiovasc. Res. 64 (2) (2004) 346–355. 5. Conclusions [14] F. Cerrato, M.E. Fernandez-Suarez, R. Alonso, M. Alonso, C. Vazquez, O. Pastor, P. Mata, M.A. Lasuncion, D. Gomez-Coronado, Clinically used selective oestrogen TAM, RAL, and TOR inhibit intracellular trafficking of LDL-derived receptor modulators increase LDL receptor activity in primary human cholesterol. Consequently, they prevent the regulatory effect of LDL lymphocytes, Br. J. Pharmacol. 172 (5) (2015) 1379–1394. [15] H. Gylling, S. Pyrhonen, E. Mantyla, H. Maenpaa, L. Kangas, T.A. Miettinen, over SREBP-2, thus increasing the processing of SREBP-2 and the Tamoxifen and toremifene lower serum cholesterol by inhibition of delta 8- expression of its target genes. This effect is synergistic when SERMs are cholesterol conversion to lathosterol in women with breast cancer, J. Clin. Oncol. combined with LOV. Furthermore, SERMs oppose to the activation of 13 (12) (1995) 2900–2905. [16] B. Kedjouar, P. de Medina, M. Oulad-Abdelghani, B. Payre, S. Silvente-Poirot, LXR in response to cholesterol overload. The molecular mechanism G. Favre, J.C. Faye, M. Poirot, Molecular characterization of the microsomal involved in these actions is independent of ERs. Our findings highlight tamoxifen binding site, J. Biol. Chem. 279 (32) (2004) 34048–34061. the impact of SERMs on lipid homeostasis regulation beyond their ac­ [17] A.M. Dnistrian, M.K. Schwartz, E.J. Greenberg, C.A. Smith, D.C. Schwartz, Effect of tamoxifen on serum cholesterol and lipoproteins during chemohormonal therapy, tions as ER modulators, and are consistent with the changes in the lipid Clin. Chim. Acta 223 (1–2) (1993) 43–52. profile of patients taking this medication. [18] P.D. Delmas, N.H. Bjarnason, B.H. Mitlak, A.C. Ravoux, A.S. Shah, W.J. Huster, M. Draper, C. Christiansen, Effects of raloxifene on bone mineral density, serum cholesterol concentrations, and uterine endometrium in postmenopausal women, Funding sources New Engl. J. Med. 337 (23) (1997) 1641–1647. [19] D. Gomez-Coronado, M.A. Lasuncion, J. Martinez-Botas, M.E. Fernandez-Suarez, Role of cholesterol metabolism in the anticancer pharmacology of selective This work was supported by Ministerio de Ciencia, Innovación y estrogen receptor modulators, Semin. Cancer Biol. 73 (2021) 101–115. Universidades (grant RTI2018-098113-B-I00), Plan Estatal de Inves­ [20] M.E. Fernandez-Suarez, J.C. Escola-Gil, O. Pastor, A. Davalos, F. Blanco-Vaca, M. tigación Científica y Técnica y de Innovación 2017–2020, Spain; Euro­ A. Lasuncion, J. Martinez-Botas, D. Gomez-Coronado, Clinically used selective pean Regional Development Fund (ERDF/FEDER); European estrogen receptor modulators affect different steps of macrophage-specific reverse cholesterol transport, Sci. Rep. 6 (2016) 32105. Cooperation in Science and Technology (COST) action BM0904, and [21] J. Leignadier, F. Dalenc, M. Poirot, S. Silvente-Poirot, Improving the efficacy of Program ALIBIRD (grant S2013/ABI-2728), Comunidad de Madrid, hormone therapy in breast cancer: the role of cholesterol metabolism in SERM- Spain. CIBEROBN and CIBERDEM are initiatives of the Instituto de Salud mediated autophagy, cell differentiation and death, Biochem. Pharmacol. 144 (2017) 18–28. Carlos III, Spain. [22] D. Calvo, D. Gomez-Coronado, Y. Suarez, M.A. Lasuncion, M.A. Vega, Human CD36 is a high affinity receptor for the native lipoproteins HDL, LDL, and VLDL, J. Lipid Res. 39 (4) (1998) 777–788. [23] U. Klein, G. Gimpl, F. Fahrenholz, Alteration of the myometrial plasma membrane Conflict of interest statement cholesterol content with beta-cyclodextrin modulates the binding affinity of the oxytocin receptor, Biochemistry 34 (42) (1995) 13784–13793. None. [24] A.J. Brown, L. Sun, J.D. Feramisco, M.S. Brown, J.L. Goldstein, Cholesterol addition to ER membranes alters conformation of SCAP, the SREBP escort protein that regulates cholesterol metabolism, Mol. Cell 10 (2) (2002) 237–245. Acknowledgements [25] S. Rodriguez-Acebes, N. Palacios, J.I. Botella-Carretero, N. Olea, L. Crespo, R. Peromingo, D. Gomez-Coronado, M.A. Lasuncion, C. Vazquez, J. Martinez-Botas, Gene expression profiling of subcutaneous adipose tissue in morbid obesity using a We thank Lorena Crespo, Alejandra Quintero, Gema de la Peña and focused microarray: distinct expression of cell-cycle- and differentiation-related Nuria Crcoles for excellent technical assistance. M.A.L. is Emeritus of the genes, BMC Med. Genom. 3 (2010) 61. [26] A.I. Saeed, V. Sharov, J. White, J. Li, W. Liang, N. Bhagabati, J. Braisted, M. Klapa, Servicio Madrileño de Salud (Comunidad Autónoma de Madrid). T. Currier, M. Thiagarajan, A. Sturn, M. Snuffin, A. Rezantsev, D. Popov, A. Ryltsov, E. Kostukovich, I. Borisovsky, Z. Liu, A. Vinsavich, V. Trush, Appendix A. Supporting information J. Quackenbush, TM4: a free, open-source system for microarray data management and analysis, BioTechniques 34 (2) (2003) 374–378. [27] B. Mlecnik, M. Scheideler, H. Hackl, J. Hartler, F. Sanchez-Cabo, Z. Trajanoski, Supplementary data associated with this article can be found in the PathwayExplorer: web service for visualizing high-throughput expression data on online version at doi:10.1016/j.biopha.2021.111871. biological pathways, Nucleic Acids Res. 33 (Web Server issue) (2005) W633–W637. [28] M.W. Pfaffl, A new mathematical model for relative quantification in real-time RT- References PCR, Nucleic Acids Res. 29 (9) (2001) e45. [29] M. Hao, S.X. Lin, O.J. Karylowski, D. Wustner, T.E. McGraw, F.R. Maxfield, Vesicular and non-vesicular sterol transport in living cells. The endocytic recycling [1] J.D. Horton, J.L. Goldstein, M.S. Brown, SREBPs: activators of the complete compartment is a major sterol storage organelle, J. Biol. Chem. 277 (1) (2002) program of cholesterol and fatty acid synthesis in the liver, J. Clin. Invest. 109 (9) 609–617. (2002) 1125–1131. [30] D. Wustner, A. Herrmann, M. Hao, F.R. Maxfield, Rapid nonvesicular transport of [2] M.S. Brown, J.L. Goldstein, Cholesterol feedback: from Schoenheimer’s bottle to sterol between the plasma membrane domains of polarized hepatic cells, J. Biol. Scap’s MELADL, J. Lipid Res. 50 (2009) S15–27. Chem. 277 (33) (2002) 30325–30336. [3] A. Chawla, J.J. Repa, R.M. Evans, D.J. Mangelsdorf, Nuclear receptors and lipid [31] D. Wustner, M. Mondal, I. Tabas, F.R. Maxfield, Direct observation of rapid physiology: opening the X-files, Science 294 (5548) (2001) 1866–1870. internalization and intracellular transport of sterol by macrophage foam cells, [4] M.B. Patel, N.A. Oza, I.S. Anand, S.S. Deshpande, C.N. Patel, Liver X receptor: a Traffic 6 (5) (2005) 396–412. novel therapeutic target, Indian J. Pharm. Sci. 70 (2) (2008) 135–144. 12 M.E. Fernández-Suárez et al. Biomedicine & Pharmacotherapy 141 (2021) 111871 [32] K.J. Moore, I. Tabas, Macrophages in the pathogenesis of atherosclerosis, Cell 145 [47] J.C. Bruning, P. Lingohr, J. Gillette, B. Hanstein, H. Avci, W. Krone, D. Muller- (3) (2011) 341–355. Wieland, J. Kotzka, Estrogen receptor-alpha and Sp1 interact in the induction of [33] N. Shibata, C.K. Glass, Macrophages, oxysterols and atherosclerosis, Circ. J. 74 (10) the low density lipoprotein-receptor, J. Steroid Biochem. Mol. Biol. 86 (2) (2003) (2010) 2045–2051. 113–121. [34] I. Tabas, Cholesterol and phospholipid metabolism in macrophages, Biochim. [48] S.D. Goldman, R.S. Funk, R.A. Rajewski, J.P. Krise, Mechanisms of amine Biophys. Acta 1529 (2000) 164–174. accumulation in, and egress from, lysosomes, Bioanalysis 1 (8) (2009) 1445–1459. [35] N. Tamehiro, Y. Shigemoto-Mogami, T. Kakeya, K. Okuhira, K. Suzuki, R. Sato, [49] C.L. Andrew, A.R. Klemm, J.B. Lloyd, Lysosome membrane permeability to amines, T. Nagao, T. Nishimaki-Mogami, Sterol regulatory element-binding protein-2- and Biochim. Biophys. Acta 1330 (1) (1997) 71–82. liver X receptor-driven dual promoter regulation of hepatic ABC transporter A1 [50] L. Liscum, J.R. Faust, The intracellular transport of low density lipoprotein-derived gene expression: mechanism underlying the unique response to cellular cholesterol cholesterol is inhibited in Chinese hamster ovary cells cultured with 3-beta-[2- status, J. Biol. Chem. 282 (29) (2007) 21090–21099. (diethylamino)ethoxy]androst-5-en-17-one, J. Biol. Chem. 264 (20) (1989) [36] K.J. Rayner, Y. Suarez, A. Davalos, S. Parathath, M.L. Fitzgerald, N. Tamehiro, E. 11796–11806. A. Fisher, K.J. Moore, C. Fernandez-Hernando, MiR-33 contributes to the [51] C.J. Shoemaker, K.L. Schornberg, S.E. Delos, C. Scully, H. Pajouhesh, G.G. Olinger, regulation of cholesterol homeostasis, Science 328 (5985) (2010) 1570–1573. L.M. Johansen, J.M. White, Multiple cationic amphiphiles induce a Niemann-Pick [37] T.J. Berrodin, Q. Shen, E.M. Quinet, M.R. Yudt, L.P. Freedman, S. Nagpal, C phenotype and inhibit Ebola virus entry and infection, PLoS One 8 (2) (2013) Identification of 5alpha, 6alpha-epoxycholesterol as a novel modulator of liver X 56265. receptor activity, Mol. Pharmacol. 78 (6) (2010) 1046–1058. [52] I. Kristiana, L.J. Sharpe, V.S. Catts, L.H. Lutze-Mann, A.J. Brown, Antipsychotic [38] G. Segala, P. de Medina, L. Iuliano, C. Zerbinati, M.R. Paillasse, E. Noguer, drugs upregulate lipogenic gene expression by disrupting intracellular trafficking F. Dalenc, B. Payre, V.C. Jordan, M. Record, S. Silvente-Poirot, M. Poirot, 5,6- of lipoprotein-derived cholesterol, Pharmacogenomics J. 10 (5) (2010) 396–407. Epoxy-cholesterols contribute to the anticancer pharmacology of tamoxifen in [53] A. Canfran-Duque, M.E. Casado, O. Pastor, J. Sanchez-Wandelmer, G. de la Pena, breast cancer cells, Biochem. Pharmacol. 86 (1) (2013) 175–189. M. Lerma, P. Mariscal, F. Bracher, M.A. Lasuncion, R. Busto, Atypical [39] N. Hamada, Y. Ogawa, T. Saibara, Y. Murata, S. Kariya, A. Nishioka, M. Terashima, antipsychotics alter cholesterol and fatty acid metabolism in vitro, J. Lipid Res. 54 T. Inomata, S. Yoshida, Toremifene-induced fatty liver and NASH in breast cancer (2) (2013) 310–324. patients with breast-conservation treatment, Int. J. Oncol. 17 (6) (2000) [54] J. Sanchez-Wandelmer, A. Davalos, G. de la Pena, S. Cano, M. Giera, A. Canfran- 1119–1123. Duque, F. Bracher, A. Martin-Hidalgo, C. Fernandez-Hernando, M.A. Lasuncion, [40] M. Matsumura, K. Tashiro, A. Miura, T. Tajima, I. Kinoshita, E. Kojima, R. Busto, Haloperidol disrupts lipid rafts and impairs insulin signaling in SH-SY5Y A. Yoshizawa, [A case of non-alcoholic fatty liver disease (NAFLD) aggravated after cells, Neuroscience 167 (1) (2010) 143–153. treatment with raloxifene], Nihon Shokakibyo Gakkai Zasshi 108 (12) (2011) [55] J. Ferno, M.B. Raeder, A.O. Vik-Mo, S. Skrede, M. Glambek, K.J. Tronstad, 2036–2041. H. Breilid, R. Lovlie, R.K. Berge, C. Stansberg, V.M. Steen, Antipsychotic drugs [41] Y. Hozumi, M. Kawano, M. Miyata, Severe hypertriglyceridemia caused by activate SREBP-regulated expression of lipid biosynthetic genes in cultured human tamoxifen-treatment after breast cancer surgery, Endocr. J. 44 (5) (1997) 745–749. glioma cells: a novel mechanism of action? Pharmacogenomics J. 5 (5) (2005) [42] F.J. Field, E. Born, S. Murthy, S.N. Mathur, Regulation of sterol regulatory element- 298–304. binding proteins in hamster intestine by changes in cholesterol flux, J. Biol. Chem. [56] M.B. Raeder, J. Ferno, M. Glambek, C. Stansberg, V.M. Steen, Antidepressant drugs 276 (20) (2001) 17576–17583. activate SREBP and up-regulate cholesterol and fatty acid biosynthesis in human [43] J.J. Repa, G. Liang, J. Ou, Y. Bashmakov, J.M. Lobaccaro, I. Shimomura, B. Shan, glial cells, Neurosci. Lett. 395 (3) (2006) 185–190. M.S. Brown, J.L. Goldstein, D.J. Mangelsdorf, Regulation of mouse sterol [57] M.B. Raeder, J. Ferno, A.O. Vik-Mo, V.M. Steen, SREBP activation by regulatory element-binding protein-1c gene (SREBP-1c) by oxysterol receptors, antipsychotic- and antidepressant-drugs in cultured human liver cells: relevance LXRalpha and LXRbeta, Genes Dev. 14 (22) (2000) 2819–2830. for metabolic side-effects? Mol. Cell Biochem. 289 (1–2) (2006) 167–173. [44] S. Rong, V.A. Cortes, S. Rashid, N.N. Anderson, J.G. McDonald, G. Liang, Y. [58] L.M. Johansen, J.M. Brannan, S.E. Delos, C.J. Shoemaker, A. Stossel, C. Lear, B. A. Moon, R.E. Hammer, J.D. Horton, Expression of SREBP-1c requires SREBP-2- G. Hoffstrom, L.E. Dewald, K.L. Schornberg, C. Scully, J. Lehar, L.E. Hensley, J. mediated generation of a sterol ligand for LXR in livers of mice, eLife 6 (2017) M. White, G.G. Olinger, FDA-approved selective estrogen receptor modulators e25015. inhibit Ebola virus infection, Sci. Transl. Med. 5 (190) (2013) 190. [45] S.J. Ellison-Zelski, N.M. Solodin, E.T. Alarid, Repression of ESR1 through actions of [59] Y. Zhao, J. Ren, K. Harlos, D.M. Jones, A. Zeltina, T.A. Bowden, S. Padilla-Parra, E. estrogen receptor alpha and Sin3A at the proximal promoter, Mol. Cell Biol. 29 E. Fry, D.I. Stuart, Toremifene interacts with and destabilizes the Ebola virus (18) (2009) 4949–4958. glycoprotein, Nature 535 (7610) (2016) 169–172. [46] M.R. Bratton, B.N. Duong, S. Elliott, C.B. Weldon, B.S. Beckman, J.A. McLachlan, [60] H. Fan, X. Du, J. Zhang, H. Zheng, X. Lu, Q. Wu, H. Li, H. Wang, Y. Shi, G. Gao, M.E. Burow, Regulation of ERalpha-mediated transcription of Bcl-2 by PI3K-AKT Z. Zhou, D.X. Tan, X. Li, Selective inhibition of Ebola entry with selective estrogen crosstalk: implications for breast cancer cell survival, Int. J. Oncol. 37 (3) (2010) receptor modulators by disrupting the endolysosomal calcium, Sci. Rep. 7 (2017) 541–550. 41226. 13