Vitamin D and lumisterol derivatives can act on liver X receptors (LXRs)

Abstract

The interactions of derivatives of lumisterol (L3) and vitamin D3 (D3) with liver X receptors (LXRs) were investigated. Molecular docking using crystal structures of the ligand binding domains (LBDs) of LXRα and β revealed high docking scores for L3 and D3 hydroxymetabolites, similar to those of the natural ligands, predicting good binding to the receptor. RNA sequencing of murine dermal fibroblasts stimulated with D3-hydroxyderivatives revealed LXR as the second nuclear receptor pathway for several D3-hydroxyderivatives, including 1,25(OH)₂D3. This was validated by their induction of genes downstream of LXR. L3 and D3-derivatives activated an LXR-response element (LXRE)-driven reporter in CHO cells and human keratinocytes, and by enhanced expression of LXR target genes. L3 and D3 derivatives showed high affinity binding to the LBD of the LXRα and β in LanthaScreen TR-FRET LXRα and β coactivator assays. The majority of metabolites functioned as LXRα/β agonists; however, 1,20,25(OH)₃D3, 1,25(OH)₂D3, 1,20(OH)₂D3 and 25(OH)D3 acted as inverse agonists of LXRα, but as agonists of LXRβ. Molecular dynamics simulations for the selected compounds, including 1,25(OH)₂D3, 1,20(OH)₂D3, 25(OH)D3, 20(OH)D3, 20(OH)L3 and 20,22(OH)₂L3, showed different but overlapping interactions with LXRs. Identification of D3 and L3 derivatives as ligands for LXRs suggests a new mechanism of action for these compounds.

Figure 1

1,25(OH)₂D3 and CYP11A1-derived secosteroids stimulate LXR activated ABCA1 gene expression in HaCaT keratinocytes. Relative expression of ABCA1 gene after treatment with 10^–7 M 1,25(OH)₂D3 and 20,23(OH)₂D3 for 6 (A) and 24 h (B). Values were retrieved from microarray data deposited at the NCBI GEO (GSE117351). ChIP analysis performed on isolated nuclei from HaCaT keratinocytes treated with 1,25(OH)₂D3 using antibody against both LXRα and β, showed significant stimulation of the LXRα/β binding to the LXR-RE of the ABCA1 (C) but not the control gene, GADPH. Means ± SD from 3 independent experiments (D). QPCR quantification of ABCA1 in RNA from HaCaT keratinocytes (n = 3) treated with 20(OH)D3, 25(OH)D3, 1,25(OH)₂D3, 20(OH)L3, 20,22(OH)₂L3 and 24(OH)L3 (E). *p < 0.5 and **p < 0.01 by student t-test. The experiment was repeated 3 times. The heatmap was prepared using ClustVis software (https://biit.cs.ut.ee/clustvis/).

Figure 2

RNAseq analysis of changes in gene expression in murine dermal fibroblasts treated with 10^–7 M 1,25(OH)₂D3, 20,23(OH)₂D3, 1,20,23(OH)₃D3 or 20(OH)cholesterol (20(OH)C, a native LXR ligand) for 24 h. Heat map of the gene expression pattern (A) with corresponding Venn diagrams shown for down (B) or upnregulated (C) protein coding genes for absolute fold change ≥ 2 cutoff. The Venn diagrams were prepared using Venny version 2.1.0: https://bioinfogp.cnb.csic.es/tools/venny/index.html. The heatmap was prepared using ClustVis: https://biit.cs.ut.ee/clustvis/.

Figure 3

Vitamin D and lumisterol hydroxyderivatives stimulate the expression of LXR-dependent genes. (A) QPCR analysis of changes in expression of genes downstream of LXR in murine dermal fibroblasts treated with 10^–7 M 1,25(OH)₂D3, 20(OH)D3, 1,20(OH)₂D3, 20,23(OH)₂D3, 1,20,23(OH)₃D3, 20(OH)7DHC, 20(OH)L3 or 20(OH)C, or ethanol (control) for 24 h. (B) Stimulation of LXR-dependent gene expression in the brain of SKH-1 (B) or Ptch^+/−/SKH-1 (C) mice (n = 3 per group) treated with 20 µg/kg of 20(OH)D3, respectively, for 6 h. Data are presented as means ± SD, n = 3. Statistical analysis was done using the t-test: *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001 versus control (ethanol).

Figure 4

L3 and D3-derivatives activate a LXR-response element (LXRE)-driven reporter in CHO cells (A,C) and human HaCaT keratinocytes (B,D). Representative dose response curves are in (A) and (B), while a summary of assays performed with 10^–7 M ligands is presented for each experiment separately for CHO cells (C) and HaCaT keratinocytes (D). Data are presented as means ± SE, (n = number of assays). Analysis was done using one-way ANOVA for dose responses and the t-test: *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001 versus control (ethanol).

Figure 5

Binding of L3 and D3 derivatives to the LBD of the LXRα (A) and β (B) in LanthaScreen TR-FRET LXRα and β coactivator assays. (A) and (B): representative binding curves with values presented as means ± SE, (n = 4). Analysis was done using one-way ANOVA with significance defined as *p < 0.05, **p < 0.01, ***p < 0.001 or ****p < 0.0001.

Figure 6

Ligand induced translocation of LXR to the nucleus. (A) Colocalization analysis of LXRα and PI (nuclear counterstain) in HaCaT cells treated with 10^–7 M of 25(OH)D3, 1,25(OH)₂D3, 20(OH)D3, 1,20(OH)₂D3, 20(OH)L3 or 20,22(OH)₂L3 or ethanol (control) for 24 h. Manders’ coefficient (0–1) (right panel) was significantly higher for cells treated with D3 and L3-hydroxyderivatives than cells treated with vehicle only. Data are presented as means ± SD, (n = 2). (B) Imaging flow cytometry analysis of HaCaT cells treated with ethanol or 10^–7 M 20(OH)D3, 1,20(OH)₂D3, 25(OH)D3, 1,25(OH)₂D3, 20(OH)L3, 20,22(OH)₂L3, 20,23(OH)₂L3 or 1,20,23(OH)₃D3 for 12 h. Fixed cells were stained with Hoechst and immunostained with antibodies against VDRR. Ratios of nuclear (co-localization with Hoechst) vs cytoplasmic localization of VDR were determined following analysis of 515 to 2339 individual cells. (C) Imaging cytometry images of individual HaCaT cells showing LXR localization in cytoplasm or nucleus following treatment with ethanol, 20,22(OH)₂L3 or 1,20,23(OH)₃D3. Bar graphs represent quantitative analysis of images acquired by imaging cytometry. HaCaT cells treated with ethanol or 10^–7 M 20(OH)D3, 1,20(OH)₂D3, 25(OH)D3, 1,25(OH)₂D3, 20(OH)L3, 20,22(OH)₂L3, 20,23(OH)₂L3 or 1,20,23(OH)₃D3 for 12 h were fixed, permeabilized cells and immunostained with Hoechst and antibodies against LXR. Ratios of nuclear vs cytoplasmic localization of LXR were determined following analysis of 515–2339 individual cells. The data in bar graphs (A–C) show significant differences between ligand -treated and control (ethanol treated) cells. Analysis was done using t-test: **p < 0.01, ***p < 0.001 or ****p < 0.0001 versus control (ethanol). For part A the slides were examined using a KEYENCE America BZ-X710 Fluorescence Microscope (Itasca, IL) and captured using KEYENCR BZ-X viewer (version 1.3.0.5, https://www.keyence.com/products/microscope/fluorescence-microscope/bz-x700/index_pr.jsp). The images were subsequently analyzed using the JACoP plugin (version 2.1.1, https://imagejdocu.tudor.lu/doku.php?id=plugin:analysis:jacop_2.0:just_another_colocalization_plugin:start) for colocalization analysis with ImageJ (version 1.52a, http://imageJ.nih.gov/ij). For part C, images were captured using an Amnis ImageStreamX Mk II Imaging Flow Cytometer (Luminex Corporation) and IDEAS software version 6.2.

Figure 7

(A) Binding modes for the selected four D3 derivatives (1,20(OH)₂D3, 1,25(OH)₂D3, 20(OH)D3, and 25(OH)D3) and two L3 derivatives (20,22(OH)₂L3, 20(OH)L3) and co-crystalized ligands in the ligand binding domain (LBD) of LXRα (PDBID:5AVI) and LXRβ (PDBID:5HJP). Docked poses of the studied ligands are shown in green and the co-crystallized ligands in LXRα and in LXRβ are shown in light brown. The mesh areas shown in the figure are hydrophobic binding regions in LXRs. (B,C) are based on last 150 ns of the equilibrated MD trajectories (B) Different L3 and D3 derivatives resulted in varied degrees of conformational fluctuation for the residues between helices in the LBDs of LXRα and LXRβ. (C) Different L3 and D3 derivatives could result in the small secondary structure changes of helix 10 to helix 12 for both LDB of LXRα and LXRβ. Image for (A) is made with PyMOL (v2.4.0, https://pymol.org/2/) based on our molecular docking results. Image for (B) is made with Microsoft Excel (v2019, https://office.microsoft.com/excel) based on the root mean square fluctuation (RMSF) analysis of our molecular dynamics simulation trajectories. Image for (C) is made with Microsoft Excel (v2019, https://office.microsoft.com/excel) based on the secondary structural analysis of our molecular dynamics simulation trajectories.

Figure 8

Varied principal dynamic motion of helix 12/AF-2 region (in brown color) and the β-sheet/helix 6 (in yellow color) in the ligand-binding pocket of LXRα (A) and LXRβ (B) by binding with different D3 and L3 derivatives. LXRα and LXRβ were shown as new cartoon and molecules were shown as licorice in green. Arrows for principal dynamic motion are shown as blue. The images were made with VMD (v1.9.2, http://www.ks.uiuc.edu/Research/vmd/) based on principal component analysis (PCA) of our molecular dynamics simulation trajectories.

Figure 9

Varied electrostatic potential distribution for the LBD of LXRα (A) and LXRβ (B) bound with different D3 and L3 derivatives, including for the ligand binding pocket and H12 helix. Positive electrostatic potential is shown in blue and negative electrostatic potential is shown in red. The images were made with PyMOL(v2.4.0, https://pymol.org/2/) based on electrostatic potential analyses of the representative complex structures in our equilibrated MD simulations using APBS (v1.2.0, http://www.poissonboltzmann.org/).