LXR activation mediates beneficial effects and unwanted lipogenesis.

LXR transcriptional complexes are RXR heterodimers that can be activated by a canonical mechanism of coregulator recruitment or an alternative de-repression mechanism of (A) a constitutively repressed complex by (B) corepressor displacement. (C) ABCA1 gene transcription leads to cholesterol efflux and mobilization, APOE lipidation and potentially Aβ clearance, in addition to other mechanisms of potential benefit in ADRD; whereas, SREBF1 transcription initiates a lipogenic program in the liver leading to unwanted LDL formation and TG elevation.

Chemical structures and compound codes of literature LXR ligands with biphenyl sulfone ligands shown in blue.

LXR isoform selectivity from CRT measurements.

Concentration-response for recruitment of SRC-1 to LXRα (dotted lines) and LXRβ (solid line), determined by CRT assay, showing mean and SD from triplicate experiments, normalized to T0 (100%) as a full LXR agonist (shown in turquoise for reference). GSK2033 (R) and SR9238 (S) titrations were run in the presence of 24HC (3 µM) and neither ligand stabilized the LXR:SRC1 complex alone (T,U).

ABCA1-luc reporter in CCF cells.

Concentration-response for LXR ligands in astrocytoma cells measured after 24 hr incubation, normalized to T0 response (100%) showing mean and SD from triplicate experiments (see Fig. S2 for biological replicates).

Correlating potency in cell-free and cell-based assays.

A-F) Correlation of potency for SRC1 coactivator stabilization by LXR ligands from CRT data with potency for ABCA1 and SRE reporter assays in CCF and HepG2 cells, respectively. Pearson and Spearman correlation data are shown with significance and Deming slope with F-test, showing poor correlation of LXRα CRT data with ABCA1 activation. Solid line and 95% confidence limits are from simple linear fit with dashed line showing Deming fit. G-H) Correlation of potency for ligand-dependent recruitment of SRC2-2, D22, and TRAP220 to LXR α and β showing correlation statistics, best-fit line and 95% confidence intervals. I) Hierarchical clustering analysis of cell-free and cell-based relative response at 1 µM ligand. J-K) Correlation of potency for ligand-dependent recruitment of SRC1 to LXR α and β isoforms showing Pearson and Spearman correlations and best-fit line and 95% confidence intervals from simple linear regression.

Lipogenic response of HepG2 cells to treatment with LXR ligands.

(A-D) Concentration-response curves for LXR ligands normalized to maximal response for T0 (EC50 = 15 nM) showing a range of potency (AZ876 EC50 = 4.2 nM to CL2-57 2.8 µM) and maximal efficacy (BE-1218 80% to GSK3987 150%). See Fig. S2 for biological replicates. (E-F) Three biphenyl sulfones gave anomalous activation of SRE at higher concentrations (Fig. S3): XL041 gave a partial agonist response at submicromolar concentrations (EC50 = 5.7 nM Emax = 22%), while SR9238 and GSK2033 had a neutral or antagonist response. (G-I) The transcriptional response to the five biphenyl sulfones was studied by RT-PCR under the same conditions as for SRE-luc measurements replicating the lipogenic induction observed in the reporter assays; both GSK2033 and SR9238 reduced response below baseline. Both DMHCA and 24HC were cytotoxic at higher concentrations. Data show mean and standard deviation for triplicate measurements. See Fig. S3 for full concentration-response curves.

Coactivator recruitment to LXR induced by LXR ligands.

CRT data was normalized to T0 maximal response (100%) for all four CoA studied: LXRα (dashed lines) LXRβ (solid lines). For clarity, only TRAP220 and D22 data are shown. As noted in the text, most ligands were β-selective or nonselective, with the exception of AZ876 that selectively stabilizes CoA binding to LXRα. GSK2033 and SR9238 were studied in the absence of agonist. Data shown mean and SD from at least triplicate measurements. (See Fig. S5 all four CoA studied).

Corepressor binding to LXR isoforms: ligand dependence.

Concentration-response from CRT measurements of: NCOR2 binding to LXRα (dashed lines) (A); NCOR2 binding to LXRβ (solid lines) (B); SMRT2 binding to LXRα (dashed lines) (C); SMRT2 binding to LXRβ (solid lines) (D). Data shown mean and SD from at least triplicate measurements. (See Fig. S6 for alternative layout).

Ligand-induced conformational change to LXR causing corepressor (NCOR2) replacement by coactivator (SRC1) measured by pCRT.

Concentration-response from CRT for SRC1 binding relative to apo-LXR (black dashed line) and NCOR2 binding relative to apo-LXR (magenta dashed line). Concentration-response from pCRT for SRC1 binding relative to apo-LXR (black solid line) and NCOR2 binding relative to apo-LXR (magenta solid line) with titration of ligand into solution of LXR, NCOR2, and SRC1. Data show mean of triplicate measurements.

Coregulator stabilization signatures.

CRT signatures and correlation with stabilization and structure of LXR:coregulator complexes (Table 1).

Categorization of LXR ligands as agonists and antagonists conforming to CRT and pCRT signatures. (A) Heat map of responses (ligand concentration at 1 µM) in cell-free and cell-based assays to LXR ligands together with theoretical response to agonist and antagonist ligand signatures. (B) Relative energy perturbations of apo and coregulator-bound LXR caused by each class of agonist and antagonist identified and classified in Table 1. (C) Co-crystal structure of LXRα:SRC1-2 co-crystal with GW3965 (PDB 3IPQ) showing key helices. (D) Superposition of LXRα:SRC1-2:GW3965 co-srystal with structure of LXRα:SMRT2 simulated using Alphafold, showing the significant structural perturbation of H11 and H12 on ligand binding (red shades) compared to the ligand-free LXRα (blue shades).