The Liver X Receptor Is Selectively Modulated to Differentially Alter Female Mammary Metastasis-associated Myeloid Cells

Abstract

Dysregulation of cholesterol homeostasis is associated with many diseases such as cardiovascular disease and cancer. Liver X receptors (LXRs) are major upstream regulators of cholesterol homeostasis and are activated by endogenous cholesterol metabolites such as 27-hydroxycholesterol (27HC). LXRs and various LXR ligands such as 27HC have been described to influence several extra-hepatic biological systems. However, disparate reports of LXR function have emerged, especially with respect to immunology and cancer biology. This would suggest that, similar to steroid nuclear receptors, the LXRs can be selectively modulated by different ligands. Here, we use RNA-sequencing of macrophages and single-cell RNA-sequencing of immune cells from metastasis-bearing murine lungs to provide evidence that LXR satisfies the 2 principles of selective nuclear receptor modulation: (1) different LXR ligands result in overlapping but distinct gene expression profiles within the same cell type, and (2) the same LXR ligands differentially regulate gene expression in a highly context-specific manner, depending on the cell or tissue type. The concept that the LXRs can be selectively modulated provides the foundation for developing precision pharmacology LXR ligands that are tailored to promote those activities that are desirable (proimmune), but at the same time minimizing harmful side effects (such as elevated triglyceride levels).

Keywords: 27-hydroxycholesterol; SLXRM; SLiMs; breast cancer; liver X receptor; myeloid cell; nuclear receptor; selective LXR modulator.

Figure 1.

27HC elicit differential gene expression profile in macrophages, compared with GW3965. (A-D) BMDMs were treated with vehicle, 27HC, or GW3965 for 6 hours, 12 hours, or 24 hours. Total RNA was subsequently isolated and purified. LXR target gene expression pattern on 2 dimensions was measured by RT-qPCR. Data are represented as mean ± SEM, and asterisks (*) represent statistical significance as indicated. (E-J) BMDMs were treated with vehicle, 27HC, or GW3965 for 24 hours. Total RNA was isolated and then sequenced after genomic DNA elimination. (E) PCA was then carried out on the sequenced expressions and each sample was plotted on PC1and PC2. (F) Heatmap plotting the expression of all DEGs between 27HC and vehicle and between GW3965 and vehicle. (G-H) Volcano plot shows the fold changes and P values of LXR target genes under 27HC or GW3965 treatments. (I,J) Venn diagram shows the overlapping and nonoverlapping DEGs between 27HC and GW3965 treatments. (K) Multidimensional scaling plot show treatments with all other ligands, in the first 2 dimensions. In A-D, data are presented as mean ± SEM and asterisks (*) indicate statistical significance (<.05) from the respective vehicle control (N = 3 in A, C, D. N = 5 in B, 1-way analysis of variance with post hoc Sidak’s multiple comparisons test). In E-J, N = 6 for vehicle, N = 5 for 27HC and GW3965. Supporting data in Figures S1 and S2 (41).

Figure 2.

Genes uniquely regulated by 27HC compared with GW3965 are associated with poor survival. Genes that were upregulated or downregulated by 27HC compared with vehicle, excluding genes that were regulated by GW3965 compared with vehicle were used to create respective gene signatures which were then used to probe the Metabric dataset. Patients were parsed into quartiles based on their signature expression, and the fourth quartile (highest expression) was compared with the first quartile (lowest expression). (A) Recurrence-free survival (RFS) and (B) overall survival (OS) were greatly reduced in tumors of patients with high expression of those genes uniquely upregulated by 27HC. (C) RFS of the high or low quartile of tumors of patients with the expression of those genes uniquely downregulated by 27HC. (D) OS was significantly increased in patients who highly expressed those genes uniquely downregulated by 27HC. N = 952. (A) P < .0001, Log-rank test. (B) P < .0001, Log-rank test. (C) P = .0839, Gehan–Breslow–Wilcoxon test. (D) P = .0014, Log-rank test, and P = .0053, Gehan–Breslow–Wilcoxon test.

Figure 3.

27HC promoted differential gene expression modules in macrophages, compared with GW3965. RNA sequenced macrophages underwent WGCNA analysis, which discovered 6 gene expression modules (A-F). The eigengene expression of each module is shown on the left, with corresponding heatmaps to the right. Note that 27HC oppositely regulated module 2, 4, and 5 compared with GW3965. The dendrogram used to generate modules is shown in Figure S3B (41).

Figure 4.

27HC differentially changed the immune cell composition of a metastatic tumor microenvironment, compared to GW3965. (A) 4T1-luc tumor cells were grafted into BALB/c mice and the indicated treatments were given 1 day postgraft. After 16 days of treatments, metastatic lungs were harvested and enriched for live CD45+ cells for single-cell RNA sequencing. (B-D) Postfiltering, the Seurat analysis pipeline identified 9 population of Cd11b+ cells, with marker gene expression highlighted in (C) and the cell populations were annotated in (D). (E) Each treatment enriched and diminished different subpopulations of Cd11b+ cells and (F) RNA velocity analysis predicts MHCII^lo population 0 to be the terminal population in the closely clustered monocytic cells. N = 3 for all treatment groups.

Figure 5.

27HC differentially regulated gene expression in distinct myeloid populations within a metastatic tumor microenvironment, compared to GW3965. (A-C) Venn diagrams show that 27HC and GW3965 share overlapping and nonoverlapping DEGs. Numbers on the bottom right corner indicate the genes that are not differentially expressed. (D-I) Heatmaps of the expression of LXR target genes show that 27HC and GW3965 regulated LXR target genes differentially in different subpopulations of myeloid cells. Supporting data in Figures S4 and S5 (41).