Inhibition of LXR controls the polarization of human inflammatory macrophages through upregulation of MAFB

Monocyte-derived macrophages contribute to pathogenesis in inflammatory diseases and their effector functions greatly depend on the prevailing extracellular milieu. Whereas M-CSF primes macrophages for acquisition of an anti-inflammatory profile, GM-CSF drives the generation of T cell-stimulatory and pro-inflammatory macrophages. Liver X Receptors (LXRα and LXRβ) are nuclear receptors that control cholesterol metabolism and regulate differentiation of tissue-resident macrophages. Macrophages from rheumatoid arthritis and other inflammatory pathologies exhibit an enriched LXR pathway, and recent reports have shown that LXR activation raises pro-inflammatory effects and impairs the acquisition of the anti-Inflammatory profile of M-CSF-dependent monocyte-derived macrophages (M-MØ). We now report that LXR inhibition prompts the acquisition of an anti-inflammatory gene and functional profile of macrophages generated within a pathological environment (synovial fluid from Rheumatoid Arthritis patients) as well as during the GM-CSF-dependent differentiation of human monocyte-derived macrophages (GM-MØ). Mechanistically, inhibition of LXR results in macrophages with higher expression of the v-Maf Avian Musculoaponeurotic Fibrosarcoma Oncogene Homolog B (MAFB) transcription factor, which governs the macrophage anti-inflammatory profile, as well as over-expression of MAFB-regulated genes. Indeed, gene silencing experiments on human macrophages evidenced that MAFB is required for the LXR inhibitor to enhance the anti-inflammatory nature of human macrophages. As a whole, our results demonstrate that LXR inhibition prompts the acquisition of an anti-inflammatory transcriptional and functional profile of human macrophages in a MAFB-dependent manner, and propose the use of LXR antagonists as potential therapeutic alternatives in macrophage re-programming strategies during inflammatory responses. Supplementary Information The online version contains supplementary material available at 10.1007/s00018-023-04745-4.

We have recently described that LXR activation with synthetic ligands prompts a pro-inflammatory transcriptional profile during the M-CSF-dependent generation of monocyte-derived M-MØ [48], which suggested LXR inhibition as a potential anti-inflammatory strategy. To directly address this issue, we have now determined the transcriptome of GM-MØ generated in the presence of an LXR inhibitor (GSK2033, GSK-GM-MØ) and revealed that LXR inhibition skews the differentiation of GM-CSF-dependent monocyte-derived GM-MØ towards the acquisition of an anti-inflammatory transcriptional and functional profile. Since LXR inhibition also impairs the monocyte inflammatory differentiation under the influence of synovial fluid from Rheumatoid Arthritis patients, our results support LXR as targets for macrophage re-programming strategies during inflammatory responses.

Quantitative real-time RT-PCR (qRT-PCR)
Total RNA was extracted using the total RNA and protein isolation kit (Macherey-Nagel, #740933.250). RNA samples were reverse-transcribed with High-Capacity cDNA Reverse Transcription reagents kit (Applied Biosystems, #10400745) according to the manufacturer's protocol. Real-time quantitative PCR was performed with LightCycler® 480 Probes Master (Roche Life Sciences) and Taqman probes on a standard plate in a Light Cycler® 480 instrument (Roche Diagnostics). Gene-specific oligonucleotides were designed using the Universal Probe Library software (Roche Life Sciences). Results were normalized to the expression level of the endogenous references genes TBP, HPRT1 or GAPDH and quantified using the ΔΔCT (cycle threshold) method.

RNA-sequencing and data analysis
RNA was isolated from three independent GM-MØ samples generated from monocytes exposed to a single dose of DMSO, 1 μM GW3965, 1 μM GSK2033, or both, at the beginning of the 7-day differentiation process. In other set of experiments, RNA was isolated from two preparations of CD14 + monocytes exposed to either DMSO (vehicle) or 1 μM GSK2033 for 1 h and then cultured for 3 days in complete medium supplemented with 20% RASF (six independent samples). Sequencing was done on a BGISEQ-500 platform (https:// www. bgite chsol utions. com). RNAseq data were deposited in the Gene Expression Omnibus (http:// www. ncbi. nlm. nih. gov/ geo/) under accession GSE156696 and GSE181313. On average, 88.04 M reads per sample were generated and clean reads were mapped to the reference (UCSC Genome assembly hg38) using Bowtie2 (average mapping ratio to reference genome, 91.82%) [50]. Gene expression levels were calculated by using the RSEM software package [51], and differential gene expression was assessed by using the R-package DESeq2 algorithms using the parameters Fold Change > 2 and adjusted p value < 0.05. Heatmaps and clustering were done using the Genesis software (http:// genome. tugraz. at/ genes iscli ent/) [52]. Differentially expressed genes were analyzed for annotated gene sets enrichment using ENRICHR (http:// amp. pharm. mssm. edu/ Enric hr/) [53,54], and enrichment terms considered significant with a Benjamini-Hochberg-adjusted p value < 0.05. For gene set enrichment analysis (GSEA) (http:// softw are. broad insti tute. org/ gsea/ index. jsp) [55], gene sets available at the website, as well as gene sets generated from publicly available transcriptional studies (https:// www. ncbi. nlm. nih. gov/ gds), were used.

Statistical analysis
For comparison of means of three or more groups, statistical significance of the generated data was evaluated using a one-way repeated measures ANOVA test with a post-hoc Tukey test for multiple comparisons between groups. In all cases, p < 0.05 was considered as statistically significant.

LXR inhibition shifts the GM-CSF-dependent differentiation of GM-MØ towards the anti-inflammatory side
GM-CSF levels are increased in the serum, synovial fluid and bone marrow of patients with RA, especially at the chronic stage [63], and its relevance in RA disease development is supported by the encouraging results yielded by clinical trials targeting GM-CSF [64]. Since macrophages from the synovium of patients with active rheumatoid arthritis (GSE10500) [65] exhibit a pro-inflammatory gene profile that resembles that of GM-CSF-dependent pro-inflammatory monocyte-derived macrophages (GM-MØ) [21] (Supplementary Fig. 2A), we next sought to determine whether LXR inhibition affected the acquisition of the pro-inflammatory profile of GM-CSF-dependent macrophages. To that end, monocytes were exposed to either 1 μM GSK2033 (LXR inverse agonist), 1 μM GW3965 (LXR agonist), or both, at the beginning of the differentiation process with GM-CSF (Fig. 3A). RNAseq on the resulting macrophages   (Fig. 3F). GSK-GM-MØ showed an over-representation of genes that distinguish control from RA synovial macrophages (GSE10500) [65] (Fig. 3G), and the genes upregulated in GSK-GM-MØ greatly coincided with those upregulated in GSK/RASF-MØ (Fig. 3H). Therefore, all these results confirm that LXR inhibition impairs the acquisition of the inflammatory gene profile that characterizes GM-CSF-dependent monocyte-derived macrophages and promotes the expression of genes that define M-CSF-dependent monocyte-derived anti-inflammatory macrophages. Of note, analysis of a validation set of samples showed that the transcriptional consequences of LXR inhibition were stronger at the start of the monocyte-to-GM-MØ differentiation process (Supplementary Fig. 3A, B) and were observed with various GM-CSF exposure and dose regimens ( Supplementary Fig. 3C,D).

LXR inhibition limits macrophage pro-inflammatory and immuno-stimulatory functions
Next, we assessed the functional significance of the transcriptional changes observed upon LXR inhibition. Comparison of the cytokine profile of CNT-GM-MØ, GW-GM-MØ and GSK-GM-MØ in resting conditions indicated that GSK-GM-MØ produced significantly lower levels of activin A and CCL17 (Fig. 4A), whose expression is a characteristic property of GM-MØ. Notably, GSK-GM-MØ also produced significantly lower levels of IL-6 and IL-1β upon activation with LPS, thus supporting its lower inflammatory nature (Fig. 4B), whereas no significant effect was seen in the LPS-induced production of IL-10 and CCL2 (Fig. 4B). While comparison of the LPS-induced production of TNF and CCL8 from GSK-GM-MØ and CNT-GM-MØ did not reach statistical significance, it is worth noting that GW/ GSK-GM-MØ produced significantly less TNF than GW-GM-MØ, further supporting the anti-inflammatory effect of GSK2033 (Fig. 4B). The lower production of pro-inflammatory cytokines from activated GSK-GM-MØ, whose lower expression of ABCA1 and ABCG1 might lead to higher intracellular cholesterol levels, agrees with the diminished expression of TNF, IL1B and CXCL8 from LPS-stimulated cholesterol-loaded GM-MØ [66]. On the other hand, analysis of the allogeneic T lymphocyte-activating effect did not reveal an alteration of the T-cell stimulatory capacity of GSK-GM-MØ (Fig. 4C) albeit GW/GSK-GM-MØ exhibited a weaker stimulatory capacity than GW-GM-MØ (Fig. 4C, Supplementary Fig. 3E). These results indicate that inhibition of LXR during GM-CSF-driven macrophage differentiation leads to diminished production of pro-inflammatory cytokines and might also affect the acquisition of the macrophage immune-stimulatory capacity. Therefore, LXR inhibition shifts monocytes towards the generation of macrophages with a weaker pro-inflammatory and cytokine profile.

Molecular mechanisms underlying the macrophage programming effect of LXR inhibition: role of MAFB
Gene ontology analysis on the transcriptome of GSK/RASF-MØ indicated an over-representation of genes regulated by MAF and/or MAFB (Fig. 1D), which are master regulators for the differentiation of anti-inflammatory M-MØ [23,24,67], and a similar enrichment was observed in the GSK-GM-MØ gene profile ("Maf Mafb-Sirnas M-Csf-Derived Macrophages GSE155719" gene set, adj p, 1.160e −7 ). Thus, we analyzed the abundance of MAFB in four independent macrophage preparations. As shown in Fig. 5A, B, MAFB protein expression was much higher in GSK-GM-MØ and GW/GSK-GM-MØ, indicating that LXR inhibition during GM-MØ differentiation increases MAFB. Similarly, and although to a lower extent, MAF protein levels were also higher in GSK-GM-MØ (data no shown). In agreement with the higher expression of MAFB, the gene expression profile of GSK/RASF-MØ was significantly enriched in MAFB-dependent genes (GSE155719) (Fig. 5C, D), as well as genes upregulated in macrophages from a patient with Multicentric Carpo-Tarsal Osteolysis (MCTO, Online allogeneic CD3 + T lymphocyte), using CNT-GM-MØ generated in the presence of DMSO as a control. Mean ± SEM of 3 H-thymidine incorporation from six independent donors is shown (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.001) Mendelian Inheritance in Man #166300), a very rare autosomal dominant disorder caused by mutations within the MAFB transcriptional activation domain [68,69] that result in higher MAFB protein levels [23] (Fig. 5E). In fact, and as shown in Fig. 5F, the expression of these MAFB-dependent genes was higher in the six independent GSK/RASF-MØ samples analyzed. Therefore, exposure to an LXR inhibitor like GSK2033 results in enhanced expression of MAFB and MAFB-regulated genes in the context of either RASF or GM-CSF. These results fully agree with the predictions of gene ontology analysis and demonstrate that the antiinflammatory outcome of LXR inhibition correlates with an enhanced expression of factors that shape the transcriptional and functional profile of anti-inflammatory M-MØ.
Having demonstrated the link between LXR inhibition and enhanced MAFB protein levels, we next determined whether MAFB mediates the re-programming activity of LXR inhibition. To that end, the effect of 1 μM GSK2033 on GM-MØ differentiation was assessed upon siRNA-mediated silencing of MAFB (Fig. 6A). In this experimental set-up, 1 μM GSK2033 increased the expression of MAFB and the GSK2033-mediated upregulation of MAFB was prevented after MAFB silencing in monocytes (Fig. 6B, C). In addition, knocked-down of MAFB significantly reduced the GSK2033-mediated upregulation of IL10, LGMN and MERTK, while the reduction did not reach statistical significance in the case of CCL2, CD163, MARCKS, and GAS6 (Fig. 6D). Conversely, reduction of MAFB levels did not significantly modify the ability of GSK2033 to blunt ABCA1 expression (Fig. 6D). As a whole, these results demonstrate that the effect of the LXR inverse agonist GSK2033 on the transcriptome of human monocyte-derived macrophages is mediated, at least partly, through the augmented expression of MAFB.

Discussion
LXR nuclear receptors are lipid sensors that control crucial aspects of sterol homeostasis and regulate the expression of inflammatory mediators [reviewed in [70]]. Numerous studies have shown, mostly in mouse models, that LXR synthetic ligands suppress inflammation when administered prior to a pathogenic stimuli [70]. However, several recent reports using human cellular models have shown a link between LXR activation and enhanced inflammatory responses, and illustrated the ability of LXR to promote macrophage pro-inflammatory re-programming [35,[38][39][40][41][42][43][44][45][46][47][48]. In the present manuscript, we provide evidences that LXR inhibition favors the acquisition of an anti-inflammatory profile in human monocyte-derived macrophages generated in the presence of GM-CSF or in the context of synovial fluid from Rheumatoid Arthritis patients. The latter is especially relevant because an enhanced activity of LXR has been shown in Rheumatoid Arthritis [39,40,42], what supports the physiological relevance of the association between LXR and a pro-inflammatory state in human macrophages and the potential therapeutic value of macrophage re-programming through modulation of LXR. Of note, we have observed that the effects of LXR inhibition are directly related to, and partly mediated by, increased expression of MAFB, a transcription factor that determines the acquisition of an anti-inflammatory, pro-fibrotic and pro-tumoral state in human monocyte-derived macrophages [23,24,27,67,71,72]. Our findings are reinforced by studies showing that LXR inhibition impedes the effects of different inflammatory stimuli on macrophage function, as it blocks the upregulation of glycolytic genes in macrophages incubated with plaque homogenates [73] and the ox-LDL-dependent trained immunity in monocytes [45]. Indeed, similar effects are observed with a distinct LXR inverse agonist (SR9238) [ [45] and not shown]. Thus, and to the best of our knowledge, our results constitute the first link between LXR inactivation and promotion of a macrophage anti-inflammatory polarization.
Like the pro-inflammatory effect of the LXR agonist on human monocyte-to-macrophage differentiation [48], the reprogramming action of LXR inhibition is also cell-specific and differs among monocytes, differentiating and fully differentiated GM-MØ. The fact that LXR inhibition has more profound effects at the monocyte stage ( Supplementary  Fig. 3B) suggests that LXR factors condition the monocyte to differentiating factors like M-CSF and GM-CSF, and that LXR activity critically determines the levels/activity of factors that drive the differentiation towards M-MØ (MAF, MAFB) or GM-MØ (e.g., activin A). Indeed, we have shown that LXR inhibition affects the expression of MAF/MAFB as well as activin A at the protein level (Figs. 4,5). Interestingly, functional LXR-binding sites have been identified . ABCA1 expression was evaluated as a readout for LXR activation. Mean ± SEM of three independent experiments are shown (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001) within GM-MØ-specific genes that are upregulated by GW3965 (http:// cistr ome. org/ db/#/; CistromeDB: 69,799) [74]. On the other hand, and although the influence of LXR on MAFB can be direct or indirect, the presence of functional LXRα-binding elements within the human MAFB gene in HT29 colorectal adenocarcinoma (http:// cistr ome. org/ db/#/, CistromeDB: 69,799) [74] suggests that LXR factors might have a direct effect on the expression of genes directly controlling the inflammatory profile of monocytederived macrophages. The identification of LXR-regulated genes in M-MØ and GM-MØ by ChIP-Seq should provide a definitive answer to this question. If this postulate holds true in the case of human macrophages, modulation of LXR would constitute a tool for re-programming myeloid cells via control of the expression of regulators of differentiation and polarization.
Interestingly, our analysis of RASF-MØ revealed an upregulation of TREM2, whose expression has been associated with a subset of macrophages with a proangiogenic and immunosuppressive profile in Hepatocellular Carcinoma patients [75]. In the tumoral context, the presence of TREM2 + macrophages correlates with a worse prognosis. However, in the context of RA, macrophages with this functional profile might be beneficial to contain the immune over-activation and inflammation of the joints, thus supporting our findings that LXR inhibition skews macrophages to acquire a more anti-inflammatory phenotype.
The partial inhibition of the GSK2033 re-programming effect upon MAFB knockdown might reflect that complete MAFB silencing is required to completely abolish the consequences of LXR inhibition. Alternatively, it is conceivable that other transcriptional programs can be modulated by LXR and thus contribute to the macrophage re-programming action of LXR inhibition. In this regard, and besides the known trans-repression action of LXR on NFκB and AP-1 [76], main drivers of inflammatory macrophage polarization [77], GSK2033 might be altering the LXR ability to promote the interaction between PU.1 and IRF8 [78], two factors that affect macrophage fate and polarization [79,80]. In addition, LXR activity directly and indirectly affects the activity of SREBP1/2 transcription factors [57], whose contribution to anti- [81] or pro-inflammatory gene expression [82] may have profound effects during monocyte differentiation and macrophage activation. Therefore, in addition to the results presented here on MAFB, LXR inhibition by GSK2033 might exert further influence on pathways that are important for macrophage differentiation and activation, including IRF8 and SREBP1/2, a hypothesis that deserves further attention.
In summary, we describe the feasibility of altering the GM-CSF-driven monocyte-to-macrophage differentiation through inhibition of LXR, which impairs the generation of pro-inflammatory GM-MØ and stimulates the acquisition of an anti-inflammatory transcriptional and functional profile. Taken together with the reverse effects reported for LXR activators [48], these results indicate that modulation of LXR might be a target for macrophage re-programming strategies in pathological conditions. Indeed, pharmacological efforts towards the design LXR isoform-specific, or tissue-restricted, agonists have been conducted in recent years and successfully tested in pre-clinical models of lipid disorders or cancer [83][84][85][86]. Thus, the design of LXR modulators for macrophage re-programming purposes in inflammatory diseases would be feasible, selecting those ligands with ability to modify the macrophage inflammatory state without compromising cholesterol metabolism. In addition, our results suggest that LXR-dependent macrophage genes might be useful prognostic/therapeutic markers for human inflammatory diseases, a property that has already been demonstrated for NR1H3 in the case of diffuse large B-cell lymphoma [47].
Author contributions All authors contributed to the study conception and design. AGA, CH, MT-T, MTS, AdC, and BA performed research and analyzed data; AGA, APK, MAV, AC, and ALC designed the research and analyzed data; AGA, AC, and ALC wrote the paper.

Data availability
Datasets generated during the current study are available in the Gene Expression Omnibus (http:// www. ncbi. nlm. nih. gov/ geo/) under accession GSE156696 and GSE181313.

Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.
Ethics approval Biological samples were obtained from the Rheumatology Unit at Hospital General Universitario Gregorio Marañón, after patients had provided informed consent. Approval was obtained from the ethics committee of Hospital General Universitario Gregorio Marañón. The procedures used in this study adhere to the tenets of the Declaration of Helsinki.

Consent to publish Not applicable.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.