Article Text
Abstract
Objectives Macrophages are central to the inflammatory processes driving rheumatoid arthritis (RA) synovitis. The molecular pathways that are induced in synovial macrophages and thereby promote RA disease pathology remain poorly understood.
Methods We used microarray to characterise the transcriptome of synovial fluid (SF) macrophages compared with matched peripheral blood monocytes from patients with RA (n=8).
Results Using in silico pathway mapping, we found that pathways downstream of the cholesterol activated liver X receptors (LXRs) and those associated with Toll-like receptor (TLR) signalling were upregulated in SF macrophages. Macrophage differentiation and tumour necrosis factor α promoted the expression of LXRα. Furthermore, in functional studies we demonstrated that activation of LXRs significantly augmented TLR-driven cytokine and chemokine secretion.
Conclusions The LXR pathway is the most upregulated pathway in RA synovial macrophages and activation of LXRs by ligands present within SF augments TLR-driven cytokine secretion. Since the natural agonists of LXRs arise from cholesterol metabolism, this provides a novel mechanism that can promote RA synovitis.
- Cytokines
- Rheumatoid Arthritis
- Inflammation
- Synovial Fluid
Statistics from Altmetric.com
Introduction
Rheumatoid arthritis (RA) is a debilitating autoimmune inflammatory condition of unknown aetiology. It affects approximately 1% of the population and is associated with increased morbidity, reduced life expectancy, a heightened degree of social burden and economic cost. The primary site of inflammation is the synovium, characterised by hyperplasia and vascularisation, inflammatory cell infiltration, hypoxia and destruction of adjacent cartilage and bone. This ultimately leads to irreversible destruction of the joint and impaired mobility. Although the use of biological therapeutics has considerably improved clinical outcomes and prognosis, they remain effective in only a proportion of patients and rates of long-term remission achieved remain low. There is therefore an ever greater need to understand the cellular and molecular processes by which the pathology is mediated to develop future treatments.
Macrophages constitute the major leucocyte population within the synovial inflammatory infiltrate (≥40%).1 Synovial macrophage numbers directly correlate with measures of disease activity and severity including C-reactive protein, erythrocyte sedimentation rate, swollen joint count, synovial lining layer vascularity and thickness, the presence of citrullinated peptides and radiological score.2 ,3 Furthermore, the number of synovial lining layer CD68 macrophages provides a useful biomarker of response upon successful clinical intervention in the context of clinical trials.4–7 As macrophages are a major source of interleukin 6 (IL-6) and tumour necrosis factor α (TNFα), the recent success of anticytokine therapeutics supports the notion that macrophage-targeted therapies may be beneficial for the treatment of RA,8 and highlights a central role for macrophages in the progression of human disease pathology.
There are multiple mechanisms by which synovial macrophages may contribute towards the inflammatory burden and joint destruction in RA.9 Proinflammatory cytokines and immune complexes present in plasma induce early activation of CD14 monocytes, upregulation of integrins and chemokine receptors (eg, CCR1 and CCR2) and transendothelial migration of monocytes into the synovium. Monocytes accumulate in the synovium where high concentrations of macrophage colony stimulating factor (M-CSF) and granulocyte-macrophage colony stimulating factor (GM-CSF) enhance monocyte to macrophage differentiation and maturation.
The advent of transcriptional array techniques allows definitive characterisation of the pathways that are active within a given cell population. We therefore adopted a microarray-based approach to elucidate the molecular pathways that are activated in RA synovial fluid (SF) macrophages to identify novel pathways that could drive disease pathology. Using this approach, we now report the unexpected prominence of the cholesterol-activated liver X receptor (LXR) pathway that integrates with Toll-like receptor (TLR) pathways to enhance synovial cytokine secretion.
Methods
Reagents
GW3965 (Merck, UK) was dissolved in dimethyl sulfoxide.
Tissues and patient samples
Blood, SF and synovial membrane samples were obtained from patients with RA after obtaining their written informed consent. Patient numbers and characteristics are shown in online supplementary table S1. All samples were processed immediately upon arrival by centrifugation on a histopaque density gradient to separate out the peripheral blood mononuclear cells followed by MACS CD14 positive selection.10 The samples were lysed in Trizol and stored at −70°C prior to RNA extraction. Cell purities were typically >95% (data not shown).
RNA purification and analysis
Single cell suspensions were lysed in buffer RLT or Trizol and the RNA extracted following the RNeasy micro kit (Qiagen). Gene expression was analysed using SYBR green quantitative real-time PCR master mix on an ABI7900HT and SDS 2.3 software (Applied Biosystems). Gene expression was normalised to TATA binding protein or GAPDH. Primer sequences are shown in online supplementary table S2.
Microarray analysis
The integrity of RNA was ensured by analysis of ribosomal 18S and 28S RNA intensity using an Agilent 2100 bioanalyser (Agilent Technologies). Following genomic DNA digestion, cDNA was synthesised, fragmented and biotin-labelled (FL-Ovation cDNA Biotin Module V2 kit, Nugen Technologies). An Affymetrix GeneChip Human Genome U133 Plus 2.0 array was performed using a GeneChip Scanner 3000. Background correction and normalisation for each probe set on the GeneChips was determined by the RMA algorithm using R and Bioconductor and quality assured using ‘arrayQualityMetrics’.11 Further analysis and differential gene expression studies were carried out using ‘oneChannelGUI’. A predictor of false positive (pfp) value of 0.05 was chosen to determine differentially expressed genes. Pathway analysis was performed using Ingenuity Pathway Analysis (IPA; Ingenuity Systems). The microarray data produced in this study is MIAME compliant (http://www.mged.org/Workgroups/MIAME/miame.html) and will be submitted to the ArrayExpress database (http://www.ebi.ac.uk/arrayexpress).
Cell culture
Cells were cultured at a density of 1.0×105/well in a 96-well plate (Corning) and matured to a macrophage phenotype as previously described.10 ,12 Cells were preincubated with GW3965 or vehicle at the indicated concentration for 48 h prior to addition of TLR ligands: 100 ng/ml ultra pure lipopolysaccharide (LPS; Calbiochem), 10 µg/ml lipoteichoic acid (LTA), 10 µg/ml Pam3CSK4 or 1 µg/ml CL097 (Autogen) for 24 h.
Cytokine analysis
The concentration of cytokines in cell culture supernatants was analysed by ELISA or luminex as previously described.10
Immunofluorescence
Five μM sections were deparaffinised, rehydrated and epitope retrieval was performed using citrate buffer followed by incubation with 20% horse serum. Primary antibodies were incubated overnight at 4°C in 5% serum/TBST at final concentration of 2.5 μg/ml LXRα (Abcam; clone PPZ0412), 2.6 μg/ml LXRβ (Santa Cruz; polyclonal), 2.5 μg/ml CD68 (Dako; clone PG-M1) or isotype controls (Dako). Biotinylated secondary antibodies were incubated for 30 min in 5% serum/TBST followed by incubation with Avidin-D fluorochrome conjugates (Vectorlabs; LXRα/β, Fluorescein or CD68, Texas Red) for 45 min in phosphate buffered saline. Slides were mounted in Vectashield containing 4′,6-diamidino-2-phenylindole (Vectorlabs) and visualised under a fluorescent microscope (Axiovert S100 and Openlab software).
Statistical analysis
Results are displayed as mean±SD. Statistical analysis was performed by the paired Student t test, Wilcoxon paired t test or two-way ANOVA using Graph Pad Prism 4 software. The significance of association between the differentially expressed genes and canonical pathways within IPA was determined using the Fisher exact test and corrected using the Benjamini–Hochberg correction for multiple testing.
Results
Analysis of the SF-derived macrophage transcriptome
To determine the gene expression profile that defines an RA SF macrophage, microarray analysis was performed on sorted SF macrophages and the transcriptome compared with that of matched peripheral blood monocytes obtained from eight subjects with RA. Only genes that were significantly differentially expressed with p≤0.05 between the blood and synovium in all eight donors were considered for further analysis. From a total of 54 675 detectable transcripts, 8303 were significantly differentially expressed between the blood and the synovium (figure 1A,B). We examined the transcriptional relationships between SF macrophages and peripheral blood monocytes by principal component analysis. Importantly, the transcriptional profile for each of the sample populations clustered together but separately from each other indicating conservation of patterns of gene expression in each group but that SF macrophages have a transcriptionally distinct profile of gene expression from that of the peripheral blood monocytes (figure 1C). It is therefore likely that the RA synovial microenvironment induces specific transcriptional changes upon entry of monocytes into the synovium that may drive disease pathology.
The LXR pathway is highly upregulated in RA SF macrophages
We next examined the canonical biological pathways most associated with this transcriptional profile using IPA. The pathway most significantly induced in SF macrophages was the LXR/retinoid X receptor (RXR) nuclear receptor activated pathway (see online supplementary table S3 and figure 2A). By microarray, LXRα expression was upregulated by approximately 2.1-fold while the expression of LXRβ was not changed (figure 2B and online supplementary table S4). LXRs are nuclear receptor transcription factors that, upon activation by oxidised cholesterol derivatives, drive the expression of a large variety of transcriptional target genes. IPA analysis also revealed increased expression of known downstream target genes, particularly ATP binding cassette (ABC) A1 and ABCG1, apolipoprotein (Apo) C1, Apo C2, lipoprotein lipase (LPL) and phospholipid transfer protein (PLTP) (figure 2A and online supplementary table S4).
Confirmation of LXR activation and induction of downstream gene expression
To confirm the microarray results, we used quantitative real-time PCR to analyse changes in the level of gene expression. Consistent with the microarray results, the expression of LXRα was significantly higher in SF macrophages than in matched peripheral blood monocytes in all donors (figure 3). In contrast, the expression of LXRβ was significantly downregulated. Furthermore, we confirmed that the expression of ABCA1, ABCG1, LPL, PLTP, ApoE, ApoC1 and ApoC2 were significantly increased (figure 3), which suggests that the level of LXR activation is increased in monocytes upon entry into the RA synovial microenvironment.
LXR protein is present in synovial macrophages
We next sought to demonstrate the presence of LXRα and LXRβ protein particularly within synovial macrophages. Sections derived from RA synovial membrane biopsies were stained with antibodies against either LXRα or LXRβ in combination with CD68. We detected LXRα (40%) and LXRβ (36%) positive macrophages in RA synovial membranes (figure 4A–C).
Macrophage differentiation upregulates LXRα expression
Migration into the synovium induces the differentiation of monocytes into macrophages.13 ,14 We therefore examined the level of LXRα expression during the differentiation of healthy peripheral blood monocytes to a macrophage phenotype. In all four donors tested, LXRα was significantly increased 10–150-fold over 6 days (figure 5A). In contrast, LXRβ was significantly downregulated by approximately twofold (figure 5B) whereas the basal level of ABCA1 expression, as a reporter of LXR activation, was not significantly different between monocytes and macrophages but could be increased by addition of the LXR agonist GW3965 (figure 5C). Together these results suggest that, whereas macrophage differentiation can alter the relative level of LXRα and LXRβ expression, it is not sufficient alone to induce activation of the LXR pathway.
Studies have previously shown that administration of GW3965 to LPS-activated monocytes or macrophages potentiates the secretion of proinflammatory cytokines.10 ,15 To determine if the difference in the level of LXR expression between monocytes and macrophages affects the level of subsequent cytokine secretion, we treated syngeneic monocytes and macrophages for 48 h with GW3965 followed by stimulation with 100 ng/ml LPS. After 24 h the concentration of TNFα in cell culture supernatants was measured by ELISA. In agreement with previous observations, LXR agonism significantly increased the secretion of TNFα from LPS stimulated monocytes (figure 5D).10 Compared with monocytes at the same concentration of GW3965, macrophages secreted significantly higher levels of TNFα in response to stimulation with LPS, suggesting that the higher level of LXR expression in macrophages enhances inflammatory cytokine secretion.
TNFα augments LXRα expression in human macrophages
Differentiation of monocytes to macrophages enhances the level of LXR expression. However, TNFα is central to the pathology of RA and has previously been shown to increase the expression of LXRα in rabbit adipocytes.16 To test whether TNFα affects LXR expression in human macrophages, we treated M-CSF matured macrophages with 25 ng/ml TNFα for 4 h. The addition of TNFα modestly but significantly increased LXRα expression twofold whereas the expression of LXRβ and ABCA1 was unchanged (figure 5E–G). Treatment of macrophages with concentrations of IL-6 up to 100 ng/ml did not change the expression of LXRα, LXRβ or ABCA1 (data not shown).
LXR activation potentiates TLR-driven cytokine secretion in human macrophages
Our microarray analysis also showed that the expression of genes known to be involved in the inhibition of RXR function were highly significantly upregulated in SF macrophages (see online supplementary table S3). This pathway consisted mainly of genes that are downstream of TLRs and that are transcriptionally upregulated upon TLR ligation. Furthermore, TLRs have been widely implicated as potential drivers of RA disease pathology. In particular TLR2, TLR4, TLR7 and TLR8 have all been shown to be expressed at higher levels in RA synovial macrophages and potential exogenous and endogenous TLR ligands have been identified in RA SF.17–21 This was of particular interest as LXR agonism has previously been shown to potentiate macrophage cytokine secretion induced by TLR4 ligation with LPS (figure 5D).10 ,15 However, the effect of LXR agonism upon cytokine secretion induced by ligation of TLR2, TLR7 and TLR8 is unknown. Human M-CSF matured macrophages were cultured in the presence of GW3965 for 48 h prior to stimulation with TLR ligands LPS (TLR4) and CLO97 (TLR7/8). TLR2 (TLR1/2 or TLR2/6) can be stimulated with either PAM3CSK4 or LTA; however, these ligands exert differential effects in macrophages and therefore both of these TLR2 ligands were used in parallel.22 In accordance with our previous findings, LXR agonism by addition of GW3965 significantly increased the secretion of TNFα from LPS stimulated human macrophages in a dose-dependent manner (figure 6A). Furthermore, LXR agonism augmented the secretion of TNFα from macrophages stimulated with ligands for TLR2 (figure 6B,C) or TLR7/8 (figure 6D). Luminex analysis of cell culture supernatants showed that this was not specific to TNFα as the secretion of other proinflammatory cytokines (IL-1β, IL-6, IL-12 and G-CSF) and inflammatory chemokines (MIP-1α (CCL3) and MIP-1β (CCL4)) that are typically secreted upon TLR ligation were also increased from macrophages stimulated with LPS (figure 6E), LTA (figure 6F), PAM3CSK4 (figure 6G) and CL097 (figure 6H). There was a significant increase in IL-1 receptor antagonist but no significant difference in the concentration of IL-10 (figure 6F–H and data not shown).
Discussion
By adopting an hypothesis-free approach, we have used microarray technology to elucidate the molecular pathways that are induced within RA SF macrophages. Importantly, we have shown that the transcriptome of SF macrophages differs considerably from peripheral blood monocytes. Furthermore, we have shown that the LXR pathway is highly induced and is a novel potential driver of RA disease pathology that is in part mediated by augmentation of TLR-induced cytokine secretion.
Our studies show that the expression of LXRα is increased during macrophage differentiation in vitro while the expression of LXRβ is downregulated. This can be further augmented by the proinflammatory cytokine TNFα which is a hallmark of the inflamed synovium. LXRs are activated by a variety of oxidised cholesterol derivatives.23 It is therefore interesting to find that a lipid-activated pathway is the most upregulated pathway in SF macrophages as RA is associated with dyslipidaemia—that is, high serum cholesterol and triglycerides and an atherosclerotic-like phenotype.24–26 Furthermore, several studies have shown that the concentration of cholesterol and the cholesterol transport lipoproteins are elevated in SF.27 Further studies are therefore required to characterise the synovial lipidome in detail. However, we speculate that the elevated levels of cholesterol in the synovium may lead to the subsequent induction of the LXR pathway, which may drive synovial inflammation. In agreement with this, we have previously shown that dual activation of LXRα and LXRβ greatly enhances the onset and severity of disease in a murine model of collagen-induced arthritis.10 ,28 Taken together, these studies demonstrate a potential proinflammatory effect of LXR activation and show for the first time that LXR-activated pathways contribute a major role in human pathology by driving inflammatory cytokine secretion which may potentiate the progression of synovitis.
The mechanism(s) by which LXR activation drives RA disease pathology are unknown. However, the microarray analysis revealed that the pathways induced by TLR ligation were highly increased in the SF macrophages, which is consistent with similar observations in the literature.18 ,29 The TLRs expressed on macrophages bind a variety of viral and bacterial derived products such as bacterial lipoproteins (TLR1/2/6), LPS (TLR4) and single-stranded RNA (TLR7/8). While bacterial cell wall fragments, peptidoglycan and double-stranded DNA have been identified within the synovium, it is now well recognised that TLRs can be activated by endogenous self proteins such as heat shock proteins and double-stranded RNA released from necrotic synoviocytes.18 ,30 LXR activation is known to potentiate cytokine secretion from LPS activated human macrophages; this is in part achieved through increased expression of TLR4.10 ,15 Here we have extended these studies and shown that activation of the LXR pathway also leads to a dramatic increase in cytokine secretion driven by TLR1/2, TLR2/6 and TLR7/8. These results are of particular interest as the expression of TLR2 is upregulated upon differentiation of monocytes to macrophages which can lead to the enhancement of Th17 cells.31 ,32 Furthermore, such studies suggest that TLR-induced cytokine secretion in RA SF macrophages is mediated mainly through TLR2. However, although we have confirmed that LXR activation increases the expression of TLR4, the expression of TLR1, TLR2, TLR6, TLR7 and TLR8 were not changed (data not shown). The mechanism by which LXR activation promotes cytokine secretion induced by ligation of TLR1/2, TLR2/6 and TLR7/8 is therefore unknown.
Overall, our results support the hypothesis that the dyslipidaemia associated with arthritis may enhance the inflammatory aspect of the disease, and that this may be mediated in part through activation of the LXRs. Furthermore, our data help to support the clinical finding that reducing the atherosclerotic burden in RA may ameliorate the inflammatory aspect of the disease and improve the long-term prognosis.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online tables
Footnotes
-
Handling editor Tore K Kvien
-
DLA and LEB contributed equally.
-
Contributors DLA, LEB, JSN, MKM, SP, PBW, JHR, SK, MK, JAG contributed towards the provision of samples, analysis and interpretation of data, study design(s), the critical revision of the article and the intellectual content. DLA, LEB, JSN, IBM also drafted the manuscript. All authors approved the final version of the manuscript.
-
Funding Funding was provided by MASTER SWITCH European community grant, Medical Research Council (UK), the Nuffield Foundation Oliver Bird Rheumatism programme, Arthritis Research UK and the Iraqi Ministry of Higher Education and Scientific Research.
-
Ethical approval This study complied with the World Medical Association Declaration of Helsinki and was approved by the Glasgow East ethics committee.
-
Competing interests SP was employed by GlaxoSmithKline.
-
Provenance and peer review Not commissioned; externally peer reviewed.