Expression and regulation of CCL18 in synovial fluid neutrophils of patients with rheumatoid arthritis

SF samples were taken from knees of nine patients with active RA, for treatment and diagnostic purposes. Aliquots of these samples and EDTA-treated blood from these patients were used in this study after informed consent had been obtained. Four patients were receiving anti-TNF-α therapy, three were being treated with conventional therapy (nonsteroidal anti-inflammatory drugs, steroids or disease-modifying anti-rheumatic drugs), and two patients were receiving no medication at the time of synovial effusion although those had been treated earlier (Table 1). Hence, all patients did not respond sufficiently to the therapeutic treatment.

Purification of mononuclear cells (peripheral blood mononuclear cells; PBMC) and PMN from blood was performed as described [21], with minor modifications. For sedimentation of erythrocytes, 1 volume of EDTA-treated blood was incubated with 1 volume of 3% dextran T500 (Roth, Karlsruhe, Germany) in PBS (Biochrom, Berlin, Germany) for 20 minutes at 4°C. The leucocyte-rich supernatant was centrifuged at 500 g for 10 minutes at 6°C in a Multifuge (Heraeus, Hanau, Germany). The pellet was resuspended in PBS, overlaid on an isotonic discontinous Percoll gradient (Amersham Biosciences, Freiburg, Germany) with densities of 1.075 g/ml and 1.09 g/ml and centrifuged at 750 g for 25 minutes at 6°C. PBMC or PMN were collected at the relevant interphase and washed twice with PBS. With the exception of dextran sedimentation, the preparation of SF PMN was performed similarly. Cell-free SF was obtained by centrifugation of SF at 750 g for 25 minutes and stored at -70°C in aliquots until use. PMN and mononuclear cells were counted in SF from patients with RA. The ratio between PMN and mononuclear cells was approximately 1:10.

The purity of PMN was routinely analyzed by flow cytometry with FACSCalibur (BD Biosciences, Heidelberg, Germany). In brief, to prevent nonspecific binding, 3 × 10⁵ cells were preincubated for 10 minutes with heat-inactivated (20 minutes, 56°C) human serum and were then incubated with a combination of fluorescein isothiocyanate-conjugated CD66b antibody (Immunotech, Hamburg, Germany) and allophycocyanin (APC)-conjugated CD14 antibody (Caltag, Hamburg, Germany). After 30 minutes at 4°C, cells were washed twice with PBS/1% FCS (Sigma, Deisenhofen, Germany) and kept on ice until analysis. Data were analyzed with CellQuest software (BD Biosciences), revealing a purity of 98 to 99% CD66b-positive PMN with a contamination of less than 0.05% CD14-positive and CD66b-negative cells. Analysis of PMN preparations of three donors with anti-CD56-APC (BD Biosciences), anti-CD3-phycoerythrin (Caltag) and anti-CD19-APC (Caltag) revealed 0.01 to 0.04% natural killer cells, 0.06 to 0.33% T cells and up to 0.06% B cells. In addition, PMN preparations were subjected to histochemistry (Diff-Quick staining; Dade Behring, Düdingen, Switzerland) showing 1 to 3% eosinophils in each preparation of blood PMN. Selected cell culture experiments were performed with PMN preparations depleted of eosinophils with anti-CD16 antibodies by means of magnetic-activated cell sorting (Miltenyi Biotec, Bergisch Gladbach, Germany) to exclude the possibility that contaminating eosinophils accounted for CCL18 production. Notably, no eosinophils were detectable in preparations of SF PMN.

For culturing PMN, RPMI 1640 containing L-glutamine and sodium bicarbonate (Sigma) was supplemented with 5 mM HEPES (Sigma), 100 IU/ml penicillin, 100 μg/ml streptomycin (Sigma) and 10% FCS. For pretreatment of PMN with SF, blood PMN from healthy donors were seeded in six-well-plates (Corning Costar, Bodenheim, Germany) at a density of 5 × 10⁷ cells in 5 ml of culture medium or 2.5 ml of culture medium plus 2.5 ml of SF from patients with RA. After 10 hours of incubation, cells were washed twice. Pretreated blood PMN as well as freshly isolated blood PMN from healthy donors or blood and SF PMN from patients with RA were seeded in 24-well plates (Greiner Bio-One GmbH, Frickenhausen, Germany) at a density of 5 × 10⁶ cells in 500 μl of culture medium with and without 20 ng/ml IL-10 (Endogen, Eching, Germany), 10^-7 M vitamin D₃ (1,25-dihydroxycholecalciferol; Calbiochem, Darmstadt, Germany) and different concentrations (10, 1, 0.5 and 0.25 ng/ml) of recombinant TNF-α (Biolegend, Eching, Germany). In addition, for co-culture experiments, 5 × 10⁵ EA-hy.926 human endothelial cells (Department of Experimental Pathology, St Bartholomew's and the Royal London School of Medicine, London) were plated in 24-well plates in DMEM with 4,500 mg/l glucose, L-glutamine, sodium bicarbonate and pyridoxine hydrochloride (Sigma) supplemented with 100 IU/ml penicillin, 100 μg/ml streptomycin and 10% FCS. After 1 hour of incubation at 37°C, 5% CO₂ and 95% humidity, 5 × 10⁵ PBMC or 5 × 10⁶ blood PMN from healthy donors were added to a final volume of 500 μl of DMEM/10% FCS. Additionally PMN and EA-hy.926 cells were cultured in Boyden chambers, separated by a transwell membrane with a pore size of 0.4 μm (Corning Costar). For control experiments, blood PMN or EA-hy.926 cells were γ-irradiated (40 Gy), washed twice after 24 hours of incubation and co-cultured with vital EA-hy.926 cells or PMN, respectively. After 24 or 48 hours of incubation, cells and supernatants were prepared for subsequent isolation of RNA or ELISA analysis.

For microarray analysis and RT-PCR, total RNA was extracted from PMN with RNeasy Mini Spin Columns (Qiagen, Hilden, Germany) in accordance with the manufacturer's instructions. From cell cultures, RNA was prepared with the acid-phenol extraction procedure [22]. All RNA samples were digested with DNAse (DNA-free™ Kit; Ambion, Huntingdon, UK). RNA yields were determined spectrophotometrically by measuring the absorbance at 260 and 280 nm. All RNA samples used for microarray analysis were analyzed with an RNA 6000 Nano Labchip (Agilent Technologies, Böblingen, Germany) for RNA degradation. Only RNA preparations with no detectable degradation were used for microarray analysis.

Experiments were performed with the Lab-Arraytor^®-60 inflammation microarray (SIRS-Lab, Jena, Germany) comprising 800 probes (each measurement being made in triplicate) addressing 780 transcripts corresponding to inflammation as well as 20 control probes. The microarray data according to Minimal Information about a Microarray Experiment (MIAME) guidelines were deposited in the database ArrayExpress [23] with the accession number E-MEXP-994. RNA samples of SF PMN from nine RA patients and of blood PMN from four healthy donors (Table 1) were amplified with BD Atlas™ SMART™ Fluorescent Amplification Kit (BD Biosciences) in accordance with the manufacturer's instructions. cDNAs were cleaned with a Promega Wizard PCR clean-up Kit (Promega, Mannheim, Germany) and labeled by use of the Dyomics DY-648-S-NHS/DY-548-S-NHS dye system (Dyomics, Jena, Germany). Dyomics DY-648-S-NHS-labeled cDNA was cohybridized with DY-548-S-NHS-labeled cDNA obtained from the same amount of total RNA isolated from the immature monocytic cell line SigM5 obtained from the German Resource Centre for Biological Material (DSMZ, Braunschweig, Germany) and subjected to culture under standard conditions. After incubation in a hybridization apparatus (HS 400; TECAN Group Ltd, Männedorf, Switzerland), for 10 hours at 42°C in a formamide-based hybridization buffer system, arrays were washed in accordance with the manufacturer's instructions and dried; hybridization signal intensities were measured immediately with an array scanner (model Gene Pix 4000B; Axon Instruments, Union City, CA, USA).

Digital images resulting from post-hybridization array scanning were quantified with GenePix Pro 4.0 software (Axon Instruments). GenePix™ Analysis Software was used for spot detection and quantification and also for spot quality flagging. Significantly regulated genes were determined by using a two-sample permutation test. The threshold of significance for multiple comparisons was defined with calculating the corresponding q value for each p value [24]. For normalization and variance-stabilized transformation of raw signals the method of Huber and colleagues [25] was used, in which the data were transformed by an arcsinh function; for example, a transformed ratio of ± 0.4 corresponds to approximately 1.5-fold change (almost identical with the natural logarithm). Genes with mean transformed ratios less than -1 and larger than +1 were significantly regulated. Genes were clustered with the Database for Annotation, Visualization and Integrated Discovery (DAVID) software [26].

Changes in gene expression assessed by microarray analysis were confirmed by semiquantitative and quantitative RT-PCR for selected genes, namely CXCL10, CCL18, CCRL2 and CXCR4. In brief, cDNA was synthesized from 0.5 μg of total RNA by using 0.5 μg of oligo(dT) 16-mer primer (Thermo, Ulm, Germany), 0.5 mM dNTP (Invitrogen, Karlsruhe, Germany) and 1 unit (U) of Omniscript reverse transcriptase (Qiagen) in a final volume of 25 μl at 37°C for 60 minutes and at 93°C for 5 minutes. PCR was performed in a 25 μl reaction mixture containing 2 μl of cDNA sample, 0.625 U of Taq polymerase (Qiagen), 0.2 mM dNTPs (Invitrogen) and sense and antisense primers (each at 0.4 μM). Amplification was performed at 95°C for 2 minutes followed by 34 cycles with a denaturing step at 94°C for 1 minute, an annealing step at 55°C for 50 s and an extension step at 72°C for 90 s. A final extension step was performed at 72°C for 7 minutes on a thermocycler (Biozym, Hess. Oldendorf, Germany). Primers for CCRL2 (5'-TGA CAA GTA TGA CGC CCA G-3' and 5'-ACC AGG ATA AGC ACA ACC AG-3'), CCL18 (5'-CTC CTT GTC CTC GTC TGC AC-3' and 5'-TCA GGC ATT CAG CTT CAG GT-3'), CXCR4 [27], CXCL10 [28] and β-actin [29] were purchased from MWG Biotech (Ebersberg, Germany). PCR samples were separated on 2% agarose gels (Roth), revealed by ethidium bromide staining (Merck, Darmstadt, Germany) and photographed on an ImageMaster VDS (Amersham Biosciences).

Quantitative PCR was performed in a 20 μl reaction mixture containing 2 μl of cDNA (diluted 1:5 in water), 1 U of Platinum-Taq-Polymerase (Invitrogen), 0.5× SYBR-Green (Roche, Mannheim, Germany), 5% dimethylsulfoxide (Sigma), 0.5 mg/ml bovine serum albumin (New England Biolabs, Frankfurt, Germany), 0.25 mM dNTP (Amersham), 4 mM magnesium chloride (Invitrogen) and sense and antisense primers (each at 0.4 μM). Amplification was performed with a real-time light cycler (Roche) at 95°C, 10 minutes, 20°C/s of preincubation followed by 50 cycles of denaturing at 95°C, 15 s, 20°C/s, annealing at 55°C (CCL18) and 56°C (HPRT), 10 s, 20°C/s and extension at 72°C, 15 s, 20°C/s. The melting curve was performed at 95°C, 0 s, 20°C/s followed by 65°C, 15 s, 20°C/s and 95°C, 0 s, 0.1°C/s. Cooling was done at 40°C, 30 s, 20°C/s. CCL18 values were normalized with HPRT values and are presented as relative gene expression ratios with the 2^-ΔΔCt method [30]. CCL18 primer (5'-GGG GGC TGG TTT CAG AAT A-3' and 5'-CTC CTT GTC CTC GTC TGC AC-3') and HPRT primer (5'-GAC TTT GCT TTC CTT GGT CA-3' and 5'-GGC TTT GTA TTT TGC TTT TCC-3') for quantitative RT-PCR were purchased from MWG Biotech. All primer sequences correspond to sequences for human cDNAs deposited in GenBank.

CCL18 and TNF-α levels in culture supernatants were quantified with human CCL18 and TNF-α ELISA (DuoSet^® ELISA Development System; R&D Systems, Wiesbaden-Nordenstadt, Germany) in accordance with the manufacturer's instructions. Substrate reagents A and B were from BD Biosciences. The sensitivity of both assays was 5 pg/ml.

All experiments were performed at least three times. The values are expressed as means ± SEM. A Wilcoxon test was used to assess the significance of differences between two conditions. All p values are two-tailed, and p < 0.05 is considered significant. Statistical analysis was performed with the software package provided by Prism 3.0^®.

RNA preparations were obtained from SF PMN from nine individual patients with RA and from peripheral blood PMN from four healthy donors (Table 1) and subjected to microarray analysis. Results summarized in Figure 1 show a highly upregulated expression of CCL3 and CCL18 mRNAs, a downregulation of CCL21 mRNA (Figure 1a) and an upregulation of CXCL8 and CXCL10 mRNAs (Figure 1b) in SF PMN from each patient in comparison with blood PMN from healthy donors. As revealed by flow cytometry with CD66b antibodies, SF PMN form one representative RA patient prepared for microarray experiments were of 99% purity. Natural killer cells, T cells, B cells and monocytes/macrophages were not detectable (Figure 1c). To validate the microarray data, semiquantitative RT-PCR was performed with the same RNAs as used in the microarray analysis of four randomly selected RA patients (namely nos 1, 2, 7 and 8) and one of the healthy donors (namely no. 2). As shown in Figure 2a, mRNA expression for two selected chemokines, CCL18 and CXCL10, was readily detectable in SF PMN but remained undetectable in peripheral blood PMN from the healthy donor. In addition, quantitative RT-PCR revealed changes in CCL18 expression in SF PMN from RA patients (n = 9; Figure 2b) similar to those observed with microarray analysis. These data indicate that the microarray analysis reflects a true transcriptional profile of chemokines in SF PMN. Differences observed in CCL18 mRNA expression levels in SF PMN between individual RA patients did not relate to the actual therapeutic treatment (p = 0.2612, none versus conventional therapy; p = 0.3798, none versus anti-TNF therapy) as determined by quantitative RT-PCR (data not shown). Notably, no significant levels of CCL18 mRNA were detectable in blood PMN form RA patients (n = 9; Figure 2b), suggesting that CCL18 gene expression in SF PMN occurs as a result of the recruitment of PMN into the inflammatory milieu of the joint.

To estimate changes in the responsiveness of SF PMN to chemokine ligands, expression levels of chemokine receptor mRNAs were determined by microarray analysis as described above. Results in Figure 3a show an upregulation of CXCR4, CCRL2 and CCR5 in SF PMN from patients with RA in comparison with blood PMN from healthy donors. For many other receptors, particularly the neutrophil receptors CXCR1 and CXCR2, mRNA levels were downregulated. When RNA from SF PMN obtained from RA patient nos 1, 2, 7 and 8 and healthy donor no. 2 was used, semiquantitative RT-PCR (Figure 3b) revealed an upregulation of two selected chemokine receptors, CCRL2 and CXCR4, in SF PMN, thereby confirming the results of the microarray analysis. These data therefore indicate that chemokine responsiveness of SF PMN is regulated, at least in part, through the differential regulation of chemokine receptor expression.

To determine whether SF PMN might contribute to CCL18 protein levels in inflamed joints, supernatants of cultured SF PMN from patients with RA were subjected to ELISA analysis. The results (Figure 4a) showed no spontaneous release of CCL18 by SF PMN. However, CCL18 could be induced by stimulation of SF PMN with IL-10; indeed, IL-10 in combination with vitamin D₃ even enhanced CCL18 release. In contrast, SF PMN from patients with RA did not release CCL18 in response to TNF-α. In addition, the effect of IL-10 in combination with TNF-α was not significantly different from that of IL-10 alone, suggesting that TNF-α may not be involved in CCL18 protein synthesis. Because IL-10 is frequently found in SF from patients with RA [31], PMN recruited into joints are likely to contribute to the local production of CCL18. Furthermore, because SF PMN from patients with RA were shown to express CCL18 mRNA in the absence of detectable CCL18 protein, it is evident from these data that the synthesis of CCL18 protein requires a secondary signal.

To identify regulatory mechanisms involved in CCL18 expression, a putative role of TNF-α as a sensitizing agent was investigated in cultures of blood PMN from healthy donors. The results shown in Figure 4b confirmed TNF-α as a potent inducer of CCL18 mRNA expression. However, ELISA analysis of CCL18 levels in supernatants of the same cell cultures revealed no detectable CCL18 protein after 24 or 48 hours of incubation (data not shown), indicating that TNF-α induced CCL18 mRNA expression in the absence of protein synthesis. Interestingly, as shown in Figure 4a, neither blood PMN from healthy donors nor blood PMN from patients with RA released significant amounts of CCL18 in response to IL-10 alone or to a combination of IL-10 and TNF-α, suggesting that TNF-α alone may not be sufficient to prime PMN for CCL18 synthesis. This conclusion is supported by data showing that CCL18 production could be induced by stimulating PMN with IL-10 and vitamin D₃ in the absence of exogenous TNF-α (Figure 4a). To investigate whether the difference between SF PMN and blood PMN in their responsiveness to IL-10 was due to soluble factors present in SF, blood PMN from healthy donors were pretreated with SF from patients with RA and subsequently stimulated with IL-10 alone or in combination with vitamin D₃. The results in Figure 4c show that preincubation with SF supported induction by IL-10 alone. Incubation with both IL-10 and vitamin D₃ produced the release of even more CCL18 by PMN preincubated with SF.

To determine whether extravasation might have a role in the induction of CCL18 release, PMN and PBMC from healthy donors were co-cultured with the endothelial cell line EA-hy.926. Results in Figure 5a revealed high levels of CCL18 in culture supernatants of PMN or PBMC co-cultured with endothelial cells, whereas no release of CCL18 was observed when PMN and EA-hy.926 cells were incubated separately from each other in Boyden chambers. Similar levels of CCL18 were released into supernatant of RA blood or SF PMN co-cultured with endothelial cells (data not shown). The data therefore indicate that cell–cell contact is crucial for the induction of CCL18. Release of CCL18 was induced by endogenous TNF-α, as revealed by blocking experiments with excess anti-TNF-α antibodies (Figure 5b). In controls, stimulation of EA-hy.926 cells with TNF-α was found to induce neither the release of CCL18 (Figure 5b) nor the expression of CCL18 mRNA (Figure 5c), suggesting that CCL18 derives from PMN. To determine the origin of TNF-α in co-cultures, either PMN or EA-hy.926 cells were pretreated with γ-radiation and then incubated with the viable cellular counterpart. As shown in Figure 5d, irradiation of PMN on their own abolished the production of TNF-α in co-culture with EA-hy.926 cells, indicating that PMN are the predominant source of TNF-α in co-cultures. The results therefore suggest that cell contact with endothelial cells induces TNF-α in PMN, which in turn induces CCL18 production by PMN in an autocrine manner.