All cell culture reagents were endotoxin-free and were purchased from Sigma (St Louis, MO, USA). Recombinant human (rh) granulocyte–macrophage colony-stimulating factor (GM-CSF), rhIL-17A and rhTNFα were from Peprotech (London, UK), monoclonal anti-human GM-CSF (MAB215) and polyclonal anti-human GM-CSF (AF215NA) were from R&D Systems (Abingdon, UK), and recombinant IFNβ was from Biosource (Paisley, UK). Type I IFN receptor (CD118) blocking antibody was obtained from Calbiochem (San Diego, CA), anti-IL-6R and anti-TNFα were from R&D (Abingdon, UK), human GM-CSF and IL-6 ELISA kits were from BDPharmingen (San Diego, CA), and protein G-conjugated agarose was from Upstate (Lake Placid, NY, USA). The 0.2 μm Transwell inserts were obtained from Falcon (BD biosciences, San Jose, CA, USA), rhIL-2 from Chiron (Middlesex, UK), phytohaemaglutinin from Murex (Dartford, UK) and 3,3'-dihexyloxacarbocyanine iodide was from Molecular Probes (Leiden, The Netherlands).

Additional reagents, including adenosine, adenosine deaminase, indomethacin, NS398 and MK886, were from Sigma. The multiplex assay used Beadmate kits (Upstate) as specified in the manufacturer's instructions, read with a Luminex 100 system (Upstate).

Rheumatoid synovial fibroblasts were isolated from the synovium of RA patients meeting the 1987 American College of Rheumatology criteria 22 undergoing joint replacement surgery. Clinical details of patients fulfilling the 1987 American College of Rheumatology criteria who donated samples are presented in Table 1. The Disease Activity Scores using 28 joint counts (DAS28) at the time of joint replacement are given for patients donating samples from which fibroblasts were cultured. Six of the eight patients undergoing joint replacement who gave samples for fibroblast culture had DAS28 counts >5.2, indicating active disease.

Peripheral blood and synovial fluid were also obtained from RA patients meeting the 1987 American College of Rheumatology criteria 22. Ethical approval for the use of patient-derived material was given by the local research ethics committee, and all patients provided written informed consent (LREC reference 5735).

Fibroblasts were maintained in fibroblast medium (consisting of 81.3% RPMI 1640, 16.3% foetal calf serum, 0.81 × MEM nonessential amino acids, 0.81 mM sodium orthopyruvate, 1.62 mM glutamine, 810 U/ml penicillin and 81 μg/ml streptomycin) at 37°C in a humidified 5% carbon dioxide atmosphere. The fibroblast phenotype was confirmed by morphology and immunofluorescence microscopy as described previously 23. All cells expressed fibronectin and prolyl-4-hydroxylase, while less than 0.5% cells stained positive for CD68, von Willebrand factor, CD31 or cytokeratin.

Synovial fibroblasts were used between passages 4 and 8. Before the addition of exogenous cytokines, fibroblasts were seeded into flat-bottomed 96-well plates at a density of 6 × 10³ cells/well and were cultured for 48 hours in coculture medium (consisting of 88.9% RPMI 1640 medium supplemented with 8.9% heat-inactivated foetal calf serum, 1.62 mM glutamine, 810 U/ml penicillin, 81 μg/ml streptomycin and 1.8 mM (pH 7.4) HEPES). The coculture medium was then refreshed and the fibroblasts exposed to 10 ng/ml IL-17A and 10 ng/ml TNFα. Following cytokine treatment or coculture, the fibroblasts were washed extensively with RPMI 1640 and were cultured with or without 10⁵ neutrophils/well for a further 24 hours. Cells or conditioned medium were then harvested.

Peripheral blood of consented healthy volunteers was taken into tubes containing the anticoagulant ethylenediamine tetraacetic acid and was then mixed with T500 Dextran to sediment the red blood cells. Mononuclear cells and granulocytes were separated by centrifugation across a saline-based Percoll gradient (upper phase, 54% vol/vol; lower phase, 79% vol/vol), and were then washed in PBS to remove residual Percoll. Neutrophil preparations were routinely >98% pure and viable, as assessed by May–Grunwald Giemsa-stained cytospins.

Flow cytometry was carried out using the 20 μl volume dump facility of the Beckman Coulter Epics XL bench-top flow cytometer (High Wycombe, UK). When used in combination with vital dyes, this technique enables accurate enumeration of live cells within a sample of defined volume. Neutrophils were gated based on their forward scatter versus side scatter profiles, and apoptotic cells were excluded by loss of 3,3'-dihexyloxacarbocyanine iodide positivity. Survival is expressed as percentage of the starting population. The percentage of cells committed to apoptosis was confirmed with May–Grunwald Giemsa-stained cytospins.

Supernatants from cytokine-treated fibroblasts, and solutions containing higher concentrations of recombinant target proteins were depleted of cytokine in three rounds of incubation with an excess of irrelevant or depleting antibody-coated protein G-conjugated agarose beads. Residual beads were removed by centrifugation before exposing neutrophils to the medium or performing ELISAs to verify that cytokine depletion was successful (the lower detection limit of the GM-CSF ELISA was 4 pg/ml). For lipopolysaccharide studies, 10 ng/ml lipopolysaccharide (Sigma) was added to neutrophils in the presence or absence of conditioned medium from IL-17A-treated and TNFα-treated cells and of 50 μg/ml polymyxin B sulphate.

Freshly isolated neutrophils were pretreated with Ly294002 (20 μM) or Bay 11-7085 (1 μM) or vehicle control for 45 minutes at 37°C, washed, and cultured for 24 hours in fibroblast-conditioned medium (FCM) from untreated rheumatoid arthritis synovial fibroblasts (RASF) or from rheumatoid arthritis synovial fibroblasts stimulated with TNFα and IL-17 (RASF_IL-17/TNF). The optimal concentration of inhibitors used was established by performing a dose–response analysis.

To measure the functionality of neutrophils, superoxide release upon stimulation with 200 nM f-Met-Leu-Phe was measured using the substrate cytochrome c. Following their 24-hour treatments, neutrophils were washed once in HBSS buffer (HBSS pH 7.3 + 25 mM HEPES and 5 mM glucose), and were resuspended in HBSS containing 1% human serum. Then 5 × 10⁴ neutrophils were added to triplicate wells of a 96-well plate in the presence or absence of f-Met-Leu-Phe. Release of superoxide was quantified by measuring the colour change of cytochrome c at 450 nm emission. The positive control was complete reduction of cytochrome c using the reducing agent sodium dithionite (100 mM).

Degradation of inhibitor of NFκB in the presence of 10 ng/ml TNFα or FCM was assessed by western blotting. Neutrophils (at least 5 million cells/sample) were lysed with 10% trichloroacetic acid (Sigma) followed by 5 minutes incubation on ice. The lysates were centrifuged for 5 minutes at 14,000 × g at 4°C and the supernatant was discarded. The pellet was washed twice with ice-cold acetone, centrifuging each time as before. The pellet was dissolved in SDS-PAGE sample loading buffer (0.125 M Tris pH 6.8 containing 20% glycerol, 2% SDS, 5% mercaptoethanol and 25 μg/ml bromophenol blue). The sample was boiled, then cooled on ice for 5 minutes each, before performing gel electrophoresis using a 12% SDS gel.

Proteins were then transferred to a polyvinylidene difluoride membrane (Flowgen Ltd, Sittingbourne, UK) for 2 hours at 0.45 A using a wet blotting system (Biorad Hemel Hempstead, UK). The blots were blocked for 1 hour with 5% skimmed milk in Tris-buffered saline containing 0.1% Tween 20 (TBST) for 1 hour. Blots were incubated with primary antibodies (IkBα Rabbit IgG; Upstate) diluted in Tris-buffered saline containing 1% skimmed milk overnight at 4°C, washed four times with TBST for 10 minutes each, and then incubated with secondary antibodies (horseradish peroxidase-donkey anti-rabbit; Amersham, Bucks, UK) diluted in TBST. Blots were washed three times with TBST for 10 minutes each and developed for 5 minutes with enhanced chemiluminescence reagents (Amersham).

Peripheral blood mononuclear cells and synovial fluid mononuclear cells were isolated by density gradient centrifugation on Ficoll-paque™-Plus (GE Healthcare, Slough, UK). Following centrifugation, peripheral blood mononuclear cells and synovial fluid mononuclear cells were isolated from the buffy coat and were added to fresh RPMI 1640 medium. Cells were washed twice with RPMI 1640 medium by centrifugation at 600 × g for 8 minutes. Cells were then counted and resuspended at 1 × 10⁶ cells/ml in fresh medium. Intracellular IL-17 was detected by flow cytometry after stimulation of cells with 50 ng/ml phorbol 12-myristate 13-acetate (PMA) (Sigma) and 250 ng/ml ionomycin (Sigma) in the presence of 10 μg/ml Breveldin A (Sigma) for 4 hours in a 24-well plate (Sarstedt, Leicester, UK).

Cells were removed from culture and resuspended in 2% BSA/PBS. Cell surface staining with fluorescein isothyocyanate (FITC)-labelled anti-CD4 (Immunotools, Friesoythe, Germany), phycoerythrin-cyanine 5-labelled anti-CD3 (Immunotools), FITC-labelled IgG₁ isotype control (Immunotools), and phycoerythrin-cyanine 5-labelled IgG₁ isotype control (Serotec, Oxfordshire, UK) was performed by incubating cells on ice with the appropriate antibody combination for 20 minutes. Cells were subsequently fixed and permeabilised with Cytoperm according to manufacturer's instructions (Caltag/Invitrogen, Towcester, UK).

Cells were incubated with either phycoerythrin-labelled anti-IL-17 (eBiosciences, San Diego, CA, USA) or phycoerythrin-labelled IgG₁ isotype control (Immunotools) for 20 minutes, and were then washed and acquired on a Beckman Coulter Epics XL bench-top flow cytometer. T cells were gated based on their forward scatter versus side scatter profiles and their CD3 expression. CD4⁺ IL-17⁺ T cells are expressed as a percentage of the CD3⁺ population.

Full thickness (5 μm) sections of human rheumatoid synovium were prepared and fixed in 100% acetone at 4°C for 20 minutes. The sections were rehydrated with PBS pH 7.4 with 5% foetal calf serum (Biosera Ltd, Ringmer, UK), and indirect immunofluorescence was performed using the following primary antibody combinations: IgG_2b mouse anti-human CD3 (UCHT-1, 17 μg/ml; gift from Peter Beverley, University College Hospital, London, UK) with rabbit anti-human von Willibrand factor (A0082, 6.2 μg/ml; Dako, Glostrup, Denmark) and IgG₁ mouse anti-human IL-17 (12-7179, 1.25 μg/ml; eBiosciences); or IgG_2b mouse anti-human CD4 (OKT4) with IgG_2a mouse anti-human CD8 (OKT8, both OKT clones used as ascitic fluid 1/100; American Type Culture Collection, Middlesex, UK) and rabbit anti-human IL-17 (AHP455G, 20 μg/ml; AbD Serotec, Oxfordshire, UK).

The secondary antibody combinations used were: goat anti-mouse IgG_2b-cyanine 5 (1090-15, 4 μg/ml; Southern Biotech Inc., Birmingham, AL, USA) with donkey anti-rabbit-tetramethyl-rhodamine (711-026-152, 6 μg/ml; Jackson Immuno-Research, Newmarket, UK) and goat anti-mouse IgG₁-FITC (1070-02, 20 μg/ml; Southern Biotech Inc.); or goat anti-mouse IgG_2b-cyanine 5 (1090-15, 4 μg/ml; Southern Biotech Inc.) with goat anti-mouse IgG_2a-FITC (1080-02, 20 μg/ml; Southern Biotech Inc.) and donkey anti-rabbit-tetramethyl-rhodamine (711-026-152 6 μg/ml; Jackson Immuno-Research).

Irrelevant control antibodies (X0931 and X0903; Dako) were used to confirm IL-17 staining, while primary antibodies were omitted as controls for other stains. Sections were counterstained with nuclear DNA dye Hoechst 33258 (20 μg/ml; Riedel De Haenag, Seelze, Hannover, Germany) and were subsequently mounted in 2.4% 1,4-diazabicyclo [2.2.2]octane (Aldrich, Gillingham, UK) in glycerol pH 8.6 (Fisons Scientific, Loughborough, UK). A Zeiss 510 laser-scanning confocal microscope (Zeiss, Welwyn Garden City, UK) was used to visualise staining with images captured and processed using the Zeiss LSM Image Examiner software (Zeiss).

A nonparametric distribution was assumed for all assays. Statistical analysis of differences in neutrophil survival induced by fibroblasts or FCM across multiple groups was performed using Kruskal–Wallis one-way analysis of variance and Dunn's post test. For the difference between paired samples (depletions, blocking experiments, experiments with and without transwells), significance was assessed using the Wilcoxon signed rank test with two-tailed P values. Results are presented as the mean ± standard deviation or standard error where appropriate.

We first determined whether T_H17 cells are present within the rheumatoid synovium. IL-17 was expressed on CD3⁺ T cells within the synovium, and in particular on T cells found in a perivascular distribution (Figure 1a). Expression of IL-17 was confined to CD4 T cells, with no expression on CD8 T cells, although there were some IL-17-expressing cells in the synovium that did not express CD4 or CD8 (Figure 1b). Equally some CD4 T cells did not express IL-17 (Figure 1a,b). IL-17-expressing CD4⁺CD3⁺ cells were also found at low frequency within rheumatoid synovial fluid (Figure 1c).

Previous studies have shown that IL-17 and TNFα synergise with each other to produce potent biological effects in leucocytes. We therefore cross-titrated recombinant human IL-17A and TNFα on RASF to determine whether these cytokines, either individually or in combination, could enhance the ability of RASF to support neutrophil viability in cocultures of neutrophils with fibroblasts.

Figure 2a shows that the basal level of neutrophil survival in the presence of RASF increases significantly and in a dose-dependent manner when RASF are pretreated with a combination of IL-17 and TNFα. As would be expected, the increase in survival is mirrored precisely by an inhibition of apoptosis (data not shown). Optimal neutrophil survival was achieved at a dose of 1 to 10 ng/ml IL-17 and 10 ng/ml TNFα. Pretreatment of RASF with the highest doses (10 ng/ml) of IL-17 or TNFα alone did not significantly enhance survival at 24 hours (Figure 2b). Neutrophils cocultured with fibroblasts that had been pretreated with IL-17 and TNFα (RASF_IL-17/TNF), however, remained viable for 24 hours. Indeed, the level of viability exceeded that obtained with a 100 ng/ml dose of rhGM-CSF – a prototypical neutrophil survival factor added directly to neutrophil monoculture (data not shown). Direct microscopic analysis of fibroblast-mediated neutrophil survival by cellular morphology and active caspase-3 immunostaining confirmed these results (Figure 2c,d and data not shown).

We next determined the kinetics of rescue from apoptosis, mediated by coculture of neutrophils with RASF_IL-17/TNF. Figure 3a shows that, in the presence of RASF_IL-17/TNF, the neutrophil lifespan is effectively doubled. Furthermore, rescued neutrophils were functionally active and capable of generating the superoxide radical to the same extent as freshly isolated neutrophils upon treatment with f-Met-Leu-Phe (Figure 3b).

We next investigated the mechanism by which cytokine-treated synovial fibroblasts prolonged neutrophil survival by comparing the ability of FCM and transwell-separated neutrophil–synovial cocultures to influence neutrophil survival (Figure 4a). In addition to exposing neutrophils to the high local concentrations of cytokines released from fibroblasts, coculture allows direct cell–cell and cell–matrix interactions to occur. Separating neutrophils from fibroblasts by a transwell filter or using FCM prohibits direct cell contact between the two cell types, but allows soluble factors to be shared.

We observed no significant differences between the rescue afforded by direct coculture, FCM or transwell-separated cultures, indicating that soluble factors released from RASF_IL-17/TNF are both necessary and sufficient to extend the functional lifespan of neutrophils in cocultures with activated synovial fibroblasts. Furthermore, at least 50% of the survival effect mediated by the soluble factor was temperature sensitive as activity was destroyed following the treatment of conditioned medium at 92°C (Figure 4b).

In an attempt to identify which soluble factor(s) might be responsible for neutrophil survival, we screened the FCM from RASF_IL-17/TNF for a range of cytokines and chemokines using multiplex bead ELISAs. This screening revealed elevated levels of GM-CSF, granulocyte colony-stimulating factor, CCL2 and CXCL8 in FCM from RASF_IL-17A/TNF compared with untreated RASF or with RASF treated with IL-17 or TNF alone (Table 1).

Since GM-CSF is a well-characterised neutrophil survival factor, we tested whether this was the factor responsible for neutrophil survival by specific immunodepletion from FCM using depleting antibodies. We consistently detected approximately 100 pg/ml GM-CSF in FCM from RASF_IL-17/TNF. Following antibody-mediated depletion, the GM-CSF concentrations fell to levels that were essentially undetectable by ELISA (Figure 5b). We found that GM-CSF only accounted for approximately 50% of the neutrophil rescue activity released by synovial fibroblasts in response to IL-17 and TNFα (Figure 5a). Combined immunodepletion of GM-CSF and other candidate survival factors detected in FCM from RASF_IL-17/TNF – such as granulocyte colony-stimulating factor, CCL2 and CXCL8 – failed to inhibit neutrophil survival any further.

Using an inhibitor of the phosphatidylinositol-3-kinase signalling pathway (Ly294002) and an inhibitor of the NF-κB pathway (Bay11-7085), we found that both phosphatidylinositol-3-kinase and NF-κB signalling pathways contributed significantly to the RASF_IL-17/TNF-mediated neutrophil survival – implying that in addition to GM-CSF, another factor that utilises phosphatidylinositol-3-kinase and/or NF-κB also plays a role (Figure 5b). We confirmed the role of the NF-κB pathway by demonstrating inhibitor of NF-κB (I-κB) degradation on a western blot of cell lysates from neutrophils treated with FCM from RASF_IL-17/TNF (Figure 5d).

Involvement of NF-κB suggested further possible neutrophil survival candidates, including TNFα, IL-6 and IFNβ. No TNFα was present, however, in the multiplex assay of cytokine-pretreated fibroblast supernatants (Table 2). IL-6 is produced in large quantities by activated fibroblasts (Table 2). IFNβ is a well described, fibroblast-derived neutrophil survival factor. Using an effective blocking antibody to the IFNβ receptor, we showed that blockade of IFNβ within neutrophil survival experiments had no effect on the enhanced survival seen in the presence of conditioned medium (Figure 6a). We found that recombinant IL-6 does not induce a significant delay in neutrophil apoptosis in this system (data not shown), On the basis that synergy between multiple survival candidates could be leading to enhanced survival, however, we performed a combined experiment in which GM-CSF and TNFα were depleted from conditioned media, before adding to neutrophils pretreated with blocking antibodies to interferon and IL-6 receptors (Figure 6b). Although recombinant TNFα induced some survival, the combined effect of all depletions and blockade was no greater than the effect of GM-CSF blockade alone.

We hypothesised that the temperature-insensitive component of neutrophil rescue might result from the presence of adenosine, arachidonic acid derivatives, or contaminating lipopolysaccharide. Neither adenosine nor adenosine deaminase, however, affected neutrophil survival. Furthermore, neither the cyclooxygenase-2 inhibitors indomethacin and NS398 nor the 5-lipooxygenase inhibitor MK-886 inhibited survival induced by FCM from RASF_IL-17/TNF (data not shown). To rule out an effect of contaminating lipopolysaccharide, we used polymyxin B to bind lipopolysaccharide, but this did not inhibit neutrophil rescue (Figure 6c).