Twenty synovial and eighteen nodule tissue samples were obtained from 31 patients with RA. All patients fulfilled American Rheumatism Association 1987 classification criteria for RA 30. Patients were classified according to their reported smoking habits as either: (i) smokers (actively smoking when the tissue sample was taken) - including those who were daily or non-daily smokers of tobacco via cigarettes or loose tobacco; or (ii) non-smokers - including ex-smokers (those who ceased smoking ≥ 3 months prior to nodule samples or ≥ 8 years prior to synovial samples being taken), and those who have never smoked tobacco in their lifetime. The demographics of RA patients contributing synovial and/or nodule tissue are summarised in Table 1. With two exceptions, all RA patients included in this study were SE⁺ (overall single copy of SE, n = 20 (65%); double copy of SE, n = 9 (29%)).

Two synovial samples were available from three patients (one smoker, two non-smokers), taken 6 months, 1 year and 5 years apart respectively. Paired synovial and nodule samples were obtained at the same time from four separate patients (one smoker, three non-smokers).

Synovia from two patients with osteoarthritis (OA) were also studied. One synovium (male patient, age 66 years) was processed commercially (Origene, Rockville, Maryland, USA); the second synovium (female patient, age 72 years) was sourced from tissue stocks. Their smoking status is unknown.

Informed consent was obtained for the use of all patient tissue, with the study approved by the New Zealand Multi-Regional Ethics Committee. The use of peripheral blood (PB) from normal healthy donors and associated work involving PB leukocytes was approved by the University of Otago Ethics committee.

Peripheral blood mononuclear cells were isolated from heparinized venous blood of healthy non-smoking human donors by Ficoll density centrifugation. CD3⁺ T-cells, CD20⁺ B-cells, and CD14⁺ monocytes were purified using appropriate antibody-coated magnetic MACS^® microbeads and an AutoMacs separator (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's instructions. Lymphocytes were used without further manipulation.

CD14⁺ monocytes were resuspended in RPMI 1640 medium supplemented with 10% fetal calf serum, 2mM glutamine, 50 U/ml penicillin, 50 μg/ml streptomycin (complete medium) and were stimulated with varied concentrations (0.001μM to 10 μM) of benzo(a)pyrene (BaP) or equivalent dimethyl sulfoxide (DMSO) alone (control) for 24 hr. Stimulation with BaP, a polycyclic aromatic hydrocarbon present in cigarette smoke mimicked the effect of smoke. Alternatively, mononcytes were stimulated with 10μM phorbol 12-myristate 13-acetate (PMA) in complete medium for 72 hr at 37ºC and 5% CO₂.

Mo-DCs were generated from CD14⁺ monocytes by culture in complete medium with recombinant human granulocyte macrophage colony-stimulating factor (rhGMCSF), and rhIL-4 (each 25ng/ml; R&D Systems, Minneapolis, Minnesota, USA) for 6 days at 37ºC and 5% CO₂. Complete medium containing cytokines was replenished on day 3. Mo-DCs were further treated with either DMSO or 0.001μM to- 10 μM BaP in DMSO, or with polyinosinic:polycytidylic acid (Poly I:C) (12.5μg/ml) for 24 hr. Purity and maturation of DCs was assessed by staining with monoclonal antibodies to CD14, CD11c and CD83 (BD Biosciences, San Jose, California, USA) and FACS analysis. Immature mo-DCs were CD14^-, CD11c⁺, CD83^low and as expected became CD14^-, CD11c⁺, CD83^high upon maturation induced by Poly I:C stimulation.

Total RNA was purified from rheumatoid nodule and synovial tissues or from cultured cells, and reverse transcribed using Superscript III (Invitrogen, Carlsbad, California, USA) as previously described 31. Gene expression was assessed by RT-PCR using SYBR Green or TaqMan gene expression assays on an Applied Biosystems ABI 7300 sequence detection system. The SYBR Green primers (Invitrogen) were AHR (sense): 5'-TTC AGT TCT TAG GCT CAG CGT -3', AHR (antisense): 5'-TGC TGC TCT ACA GTT ATC CTG G -3' and GAPDH (sense): 5'- TGC ACC ACC AAC TGC TTA GC -3', GAPDH (antisense): 5'- GGC ATG GAC TGT GGT CAT GAG -3'. Commercially available Taqman assays (Applied Biosystems, Foster City, California, USA) were used to measure CYP1A1 (Hs00153120_m1), AHRR (Hs01005075_m1), IL17A (Hs00174383_m1), IL17F (Hs00369400_m1), IFNG (Hs00174143_m1), IL22 (Hs01574154_m1), IL23A (Hs00372324_m1), IL23R (Hs00332759_m1), FOXP3 (Hs00203958_m1), IL6 (Hs00985639_m1), IDO1 (Hs00984148_m1), TBX21 (Hs00203436_m1) and GAPDH (Hs99999905_m1) gene expression. All samples were assessed in triplicate. Samples were considered negative for gene expression when threshold cycle (Ct) values were <40. Positive, test sample Ct values were extrapolated to standard curves obtained from human tonsil or lung standards, to calculate the mean amount of gene-specific RNA in each sample. Results are expressed as the mean ± SEM ng RNA for each gene of interest, relative to the expression of glyceraldehede-3-phosphate dehydrogenase (GAPDH) RNA.

Seven-μm sections of frozen synovial tissues were cut and fixed in acetone for 10 minutess at 4ºC. After blocking with 5% goat immunoglobulins (IgGs; Sigma, Saint Louis, Missouri, USA) for 30 minutes at room temperature, the sections were incubated separately with cell-specific mouse monoclonal anti-CD3 (T cells; clone UCHT1), anti-CD20 (B cells; clone L26; DAKO, Glostrup, Denmark), anti-CD14 (monocytes/macrophages; clone FMC-17), anti-DCs (clones CMRF44 32 and CMRF56 3334), anti-CD303(BDCA-2) (pDCs; clone AC144; Miltenyi Biotec), anti-CD1 (imDC; clone Na134), or anti-prolyl 4-hydroxylase (fibroblast; clone 5B5; Abcam, Cambridge, UK) overnight at 4ºC, followed by incubation with AlexaFluor568-conjugated goat anti-mouse antibody (diluted 1:1500; Invitrogen) for 1.5 hr at 4ºC. The sections were then blocked with 5% mouse IgGs for 30 minutes at room temperature following by incubation with biotinylated mAbs either against AHR (clone RPT9; Abcam) or CYP1A1 (clone b-2; Santa Cruz Biotechnology, Santa Cruz, California, USA) overnight, at 4ºC with subsequent incubation with AlexaFluor488-conjugated streptavidin (10μg/ml; Invitrogen) for 30 minutes, at RT. Cell nuclei were identified by counterstaining with Hoechst 33342 (2.5 μg/ml). Negative controls included the omission of primary antibodies and substitution of primary antibodies with mouse-named IgGs (DAKO). Co-localization of fluorescent staining was assessed by epifluorescence microscopy using a Zeiss microscope fitted with SpotRT digital camera and imaging software (Diagnostic Instruments, Sterling Heights, Michigan, USA). All samples were assessed by a single observer (MK) in a blinded fashion and verified by a second observer (PAH).

Differences in gene-expression levels were analyzed by the Mann-Whitney U-test. Where appropriate data was log transformed to stabilise the variance. The one-way analysis of variance (ANOVA) multirange post hoc Tukey's test was used when multiple comparisons were analyzed. Fisher's exact probability test was used to measure association. Correlation coefficients were determined by rank correlation using a nonparametric Spearman test. Analyses were done using Prism for Windows v4.03 (GraphPad Software Inc, San Diego, California, USA) or STATA11 statistical software. A probability level of 5% was considered significant for all statistical analysis.

AHR expression was present in both synovial tissues and nodules. Expression was higher in synovia (7.00 ± 1.12 ng RNA, n = 20) compared to nodules (2.76 ± 0.37 ng RNA, n = 18, P < 0.0006) (Figure 1A), but these levels of AHR expression in inflamed rheumatoid tissues were not affected by smoking (Figure 1B).

Expression of the prototypic AHR responsive gene, CYP1A1 was used to establish the activation status of AHR in synovial and nodule tissues. CYP1A1 transcript was detected in 7/20 synovial membranes (35% CYP1A1⁺) and in 7/18 nodules (39% CYP1A1⁺). We considered whether smoking was a factor influencing the activation of AHR, reflected by the CYP1A1 expression. We found six of the 7 CYP1A1⁺ synovia were from smokers, whereas 12/13 CYP1A1^- synovia were from patients who were non-smokers. The association between patients who smoke and the activation of AHR was significant in inflamed synovium (Fisher's exact test, P = 0.005) (Table 2) but was not evident in the separate cohort of 18 patients who provided nodule tissues (Table 2).

Levels of CYP1A1 expression were also significantly higher in synovial tissues from RA patients who smoked compared to non-smokers (0.048 ± 0.011 vs. 0.003 ± 0.003 ng RNA respectively, P = 0.004). There was no such difference in nodule tissues (0.21 ± 0.12 vs. 0.27 ± 0.25 ng RNA, Figure 1C). Similarly, expression of the AHRR gene, which also depends on AHR activation and consequent transcription activity 35, was significantly higher in synovial tissues from RA patients who smoked when compared to non-smokers (1.79 ± 0.48 vs. 0.53 ± 0.21 ng RNA, P = 0.036); this was not the case in nodule tissues (4.82 ± 3.41 vs. 1.78 ± 0.95 ng RNA respectively, Figure 1D). There was a statistically significant positive correlation between CYP1A1 and AHRR expression in synovial tissues (Spearman r = 0.71, P = 0.0005) that also occurred in nodules (Spearman r = 0.62, P = 0.0062, data not shown).

Variable expression of AHR was also observed in synovia from separate patients with OA. Expression levels overlapped with the range established for rheumatoid synovia (Figure 1E). Irrespective of the extent of AHR expression, CYP1A1 expression was not detected in OA synovia.

Possible long-term effect of smoking on AHR expression and activation within synovial tissue was considered. Three patients providing four synovial samples were ex-smokers, having ceased smoking ≥ 8 years prior. Levels of synovial AHR expression were equivalent between patients who were smokers (n = 7), ex-smokers (n = 4) and those who had never smoked (n = 9), consistent with a lack of effect from smoking on AHR expression (Additional file 1). CYP1A1 expression was not detectable in ex-smokers, suggesting no long-term effect from smoking on AHR activation (Additional file 1).

We further considered the implications of smoking for the expression of a number of immuno-inflammatory genes implicated in RA. Experimentally, Th17 cells are a direct cellular target of AHR agonists 26. Their signature cytokine, IL-17A has been implicated in the pathogenesis of RA 36. As such, Th17 cells provide a possible link between smoking, AHR activation and the exacerbation of synovial inflammation in RA. However, we found significantly less IL17A gene expression in synovial tissue from smokers when compared to non-smokers (Figure 2) and a negative association between IL17A and CYP1A1 gene expression (Spearman r = -0.51, P = 0.022).

Amongst other Th17 cell cytokines, expression of the IL17F gene was not affected by smoking but was restricted to CYP1A1^- synovia, whereas IL22 gene expression was not detected in any synovia (data not shown). We also considered the potential for smoking to impact upstream of IL17A but found no evidence of an impact on gene expression of the critical cytokine IL23, nor of the IL23 receptor (IL23R). Similarly, smoking had no effect on Th1-cell mediated inflammation via interferon-γ (IFNG), T-bet (TBX21) or FOXP3 gene expression in synovia (Figure 2).

Next we sought to identify the cell type(s) expressing AHR and showing AHR activation in inflamed synovial tissues. Using double immunofluorescence staining, CYP1A1 protein was observed only in CYP1A1⁺ synovia from RA patients who were smokers. In these synovia, both AHR and CYP1A1 proteins were produced by early differentiating CMRF44⁺ and CMRF56⁺ DCs (Figure 3A, B, C). These AHR⁺/CYP1A1⁺ DCs were mainly observed in close proximity to T and B cells (data not shown). Staining for AHR was cytoplasmic and in some DCs more obviously perinuclear, with small amounts of AHR protein localized within the nucleus, consistent with AHR activation (Figure 3B). Amongst the other cell types examined, rare (much less than 1%) CD3⁺ T cells produced AHR and CYP1A1 proteins (Figure 3D, E). Although most CD303⁺ plasmacytoid DCs (pDCs) produced AHR protein, only occasional CD20⁺ B cells were AHR⁺ and neither pDCs nor B cells produced detectable CYP1A1 protein. In synovium, immature CD1⁺ DCs, CD14⁺ monocyte/macrophages and fibroblasts expressing prolyl-4-hydroxylase did not express either AHR or CYP1A1 protein (data not shown).