Fibroblasts from different sites may promote or inhibit recruitment of flowing lymphocytes by endothelial cells

Abstract

We examined the hypothesis that stromal fibroblasts modulate the ability of endothelial cells (EC) to recruit lymphocytes in a site-specific manner. PBL were perfused over HUVEC that had been cultured with fibroblasts isolated from the inflamed synovium or the skin of patients with rheumatoid arthritis or osteoarthritis, or from normal synovium, with or without exposure to the inflammatory cytokines TNF-alpha+IFN-gamma. Fibroblasts from inflamed synovium, but no others, caused unstimulated HUVEC to bind flowing lymphocytes. This adhesion was supported by alpha(4)beta(1)-VCAM-1 interaction and stabilised by activation of PBL through CXCR4-CXCL12. Antibody neutralisation of IL-6 during co-culture effectively abolished the ability of EC to bind lymphocytes. Cytokine-stimulated EC supported high levels of lymphocyte adhesion, through the presentation of VCAM-1, E-selectin and chemokine(s) acting through CXCR3. Interestingly, co-culture with dermal fibroblasts caused a marked reduction in cytokine-induced adhesion, while synovial fibroblasts had variable effects depending on their source. In the dermal co-cultures, neutralisation of IL-6 or TGF-beta caused partial recovery of cytokine-induced lymphocyte adhesion; this was complete when both were neutralised. Exogenous IL-6 was also found to inhibit response to TNF-alpha+IFN-gamma. Normal stromal fibroblasts appear to regulate the cytokine-sensitivity of vascular endothelium, while fibroblasts associated with chronic inflammation bypass this and develop a directly inflammatory phenotype. Actions of IL-6 might be pro-inflammatory or anti-inflammatory, depending on the local milieu.

Figure 1

Lymphocyte adhesion to EC cultured alone or with fibroblasts, with or without cytokine treatment. (A) Adhesion to unstimulated EC cultured alone (None) or co-cultured with dermal or synovial fibroblasts from patients with RA or OA, or with synovial fibroblasts from a patient suffering from mechanical trauma. (B) Adhesion to endothelial mono-cultures unstimulated or stimulated with TNF-α or TNF-α+IFN-γ. (C) Behaviour of lymphocytes adherent to endothelial mono-cultures stimulated TNF-α or TNF-α+IFN-γ (analysed 10 min after washout). (D) Endothelial mono- or co-cultures stimulated with TNF-α+IFN-γ. Data are mean±SEM from 3 to 11 independent experiments, using at least 3 different donors for each cell type, except for ‘trauma’ where fibroblasts from a single donor were tested on 2 occasions. In all experiments, co-cultures were compared with mono-cultures (None), and data are expressed as a percentage of that observed in the paired control (None), which itself was normalised to 100%. ^*p<0.05, ^**p<0.01 for comparison to ‘None’ by paired t-test.

Figure 2

Effects of blocking adhesion receptors and chemokines on lymphocyte adhesion to (A) unstimulated endothelial-RA synovial fibroblast co-cultures or (B) endothelial mono-cultures treated with TNF-α+IFN-γ. HUVEC were treated with antibodies against E-selectin (ENA-2) and/or VCAM-1 (4B2 and GH12, against domains 1 and 4, respectively) for the last 30 min of the 24 h cytokine treatment. Lymphocytes were incubated for 15 min with antibodies against α₄-integrin (Max68P), CXCR3 (1C6), β₂-integrins (R6.5E) or with the CXCR4 inhibitor, AMD3100, before perfusion. Alternatively, Lymphocyte adhesion was expressed as a percentage of control (untreated), which itself was normalised to 100%. Data are mean±SEM from at least three independent experiments. Not all antibodies were used in all experiments, but treated cells were always compared with paired untreated controls. ^**p<0.01 for comparison to untreated by paired t-test.

Figure 3

Effect of RA co-culture on the mRNA expression of E-selectin or VCAM-1 by EC. Endothelial mRNA was isolated from (A and B) unstimulated or (C and D) cytokine (TNF-α+IFN-γ) treated co-cultures and gene expression of (A and C) E-selectin or (B and D) VCAM-1 was assessed by qPCR. In (A) and (B), ANOVA shows a significant effect of co-culture on E-selectin expression (p<0.05) and borderline significance of co-culture on VCAM-1 expression (p=0.08). Values were expressed as REU compared with β-actin. Data are mean±SEM from three independent experiments, using three different donors for each cell type. ^*p<0.05 compared with None by Dunnett's test.

Figure 4

Effect of RA co-culture on the surface expression of E-selectin or VCAM-1 by EC. EC from cytokine treated co-cultures were stained with antibodies against (A) E-selectin or (B) VCAM-1 and surface expression was assessed by flow cytometry (expressed as MFI). Data are mean±SEM from three independent experiments, using three different donors for each cell type.

Figure 5

Effect of neutralising IL-6 or TGF-β, or of adding IL-6 on lymphocyte adhesion. (A) Neutralisation in unstimulated endothelial-RA synovial fibroblasts co-cultures; (B) neutralisation in cytokine-stimulated endothelial-RA dermal fibroblasts co-cultures; (C) addition of IL-6 to unstimulated or cytokine-stimulated EC mono-cultures. Data are mean±SEM from at least three experiments, using at least three different donors for each cell type. In all cases, ANOVA showed a significant effect of treatment on lymphocyte adhesion (p<0.01). ^*p<0.05 and ^**p<0.01 by Bonferroni test compared with co-cultures without neutralisation in (A) and (B), or to cytokine-treated mono-culture without IL-6 in (C).

Figure 6

Effect of co-culture on the gene transcriptional profiles of EC. EC were cultured alone or with human RA dermal or synovial fibroblasts (HDF or HSF respectively) for 24 h in the absence of exogenous cytokines. Endothelial mRNA was isolated and analysed using microarray technology on an 850 human gene array. Data was normalised (sample/reference) and then analysed by TIGR Multiple Experiment Viewer allowing statistical analysis of microarrays (SAM) and hierarchical cluster analysis (HCA). The SAM false-positive level was set at 0%. (A) List of 18 genes identified by SAM to vary significantly between the three culture-conditions. (B) HCA based on the 18 genes identified. The colour scale from green to red for each gene represents the natural logarithm of the gene expression, relative to the value for the reference sample tested with it. The different culture conditions showed distinct segmentations, as indicated by the coloured tree (blue-brown-yellow) across the top, associated with them. Comparisons between the different culture conditions were done on three occasions, each using different primary EC isolates and different arrays. In each array, the individual genes were tested in triplicate.