Background
Rheumatoid arthritis (RA), a chronic inflammatory disease of the joint, is one of the most common autoimmune diseases in the UK, affecting over 1% of the population [
1]. In RA, high levels of pro-inflammatory cytokines are maintained within the joint, chemokine gradients continue to recruit leukocytes and the cells within the synovium are retained and are prevented from undergoing apoptosis. Further, the tissue-resident stromal cells become hyperplastic, leading to the growth of an invasive pannus, which narrows the joint space and degrades both articular cartilage and bone.
One of the predominant cell types of the synovial stroma is the fibroblast-like synoviocyte (FLS). It has become increasingly apparent that the FLS is crucial to normal and aberrant inflammatory responses in the joint [
2‐
4]. FLS can release large quantities of pro-inflammatory cytokines. IL-6, for example, is secreted at higher basal levels in FLS in RA than in osteoarthritis (OA) [
5]. FLS in RA increase leukocyte adhesion to [
6] and transmigration through [
7] endothelial layers, and the high concentration of CXCL12 produced by FLS in RA restricts the egress of leukocytes from the joint [
8]. Further, FLS release of granulocyte macrophage colony stimulating factor (GMCSF) keeps neutrophils alive for twice the normal span in vitro, and this growth factor is produced at higher levels by FLS in RA than in healthy controls [
9‐
11]. FLS in patients with RA have an increased rate of proliferation [
12], are more resistant to apoptosis than FLS in their healthy counterparts [
13], and also have increased invasive capabilities and matrix metalloproteinase (MMP) production [
14]. Finally, the strongest evidence for the pathogenic role of FLS in RA comes from the experiments of Muller-Ladner et al., who showed that FLS in RA could be cultured in vitro and then implanted in mice with severe combined immunodeficiency (SCID) to induce RA-like disease in the absence of immune cells [
15]. Thus, FLS in RA are now viewed as “imprinted aggressors” rather than the traditional “passive responders” [
16].
It is well-known that inflammation does not passively end, but is actively switched off [
17‐
21]. It is therefore possible that chronic inflammatory diseases such as RA involve failure of mechanisms in the resolution of inflammation. The negative regulation of inflammation has been well-characterised in monocytes/macrophages (reviewed in [
22]). Macrophages respond to a variety of inflammatory stimuli, both endogenous, such as TNFα and IL-1, and exogenous, such as lipopolysaccharide (LPS). Whilst these stimuli elicit strong inflammatory responses, they also induce anti-inflammatory pathways that curtail macrophage inflammation [
19,
22,
23]. Negative regulation of inflammation occurs at a range of levels, both within and without the cell. Pro-inflammatory stimuli often elicit anti-inflammatory mechanisms to curtail their own action [
22]. This is exemplified by the nuclear factor (NF)-κB-IκBα negative feedback loop. Once a stimulatory signal causes degradation of its inhibitor IκBα, NF-κB is free to translocate to the nucleus and facilitate transcription of pro-inflammatory mediator genes. Another gene also induced by NF-κB is NF-κB-inhibitor-α (Nfkbia), which encodes IκBα. Nascent IκBα shuttles NF-κB out of the nucleus to once again be held inactive, thereby curtailing the transcriptional response to stimulus [
24,
25].
Despite the abundant literature on negative regulation of inflammation in macrophages, there is a comparative paucity of information on fibroblasts. Evidence from Lee et al. suggests that FLS in RA lack the inherent negative regulation seen in macrophages [
26]. Rather, FLS continued to produce pro inflammatory mediators for the duration of incubation with TNFα, even after four days of stimulation. This paper also showed significantly lower expression of negative regulators such as A20-binding inhibitor of NFκB activation 3 (ABIN3), activating transcription factor 3 (ATF3), IL-1 receptor-associated kinase M (IRAK-M), and suppressor of cytokines signalling 3 (SOCS3) in FLS in RA compared to macrophages. The implication is that FLS are deficient in the negative feedback regulation of inflammation in RA. However, no comparison was made with FLS obtained from non-inflamed joints. A similar study showed that gingival fibroblasts (HGF) lack SOCS1, IRAKM and SH2 domain-containing inositol phosphatase 1 (SHIP1) proteins, which are all used in negative regulation of inflammation [
27].
We sought to compare the inflammatory responses of FLS obtained from inflamed (RA) and non-inflamed joints, with the aim of elucidating which (if any) mechanisms are used by FLS to limit their inflammatory responses, and to test the hypothesis that such mechanism(s) would be aberrant or absent in FLS in RA. Experimental design has been based upon the model of endotoxin tolerance in macrophages, which has been extensively used to elucidate mechanisms of negative regulation of inflammatory responses in those cells [
28‐
31]. Exposure of macrophages to a single dose of LPS reprogrammes their responses to a second dose, such that expression of pro-inflammatory mediators is attenuated, whilst expression of anti-microbial products is spared [
30,
31]. In a similar fashion, we exposed fibroblasts to two doses of TNFα, with an intervening rest period in the absence of the pro-inflammatory cytokine. We then compared the first and second responses. Contradictory to our initial hypothesis, we found FLS to augment their cytokine-induced IL-6 secretion when primed with either TNFα or IL-1α. We found this priming phenomenon to occur in FLS derived from both inflamed and non-inflamed joints, but not fibroblasts from adult skin. The augmented second response was gene-specific and transient. Sustained activation of NF-κB in response to the second stimulation may play a role in fibroblast priming.
Methods
Study participants
Patients with RA who were involved in this study were diagnosed according to the 1987 American College of Rheumatology (ACR) criteria [
32]. Synovial tissue was collected during joint replacement surgery. Ultrasound-guided synovial biopsies were collected during arthroscopic examination of unexplained joint pain. Where there was no evidence of inflammatory joint pathology, and no diagnosis of RA was subsequently made, these samples were designated as “healthy” (non-inflamed) controls. Dermal fibroblasts were derived from skin samples collected at the time of joint replacement surgery in patients with OA and from patients with RA. Synovial and dermal fibroblasts were cultured as previously described [
9]. The study (National Research Ethics Service (NRES) Committee West Midlands - The Black Country Ref. 07/H1204/191) and all participants in this study gave written, informed consent.
Cells
Fibroblasts were isolated from synovium and skin as previously described [
9]. Cells were grown in Roswell Park Memorial Institute (RPMI) medium supplemented with 10% foetal calf serum, 0.81 × minimum essential medium (MEM) non-essential amino acids, 0.81 mM sodium orthopyruvate, 1.62 mM glutamine, 810 U/mL penicillin and 81 μg/mL streptomycin. Cells were used at passages 3–8. A dermal fibroblast line from neonatal foreskin, BJ (ATCC CRL-2522), was purchased from ATCC. Cells were grown in Eagle’s minimum essential medium (EMEM) (ATCC) supplemented with 10% foetal calf serum, 810 U/mL penicillin and 81 μg/mL streptomycin.
Experimental design
The repeat dose experiments involved seeding cells and allowing them to adhere overnight, before stimulating cells with vehicle (growth medium), TNFα at 10 ng/mL, or IL-1α at 10 ng/mL (both Peprotech) for 24 h. Conditioned medium was removed and cells were washed thoroughly before being rested for one, three or seven days in fresh medium. In some instances, cells were treated with 100 nM MLN4924 or vehicle control (0.1% dimethyl sulfoxide (DMSO)) during the first stimulation with TNFα, or during the one-day rest period. Cells were washed again after the rest period and stimulated with vehicle, TNFα or IL-1α at the same concentration as above. Conditioned medium was removed after 24 h for analysis.
Assessment of NF-κB activity or localization was based on cells seeded into 6-cm dishes for western blots, and eight chamber glass slides for immunofluorescence studies. Cells were left overnight to adhere, and then stimulated with TNFα (10 ng/mL) at the time points described in the appropriate figures.
Analysis
Enzyme-linked immunosorbent assay (ELISA) was used to measure the secretion of cytokines. IL-6 (cat #88-7066-88), IL-8 (cat #88-8086-88), and CCL5 (cat #BMS-287/2INST) ELISA kits were purchased from E Bioscience. The latter was used according to manufacturer’s instructions, the two former were used with a 1-ng/mL top standard, in order to capture a greater range of responses to various stimuli.
Western blot analysis was conducted to assess internal protein abundance and activation. BJ cells were lysed in radioimmunoprecipitation assay (RIPA) buffer (150 mM sodium chloride, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris pH8). Assessment of nuclear and cytosolic protein localization was performed using a hypotonic lysis buffer and centrifugation to isolate the cytosolic fraction, and a nuclear extraction buffer and centrifugation to isolate the nuclear fraction. Gel blotting was performed using 12-well pre-cast gels and membranes from BioRad. The following western blotting antibodies were all from Cell Signaling Technology: α Tubulin, Lamin A/C, phospho-p38 (T180/Y182), RelA, phospho-RelA (S536), phospho-ERK (T202/Y204) and phospho-JNK (T183/Y185). All antibodies were used at 1:1000 dilution in 5% BSA as blocking reagent, except α Tubulin and RelA, which were at 1:2000 dilution in 5% milk.
Immunofluorescence
Cells on chamber slides were fixed using 4% paraformaldehyde for 20 minutes, washed three times for 5 minutes in PBS, then permeabilized using 0.1% Triton for 2 minutes before repeat washing with PBS. Slides were then blocked in 10% horse serum for 1 h. NF-κB cellular localization was assessed using a RelA antibody (#H0714 Santa Cruz) used at 1:200 dilution in 1% horse serum and incubated overnight in the dark at 4 °C. Slides were then washed three times for 5 minutes and incubated with a goat anti-rabbit IgG fluorescein isothiocyanate (FITC)-conjugated antibody (#4050-02 Southern Biotech) (1:200) in the dark for 1 h at room temperature. Slides were then mounted in 1,4-diazabicyclo[2.2.2]octane (DABCO) for microscopy analysis. Control chambers received no primary antibody, but received secondary antibody at the same concentration at stated above. As an isotype control, FLS received FITC-conjugated rabbit IgG. Cells were imaged on a Zeiss AxioCam ERc5s with supporting Zen blue edition software.
Statistical analysis
Repeat dose data are represented either as raw values, or as fold changes from response to the first dose, which is normalised to one. Data are presented as mean with SEM error bars. Comparison of the first and second response to stimulus was conducted after normalisation. The Wilcoxon signed rank test was used, and significance is designated as p = 0.05 (*) or p = 0.01 (**). Comparison of the first and second response at equivalent time points was analysed by the Mann-Whitney U test, with p = 0.05 (*). Statistical analyses were performed using GraphPad Prism 6.0 (GraphPad Software).
Discussion
LPS tolerance involves a range of epigenetic and signalling mechanisms that extensively reprogramme the responses of macrophages to re-stimulation, suppressing the induction of overtly pro-inflammatory genes whilst sparing those with anti-microbial properties [
28‐
31]. This may be viewed as a form of innate immune memory that guards against excessive inflammation without compromising defence against pathogens. The present study was devised to identify equivalent negative regulatory mechanisms by which synovial fibroblasts constrain their responses to repeated pro-inflammatory challenge. Surprisingly, we instead found that synovial fibroblasts were primed by a first exposure to either TNFα or IL-1α, such that they secreted significantly greater amounts of IL-6 when challenged a second time, after a 24-h rest period.
Fibroblasts derived from RA synovium retain an imprinted, invasive phenotype throughout prolonged culture in vitro [
15,
16,
34]. Nevertheless, we found no evidence that the priming phenomenon was specific to RA-derived synovial fibroblasts. Similar priming responses were demonstrated by synovial fibroblasts from ostensibly healthy, non-inflamed joints of patients who did not have RA, and who did not later develop RA. A neonatal foreskin fibroblast line was primed by TNFα or IL-1α in a manner very similar to synovial fibroblasts, and was used in this study as a tool to explore mechanisms. In contrast, adult dermal fibroblasts (obtained from patients with OA and patients with RA at the time of joint surgery) showed no evidence of priming.
Synovial fibroblasts from joints affected by RA were previously shown to mount unremitting inflammatory responses, continuing to express IL-6 and other inflammatory mediators for at least four days in the presence of TNFα [
26]. The same group also demonstrated priming by TNFα of synovial fibroblasts in RA, resulting in enhanced responses to later challenge with interferons [
35]. In those studies no comparison was made with “normal” synovial fibroblasts. It is therefore possible that the properties described are characteristic of synovial fibroblasts as a whole, rather than acquired during the pathogenesis of RA. It is not a novel concept that fibroblasts differ in their patterns of gene expression and response to stimulation, according to the anatomical location from which they are obtained [
11,
36‐
38]. The important questions are why synovial fibroblasts should possess “inflammatory memory” [
35], and whether such memory could play a role in the development of chronic joint disease.
One argument is that fibroblast responses to pro-inflammatory stimuli are dictated by their usual pattern of exposure to such stimuli [
27]. In mucosal or dermal fibroblasts that are repeatedly exposed to pro-inflammatory insults, priming responses would be severely maladaptive. In contrast, fibroblasts from privileged sites such as the joint or the eye presumably seldom face sustained or repeated exposure to pro-inflammatory cytokines in healthy individuals, and may have evolved different responses to such danger signals. This hypothesis will be tested by investigating priming responses of fibroblasts from different anatomical locations of healthy and diseased individuals. Another possibility is suggested by the apparently gene-specific nature of the phenomenon, in which primed responses of IL-6 and CCL5 were enhanced, whereas that of IL-8 was not. Perhaps, as an inflammatory response evolves, differential priming allows the stroma to influence leukocyte recruitment and activation by modifying the profile of cytokines and chemokines expressed. This possibility is also under investigation.
Finally, the molecular basis of synovial fibroblast priming remains unknown. Increased expression of cytokine receptors is unlikely to provide an explanation, because of the gene-specific and signaling-pathway-specific effects described. Sustained activation of NF-κB has been implicated in the unremitting inflammatory response of synovial fibroblasts to prolonged TNFα exposure in RA [
26]. Although we found NF-κB activity to decline after withdrawal of TNFα, the dynamics of activation in response to the second challenge were altered, with sustained nuclear localization of the RelA subunit and phosphorylation of its serine 536. It appears that the first exposure of FLS to TNFα induces a transient burst of
IL6 gene expression, whereas the second exposure induces more sustained expression, dependent on prolonged activation of NF-κB. We speculate that priming influences the expression or function of negative feedback regulators required for the termination of the NF-κB activation signal.
Acknowledgements
We thank Jason Turner for helpful discussions. This report is independent research supported by the National Institute for Health Research/Wellcome Trust Clinical Research Facility at University Hospitals Birmingham NHS Foundation Trust. The views expressed in this publication are those of the author(s) and not necessarily those of the National Health Service (NHS), the National Institute for Health Research, or Arthritis Research UK.