Background
Genetic studies have shed light on our understanding of the causes of autoimmune diseases by identifying shared and unique risk loci among these diseases. However, in rheumatoid arthritis (RA), only a fraction of disease susceptibility can be explained by genetic variation [
1], and the temporal link between the break of self-tolerance and development of clinical disease remains elusive because circulating autoantibodies are detectable long before the onset of arthritis [
2]. In RA, stromal cells in the joint, fibroblast-like synoviocytes (FLS), exhibit an imprinted and epigenetically maintained aggressive phenotype, predisposing them to participate in an inflammatory positive feedback loop in response to the cues from the synovial environment [
3]. Identifying local tissue conditions able to initiate and perpetuate the ensuing inflammatory cycle is therefore of critical importance to understanding and intervening in the disease process.
Endoplasmic reticulum (ER) stress is a common cellular response to many of the conditions RA FLS encounter in the inflamed synovium [
4], and it occurs when the amount of newly synthesized proteins in the ER exceeds the organelle’s capacity to ensure their proper folding. The resulting accumulation of misfolded proteins in the ER triggers a set of signals collectively referred to as the
unfolded protein response (UPR), aimed at relieving the burden by slowing down the global translation rate while increasing production of a selected set of proteins, particularly ER chaperones [
5]. The UPR depends upon the triggering of inositol-requiring enzyme 1α (IRE1α), protein kinase R-like endoplasmic reticulum kinase (PERK), and activating transcription factor 6, sensors embedded in the ER membrane, by unfolded protein aggregates in the lumen. In response, IRE1α homodimerizes and causes the unconventional splicing of X-box binding protein 1 (
XBP1) messenger RNA (mRNA). This causes a frameshift mutation in
XBP1, making it a powerful transcription factor instrumental in restoring homeostasis. Additional transcription factors are activated by the two other sensors [
6].
Although primarily a safeguard for protein folding homeostasis, ER stress is tightly associated with immunological processes via crosstalk occurring between the UPR and inflammatory signaling pathways. For example, the decrease in translation rate caused by PERK activity limits expression of proteins with a short half-life, such as nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, alpha (IκBα), resulting in enhanced activation of the nuclear factor (NF)-κB pathway [
7]. Autophosphorylated IRE1α interacts with the tumor necrosis factor receptor-associated factor 2 adaptor molecule, facilitating activation of NF-κB and mitogen-activated protein (MAP) kinase pathways [
8]. Transcription factors involved in ER stress can directly drive expression of inflammatory gene products such as interleukin (IL)-6 and tumor necrosis factor (TNF) [
9], and elements of the UPR are necessary for maturation of several immune cell populations [
10,
11]. Consequently, ER stress has been linked to a number of human disease conditions, including autoimmunity, where it has been postulated to drive inflammatory activation, act as the source of or as an adjuvant for autoantigens, or contribute to pathology by modulating apoptotic pathways [
12,
13].
Despite this, understanding of the relevance of ER stress to pathology in RA is largely incomplete. Analysis of publicly available datasets of microarrays performed on synovial tissue has identified genes related to ER stress and protein processing in the ER as those most significantly differentiating between RA and osteoarthritis (OA) synovia, whereas no such difference was observed between OA and normal synovia [
14]. Prominent staining for ER stress markers was observed throughout RA synovial tissue, particularly in the lining layer, indicating that these differences were unlikely to reflect changes in numbers of minor cell populations. A similar enhancement of ER stress and ER stress signaling to the nucleus in synovial fluid macrophages has been observed, and the ER chaperone binding immunoglobulin protein (BiP) is an important regulator of synovial angiogenesis, synoviocyte proliferation and survival, and disease severity in animal models of RA [
14]. In experimental arthritis, strong expression of ER stress markers is observed during disease development in both synovial macrophages and fibroblasts [
15]. Whereas myeloid-specific targeting of UPR pathways resulted in decreased cytokine expression and ameliorated disease in K/BxN serum-induced arthritis [
16], studies involving RA FLS focus predominantly on changes in cellular viability and their potential consequences for synovial hyperplasia [
17]. Unlike other cell types, RA FLS are resistant to apoptosis induced by ER stress, likely due to enhanced rates of autophagy and proteasomal activity [
18,
19]. However, little is known about how ER stress changes the potential of FLS to directly modulate synovial inflammation, and recent studies have indicated that splicing of
XBP1 may be associated with the activation of RA FLS by Toll-like receptor (TLR) signaling, IL-1β, and TNF [
20]. The aim of this study was therefore to examine if ER stress could regulate inflammatory gene expression in RA FLS.
Methods
Patients and cells
FLS were derived from synovial biopsies obtained by needle arthroscopy from patients fulfilling the 2010 American College of Rheumatology/European League Against Rheumatism classification criteria for RA [
21,
22] and isolated as previously described [
23]. Healthy skin biopsies were obtained as resected material after cosmetic surgery, and dermal fibroblasts (DF) were isolated using the Whole Skin Dissociation Kit (Miltenyi Biotec, Leiden, The Netherlands) following the manufacturer’s instructions. FLS and DF were cultured in DMEM (Gibco/Thermo Fisher Scientific, Waltham, MA, USA) containing 10% FBS (Invitrogen/Thermo Fisher Scientific) and used for experiments between passages 5 and 10. Prior to stimulations, cells were incubated in medium containing 1% FBS overnight.
Monocytes were isolated from healthy donor buffy coats (Sanquin, Amsterdam, The Netherlands) using Lymphoprep (AXIS-SHIELD; Alere Technologies, Oslo, Norway) density gradient centrifugation followed by standard isotonic Percoll gradient centrifugation (GE Healthcare, Eindhoven, The Netherlands). Monocytes were plated in Iscove’s modified Dulbecco’s medium (IMDM; Invitrogen/Thermo Fisher Scientific), supplemented with 1% FBS, for 30 minutes at 37 °C, followed by the removal of nonadherent cells. Monocytes were differentiated into macrophages by 7 days of culture in IMDM containing 10% FBS, 100 μg/ml gentamicin, and 800 U/ml granulocyte-macrophage colony-stimulating factor (Tebu-Bio, Heerhugowaard, The Netherlands).
Cell stimulation
Escherichia coli 0111:B4 lipopolysaccharide (LPS) was ordered from Sigma-Aldrich (Zwijndrecht, The Netherlands) and used at 1 μg/ml. ER stress was induced by tunicamycin from Streptomyces sp. (10 μg/ml; Sigma-Aldrich) or thapsigargin at varying concentrations (Calbiochem/Merck, Amsterdam-Zuidoost, The Netherlands). Other stimulants used included IL-1β (1 ng/ml; R&D Systems, Minneapolis, MN, USA), polyinosinic:polycytidylic acid (pI:C; TLR3 agonist, 25 μg/ml; InvivoGen, San Diego, CA, USA), Pam3CSK4 (TLR1/2 agonist, 5 μg/ml; InvivoGen), flagellin (TLR5 agonist, 200 ng/ml; InvivoGen), SB202190 (p38 inhibitor,10 μM; Tocris Bioscience, Bristol, UK), U0216 (extracellular signal-regulated kinase [ERK] inhibitor, 10 μM; Tocris Bioscience), and c-Jun N-terminal (JNK) inhibitor IX (20 μM; Calbiochem/Merck).
Gene expression measurement
Total RNA was isolated using an RNeasy Micro Kit (QIAGEN, Venlo, The Netherlands) according to the manufacturer’s instructions and reverse-transcribed using a RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). Quantitative polymerase chain reaction (qPCR) reagents were purchased from Thermo Fisher Scientific, and reactions were performed using TaqMan probes and Master Mix (for detection of
HSPA5,
DDIT3,
ERN1) or SYBR Select Master Mix (all other targets) (Applied Biosystems/Thermo Fisher Scientific, Foster City, CA, USA). Alternatively, gene expression was measured using qPCR-based low-density arrays (QIAGEN). The custom array in use was previously designed to cover 84 genes relevant to joint pathology and regulated by proinflammatory stimuli in RA FLS [
24,
25].
Cell viability and apoptosis detection
RA FLS were exposed to thapsigargin at concentrations ranging from 10 nM to 1 μM for 4–24 h. Apoptosis induction was analyzed using the Cell Death Detection ELISA (enzyme-linked immunosorbent assay; Roche Diagnostics/Sigma-Aldrich, Mannheim, Germany) according to the manufacturer’s instructions. Viability was assessed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Following treatment, cells were incubated with 1 mg/ml thiazolyl blue tetrazolium bromide (Sigma-Aldrich) for 1 h at 37 °C. The water-insoluble reaction product was dissolved with isopropanol containing 5 mM HCl and 0.1% Nonidet P-40 and quantified by measuring absorbance at 595 nm.
ELISA
Cells were stimulated with 10 nM thapsigargin or 1 μg/ml LPS, alone or in combination, for 24 h. Cell-free supernatants were collected, and the concentrations of IL-6 and IL-8 were measured using PeliKine Compact human IL-6 and IL-8 ELISA kits (Sanquin) according to the manufacturer’s instructions.
Immunoblotting
Cells were stimulated with 10 nM thapsigargin or 1 μg/ml LPS or a combination thereof for 30 minutes and 1, 2, 4, and 8 h, and then they were lysed in modified Laemmli buffer (120 mM Tris-HCl, pH 6.8, 4% SDS, 4% glycerol). Lysates were combined with loading buffer containing β-mercaptoethanol, heat-denatured at 95 °C, resolved by SDS-PAGE electrophoresis, and blotted onto polyvinylidene fluoride membranes. Membranes were blocked with 4% nonfat dry milk for 1 h, followed by overnight probing with primary antibodies recognizing histone 3, IκBα, and phosphorylated forms of JNK, p38 (all from Cell Signaling Technologies, Leiden, The Netherlands), and ERK (Santa Cruz Biotechnology, Dallas, TX, USA). HRP-conjugated secondary antibodies were purchased from Dako/Agilent Technologies (Santa Clara, CA, USA), and proteins were detected using Lumi-Light enhanced chemiluminescence substrate (Roche/Sigma-Aldrich) and the ChemiDoc imaging system (Bio-Rad Laboratories, Hercules, CA, USA). Densitometric analysis of bands was performed with ImageJ software (
https://imagej.nih.gov/ij/). Band intensities were normalized to histone 3 signal in the sample and expressed relative to the unstimulated cells.
Statistical analysis
Data are presented as mean ± SEM. Statistical analysis was performed using Prism 6.02 software (GraphPad Software Inc., La Jolla, CA, USA). The number of replicates in the figure description refers to the number of different FLS donors included in the analysis. For comparison of multiple datasets with a single reference set, repeated measures analysis of variance followed by Dunnett’s post hoc test was used. For comparison between two datasets only, a paired t test was used unless otherwise indicated. All tests were two-tailed, and p values <0.05 were considered significant.
Discussion
ER stress plays an important role in both physiological and pathological immune responses, and RA and OA synovia are distinguished by a strong UPR signature [
14]. Similarly, macrophages isolated from RA patient synovial fluid were demonstrated by several groups to bear signs of ER stress, as compared with both peripheral blood monocyte-derived macrophages or macrophages isolated from OA patient synovial fluid [
14,
16]. In the present study, we demonstrate that also stromal cells, FLS, are readily responsive to ER stress induction, but by itself this has little significant effect on cellular activation, nor does it affect viability. Instead, ER stress primes stromal cells for enhanced cytokine and chemokine production in the presence of other agonistic signals. The magnitude and range of this synergistic response are in stark contrast with the negligible changes in cytokine expression induced by thapsigargin alone, and they are a result of effects on transcriptional rates, to some extent, and primarily on mRNA decay rates of inflammatory genes.
In regard to cell survival, RA FLS were previously described as more resistant to such challenge than OA FLS, with altered expression of CHOP and synoviolin postulated as possible mechanisms [
17,
18]. Similarly, the ER chaperone BiP has been identified as an important survival factor for stressed synoviocytes, and its expression regulates joint destruction in animal models [
14]. On the other hand, inflammatory responses to ER stress in FLS have been scarcely studied so far. Contrary to freshly isolated synovial fluid macrophages, cultured RA FLS do not show signs of increased ER stress [
20], although they upregulate UPR-related genes more readily than OA FLS in response to a variety of stimuli [
14]. Analogously to similar observations in macrophages [
9], a possible effect of TLR-dependent XBP1 activation on gene expression has been proposed in RA FLS [
20]. However, following stimulation with LPS alone, we have observed only minor differences in the amount and fraction of
XBP1 existing in the spliced form, indicating no significant changes in UPR signaling (data not shown).
Our data suggest that enhanced mRNA stability is a major contributor to the increased level of gene expression during ER stress. A growing number of reports underscore the importance of mRNA stability regulation during chronic synovitis in RA. In particular, Loupasakis et al. [
27] recently demonstrated mRNA stabilization as a crucial factor shaping the FLS transcriptome during long-term exposure to TNF, with a strong influence on
IL6,
IL8,
CCL2,
PTGS2, and other genes with pathogenic potential.
Intriguingly, ER stress has long been known to impact the mRNA stability of certain genes via regulated IRE1α-dependent degradation [
28]. In such cases, activated IRE1α was shown to splice not only
XBP1 but also several other mRNAs, resulting in their accelerated decay. However, the idea that ER stress might conversely contribute to inflammation by stabilizing cytokine mRNA has not previously been explored. Regulation of cytokine expression through changes in mRNA stability depends primarily on the presence of adenylate- and uridylate-rich elements in their sequences. These are recognized by adenylate- and uridylate-rich element-binding proteins (ABPs) whose expression and activity are tightly regulated and can lead to both positive and negative regulation of mRNA half-life [
29]. We have screened possible candidate ABPs, including BRF1 (
ZFP36L1), BRF2 (
ZFP36L2), AUF1 (
HNRNPD), TTP (
ZFP36), HuR (
ELAVL1) and
KHSRP, using small interfering RNA-mediated knockdown (data not shown), but we were unsuccessful in mimicking or significantly modulating the effects of combined LPS and thapsigargin stimulation by their independent targeting. Additionally, inhibition of conventional pathways involved in ABP-mediated decay [
26], such as p38, ERK, and JNK, did not block a positive effect of ER stress on mRNA stability. These observations indicate that additional regulatory layers, such as microRNAs or components of nonsense-mediated decay, may be implicated.
The observation that ER stress regulates mRNA stability in DF is similarly novel, suggesting a shared mode of stromal cell response to suspected injury by preparing to mount a rapid inflammatory response if further danger signals appear in the environment. This may be relevant to rheumatic diseases other than RA characterized by skin involvement, such as psoriatic arthritis and systemic sclerosis. In this regard, the role of TLR ligands and ER stress in systemic sclerosis has been described extensively (reviewed in [
30,
31]), and it will be of interest to determine whether ER stress-dependent regulation of gene expression contributes to the acquisition of the profibrotic phenotype in these patients.
Our inability to observe a similar effect of ER stress on mRNA stability in macrophages was surprising, given the available literature. For example, the IRE1α-XBP1 signaling pathway was shown to be a critical element of macrophage responses to TLR ligation [
9], and myeloid-specific knockout of IRE1α ameliorated disease severity in the K/BxN serum-induced arthritis model [
16]. Although the primary focus of these previous studies was the role of IRE1α during TLR stimulation alone, an enhancement of LPS-induced cytokine expression during ER stress in murine bone marrow-derived macrophages was noted, and a similar finding was observed in human macrophages [
9]. The discrepancy in macrophage responses to ER stress between these studies and ours, where we also noted a sensitivity of macrophages to LPS and ER stress-induced apoptosis, may be a result of differences in tissue- and polarization-specific macrophage responses. In line with this, it was previously observed that resident and thioglycolate-elicited peritoneal macrophages show opposite patterns of regulation of
CXCL1 during stimulation with LPS and thapsigargin [
32]. Our results suggest that in RA synovial tissue, the IRE1α-XBP1 axis might contribute to macrophage responses to TLR signaling in the absence of induction of ER stress, whereas in stromal cells, TLR stimulation in the presence of ER stress amplifies cytokine and chemokine production.
Ethics approval and consent to participate
FLS isolation was approved by the medical ethics committee of the Academic Medical Center, Amsterdam (METC 2013_069), and prior, informed, written consent was obtained from all patients. Healthy human skin samples were collected as discarded tissue after cosmetic surgery from anonymous donors who gave prior informed consent for the use of material in research. In accordance with local regulations, the use of this material is exempted from the separate ethical review process.