Nerve growth factor and receptor expression in rheumatoid arthritis and spondyloarthritis

Introduction We previously described the presence of nerve growth factor receptors in the inflamed synovial compartment. Here we investigated the presence of the corresponding nerve growth factors, with special focus on nerve growth factor (NGF). Methods mRNA expression levels of four ligands (NGF, brain derived growth factor (BDNF), neurotrophin (NT)-3, NT-4) and their four corresponding receptors (tyrosine kinase (trk) A, trkB, trkC, NGFRp75) were determined in the synovial fluid (SF) cells of 9 patients with rheumatoid arthritis (RA) and 16 with spondyloarthritis (SpA) and compared with 7 osteoarthritis (OA) patients. NGF was also determined in synovial tissue (ST) biopsies of 10 RA and 10 SpA patients. The production of NGF by monocytes and lymphocytes was assessed by flow cytometry of SF cells, synovial tissue derived fibroblast-like synoviocytes (FLS) were assessed by ELISA on culture supernatant. Results SF cell analysis revealed a clear BDNF and NGF mRNA expression, with significantly higher NGF expression in RA and SpA patients than in the OA group. NGF expression was higher in ST samples of RA as compared to SpA. Using intracellular FACS analysis, we could demonstrate the presence of the NGF protein in the two inflammatory arthritis groups on both CD3+ T lymphocytes and CD14+ cells, i.e. monocytes/macrophages, whereas cultured FLS did not produce NGF in vitro. Conclusions Neurotrophins and especially NGF are expressed in the synovial fluid and tissue of patients with peripheral synovitis. The presence of neurotrophins as well as their receptors, in particular the NGF/trkA-p75 axis in peripheral synovitis warrants further functional investigation of their active involvement in chronic inflammatory arthritis.


Introduction
There is increasing evidence for the presence of neuronal growth factors in chronic inflammatory arthritis. Neurotrophins (nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and the neurotrophins NT-3 and NT-4) constitute a family of growth factors essential for the development, proliferation, differentiation, and survival of neuronal as well as various non-neuronal cells. Neurotrophins bind to their specific high-affinity receptors tyrosine kinase (trk) A (NGF), trkB (BDNF, , and trkC (NT-3), and to one low-affinity receptor p75 (or NGFRp75) that binds to all ligands. This p75 receptor is a member of the TNF receptor superfamily [1]. In particular, the NGF-trkA/p75 axis arouses increasing interest due to its role in chronic inflammatory arthritis in which a pathogenic function of this system has been postulated [2][3][4][5][6].
Using immunohistochemistry, we previously performed a detailed analysis on the synovial expression of neurotrophins showing convincingly high levels of both the trkA and the p75 NGF receptors in peripheral synovitis of patients with spondyloarthritis (SpA). Their expression correlated with signs of inflammation and was modulated by effective treatment with anti-TNF [7]. However, apart from BDNF we were unable to demonstrate the presence of the ligands at the protein level. Accordingly, this study was designed to assess the expression of NGF and all other known neurotrophic ligands (BDNF, NT-3, NT-4), as well as their receptors (trkA, NGFRp75, trkB, trkC), in order to provide evidence that all actors of this system are present and actively upregulated in the inflamed synovial compartment.

Patients
Synovial fluid (SF) samples were collected from nine patients with rheumatoid arthritis (RA) fulfilling the American College of Rheumatology classification criteria and from 16 patients with SpA fulfilling the European Spondyloarthropathy Study Group classification criteria [8,9]. Also, seven patients with osteoarthritis (OA) served as non-inflammatory controls. All patients had active synovitis of the knee. Patient characteristics and disease activity parameters are listed in Table 1. OA patients were graded according to the Kellgren and Lawrence classification [10]. The majority of patients had OA of the knee grade 2 (mean 2.3 ± standard deviation (SD) 1.0).
Synovial tissue (ST) biopsies were obtained from another panel of patients including 10 SpA and 10 RA patients with early and active arthritis of the knee (disease duration <6 months, only NSAID treatment, but no steroids, no diseasemodifying anti-rheumatic drugs, or biologics).
All subjects gave their written informed consent before inclusion in the study, which was approved by the local ethics committee of the involved institutions.
Synovial fluid and synovial tissue biopsy samples and extraction of total RNA SF was obtained by a conventional puncture of an actively inflamed knee joint. Samples were centrifuged for 15 minutes at 1000 g. Supernatants were removed, whole SF cell pellets were resuspended in RNAlater solution (Ambion, Austin, TX, USA), and frozen at -80°C until use. For intracellular fluorescence-activated cell sorting (FACS) analysis, mononuclear cells derived from SF samples (SFMC) were obtained by standard Ficoll histopaque procedure and conserved in FCS and 10% dimethyl sulfoxide. They were kept in liquid nitrogen until use. ST biopsies were obtained by a standard procedure as previously described [11]. The Ficoll procedure was also used in order to obtain peripheral blood mononuclear cells (PBMC) from four healthy individuals used as controls for PCR. Total RNA was extracted from SF cell pellets, ST biopsies, and PBMC using TRizol reagent (Invitrogen, Karlsruhe, Germany) and precipitation with isopropyl alcohol. All procedures were previously described in detail [12][13][14].

Quantitative real-time RT-PCR (TaqMan assay)
Generation of cDNA by reverse transcription and utilization of TaqMan ® assay followed the protocol of the manufacturer Table 1 Clinical characteristics of neurotrophins and receptors in the synovial fluids of SpA, RA, and OA patients

Delta delta Ct method and statistical analysis
All PCR data were normalized to the expression of the glyceraldehyde 3-phosphate dehydrogenase (G3PDH) housekeeping gene used as an internal control; SF data were also normalized with a set of four healthy PBMC used as an external control. The ST data were compared with each other. PCR data were obtained as cycle threshold (Ct) values. The Ct value is defined as the cycle with a fluorescence intensity significantly above the background fluorescence but within the exponential phase of the amplification [15,16]. The mean of two Ct measurements of one sample was calculated for both the given target gene and the G3PDH gene. Delta Ct was determined as the mean of the duplicate Ct values for the target gene subtracted by the mean of the duplicate Ct values for the G3PDH gene. For each target gene, delta Ct measurements were performed separately for both the SF samples and the healthy PBMC. The delta delta Ct method represents the difference between the two cell types for a given target gene [17]. Expression levels are given as fold of expression and were compared between SpA, RA, and OA groups using the non-parametric Mann Whitney U test as appropriate.
Fluorescence-activated cell sorting for detection of NGF SF mononuclear cells (SFMC) of three RA and three SpA patients from the panel described in Table 1, as well as PBMC from two healthy controls were prepared by density gradient centrifugation using biocoll (1.077 g/ml; Biochrom, Berlin, Germany). SFMC were harvested from the interphase and washed twice at 1000 g and 300 g, respectively.

Culture of fibroblast-like synoviocytes to determine NGF production
Fibroblast-like synoviocytes (FLS) were isolated from synovial biopsies of one RA and one SpA patient as described previously [18]. After three passages, cells were resuspended at 10,000 cells/ml Dulbecco's modified eagle's medium (DMEM) with 10% FCS and plated at 2 ml/well in 24 well plates. Cells were grown for three days until confluence, then the normal medium (DMEM + 10% FCS) was replaced by starvation medium (DMEM + 1% FCS). After 24 hours, cells were stimulated with either medium alone (DMEM 1% FCS); TNF-alpha at 10 ng/ml; or IL-1 beta at 10 ng/ml, or lipopolysaccharide at 1 ug/ml. After 72 hours of culture with stimulation, supernatants were collected and used undiluted for measuring NGF by ELISA (R&D Systems, Minneapolis, MN, USA; lowest detection level: 30 pg/ml).

Results
The mRNA expression levels of the four neurotrophic ligands and the four receptors as determined in the SF cells are depicted in Table 2 and outlined in detail by scatter plots in Figure 1 (ligands) and Figure 2 (receptors).
As for the transcripts encoding the ligands, BDNF revealed high expression levels in all three groups (RA median: 163, SpA median: 92 with a high range from 71 to 444, OA median: 137). The highest mRNA expression in SF samples was found for NGF revealing significantly higher levels in RA and SpA (median 418 and 323, respectively) as compared with OA (median 49; P = 0.001 for both comparisons). Transcripts encoding NT-3 and NT-4 were expressed on lower levels and slightly higher by trend in RA as compared with SpA and OA. The NGF mRNA expression levels as determined in ST biopsies were found to be significantly higher in RA (mean expression level 1.6 ± 1.2 SD) as compared with SpA (mean expression level 0.7 ± 0.3 SD) patients (P = 0.02) indicating that NGF is produced locally, at least in the arthritic synovium of RA patients (Figure 3).
In agreement with our previous immunohistochemistry data on ST samples [7], the mRNA transcripts encoding both the highaffinity NGF receptor trkA and the common low-affinity receptor p75 revealed the highest expression levels in SF of the SpA and the RA group being significantly higher expressed as compared with the OA group. We also measured NGF expression by staining on a single cell level using flow cytometry. Concomitant staining of cell surface markers and intracellular NGF revealed the presence of NGF in T lymphocytes (CD3+) and monocytes (CD14+). In contrast, B lymphocytes (CD19+) and nearly all natural killer cells (CD16+) were NGF negative in patients as well as in healthy controls. In both the SpA and the RA group, percentages of NGF expressing T lymphocytes and monocytes were considerably higher as compared with healthy controls ( Figure  4). As we did notice clear mRNA expression for NGF not only in SF but also in ST, we additionally investigated the production of NGF by FLS. In vitro cultured FLS from RA as well as SpA did not secrete detectable levels of NGF, even upon stimulation with various proinflammatory cytokines (data not shown). Taken together, these data suggest that infiltrating T lymphocytes and myeloid cells are the main source of NGF in the inflamed peripheral joint.

Discussion
The present study is a descriptive comprehensive quantitative expression analysis of mRNA transcripts encoding the four known human neurotrophins and their four corresponding receptors in the synovial compartment of arthritis patients.
The presence of neurotrophic factors in the inflamed joint has been described earlier [2,3]. Focusing on the NGF/trkA-p75 axis in our own and other work, we could previously demonstrate high trkA and p75 NGF receptor expression at the protein level in the inflamed ST in peripheral SpA synovitis. This expression was correlated with inflammatory disease activity and was downregulated by TNF-blocking treatment indicating that their expression is not constitutive but actively modulated in inflammation [7]. However, the high-affinity receptors trkB and trkC as well as the ligands NGF, NT-3, and NT-4 were expressed in the minority of patients or not detectable by immunohistochemistry. In order to investigate the NGF/trkA-p75 axis as well as all other neurotrophic ligands and receptors at the transcript level, we used quantitative real-time RT-PCR to determine their expression in a larger panel of SpA and RA patients with active peripheral synovitis of the knee. Our data confirm the high expression of both the trkA and p75 NGF receptors at the transcript level in the synovial compartment of SpA and RA patients. Of note, we now provide evidence that the NGF ligand is also expressed in the SF and tissue biopsy samples of peripheral synovitis indicating that this system is active in chronic inflammatory arthritis. However, we need to state, that the high NGF transcript expression is in contrast to our previous ELISA data in SF samples [7]. This discrepancy between mRNA and protein expression has been reported earlier in studies on brain tissue [19]. Reasons for this phenomenon might involve post-transcriptional modifications of NGF [1,20]. We also can not exclude technical reasons such as the NGF antibody used for the previous quantitative immunoassay.
The cellular source of NGF in humans has been investigated in several studies. Under unstimulated conditions, NGF is produced mainly by CD4+ T and B lymphocytes [1,21]. Under inflammatory conditions such as allergy and arthritis, NGF can be produced, stored, and released by eosinophils, mast cells, lymphocytes, and synovial fibroblasts, as well as monocytes and macrophages [22]. Using intracellular FACS analysis, we could demonstrate the presence of the NGF protein in the two inflammatory arthritis groups on both CD3+ and CD14+ cells, that is, T lymphocytes and monocytes/macrophages, which are known to be involved in the major pathways of both SpA and RA. However, ST-derived FLS do not seem to produce NGF as measured by ELISA. This finding might indicate the mere pro-inflammatory potential of NGF as opposed to factors released by fibroblasts, which are predominantly involved in structural damage. On the other hand, we can not definitely rule out the production of NGF by FLS. One explanation would be, that FLS loose their ability to produce NGF when cultured    over three passages in vitro. Another explanation would be that NGF is produced but not secreted, at least not in large amounts. Nevertheless, FLS are most likely one of the targets of NGF.
To date, the functional role of neurotrophins in inflammatory joint disorders is unclear. A pathogenic role for the NGF/trkA-p75 axis and other neurotrophins has been postulated for airway inflammation [23], atopic dermatitis [24], psoriasis [25], inflammatory bowel disease [26], and arthritis [2][3][4][5][6][7]. In inflammatory syndromes, NGF has been attributed to upregulating TNF-alpha, promoting the differentiation of B cells to plasma cells, enhancing chemotaxis and production of superoxide by neutrophils [22]. NGF is also involved in humoral immune responses by acting as an autocrine survival factor maintaining the viability of memory B-cells and macrophages [27]. NGF and its receptors have also tissue remodeling capacities exerting a strong fibrotic stimulus on skin and lung fibroblasts [28]. Upon binding to trkA, NGF induces its auto-phosphorylation and subsequently the activation of both phospholipase PLCγ and protein kinase C, which in turn activates the mitogen-activated protein kinase pathway involving the c-jun N-terminal, the p38, and the extracellular-regulated protein kinases (ERK1/2) all of which have been identified in arthritis as well. Interestingly, the wnt proteins, which have been described as regulating neurotrophin expression [29], have recently also been found to play a pathogenic role in spondyloarthritis [30]. In addition, NGF has been identified as a proangiogenic factor, another significant pathogenic pathway in chronic inflammatory arthritis such as SpA and RA [31,32].

Conclusions
Taken together, this comprehensive analysis demonstrates the expression of the pleiotropic NGF and its two receptors in peripheral synovitis of SpA and RA. The knowledge of neurotrophin expression on cells from the inflamed synovial compartment in arthritis patients adds to the potential evaluation of pathogenic mechanisms and the development of new therapeutic strategies (e.g. the pharmacological blockade of the NGF receptor and their signaling pathway by using receptor antagonists). Our findings prompt further functional as well as clinical studies on the role of neurotrophins and their therapeutic potential in arthritis.