1. Introduction
Primary knee osteoarthritis (OA) is the most common joint disorder, affecting more than 10% of the Western European population aged 60 years or older [
1]. Joint pain and functional disability decrease quality of life in OA patients [
2]. OA affects hands, feet, knees, hips and spine [
3] with structural damage in the articular cartilage, subchondral bone, ligaments, tendons, menisci, muscles, synovium and nerve tissues [
4]. The ethiopathogenesis of joint pain in OA is not well understood [
5] and the therapeutic strategies to relieve OA pain are therefore limited. The correlation between the severity of pain and the degree of radiographic changes in OA knees is poor [
6,
7].
A number of studies substantiate the role of the central nervous system in the ethiopathogenesis of chronic pain in OA [
8,
9], nevertheless the sources of peripheral nociception are increasingly explored [
6,
10]. Peripheral pain can originate in any of the innervated joint structures, including the synovium [
5]. Synovial inflammation in OA causes joint pain [
11]. Inflammation decreases the activation threshold of local afferent nerve fibers in response to mechanical stimuli resulting in peripheral sensitisation. Additionally, interactions between a damaged joint and the sensory nervous system cause pain [
11]. Most likely, there are two interconnected mechanisms with reverse causality: Joint damage causes synovial inflammation, whereas neurogenic inflammation contributes to joint damage, creating a positive feedback loop [
12].
Genes involved in pain pathways in the nervous system are also expressed in non-neuronal cells of the joint [
13]. Epidemiology studies identified a panel of candidate genes and functional genetic variants associated with OA pain [
14]. To date, no hypothesis-free study with a genome-wide approach has been conducted to interrogate the molecular basis of OA pain.
Here we analyse by RNA sequencing the transcriptomes of knee synovial tissues from OA patients with high and low intensity knee pain as assessed by visual analogue scale (VAS). We uncover the molecular complexity of OA joint pain and reveal novel candidate genes associated with knee OA pain.
4. Discussion
Pain is a subjective and complex sensory experience and evaluation of pain is rather difficult. Among several methods for pain assessment [
28], we used VAS since it is a clinical routine, well accepted in scientific pain literature and it influences functional scores [
29]. Patients with contrast pain phenotypes were matched for gender (females), smoking status (non-smokers) and ethnic origin (Europeans) which can confound experiencing the pain [
6]. Additionally, patient groups with contrast pain phenotypes did not differ significantly in age (70.6 ± 6.5 vs. 71.0 ± 6.0 in high vs. low pain group), Body Mass Index (BMI) (31.1 ± 8.5 vs. 29.0 ± 5.1 in high vs. low pain group) or the extent of joint damage (3 and 4 in Kellgren and Lawrence grade system). The use of analgesics could decrease pain in the low pain group patients 021 and 022, but appeared inefficient in controlling the pain in the high pain group patients 016, 018, 019.
We uncovered a signature of genes that are differentially expressed in the synovial tissues from knees of OA patients with contrast pain phenotypes. The synovial inflammatory mediators that are known to induce and/or respond to pain [
12] were not enriched in the transcriptome profiles in synovial tissues from patients with high compared to patients with low knee pain. The deconvolution analysis of RNA-seq data showed no differences in the composition of immune cells or synovial fibroblasts between patients. Instead, the three top differentially expressed genes in patients with high pain are neuronal proteins. Two of them, specifically
SDIM1 and
CPE, are known for their roles in cellular responses to stress. This suggests a potential molecular connection between chronic pain and stress, as also recently demonstrated by Descalzi et al. [
30].
SDIM1 is involved in cellular response to stress and inhibition of cell death [
31]. In NT2 neurons SDIM1 exhibits a bi-phasic response to cell death-inducing injuries; initial down-regulation of SDIM1 is followed by an up-regulation in surviving cells [
31]. Overexpression of
SDIM1 improves survival of neuro-progenitor cells after injury, substantiating the pro-survival effects of SDIM1 under stress conditions [
31]. We show an increased expression of
SDIM1 in knee synovial tissues from OA patients with high pain. Our data suggest that acting as a stressor, the chronic pain might increase the expression of
SDIM1 in the synovial microenvironment of OA knees to activate its protective functions.
Besides
SDIM1,
CPE can steer cellular responses to stress and we show that also CPE is significantly up-regulated in the synovium of OA patients with high pain. CPE acts as a trophic factor and promotes neuronal survival upon various stressors via increasing anti-apoptotic protein BCL-2 Bcl-2 in a MEK/ERK and/or PI3-K/AKT-dependent manner [
32]. Stress-driven increase of CPE in brain has pro-survival effects in neurons [
32]. Considering the sensitivity of CPE in responding to different stressors, pain as a stressor might induce CPE expression in OA knees. CPE acts through a similar molecular mechanism like brain derived neurotrophic factor (BDNF) [
33] which is expressed in knee synovium of OA patients [
13], its expression however does not differ between OA patients with contrast pain phenotypes. This suggests that CPE and BDNF may not function together to promote cell survival in painful OA joints; a similar scenario proposed for hippocampal neurons [
32]. Instead, we show increased expression of
IGF1, which is known to have neuroprotective and pro-survival effects by activating ERK and Akt [
34].
Collectively, these results suggest that the cell protective functions of SDIM1 and CPE might be needed within the stress microenvironment of the OA synovium in patients with high pain. CPE and SDIM1 appear striking molecular candidates for future studies due to its potential joint protective effects.
OTOF was the second of the top three up-regulated genes in the synovium of OA patients with high pain. OTOF is a transmembrane protein required for calcium-dependent synaptic exocytosis in cochlear sensory cells [
35] and mutations in OTOF cause deafness [
35]. Besides, OTOF can modulate the GABAergic activity in the GAD65-dependent manner in neuronal and non-neuronal cells [
36]. Future studies might reveal whether OTOF can change the secretion of GABA in the OA synovium in response to pain, thereby modulating the pain transmission.
In the gene networks enrichment analysis using GO terms and Reactome pathways, the networks that include
SDIM1 and
OTOF did not appear enriched, possibly due to a rather limited published data on the function of these genes. Meanwhile, CPE was linked to the anatomical structure morphogenesis. Several other up-regulated genes in the synovium of OA patients with high pain did not appear in enriched pathways and the knowledge about most of these genes is rather limited. Nevertheless, some of these genes participate in nervous system development (
CLSTN2,
TUBB2B) [
37,
38], inflammation including arthritis (
AOC3) [
39] and stress-induced responses (
C10orf10) [
40], making them promising molecular candidates in OA-driven pain.
On the other hand, the neurotrophic tyrosine kinase receptor type 2 (NTRK2), also known as TrkB was present in every enriched GO term and in both enriched Reactome pathways. This might suggest that NTRK2 is one of the key drivers, involved in the pain-related transcriptional changes in the OA synovium. The expression of the neurotrophic factors that function via TrkB including BDNF, nerve growth factor (NGF), was not altered in OA patients with high knee pain. This suggests that other molecular candidates might function via TrkB to activate downstream signaling pathways in the synovium of painful OA joint.
To date no studies have explored the global alterations in the synovial noncoding RNA expression in relation to OA joint pain. We identified two differentially expressed lncRNAs and 35 differentially expressed miRNAs in the synovium of OA patients with contrast pain phenotypes, but the differences in the miRNA expression were not significant when correcting for FDR. In all this suggests that pain-related transcriptional changes primarily affect the protein coding transcriptome. Pain-associated changes in miRNAs in OA appear less frequent or smaller, thus larger sample sizes might be required for miRNA studies.
The two identified lncRNAs (
RN7SL3 and
RP11-195E2.1), which were strongly down-regulated in OA patients with high knee pain, have largely unknown function. In contrast, miR-146a-3p, the top differentially expressed miRNA in the synovium of OA patient with contrast pain phenotypes, has already been linked to the pain-related pathophysiology of knee OA [
41]. MiR-146a-3p, which regulates the cell repair responses to tissue damage. It was also shown to be up-regulated in the cartilage and synovium of OA patients compared with healthy controls, which directly links this miRNA to OA pathogenesis [
41]. We show that miR-146a-3p is down-regulated in the synovium of OA patients with high pain. This is in line with the observed down-regulation of miR-146a-3p in peripheral and central neurons in an animal model of OA-pain [
41], making miR-146a-3p a prominent molecular candidate in pain-related pathology in OA.