Toll‐like receptor involvement in adolescent scoliotic facet joint degeneration

Abstract Facet joint osteoarthritis is prevalent in young patients with adolescent idiopathic scoliosis (AIS) and might contribute to back pain. Toll‐like receptors (TLR) have been linked to cartilaginous tissue degeneration but their involvement in facet joint osteoarthritis in AIS patients is still unknown. We compared baseline gene expression levels of TLRs ‐1, ‐2, ‐4, and ‐6 in scoliotic and non‐scoliotic chondrocytes and found higher expression levels in scoliotic chondrocytes with significantly higher TLR2 levels. Furthermore, TLR expression correlated strongly and significantly with inflammatory and catabolic markers in scoliotic but not in non‐scoliotic chondrocytes. TLR activation with a synthetic TLR2/6 agonist resulted in a robust induction and release of pro‐inflammatory and catabolic factors which exacerbated proteoglycan loss in scoliotic but not in non‐scoliotic cartilage. We also detected a higher abundance of alarmins including S100A8/9 and biglycan in scoliotic cartilage. Finally, the small‐molecule antagonists Sparstolonin B and o‐Vanillin reduced catabolism following induction with naturally occurring alarmins and the synthetic TLR2/6 agonist. The high baseline expression, robust responsiveness and strong and significant correlation with proteases and pro‐inflammatory cytokines suggest that TLRs are key regulators of facet joint degeneration in AIS. Blocking their activity could therefore potentially modify disease progression.


| INTRODUC TI ON
Zygapophyseal also known as facet joints are synovial joints located on each side of the posterior spinal column of each motion segment.
They play a primary role in spinal stabilization and load bearing.
Studies show facet joints can carry up to 25% of the axial compression and 40%-65% of torsional and shearing forces in a healthy spine. [1][2][3] Balanced load dispersion between the intervertebral discs (IVD) and both facet joints is important to prevent degeneration of these cartilaginous tissues. Many studies have confirmed the detrimental effects of non-physiological loading on both facet joints and IVDs, as well as segmental degeneration following injurious loading. One such example is the presence of IVD and facet joint degeneration in adolescent idiopathic scoliosis (AIS). [4][5][6] AIS is the most

O R I G I N A L A R T I C L E
Toll-like receptor involvement in adolescent scoliotic facet joint degeneration prevalent orthopaedic condition (up to 4%) in young individuals aged from 10 to 16 worldwide that involves a progressive curvature of the spine. 7 The imbalanced load results in wedging that causes disc degeneration, which is thought to contribute to coronal deformation. 8 Similarly, we have recently reported facet joint osteoarthritis (OA) in AIS patients, although, how the altered biomechanics affect the facet joints is still unclear. Histologically, we found that scoliotic and non-scoliotic facet joint OA share many characteristics such as fibrillation of the cartilage surface, erosion, fissuring, osteophyte formation, increased cell density and elevated expression of pro-catabolic and pro-inflammatory factors which suggests that they are closely related. 4,9,10 The underlying pathology of facet joint OA is a catabolic and inflammatory environment that leads to degradation, loss of function and potentially pain. Pro-inflammatory cytokines are secreted by chondrocytes and synovial fibroblasts and cause catabolism by triggering the production of proteases degrading the extracellular matrix (ECM). [11][12][13] Interestingly, a growing body of literature links toll-like receptors (TLR) to cartilaginous tissue degeneration. [14][15][16] TLRs are a family of pattern recognition receptors with 10 members in human cells. They are activated by a variety of molecules derived from pathogens, endogenous danger related proteins released by stressed cells or fragmented ECM components. These endogenous proteins such as HSP60, HSP70, S100A8/9, biglycan, HMGB1, fibronectin and aggrecan-fragment belong to a group called alarmins and have been measured in higher abundance in degenerating cartilage. [17][18][19][20][21][22][23] TLR ligands selectively bind specific TLR hetero-or homo-dimers and generate an inflammatory downstream response. 24 Although TLR1-9 protein expression was found to correlate with degenerative changes in adult OA chondrocytes, most of the alarmins bind to either TLR2 which dimerizes with TLR1 and 6, or to TLR4-heterodimers. 25 TLR homo-or hetero -dimerization generally results in recruitment and activation of MyD88 through the Toll/IL-1R homology domain. This leads to the activation of the transcription factor NF-kB and downstream expression of pro-inflammatory, catabolic and pro-nociceptive factors. 16,24,[26][27][28] There has been a recent interest in using small-molecule inhibitors to block TLR activation. Two such molecules are o-Vanillin and Sparstolonin B which interfere with MyD88 recruitment preventing downstream signalling. [29][30][31] This study is focused on evaluating a potential involvement of the TLRs in AIS facet joint OA and evaluating the potential of inhibiting TLRs using small-molecule inhibitors as potential disease-modifying therapeutics.

| Gene expression analysis
Freshly isolated scoliotic (n = 18 from 9 donors) and non-scoliotic  (v/w)) basis for 4 days while the other half served as untreated control in 20× (v/w) explant media. After the treatment period, media were collected, and samples taken for histology as described below.

| Secreted protein analysis
Secreted S100A8/9 was measured in explant culture media using a Quantikine ELISA (R&D Systems). Secreted MMP3, MMP13, IL-6 and IL-8 were measured in explant culture media from tissue with or without TLR2 activation (Raybiotech). Measures were obtained using a VICTOR Nivo microplate reader (Perkin Elmer).   Culture media were collected after 48 hours.

| Statistical methods
For each comparison between scoliotic and non-scoliotic groups or control and treated groups, a parametric Student's t test was used. For correlations between TLRs and degenerative factors, the Pearson method was used to assess the degree of relation and significance of P < .05. Each statistical method and correlation was calculated using GraphPad Prism.

| Baseline gene expression of TLRs and degenerative factors
The proteinases MMP3 and MMP13 and the cytokines IL-1ß, IL6 and IL8 are often linked to OA in adults. 35 We have demonstrated that young patients with AIS show evidence of facet joint OA but Significance is evaluated at P < . 05   MMP3  MMP13  IL-1b  IL6  IL8  TLR1  TLR2  TLR4  TLR6  MMP3  MMP13  IL-1b  IL6  IL8  TLR1  TLR2  TLR4  TLR6   MMP3   MMP13   IL-1b   IL6   IL8   TLR1   TLR2   TLR4   TLR6   MMP3   MMP13   IL-1b   IL6   IL8   TLR1   TLR2   TLR4   TLR6   MMP3   MMP13   IL-1b   IL6   IL8   TLR1   TLR2   TLR4   TLR6   MMP3   MMP13   IL-1b   IL6   IL8   TLR1   TLR2   TLR4   TLR6   the expression of cytokines and proteases has not been well established. Here, we found the transcripts to be more abundant in scoliotic chondrocytes compared with non-scoliotic ( Figure 1A). MMP3 had a modest 1.28-fold increase while MMP13 showed a 2.37-fold higher expression in scoliotic chondrocytes. The pro-inflammatory cytokine IL-1ß had elevated mRNA levels in scoliotic chondrocytes, with a 2.97-fold higher expression compared to non-scoliotic samples ( Figure 1B). A 1.98-fold higher IL-6 expression was seen scoliotic tissues and IL-8 showed no difference between groups. TLR 1, -2, -4, -6 are the main members involved in alarmin and danger signal recognition and the subsequent induction of pro-catabolic and inflammatory mediators can contribute to tissue degeneration. 36,37 To evaluate the presence of this pathogenic mechanism in degenerating scoliotic cartilage, baseline mRNA expression of TLR receptors was evaluated from isolated chondrocytes ( Figure 1C).
Interestingly, all four TLRs evaluated were more abundant in scoliotic cartilage compared to non-scoliotic, with TLR2 having the largest difference with a significant 2.84-fold (P = .03) higher expression.
In the scoliotic samples, TLR6 expression was 1.71-fold higher, TLR1 expression was 1.49-fold higher, and TLR4 expression was 1.15-fold higher compared with non-scoliotic samples. These data reinforce the knowledge of a catabolic and pro-inflammatory state of degenerating facet joint chondrocytes with the added insight of an elevated TLR expression. For the non-scoliotic samples, only 3 out of 6 pairs correlated significantly (TLR1-TLR4, TLR1-TLR6 and TLR4-TLR6). Notably, TLR2 only had weak correlations with TLR1, -4 and -6, in non-scoliotic cells. The strongest correlation in non-scoliotic cells was between TLR4-TLR6 (Pearson = .864, P = .0002) Together, the data indicate a strong relationship between TLR expression with inflammation and catabolism in degenerating facet joints.

| TLR activation-induced production of catabolic and pro-inflammatory factors
To evaluate the effect of TLR activation, we subjected facet joint chondrocytes to the potent synthetic TLR2/6 agonist Pam2CSK4.
Gene expression analysis revealed a significant increase of IL-6, IL-8, activation. Scoliotic explants produced higher levels than non-scoliotic, disregarding the initial amount measured in untreated controls.

| TLR activation-induced cartilage ECM degradation
Proteoglycan loss is a key feature of OA. 38 To evaluate the effect of TLR2/6 activation on proteoglycan content, safranin-O fast green staining was performed on ex vivo treated scoliotic and non-scoliotic cartilage explants ( Figure 2C). Proteoglycan content was measured semi-quantitatively ( Figure 2D). In scoliotic tissues, the cartilage had a lower baseline proteoglycan content compared with the nonscoliotic cartilage and lost significantly more (P = .003) following TLR2/6 activation. In contrast, TLR2/6 activation did not cause an as robust proteoglycan loss in non-scoliotic samples. Proteoglycan release in the culture media was measured with the DMMB assay and confirmed histological findings with increased release from scoliotic cartilage after TLR2/6 activation ( Figure 2E). There was no significant difference in proteoglycan release in non-scoliotic cartilage after TLR2/6 activation. Together, these findings support a role for TLR2/6 in the degenerative cycle of scoliotic facet joint cartilage.

| Alarmins in scoliotic cartilage
An increase in abundance of alarmins activating TLRs has been described in joints affected by OA in adults. 38 Mass spectrometry and F I G U R E 2 A, Gene expression analysis of TLR2/6 agonist (TLR2/6A) treated scoliotic and non-scoliotic chondrocytes compare to nontreated controls (n = 5). B, MMP3, MMP13, IL-6 and IL-8 secretion analysis of ex vivo facet joint cartilage cultured media from scoliotic (n = 10) and non-scoliotic groups (n = 10) after 4 d of TLR2 activation with Pam2csk4. C, Safranin-O fast green histology of scoliotic (n = 30) and non-scoliotic (n = 13) facet joint cartilage before (Control) and after (Pam2csk4) TLR2 activation for 4 d. D, Red staining quantification for proteoglycan content using a MATlab script. E, GAG content analysis in the cultured media before and after Pam2CSK4 treatment. Paired Student t test were performed to assess significance defined by *P < .05, **P < .01, ****P < .001 ELISA assays were used to reveal the alarmin secretion profile of scoliotic and non-scoliotic cartilage ( Figure 3A). The z-score heat map shows an increased level of alarmins released from scoliotic cartilage. In contrast, protein levels were frequently below the detection threshold non-scoliotic cartilage. Notably, cytoplasmic alarmins such as heat shock proteins (HSP) were found in all scoliotic samples but were mostly undetected in the non-scoliotic group. S100A8/A9 has been strongly linked to adult OA. 39 Although not detected with mass spectrometry analysis we set out to quantitively measure the concentration. Scoliotic cartilage released 2.9-fold (P = .03) more S100A8/A9 than non-scoliotic. The concentration was 10.33 ng/ mL in scoliotic and 3.57 ng/mL in non-scoliotic cartilage ( Figure 3B).
These results indicate an increased release of alarmins from degenerating scoliotic cartilage that could activate TLRs.

| Alarmins-induced pro-inflammatory and catabolic factor production
Alarmins and TLR activation are believed to contribute to the chronic low-grade inflammatory state present in adult OA. 40 To evaluate the potential of naturally occurring alarmins to participate in facet joint cartilage degeneration, isolated scoliotic chondrocytes were subjected to S100A8/9 and biglycan, which are both released by scoliotic cartilage. A lowered concentration of the TLR2/6 agonist was used in these experiments to more accurately compare to the alarmin-induced responses. As expected, the agonists caused an increase in IL-6 an IL-8 gene expression ( Figure 3C). S100A8/A9 had the strongest effect on IL-6 with a significant 5.14-fold (P = .0009) increase. MMP3 and MMP13 gene expressions were only affected by S100A8/A9, with a 2.34-fold increase. Biglycan treatment increased TLR2 expression with a modest 1.829-fold increase. As expected, the low-dose TLR2/6 agonist increased gene expression of the 4 genes to a similar level as found with S100A8/9. These results suggest a role for alarmins in scoliotic facet joint degeneration.

| Sparstolonin B and o -Vanillin reduced alarmininduced TLR activation
As TLR expression is elevated and the chondrocytes are strongly responsive to alarmins in scoliotic cartilage blocking their activity could potentially modify disease progression. To assess the effect of blocking TLR activation, two naturally derived compounds that prevent the recruitment of MyD88 to the TIR domain of TLRs, and thus TLR signalling, were used. 31,41 The antagonists did not significantly modulate gene expression in scoliotic chondrocytes challenged with biglycan ( Figure 4A,B).
In contrast, MMP3 expression was reduced by 2.14-fold (P = .0008) with Sparstolonin B treatment, and MMP13 gene expression was unchanged by both antagonists, although a small decreasing trend after o-vanillin treatment was found. This indicates that TLR inhibition overall reduces the alarmin-induced pro-inflammatory response.

TLR activation has been linked to an increased expression of
TLRs. This mechanism can drive a vicious cycle in the presence of F I G U R E 3 A and B, Mass spectrometry and ELISA of alarmin and S100A8/9 abundance in scoliotic and nonscoliotic cartilage explant conditioned media (n = 6). C, Gene expression analysis of IL-6, IL-8, MMP3, MMP13 after stimulation by alarmins S100A8/9 and Biglycan and low-dose TLR2/6 agonist treatment (TLR2/6A) in scoliotic facet joint chondrocytes (n = 4). Significance was evaluated by Student t tests (*P < .05, ***P < .001)  Fold change compared to untreated control Alarmins S100A8/9 S100A8/9 Biglycan TLR2/6A C abundant alarmins such as in degenerating cartilage. 42 We therefore evaluated TLR gene expression in response to activation in scoliotic cartilage. TLR4 gene expression was unchanged following biglycan, S100A8/9 and TLR2/6 agonist treatment. (Figure 5A). S100A8/9 had the strongest effect and induced TLR1, -2 and -6 expression by

| D ISCUSS I ON
Our findings suggest that TLR activation is contributing to scoliotic facet joint degeneration. We found higher TLR gene expres- A current hypothesis of the catabolic shift and chronic inflammation in osteoarthritic cartilage implicates the formation of alarmins through tissue degradation with a subsequent TLR activation. 43 We confirmed that this mechanism is active in facet joints of young patients with scoliosis. cartilage with increasing OA severity in adults. 25 Notably, TLR2 had the largest and most significant difference between the two groups which is in accordance with data from osteoarthritic knee chondrocytes in adults. 26 The amount of highly significant and strong cor- in OA, where overexpression of MMP13 was shown to be sufficient to cause osteoarthritis in mice. 44 Another plausible explanation is a modulation of TLR signalling by a yet unknown mechanism in scoliotic facet joint OA, which should be investigated further.
To better assess the potential of TLR activation in degenerating scoliotic facet joints, we quantified the alarmin profile. The observable trend of increased abundance of alarmins in scoliotic samples is corroborated by studies showing increased alarmins in degenerating cartilage of adults. 38 The alarmins detected has been described in the literature to activate TLR1,-2,-4, or -6. For example, the extracellular matrix molecule biglycan interact with TLR2 and TLR4 and induces pro-inflammatory molecules in macrophages. 20 S100A8/9, which was released at a significantly higher level from scoliotic cartilage, has been described to induce pro-inflammatory responses in macrophages. 19 Here, we evaluated the effect of biglycan and S100A8/9 in scoliotic chondrocytes where treatment increased gene expression of IL-6, IL-8, MMP3 and MMP13, which is similar to experiments done OA chondrocytes. 39 The alarmins had a similar response to a low concentration of the synthetic TLR2/6 agonist, mimicking a chronic low-grade inflammation found in OA.
The difference in gene expression induction between high and low concentration of the synthetic TLR2/6 agonist reflect an increase proportional to concentration. Further studies are needed to fully elucidate the detrimental role of TLR activation following exposure to alarmins, as every alarmin has a different binding potential that could result in varying downstream effects.
Although this study suggests the involvement of TLRs in scoliotic facet joint OA, the clinical relevance of blocking TLRs to slow disease progression still needs to be evaluated. In support of our suggestion, a joint-saving outcome of blocking TLRs has been described in an experimental osteoarthritis models. 45 The successful reduction of alarmin-induced protease and pro-inflammatory cytokine production by Sparstolonin B and o-Vanillin opens the possibility of using small-molecule inhibitors to suppress chronic inflammation and catabolism in scoliotic facet joints. Therefore, blocking TLR signalling should be studied further to explore a potential disease-modifying effect of such inhibitors.
A limitation in this study was the age differences between scoliotic and non-scoliotic subjects. The scoliotic cohort had a lower and homogeneous age range whereas non-scoliotic organ donors had a wider age range and a higher average age. This is explained by the window at which patients with AIS undergo corrective surgery and the variable age of organ donors. However, there was a variability in curve severity and degree of OA in AIS patients whereas all non-scoliotic organ donors had healthy spines with no signs of OA and no prior spine deformities. Therefore, we believe that the heterogeneity in age is counterbalanced by the clear differences in cartilage health between the groups.

| CON CLUS IONS
In conclusion, our results suggest that TLRs are integral to cartilage degeneration in scoliotic facet joints. The higher baseline expression F I G U R E 5 A, TLR gene expression after stimulation with alarmins S100A8/9 and Biglycan B) in conjunction with antagonists Sparstolonin B and o-Vanillin. Significance between control and treatment was evaluated at *P < .05, **P < .01 of TLRs in scoliotic cartilage and the strong and significant correlation with proteases and pro-inflammatory cytokines suggests that they are key regulators of tissue degradation. Taken together, these findings provide an insight into a potential target for future molecular therapies aiming at restoring tissue homoeostasis and prevent tissue degradation and loss of function.

ACK N OWLED G EM ENTS
The authors would like to acknowledge funding support for the study from the Shriners Hospitals for Children and from the RSBO (Réseau de recherche en Santé Buccodentaire et Osseuse). We would also like to thank Dr Derek Rosenzweig for reading the manuscript and providing comments on the experimental procedure.

CO N FLI C T S O F I NTE R E S T
The authors have no conflict of interest to disclose. Dr Lisbet

DATA AVA I L A B I L I T Y S TAT E M E N T
The data generated in this study are available from the corresponding author on reasonable request.