Periostin increased by mechanical stress upregulates interleukin‐6 expression in the ligamentum flavum

Ligamentum flavum (LF) hypertrophy is a major cause of lumbar spinal canal stenosis. Although mechanical stress is thought to be a major factor involved in LF hypertrophy, the exact mechanism by which it causes hypertrophy has not yet been fully elucidated. Here, changes in gene expression due to long‐term mechanical stress were analyzed using RNA‐seq in a rabbit LF hypertrophy model. In combination with previously reported analysis results, periostin was identified as a molecule whose expression fluctuates due to mechanical stress. The expression and function of periostin were further investigated using human LF tissues and primary LF cell cultures. Periostin was abundantly expressed in human hypertrophied LF tissues, and periostin gene expression was significantly correlated with LF thickness. In vitro, mechanical stress increased gene expressions of periostin, transforming growth factor‐β1, α‐smooth muscle actin, collagen type 1 alpha 1, and interleukin‐6 (IL‐6) in LF cells. Periostin blockade suppressed the mechanical stress‐induced gene expression of IL‐6 while periostin treatment increased IL‐6 gene expression. Our results suggest that periostin is upregulated by mechanical stress and promotes inflammation by upregulating IL‐6 expression, which leads to LF degeneration and hypertrophy. Periostin may be a pivotal molecule for LF hypertrophy and a promising therapeutic target for lumbar spinal stenosis.


| INTRODUCTION
Lumbar spinal stenosis (LSS) is a common disease in the elderly that causes low back pain, leg pain, and gait disturbance, leading to the impairment of their quality of life and activities of daily living. It is estimated that more than 10% of people over the age of 70 suffer from LSS symptoms. 1 Ligamentum flavum (LF), the so-called yellow ligament, connects the laminae of two adjacent vertebrae and stabilizes spinal motion. LF hypertrophy is a major factor contributing to LSS development and is highly associated with the pathogenesis of LSS. 2 The molecular mechanisms of LF hypertrophy are not fully elucidated, and there is no effective treatment to prevent or thin the hypertrophied LF without surgical resection.
The number of studies focusing on LF hypertrophy has recently increased. 2 Using human hypertrophied LF tissue derived from patients with LSS, several molecules were found to be related to LF hypertrophy, 2 and gene expression profiles were also investigated using microarrays. 3 However, the occurrence of human LF hypertrophy is highly diverse in terms of age, sex, severity of stenosis, disease duration, and comorbidities, and there is great difficulty in unifying the background related to the causative factors to clarify the effect of one factor. A standardized laboratory animal model is therefore advantageous for the study of LF hypertrophy. Although mechanical stress is thought to play an important role in the pathogenesis of LF hypertrophy, the underlying mechanism is not well understood. Thus, in a previous study, we established a novel rabbit model of LF degeneration and hypertrophy. 4 In this model, spinal fusion of two adjacent segments increases intervertebral mechanical stress and causes the disruption of elastic fibers, and an increase in collagen synthesis and LF hypertrophy, similar to the findings of human hypertrophic LF, is observed over time. 5 Furthermore, DNA microarray analysis, which was performed in a shortterm (16 weeks) rabbit model, confirmed that 680 genes respond to mechanical stress. 6 Although it is established that aging is an important factor in LF hypertrophy, a single short-term time point was investigated in our previous study. Therefore, it is necessary to investigate the effects of long-term mechanical stress to further elucidate the underlying mechanism.
RNA-seq can detect transcripts with a wide and dynamic range of expression levels and a very low background signal. 7 Thus, in this study, we aimed to evaluate the effect of long-term (1 year) mechanical stress on gene expression profiles using RNA-seq. Furthermore, we referred to microarray data (short-term rabbit model 6 and human LF 3 ) and focused on genes that were co-upregulated in all the datasets. An overlap of three genes was detected and the upregulation of these genes was confirmed in human LF tissues. Among these genes, we focused on periostin, which is reportedly involved in the response to mechanical stress in various tissues 8 and investigated the effect of mechanical stress on the production of periostin in LF cells and its involvement in the pathophysiological mechanism underlying LF hypertrophy.

| Animal experiments
All experimental animal procedures were approved by the Osaka Metropolitan University Graduate School of Medicine Committee on Animal Research (approval number: 15007) and conducted in accordance with the regulations of the approving body.
Eighteen-week-old male New Zealand white rabbits were used in this study. Based on the surgical procedure, the rabbits were randomly divided into two groups ( Figure 1). The mechanical stress group underwent resection of the L3-4 supraspinal muscle and both L2-3 and L4-5 posterolateral fusion with metal implant to increase the mechanical stress at the L3-4 level ( Figure 1A,B; n = 7). All surgical interventions were performed according to previously reported methods. [4][5][6] Briefly, after a dorsal midline skin incision, the lumbosacral fascia to the left of the mammillary process was cut. The postero-lateral side of the vertebral body was exposed from the intermuscular plane between the multifidus and longissimus. Then, to generate mechanical stress increase at the L3-4 level, the L3-4 supraspinal muscle was resected and a four-hole titanium locking plate (Universal Mandibular System; Leibinger, Stuttgart, Germany) was placed on the posterolateral side of L2-3 and L4-5 and fixed with a 2.0 mm titanium locking screw. The rabbits which underwent surgical exposure without fixation acted as the control group (n = 6). All rabbits were housed in separate cages (350 mm * 350 mm * 527 mm) with free access to water and food. All rabbits were sacrificed one year after surgery.
To investigate the effect of mechanical stress, the LFs obtained at the intervertebral stress concentration (ISC; L3-4 level of the mechanical stress group), fused (L2-3 and L4-5 levels of the mechanical stress group), and control (L2-3, L3-4, and L4-5 levels of the control group) levels were compared.  ) with lumbar disc herniation (LDH), cauda equina tumor, or lumbar spondylolysis. LF thickness at the facet joint level was evaluated using T2-weighted magnetic resonance (MR) images following a previously reported method 9 ( Figure 2). The demographic data of the patients are presented in Table 1

| Integrated analysis of transcriptomic data
The gene expression profile in the LF at the ISC level was compared with that at the control level, and that at the fused level was compared with that at the control level (three samples in each group). The genes showing a significant difference in both comparisons were considered as important genes closely related to mechanical stress that causes degenerative changes in the LF. A twofold change was considered a significant difference. Principal component analysis was performed using the iDEP platform, and gene ontology analyses were performed using DAVID Bioinformatics Resources 6.7.
We had previously performed DNA microarray analysis on a short-term rabbit model (GSE204798) and performed the comparisons described above, confirming that 680 genes respond to mechanical stress. 6 In contrast, analysis of the human hypertrophic and non-hypertrophic LF gene expression profile (GSE113212) identified 197 upregulated genes involved in LF hypertrophy. 3 In this study, we focused on the genes that were upregulated in all datasets.

| Reverse transcription-polymerase chain reaction
For reverse transcription-polymerase chain reaction (RT-PCR) analysis, first-strand cDNA was synthesized from total RNA using the PrimeScript IV 1st strand cDNA Synthesis Mix (Takara Bio, Shiga, Japan). RT-PCR was performed using RT-PCR kits (SYBR Premix Ex Taq, Takara Bio, Shiga, Japan) and a 7500 Fast Real-time PCR system (Applied Biosystems, Foster City, CA) according to the manufacturer's instructions. Gene expression

| Histological examination
Spinal motion segments with the rabbit LF from L2-3 to L4-5 were harvested and fixed with a 4% paraformaldehyde phosphate buffer solution for 48 h. The specimens were then dehydrated and embedded in paraffin after demineralization in Morse's solution (Wako Pure Chemical Industries, Osaka, Japan) for 3 weeks. To standardize the sliced section in each specimen, sections from the L2-3 to L4-5 segments with facets and LFs were sliced along the axial plane at the junction of the right and left inferior articular processes (4 mm thick).
Human hypertrophied LF samples from the 14 LSS group (seven females, mean age 74.2 years [range, 53-84]) and 10 non-hypertrophied LF samples from control patients (two females, mean age 34.9 years [range, ) were fixed in 10% neutral-buffered formalin for 24 h, followed by immersion in 70% ethyl alcohol for 48 h. The samples were embedded in paraffin. From each sample, axial 4-mm slices were obtained.
Hematoxylin and eosin (HE) staining was performed to evaluate the general conditions and features of the tissue. Elastic and collagen fibers were assessed using Elastica van Gieson (EVG) staining.

| Immunohistochemistry
After deparaffinization and rehydration, to retrieve the antigens, sections for immunohistochemical (IHC) analysis T A B L E 2 Primers used for quantitative reverse transcription polymerase chain reaction (rabbit)

| Isolation and culture of LF cells
LF cells were isolated from six patients with LSS (one female, average age, 79.2 years [range, 71-87]) for primary cell culture. Isolation was performed according to previously reported methods. 10,11 Briefly, after the removal of excess tissue, LF tissue was cut into pieces of approximately 0.5-1 mm 3 . The chopped LF tissues were then digested using type I collagenase (Worthington Biochemical Corp, Lakewood, NJ, USA) at 37°C for 1 h. Subsequently, digested LF tissues were cultured with Dulbecco's modified Eagle Medium/F12 (DMEM; Life Technologies Corp., Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, Life Technologies, NY, USA) in a 5% CO 2 humidified incubator. LF cells from each patient were cultured separately, and further in vitro experiments were performed on six lineages of LF cells. All experiments were performed at the 4th cell passage.

| siRNA transfection
LF cells were cultured in six-well plates at 50-70% confluence. LF cells were transfected with periostin siRNA or control siRNA (Dharmacon, Lafayette, CO, USA) using Lipofectamine RNAiMAX reagent (Invitrogen, Carlsbad, CA, USA), according to the manufacturer's recommendations. Twenty-four hours later, the composite transfection mixture was removed and replaced with serum-starved DMEM.

| Application of fluid flow shear stress (FFSS)
LF cells (1.0 × 10 5 per well) were seeded into six-well plates with 2 ml of DMEM supplemented with 10% FBS. After cell attachment to the wells, LF cells were cultured in serum-free DMEM and pretreated with siRNA or integrin blocker (0.5 μM cilengitide [MedChem Express, Monmouth Junction, NJ, USA]), nuclear factor kappa B (NF-κB) blocker (1 μM Bay11-7082 [MedChem Express, Monmouth Junction, NJ, USA]). Both integrin blocker and NF-κB blocker were dissolved in dimethyl sulfoxide (DMSO), the same amount of DMSO was added to the controls for both blockers. After pretreatment, the six-well plates were placed on the platform of an orbital shaker (SK-O330-Pro, Dlab, DLAB Scientific Co., Ltd., Beijing, China) and housed in an incubator. The orbit of the platform was circular, with a radius of 10 mm and a rotation rate of 210 rpm. LF cells were subjected to shear stress for 48 h. LF cells cultured in a similar 37°C environment without disturbance were used as static (no flow) controls.

| Stimulation of LF cells with transforming growth factor-β1
To analyze the biological response of LF cells to transforming growth factor-β1 (TGF-β1), LF cells (1.0 × 10 5 per well) were seeded into six-well plates with 2 ml DMEM supplemented with 10% FBS. After cell attachment to the well, the cells were cultured in serum-free DMEM and treated with 10 ng/ml recombinant TGF-β1 (Wako Pure Chemical Industries, Osaka, Japan) for 48 h.

| Stimulation of LF cells with periostin
To analyze the biological response of LF cells to periostin, LF cells (2.0 × 10 4 per well) were seeded into 24well plates with 0.5 ml DMEM supplemented with 10% FBS. After cell attachment to the well, the cells were cultured in serum-free DMEM and treated with 10 μg/ ml recombinant periostin (R&D Systems, Minneapolis, MN, USA) for 24 h.
2.14 | Enzyme-linked immunosorbent assay IL-6 protein levels in the culture medium were detected using a human IL-6 enzyme-linked immunosorbent assay (ELISA) kit according to the manufacturer's protocol (D6050, R&D Systems, Minneapolis, MN, USA). The IL-6 assay sensitivity was 0.7 pg/ml, and the assay range was 3.13-300 pg/ml.

| Western blot
Total cell lysates were extracted using RIPA lysis buffer (Wako Pure Chemical Industries, Osaka, Japan) and centrifuged. Supernatants were collected, and protein concentration was determined using a BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA, USA). Proteins were loaded on a 12.5% polyacrylamide gel, transferred electrophoretically to a polyvinyl difluoride membrane (Bio-Rad, Hercules, CA, USA), and blocked in 5% skim milk in Tris-buffered saline containing 0.1% Tween 20 (TBS-T) for 10 min at room temperature. Membranes were incubated with rabbit polyclonal anti-periostin (1:500, ab14041, RRID: AB_2299859, Abcam, Cambridge, UK) and rabbit monoclonal anti-GAPDH (1:1000, ab181602, RRID: AB_2630358, Abcam, Cambridge, UK) antibodies at 4°C overnight. Subsequently, membranes were washed with TBS-T and incubated with anti-rabbit horseradish peroxidaseconjugated secondary antibodies (1:5000, 7074, RRID: AB_2099233, Cell Signaling Technology, Danvers, MA, USA) for 2 h at room temperature. A chemiluminescence reagent (ImmunoStar LD, Wako Pure Chemical Industries, Osaka, Japan) was used for signal enhancement. An imaging system (Fusion Solo System, Vilber Lourmat, Collégien, France) was used for signal detection. The band signal intensities were quantified using Fusion software (Vilber Lourmat). The value for periostin was adjusted to the value for GAPDH, and data are presented as relative intensity to that of the control.

| Statistical analysis
To compare continuous variables, data were statistically analyzed using the Kruskal-Wallis test followed by the Mann-Whitney U test. Differences in categorical variables were examined using the chi-square test. Statistical significance was set at a p value of <.05. Spearman's rank correlation coefficient was used to assess the correlation between mRNA levels and LF thickness. All p values were two-sided. R (version 3.5.1, patched, http://www.r-proje ct.org; The R Foundation, Vienna, Austria) was used for all the statistical analyses and GraphPad Prism 9.4.0 Software (GraphPad Software, CA) was used for generating graphs.

| Mechanical stress concentration induces LF hypertrophy and degeneration in a rabbit model
Long-term (1 year) mechanical stress concentration by adjacent segment fusion ( Figure 1A,B) causes LF degeneration and hypertrophy in rabbits, as previously reported. 4,5 HE staining showed a hypertrophied LF with disrupted fiber alignment at the ISC level (level; L3/4) compared with that at the fusion level (L2/3 or L4/5) and control level in the sham group ( Figure 1C). In EVG staining, elastic fibers are stained black and collagen fibers are stained pink. The fusion and control levels showed well-aligned rich elastic fibers, whereas the ISC level showed a decrease in elastic fiber density ( Figure 1D).

| Integrated analysis of transcriptomic data identify periostin, biglycan, and HAPLN1 as differentially expressed genes
To identify mechanical stress-caused changes in the gene expression profile of the LF, RNA-seq analysis was performed in a long-term rabbit model. Differential gene expression analysis among rabbit LF tissues harvested from the control, fused, and ISC levels ( Figure 3A) revealed that 942 genes at the ISC level were significantly upregulated compared with those at the control level. In contrast, 617 genes at the fused level were significantly downregulated compared with those at the control level. A total of 33 genes were identified in the overlap between those upregulated by ISC and downregulated by fusion ( Figure 3B). Among the fluctuating genes, extracellular matrix (ECM)-related genes were enriched ( Figure 3C). Furthermore, microarray data (short-term rabbit model and human LF) identified that the following three genes overlapped among all datasets: periostin, biglycan, and HAPLN1 ( Figure 3D).
To validate the change in the expression of periostin, biglycan, and HAPLN1 in the LF, RNA was extracted from rabbit LF tissues for quantitative RT-PCR (qRT-PCR) analysis. As in the RNA-seq analysis, the expression of these three genes was increased at the ISC level and conversely decreased at the fused level compared with that at the control level ( Figure 3E).

| Periostin, biglycan, and HAPLN1 expression in human LF tissue
To confirm the gene expression levels of periostin, biglycan, and HAPLN1 in human LF tissue, hypertrophied LF tissue was harvested from patients with LSS (Figure 2A,B) and non-hypertrophied LF was harvested from patients with LDH ( Figure 2C,D), lumbar spondylolysis, or cauda equina tumor during surgery. Similarly, the expression of these genes in hypertrophied LF tissue from the LSS group was significantly higher compared with that in LF tissue from the control group ( Figure 4A). Furthermore, correlation analysis demonstrated that the mRNA expression level of these genes was positively correlated with LF thickness ( Figure 4B). The correlation coefficient of periostin was the highest (periostin: R = 0.646, p < .01, biglycan: R = 0.635, p < .01, HAPLN1: 0.047, p < .01) among the three; therefore, further analyses were focused on periostin.
The LF tissue of the control group was rich in elastic fibers and the fibers were arranged relatively regularly; however, in the LSS group, the elastic fibers were significantly lost, and the arrangement was uneven and disordered ( Figure 4C). These findings were similar to those in the LF at the ISC level in the rabbit model ( Figure 1C,D). Representative IHC images for periostin in the LF of LSS and control group are shown in Figure 4C. The mean percentage of periostin-positive cells was higher in the LF of the LSS group compared to that of the control group ( Figure 4D).

| Periostin expression is induced by mechanical stress in LF cells
To investigate whether mechanical stress induces periostin expression in LF cells, FFSS was applied to LF cells and periostin expression was examined in primary cell cultures. After 48 h of FFSS exposure, the mRNA expression level of periostin in LF cells was significantly higher than that in control cells ( Figure 5A). Western blotting also showed that FFSS increased the periostin protein levels in LF cells compared with those in control cells ( Figure 5B).

| FFSS increases gene expression of fibrosis-related factors
Upregulation of TGF-β1, 12 αsmooth muscle actin (α-SMA), 13 Col1a1, 14 Col3, 14 IL-6, 15 and matrix metallopeptidase 2 (MMP2) 16 as fibrosis-related factors has been reported in hypertrophied LF. Therefore, we examined the influence of FFSS on the activation of these genes in LF cells. qRT-PCR showed that the expression of TGF-β1, αSMA, Col1a1, and IL-6 was significantly higher in the FFSS group compared with that in the control group ( Figure 5A). No significant differences were found in other genes (data not shown). To determine the amount of IL-6 secreted, ELISA was performed using cell culture supernatants. The IL-6 protein levels were also increased by FFSS ( Figure 5C). The mRNA expression of IL-6 in LF of LSS group was significantly increased relative to that in LF of Control group ( Figure 5D).

TGF-β1 in LF cells
To determine whether TGF-β1 could induce periostin expression, LF cells were treated with or without recombinant TGF-β1. qRT-PCR analysis showed that the periostin mRNA levels in LF cells were upregulated by 10 ng/ml TGF-β1 ( Figure 5E).

| FFSS or TGF-β1 causes transdifferentiation of LF cells into myofibroblasts
The transdifferentiation of fibroblasts into myofibroblasts is important in fibrosis. It affects the ECM production and the secretion of fibrosis-related factors. The mRNA expression of α-SMA, which is regarded as a marker of activated myofibroblasts, was upregulated by FFSS or treatment with 10 ng/ml TGF-β1 ( Figure 5A,E). Stimulation with   FFSS or TGF-β1 induced cytoskeletal changes, resulting in enlarged cell bodies compared with control cells ( Figure 5F). Consistently, immunocytochemical analysis showed increased expressions of periostin and α-SMA protein in LF cells stimulated with FFSS or TGF-β1 compared with that in control cells ( Figure 5G).

| Periostin blockade reduces the gene expression levels of α-SMA, Col1a1, and IL-6 in shear stressed LF cells while periostin treatment upregulates the IL-6 gene expression levels
To evaluate the effect of periostin blockade on FFSSinduced gene expression of fibrosis-related factors, FFSS was applied to LF cells treated with control siRNA or siRNA against periostin. The FFSS-induced gene expressions of α-SMA, Col1a1, and IL-6 were significantly decreased in LF cells treated with periostin siRNA. However, there was no significant difference in the FFSS-induced TGF-β1 gene expression between LF cells treated with periostin siRNA and control siRNA treatment ( Figure 6A). On the other hand, when LF cells were treated with recombinant periostin, the gene expression of IL-6 was significantly upregulated, but other genes showed no significant differences ( Figure 6B). These results indicate that periostin increases IL-6 gene expression without affecting the expression of TGF-β1.

IL-6 is attenuated by inhibition of the integrins and NF-κB signaling
Integrins α V β 3 and α V β 5 are known as receptors for periostin. 17 NF-κB has been reported to play an important role in the production of inflammatory mediators in LF tissues. 9,15 Furthermore, it has been reported that periostin induces IL-6 expression through integrin and NF-κB signaling in chondrocytes 18 and tumor cells. 19 Therefore, to further investigate the signal transduction pathway involved in periostin-induced IL-6 expression in LF cells, the effects of inhibiting integrin α V β 3 and α V β 5 and NF-κB pathways on the gene expression of IL-6 were analyzed. The FFSS-induced gene expression of IL-6 was significantly attenuated in LF cells treated with integrin α V β 3 , α V β 5 , or NF-κB pathway inhibitor ( Figure 7A,B).

| DISCUSSION
In this study, a comprehensive genetic analysis was performed, and periostin, biglycan, and HAPLN1 were identified as genes involved in mechanical stress-caused LF hypertrophy. Interestingly, these genes encode proteins that constitute the ECM. The expression of three genes showed a significant positive correlation between LF thickness, with periostin showing the highest correlation. Thus, further analyses were focused on periostin. To our knowledge, the present study is the first to demonstrate a possible role of periostin in the pathomechanism of LF hypertrophy.
The pathomechanism of LF hypertrophy has been studied for several decades. LF hypertrophy progresses with age, and it is widely accepted that various factors, such as mechanical stress, 20,21 inflammation, 22 and angiogenesis, 13 contribute to LF hypertrophy. In addition, it has been reported that many cytokines, such as IL-6, 9 MMP family cytokines, 16 TGF-β1 12,21,23,24 and fibroblast growth factor family cytokines 10 are upregulated in LF hypertrophy. Although the above studies have partially elucidated the pathological change and discrete contribution of these molecules to LF hypertrophy, the sequential pathomechanism is still unclear. The lumbar spine is subjected to three-dimensional mechanical stresses, such as compressive, shear, and torsional forces. In clinical practice, segmental instability resulting from degeneration of the facet joint and intervertebral disc is known to cause LF hypertrophy, 13 while other studies have reported LF atrophy over time by interbody fusion, which results in stress shielding. 25,26 At the cellular and molecular levels, mechanical stress induces a marked increase in the expression of collagen types I, III, and V in LF cells. 21 Although mechanical stress is considered the main factor involved in LF hypertrophy, the key molecular biological mechanisms remain unknown due to the lack of a standardized laboratory animal model. Therefore, we established a rabbit model in which mechanical stress is applied to the LF by generating intervertebral mechanical stress concentration and performed a comprehensive genetic analysis of molecules related to LF degenerative changes caused by mechanical stress concentration using microarrays (short-term rabbit model) 6 and RNA seq (long-term rabbit model). Furthermore, these results were collated with previously reported gene expression profiles associated with human LF hypertrophy investigated using microarrays. 3 Periostin, biglycan, and HAPLN1 were identified as upregulated genes in all data sets. Biglycans are members of the Class I family of small leucine-rich proteoglycans that bind to cells via Toll-like receptors and cause inflammation. 27 We have previously reported that biglycans play an important role in the pathophysiology of LF hypertrophy through cell proliferation, myofibroblast differentiation, and cell migration. 11 HAPLN1 is a hyaluronan-binding protein that stabilizes the aggregates of proteoglycan monomers with hyaluronic acid that form the chondrocyte pericellular matrix. In addition, HAPLN1 promotes inflammation and synoviocyte proliferation in rheumatoid arthritis. 28 There have been few studies on the expression and the function of periostin and HAPLN1 in LF. However, periostin expression was more strongly associated with LF thickness than HAPLN1 (Figure 3), and we selected periostin for further analysis in this study.
Periostin, an ECM protein, plays an important role in fibrillogenesis by interacting with tenascin-C and fibronectin. Periostin functions as a matricellular protein, defined as an ECM protein that regulates cell function by interacting with cell surface receptors. It is found in various tissues involved in mechanical stress conditions, such as the periosteum, ligaments, blood vessels, heart, lungs, and the skin. Furthermore, periostin is recognized as an important molecule in various diseases, such as fibrosis in asthma, 29 scleroderma, 30 scar formation in myocardial infarction, 31 and migration of cancer cells. 32 Interestingly, the present study revealed that periostin is upregulated with mechanical stress in LF cells and also highly expressed in hypertrophied LF tissues, suggesting that it may be involved in the pathogenesis of LF hypertrophy.
TGF-β1 plays an important role in the pathogenesis of conditions, such as fibrosis, in many organs. TGF-β1 is significantly upregulated even in the hypertrophied LF. 12,24 In LF cells, mechanical stress increases TGF-β1 production, 21 and furthermore, TGF-β1 increases the synthesis of ECM proteins containing collagen. 23 In addition, mechanical stress is involved in the activation of TGF-β1. When the latent TGF-β1 complex is subjected to shear stress, TGF-β1 is released from the complex. 33 TGF-β1 has also been reported to increase periostin expression in various cell lines. [34][35][36] The present results combined with the previous findings suggest that mechanical stress increases and activates TGF-β1, followed by the upregulation of periostin in LF cells. Fibrosis can be prevented by administering a TGF-β inhibitor, but since the TGF-β family has other important functions, there is concern that inhibiting it to prevent LF hypertrophy as a treatment for LSS may cause serious side effects. Inhibition of TGF-β receptor I triggers overt cardiac toxicity in experimental models. 37 LF cell differentiation into myofibroblasts and increased collagen synthesis by myofibroblasts are key F I G U R E 7 Effect of integrin α V β 3 , α V β 5 , or nuclear factor kappa B (NF-κB) pathway inhibitor on shear stressed ligamentum flavum (LF) cells. (A) Relative mRNA levels of periostin and interleukin-6 (IL-6) in shear stressed LF cells treated with 0.5 μM cilengitide (integrin α V β 3 , α V β 5 inhibitor). (B) Relative mRNA levels of periostin and IL-6 in shear stressed LF cells treated with 1 μM Bay11-7082 (NF-κB pathway inhibitor). All data are expressed as mean ± SEM (n = 6 per group). *p < .05 (Mann-Whitney U test). events in fibrosis. 38 Fibroblasts transdifferentiate to myofibroblasts via proto-myofibroblasts. Fibroblasts do not contain stress fibers, but proto-myofibroblasts form cytoplasmic actin-containing stress fibers. Myofibroblasts express α-SMA in more extensively developed stress fibers. Periostin has been proposed as a marker for protomyofibroblast and αSMA as a marker for myofibroblasts. 39 Mechanical stress and the TGF-β pathway impact the transition of fibroblasts to myofibroblasts. Our results, combined with the previous findings 40 suggest that LF cells exposed to mechanical stress or TGF-β1 transdifferentiate to myofibroblasts and contribute significantly to LF hypertrophy. It has also been reported that periostin is involved in differentiation into myofibroblasts. Periostin knockout mice showed decreased αSMA and Col1a1 expression in in vivo models of scleroderma 30 and palatal wounds. 41 Similarly, periostin knockdown decreased the expression of αSMA and Col1a1 in fibroblasts treated with TGF-β1. 30,42 On the other hand, some reports indicate that administration of periostin increases αSMA expression in fibroblasts, 43 while in other studies, periostin administration had no effect on αSMA expression in fibroblasts, 30,44 and a certain consensus has not been reached. The present study showed that FFSS-induced gene expression of both α-SMA and Col1a1 was suppressed by periostin knockdown, while periostin treatment did not increase the expression of these genes. These findings suggest that periostin may be necessary but not sufficient in promoting the differentiation of LF cells into myofibroblasts and in increasing collagen synthesis.
Mechanotransduction, which converts mechanical stress into biochemical signals that result in intracellular changes, such as activation of signaling pathways, is important in development, physiology, and pathology. Integrin is a heterodimer transmembrane receptor composed of α and β subunits which acts as a mechanosensor that is normally inactive, but is activated by an outside-in signaling that follows mechanical stress to allow the binding of various ligands. Integrins α V β 3 and α V β 5 are known receptors for periostin. 17 Integrins α V β 3 and α V β 5 also act as receptors for other molecules, and it has been reported that thrombospondin2 and connective tissue growth factor upregulate IL-6 via integrin α V β 3 or α V β 5 in osteoarthritis synovial fibroblasts. 45,46 Thus, we hypothesized that periostin upregulate IL-6 expression in LF cells. Our in vitro experiments revealed that periostin, which is a component of the ECM, is upregulated in LF cells in response to mechanical stress, leading to IL-6 expression. Our results also showed that blockade of integrins and NF-κB signaling inhibited mechanical stress-induced gene expression of IL-6. We also attempted to examine the effect of integrin or NF-κB pathway inhibitor for IL-6 expression in LF cells treated with recombinant periostin. However, LF cells were detached under the condition in which IL-6 expression increased with recombinant periostin without FFSS, and we were unable to determine whether periostin regulates IL-6 secretion via integrin/NF-κB signaling. According to the previous reports using other cell types, 12,19,47 integrins and NF-κB may play an important role in periostin induced IL-6 expression, including in LF cells. Further study will be necessary to elucidate the exact signaling pathway between periostin and IL-6 expression.
Recently, it has been proposed that age-related inflammation, "inflammaging," plays an important role in the onset and progression of age-related diseases. 48 The prevalence of LSS increases with increasing age. 49 The ECM is important for maintaining tissue homeostasis; however, during aging, many pathological conditions occur due to aberrant matrix remodeling caused by inflammation. A normal LF is an elastic structure that is rich in elastic fibers. In contrast, hypertrophic LF shows a decrease in elastic fiber content and collagen accumulation. IL-6 is an important modulator of the inflammatory process involved in fibroblast differentiation, activation, and proliferation, and affects ECM remodeling in many diseases. 50 The increased production of IL-6 may also play an important role in the process of LF degeneration and hypertrophy. IL-6 was significantly increased in hypertrophied LF tissue, 15 and IL-6 can elevate collagen expression in LF cells. 51 In addition to changes in ECM composition, chondrogenesis, 52 and infiltration of macrophages 12 or lymphocytes 12 are important histological observations in the hypertrophied LF. Fibroblasts, mesenchymal stem cells (MSCs), macrophages, lymphocytes, and vascular endothelial cells are examples of LF cells derived from hypertrophied LF. 12,23,41 It has been reported that IL-6 promotes the differentiation of MSCs into chondrocytes 53 and activates lymphocytes, 54 and IL-6 induced by periostin may contribute to LF hypertrophy through chondrogensis and inflammation. However, we used 4th passage LF cells, whereas most of the LF cells from 2nd passage onward were considered to be fibroblasts 11,55,56 Further studies including various cell types are needed to clarify the role of IL-6 in the process of LF hypertrophy.
This study had several limitations. First, an animal model of LF degeneration and hypertrophy was established using quadrupedal rabbits. However, the kinematics of the spine may differ from that of bipedal humans. An LSS model using bipedal animals might be better but would entail higher costs. Our rabbit model showed the same tissue degeneration observed in humans with hypertrophied LF. An animal study for LF hypertrophy with an increase in mechanical stress for such a long period replicating the aged human condition has not been previously published. Second, immunostaining for periostin in rabbit LF tissue could not be performed because an appropriate primary antibody was not available. Third, we showed that periostin may be a therapeutic target in vitro, but it is unclear whether it can actually suppress LF hypertrophy. Further in vivo studies using neutralizing antibody 57,58 or siRNA for periostin are required to clarify this issue.
In summary, this study demonstrated that mechanical stress induced periostin expression in LF. Furthermore, periostin promoted LF inflammation by increasing IL-6 expression. These results suggest that anti-periostin treatment may serve as a target for new strategies in the prevention and treatment of LSS.