Protective Effects of Growth Differentiation Factor-6 on the Intervertebral Disc: An In Vitro and In Vivo Study

Growth differentiation factors (GDFs) regulate homeostasis by amplifying extracellular matrix anabolism and inhibiting pro-inflammatory cytokine production in the intervertebral disc (IVD). The aim of this study was to elucidate the effects of GDF-6 on human IVD nucleus pulposus (NP) cells using a three-dimensional culturing system in vitro and on rat tail IVD tissues using a puncture model in vivo. In vitro, Western blotting showed decreased GDF-6 expression with age and degeneration severity in surgically collected human IVD tissues (n = 12). Then, in moderately degenerated human IVD NP cells treated with GDF-6 (100 ng/mL), immunofluorescence demonstrated an increased expression of matrix components including aggrecan and type II collagen. Quantitative polymerase chain reaction analysis also presented GDF-6-induced downregulation of pro-inflammatory tumor necrosis factor (TNF)-α (p = 0.014) and interleukin (IL)-6 (p = 0.016) gene expression stimulated by IL-1β (10 ng/mL). Furthermore, in the mitogen-activated protein kinase pathway, Western blotting displayed GDF-6-induced suppression of p38 phosphorylation (p = 0.041) under IL-1β stimulation. In vivo, intradiscal co-administration of GDF-6 and atelocollagen was effective in alleviating rat tail IVD annular puncture-induced radiologic height loss (p = 0.005), histomorphological degeneration (p < 0.001), matrix metabolism (aggrecan, p < 0.001; type II collagen, p = 0.001), and pro-inflammatory cytokine production (TNF-α, p < 0.001; IL-6, p < 0.001). Consequently, GDF-6 could be a therapeutic growth factor for degenerative IVD disease.


Introduction
Lower back pain is experienced by 85% of adults, notably affecting the global workforce owing to disability [1] which increases the socioeconomic burden [2]. Lower back pain can be caused by multiple factors, including intervertebral disc (IVD) degeneration [3,4]. The IVD comprises a complex discoid structure with the nucleus pulposus (NP) enclosed by the annulus fibrosus (AF) and endplates [4][5][6]. Disc cells have a chondrocytic phenotype to produce extracellular matrix (ECM) components comprising proteoglycans (principally aggrecan) and collagens (primarily type I in the AF and II in the NP) [7].
During IVD degeneration, inflammatory stimuli, primarily involving tumor necrosis factor (TNF)-α, interleukin (IL)-1, and IL-6, modulate the degradation of the ECM through the release of matrix-degrading enzymes, such as matrix metalloproteinase

Ethics Statement
All human and animal experiments were performed under the approval and guidance of the Institutional Review Board (160004) and Institutional Animal Care and Use Committee (P200505) at Kobe University Graduate School of Medicine. Written informed consent was obtained from each patient in accordance with the principles of the Declaration of Helsinki and the laws and regulations of Japan.

Protein Extraction
To compare protein expression in IVD NP tissues by Western blotting, proteins were extracted from IVD NP tissues. Harvested tissues were homogenized using the MS-100R bead-beating disrupter (Tomy Seiko, Tokyo, Japan) for 30 s twice at 4 • C in the T-PER tissue protein extraction reagent (78510; Thermo Fisher Scientific, Waltham, MA, USA) with protease and phosphatase inhibitors (Protease inhibitor cocktail [25955-11; Nacalai Tesque], Phosphatase inhibitor cocktails 2 [P5726; Sigma-Aldrich], and Phosphatase inhibitor cocktails 3 [P0044; Sigma-Aldrich]). Finally, soluble proteins were collected by centrifugation at 20,000× g for 15 min at 4 • C, and the protein concentration was determined using a bicinchoninic acid assay (23227; Thermo Fisher Scientific). Samples were stored at −80 • C.

RNA Isolation and Real-Time Reverse Transcription-Polymerase Chain Reaction
To verify the effects of GDF-6 on human IVD NP cells, we used real-time reverse transcription-polymerase chain reaction (RT-PCR) for the messenger RNA (mRNA) expression levels of anabolic factors and pro-inflammatory cytokines. Initially, cells (Pfirrmann degeneration grades II, n = 2; III, n = 4; IV, n = 6) were divided into four groups, cultured in TASCL microplates, and treated with GDF-6 and IL-1β (control, group C; 100-ng/mL GDF-6, group G; 10-ng/mL IL-1β, group I; 100-ng/mL GDF-6 + 10-ng/mL IL-1β, group G + I). Total RNA was extracted using the RNeasy mini kit (74104; Qiagen, Hilden, Germany), and RNA (0.2 µg) was reverse transcribed with High Capacity cDNA Reverse Transcription Kit (4368814; Thermo Fisher Scientific). The mRNA expression of anabolic factors (ACAN encoding aggrecan and COL2A1 encoding type II collagen) and pro-inflammatory cytokines (TNFA encoding TNF-α, IL-6, MMP-3, and ADAMTS-4) relative to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was assessed through RT-PCR using SYBR Green (4367659; Thermo Fisher Scientific). The primer sequences are listed in Table 1, obtained from published reports [35,36] and purchased from Takara Bio (Shiga, Japan) and Thermo Fisher Scientific. Measurements were performed using the ABI Prism 7500 RT-PCR system (Thermo Fisher Scientific). Melt curve analysis was performed using the Dissociation Curves software to ensure that only a single product was amplified. The relative mRNA expression was analyzed using the 2 −∆∆Ct method, and the vehicle control value was set at 1.0. In quantitative RT-PCR analysis, the effects of treatment on target gene expression were calculated as relative values of the vehicle control. Sample analysis for each patient was performed in duplicate, and single data values were obtained by averaging the values of these replicates. Experiments were conducted six times using six patient samples (n = 12) (Figure 1a).

p38 Mitogen-Activated Protein Kinase Phosphorylation Assay
Activation of the p38 mitogen-activated protein kinase (MAPK) pathway owing to increased IL-1β levels during the progression of IVD degeneration is an early event [31]. Hence, we investigated the phosphorylation of MAPK pathways by GDF-6 and proinflammatory IL-1β stimulation in cultured moderately degenerated human IVD NP cells (Pfirrmann degeneration grades II, n = 2; III, n = 4). Initially, we selected five time points (0, 15, 30, 45, and 60 min) to determine the time point at which the activation of the MAPK signaling pathway peaked following the stimulation of cultured cells with 10-ng/mL IL-1β using Western blotting. Furthermore, we co-treated the cultured cells with 100-ng/mL GDF-6 and 10-ng/mL IL-1β and evaluated the effects of GDF-6 on the p38 MAPK signaling pathway at the peak time point. In addition, Western blotting was performed six times using six patient samples to evaluate the phosphorylation of p38 MAPK (#9926; MAPK Family Antibody Sample Kit, Cell Signaling, Danvers, MA, USA). Representative immunoblots of six similar results are presented (n = 6) ( Figure 1a).

Animals
Twelve-week-old male Sprague-Dawley rat tails were used to develop a model of IVD degeneration (n = 24: weight, 412.2 ± 15.4 (380-444) g) [37]. Under general anesthesia, we used a 20-gauge (G) needle to puncture the skin of rat caudal vertebra to the center of the NP at a depth of 5 mm. The needle was rotated 360 • and held in that position for 30 s before removal. Moreover, we adjusted the AC to be injected into the IVD and prepared 8 mL of cooled atelocollagen (621092; Koken, Tokyo, Japan) solution composed of 0.3% type II collagen, as described in a published report [26], by adding 1 mL of low-concentration glucose DMEM, 0.1 mL of 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (H0887; Sigma-Aldrich), and 0.1 mL of 2.2% NaHCO 3 (28-1850-5; Sigma-Aldrich). Punctures at C8-C9, C9-C10, and C10-C11 were performed to assess effects of GDF-6 and AC [26].
Immediately following the initial puncture, either phosphate-buffered saline (PBS) vehicle (2 µL per disc, group P), AC (2 µL per disc, group A), or GDF-6 (20 µg in 2-µL AC gel per disc, group G + AC) was randomly injected into 12 rats at the center of the IVD NP, using a 27-G needle (MS-N05; Ito Corporation, Shizuoka, Japan). The amount of GDF-6 administered per rat was set to 20 µg using the body weight ratio from a prior study on rabbits (100 µg/3.3-4.5-kg rabbit) [21]. Disc height evaluation, histomorphology, and immunofluorescence were performed at 14 and 28 days following the puncture (model 1, n = 12).

Histomorphology
Safranin-O (S-0145; Tokyo Chemical Industry, Tokyo, Japan), fast green (10720; Chroma Gesellschaft Schmidt, Munster, Germany), and hematoxylin (30002; Muto Tokyo, Japan) staining were used to visualize the distribution of proteoglycans. Images of the sections were captured with the BZ-X700 microscope and graded by three individual examiners from 0 (non-degenerative) to 16 (severely degenerative) for NP morphology, NP cellularity, NP-AF boundary, AF morphology, and endplate, according to a reported histological degeneration scale (n = 6 per group) [39].

Statistical Analysis
Statistical analysis was performed using IBM SPSS Statistics 23.0 software (IBM Corp., Armonk, NY, USA). Data were expressed as the mean ± standard deviation, except for histological grading scores, which were reported as the median with the range. To assess the effects of changes in culture durations and experimental conditions, one-way analysis of variance (ANOVA) with the Tukey-Kramer post-hoc test was performed with a significance level of p < 0.050 with the normality assumption.

In Vivo Human IVD Experiments
Western blotting was used to evaluate GDF-6 expression in surgically collected human IVD NP tissues based on the patient's age and Pfirrmann degeneration grade (n = 12). The expression of GDF-6 tended to decrease with increasing age and degeneration grade. Additionally, BMPRII and notochordal brachyury and CD24 diminished in protein expression with aging and degeneration. Levels of GDF-6 were lower in Pfirrmann grade-IV IVDs than in grade-II and grade-III tissues. These results indicate that the roles of GDF-6 could differ depending on the degeneration grade. Therefore, we divided these human tissue samples into two experimental groups: grades II and III as the moderate degeneration group and grade IV as the severe degeneration group (Figure 2a). IVD nucleus pulposus (NP)-associated notochordal brachyury, CD24, and actin (used as a loading control) were examined by Western blotting in protein extracts from surgically obtained human IVD NP tissues. Samples were randomly selected (n = 12). (b) Validation of the IVD NP phenotype in cell clusters. Levels of IVD NP-associated notochordal brachyury and CD24 were evaluated by Western blotting in protein extracts from human IVD NP cells. Immunoblots shown are representative of in vitro human IVD experiments, as described below, with similar results. (c) Effects of growth differentiation factor-6 (GDF-6) treatment on CD24, aggrecan, and type II collagen expression in human IVD NP cells (Pfirrmann degeneration grades III and IV, n = 4). Cells were cultured in a monolayer medium, followed by seeding in a three-dimensional culturing system (Tapered Stencil for Cluster Culture [TASCL]) at 1.0 × 10 6 cells/TASCL to form cell clusters. After 48-h culture, cell clusters were treated with solvent as a control (group C, set as 1.0), GDF-6 treatment (100 ng/mL, group G), pro-inflammatory interleukin (IL)-1β (10 ng/mL, group I), or both GDF-6 and IL-1β (group G + I). Then, CD24 (purple), aggrecan (green), type II collagen (red), nuclear 4 ,6-diamidino-2-phenylindole (DAPI) (blue), and merged signals of cell clusters on TASCL were obtained using a fluorescence microscope. Relative fluorescence intensity of 10 randomly selected cell clusters per image was measured. Data are the mean ± standard deviation. (n = 4). One-way analysis of variance with the Tukey-Kramer post-hoc test was used. Significant differences were set as * p < 0.050 and ** p < 0.010.

Effects of GDF-6 Treatment on CD24, Aggrecan and Type II Collagen
Following Western blotting in human IVD NP tissues, multi-color immunofluorescence was performed in both moderate and severe degeneration groups to confirm the levels of IVD NP-associated notochordal CD24 and anabolic aggrecan and type II collagen in GDF-6-treated IVD NP cell clusters under pro-inflammatory IL-1β stimulation. The phenotype of human IVD NP cells used in all in vitro experiments of this study was confirmed by Western blotting for cell markers (Brachyury and CD24) (Figure 2b). In the moderate degeneration group, immunofluorescence identified significantly higher immunoreactivity for CD24 (p = 0.028), aggrecan (p = 0.001), and type II collagen (p < 0.001) in group G than in group C. Although IL-1β supplementation resulted in significantly lower expression of CD24 (p = 0.029), aggrecan (p = 0.013) and type II collagen (p = 0.006) in group I, IL-1β-induced suppression of CD24 (p = 0.030), aggrecan (p = 0.001), and type II collagen (p = 0.002) levels was significantly rescued by the supplementation of GDF-6 in group G + I. However, no significant difference was observed between groups G, I, and C in the severe degeneration group (Figure 2c).

Effects of GDF-6 Treatment on TNF-α, IL-6, MMP-3, and ADAMTS-4 Gene Expression
To analyze the mRNA expression levels of anabolic factors and pro-inflammatory cytokines, real-time RT-PCR was performed. Quantitative RT-PCR identified significantly higher gene expression of ACAN encoding aggrecan (p = 0.031) and COL2A1 encoding type II collagen (p = 0.024) in group G than in group C with the moderate degeneration group (Figure 3a). However, no significant difference was observed between groups G and C with the severe degeneration group (Figure 3b). Furthermore, significantly lower gene expression of TNFA encoding TNF-α (p = 0.014), IL-6 (p = 0.016), MMP-3 (p = 0.038), and ADAMTS-4 (p = 0.017) was found in group G + I than in group I with the moderate degeneration group. No significant difference was observed between groups I and G + I with the severe degeneration group. These results suggest that GDF-6 has antiinflammatory effects under pro-inflammatory conditions. Thus, we further investigated the involvement of the MAPK pathway.   and metalloproteinase with thrombospondin motifs-4 (ADAMTS-4) gene expression in intervertebral disc nucleus pulposus cells. Cells of the moderate degeneration group (a) and severe degeneration group (b) were seeded on Tapered Stencil for Cluster Cultures and treated with solvent as a control (group C; set as 1.0), 100-µg/mL GDF-6 (group G), 10-ng/mL pro-inflammatory IL-1β (group I), or both (group G + I) for 48 h. Gene expression of ACAN, COL2A1, TNFA, IL-6, MMP-3, and ADAMTS-4 relative to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was assessed through real-time reverse transcription-polymerase chain reaction using SYBR Green. Data are the mean ± standard deviation as the fold change relative to the control (group C) (n = 6). One-way repeated-measures analysis of variance with the Tukey-Kramer post-hoc test was used. Significant differences were set as * p < 0.050 and ** p < 0.010.

Effects of GDF-6 on p38 Phosphorylation under Pro-Inflammatory Conditions
Based on real-time RT-PCR results, we further investigated the involvement of GDF-6 in the MAPK pathway. In human IVD NP cells of the moderate degeneration group, Western blotting found increased p38 phosphorylation with a peak at 15 min after IL-1β supplementation (Figure 4a). Then, after the 15-min stimulation, Western blotting further identified a significant partial reduction in elevated p38 phosphorylation in group G + I compared with group I (p = 0.041) (Figure 4b).
Cells 2022, 11, x FOR PEER REVIEW 12 of 26 nucleus pulposus cells. Cells of the moderate degeneration group (a) and severe degeneration group (b) were seeded on Tapered Stencil for Cluster Cultures and treated with solvent as a control (group C; set as 1.0), 100-μg/mL GDF-6 (group G), 10-ng/mL pro-inflammatory IL-1β (group I), or both (group G + I) for 48 h. Gene expression of ACAN, COL2A1, TNFA, IL-6, MMP-3, and ADAMTS-4 relative to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was assessed through real-time reverse transcription-polymerase chain reaction using SYBR Green. Data are the mean ± standard deviation as the fold change relative to the control (group C) (n = 6). One-way repeated-measures analysis of variance with the Tukey-Kramer post-hoc test was used. Significant differences were set as * p < 0.050 and ** p < 0.010.

Effects of GDF-6 on p38 Phosphorylation under Pro-Inflammatory Conditions
Based on real-time RT-PCR results, we further investigated the involvement of GDF-6 in the MAPK pathway. In human IVD NP cells of the moderate degeneration group, Western blotting found increased p38 phosphorylation with a peak at 15 min after IL-1β supplementation (Figure 4a). Then, after the 15-min stimulation, Western blotting further identified a significant partial reduction in elevated p38 phosphorylation in group G + I compared with group I (p = 0.041) (Figure 4b). Effects of growth differentiation factor-6 (GDF-6) on p38 phosphorylation under pro-inflammatory condition. (a) Time-course change in p-p38 protein levels relative to p38. We stimulated human intervertebral disc (IVD) nucleus pulposus (NP) cells with 10-ng/mL interleukin (IL)-1β and measured the protein levels of p-p38 and p38 at different time points, i.e., 0 (set as 1.0), 15, 30, 45, and 60 min. (b) p-p38 protein levels relative to p38 in IVD NP cells treated with phosphate-buffered saline (group C; set as 1.0), GDF-6 (100 μg/mL, group G), IL-1β (10 ng/mL, group I), and both (group G + I) for 15 min. Data are the mean ± standard deviation. Results from six independent experiments were analyzed using one-way repeated-measures analysis of variance with the Tukey-Kramer posthoc test (n = 6). Significant differences are set as * p < 0.050 and ** p < 0.010. Figure 4. Effects of growth differentiation factor-6 (GDF-6) on p38 phosphorylation under proinflammatory condition. (a) Time-course change in p-p38 protein levels relative to p38. We stimulated human intervertebral disc (IVD) nucleus pulposus (NP) cells with 10-ng/mL interleukin (IL)-1β and measured the protein levels of p-p38 and p38 at different time points, i.e., 0 (set as 1.0), 15, 30, 45, and 60 min. (b) p-p38 protein levels relative to p38 in IVD NP cells treated with phosphate-buffered saline (group C; set as 1.0), GDF-6 (100 µg/mL, group G), IL-1β (10 ng/mL, group I), and both (group G + I) for 15 min. Data are the mean ± standard deviation. Results from six independent experiments were analyzed using one-way repeated-measures analysis of variance with the Tukey-Kramer post-hoc test (n = 6). Significant differences are set as * p < 0.050 and ** p < 0.010.

Effect of Co-Administration of GDF-6 and AC on Radiographic Disc Height in a Rat Tail IVD Annular Puncture Model
Following the identification of anti-inflammatory characteristics of human IVD NP cells under GDF-6 treatment, we designed an in vivo study using a well-established rat tail IVD annular puncture model characterized by its high accessibility and reproducibility. Group P showed a significantly lower DHI on days 14 (90.9 ± 3.2%, p = 0.011) and 28 (75.9 ± 5.9%, p < 0.001) after surgery than group C. Moreover, group AC exhibited a significantly lower DHI on days 14 (91.7 ± 5.0%, p = 0.022) and 28 (76.6 ± 4.6%, p < 0.001) than group C. Similarly, group G + AC also had significantly lower DHI on days 14 (92.0 ± 5.9%, p = 0.027) and 28 (85.7 ± 2.8%, p < 0.001) than group C. However, DHI at postoperative 28 days in group G + AC was significantly higher than that in groups P (p = 0.005) and AC (p = 0.009) (Figure 5a). Groups P , G , and G + AC presented significantly lower DHI on days 14 and 28 than group C . At postoperative 28 days, DHI in group G + AC was significantly higher than in groups P (p < 0.001) and G (p = 0.012). In addition, DHI in group G was significantly higher compared with that in group P (p = 0.037) (Figure 5b).

Effect of Co-Administration of GDF-6 and AC on Histological IVD Degeneration in a Rat Tail Puncture Model
Safranin-O, fast green, and hematoxylin staining were performed to assess the general IVD morphology with proteoglycan distribution. Control IVDs had round NPs at 14 and 28 days after the start of the experiment, with a clear boundary observed between the AF and NP.

Effects of Co-Administration of GDF-6 and AC on Aggrecan and Type II Collagen Expression in a Rat Tail Puncture Model
Immunopositivity was calculated as a percentage relative to DAPI-positive cells. In the control, aggrecan-positive cells were widely observed at both time points (C: 14 days, 87.1 ± 5.1%; 28 days, 85.9 ± 7.4%) and C : 14 days, 87.6 ± 6.4%; 28 days, 86.0 ± 4.6%) (Figure 7a). Type II collagen-positive cells were also abundant at both time points (C: 14 days, 85.0 ± 6.3%; 28 days, 84.5 ± 6.3%) and C : 14 days, 87.9 ± 2.4%; 28 days, 86.1 ± 7.1%) (Figure 7b). Therefore, approximately 85-90% of IVD NP cells had abundant ECM components. p = 0.027) and 28 (85.7 ± 2.8%, p < 0.001) than group C. However, DHI at postoperative 28 days in group G + AC was significantly higher than that in groups P (p = 0.005) and AC (p = 0.009) (Figure 5a). Groups P', G', and G + AC' presented significantly lower DHI on days 14 and 28 than group C'. At postoperative 28 days, DHI in group G + AC' was significantly higher than in groups P' (p < 0.001) and G' (p = 0.012). In addition, DHI in group G' was significantly higher compared with that in group P' (p = 0.037) (Figure 5b). Figure 5. Effects of co-administration of growth differentiation factor-6 (GDF-6) and atelocollagen (AC) on radiographic disc height in a rat tail intervertebral disc annular puncture model. A 20-G Figure 5. Effects of co-administration of growth differentiation factor-6 (GDF-6) and atelocollagen (AC) on radiographic disc height in a rat tail intervertebral disc annular puncture model. A 20-G needle was used to puncture the rat caudal disc at a depth of 5 mm from the skin to the center of nucleus pulposus. The needle was rotated 360 • and held in that position for 30 s before extraction to establish an intervertebral disc degeneration model. Following the initial puncture, (a) either vehicle (phosphate-buffered saline [PBS]; 2 µL per disc, group P), atelocollagen (AC; 2-µL AC per disc, group AC), or GDF-6 (20 µg in 2-µL AC gel per disc, group G + AC) or (b) vehicle (PBS; 2 µL per disc, group P ), GDF-6 (20 µg in 2-µL PBS per disc, group G ), or GDF-6 (20 µg in 2-µL AC gel per disc, group G + AC ) was randomly injected. Lateral radiographs of the rat tail were obtained on days 14 and 28 postoperatively, and the disc height index (DHI) was calculated by measuring, averaging, and normalizing the disc height compared to the adjacent vertebral body heights in the anterior, middle, and posterior regions. We designated the control without puncture as 100% (groups C and C ), and compared the four groups (C, P, A, G + AC) and (C , P , G , G + AC ). Data are the mean ± standard deviation (n = 6). One-way analysis of variance with the Tukey-Kramer post-hoc test was used. Significant differences are set as * p < 0.050 and ** p < 0.010. Cells 2022, 11, x FOR PEER REVIEW 15 of 26 Figure 6. Effects of co-administration of growth differentiation factor-6 (GDF-6) and atelocollagen (AC) on histological intervertebral disc (IVD) degeneration in a rat tail puncture model. After establishment of the IVD degeneration model, (a) either vehicle (phosphate-buffered saline [PBS]; 2 μL per disc, group P), AC (2-μL AC per disc, group AC), or GDF-6 (20 μg in 2-μL AC gel per disc, group G + AC) or (b) vehicle (PBS; 2 μL per disc, group P'), GDF-6 (20 μg in 2-μL PBS per disc, group G'), or GDF-6 (20 μg in 2-μL AC gel per disc, group G + AC') was randomly injected. The IVD degeneration was assessed by Safranin-O and fast green staining of the IVDs at 14 and 28 days after injection in low-power fields (×100). We designated the control without puncture (groups C and C'), and compared the four groups (C, P, A, G + AC) and (C', P', G', G + AC'). Data are the mean ± standard deviation (n = 6). One-way analysis of variance with the Tukey-Kramer post-hoc test was used. Significant differences are set as * p < 0.050 and ** p < 0.010. 2 µL per disc, group P), AC (2-µL AC per disc, group AC), or GDF-6 (20 µg in 2-µL AC gel per disc, group G + AC) or (b) vehicle (PBS; 2 µL per disc, group P ), GDF-6 (20 µg in 2-µL PBS per disc, group G ), or GDF-6 (20 µg in 2-µL AC gel per disc, group G + AC ) was randomly injected. The IVD degeneration was assessed by Safranin-O and fast green staining of the IVDs at 14 and 28 days after injection in low-power fields (×100). We designated the control without puncture (groups C and C ), and compared the four groups (C, P, A, G + AC) and (C , P , G , G + AC ). Data are the mean ± standard deviation (n = 6). One-way analysis of variance with the Tukey-Kramer post-hoc test was used. Significant differences are set as * p < 0.050 and ** p < 0.010.  After establishment of the intervertebral disc degeneration model, vehicle (phosphate-buffered saline [PBS]; 2 µL per disc, group P), AC (2-µL AC per disc, group AC), or GDF-6 (20 µg in 2-µL AC gel per disc, group G + AC) was randomly injected. Extracellular matrix components were assessed using immunofluorescence (aggrecan: green, type II collagen: red, CD24: purple, 4 ,6-diamidino-2-phenylindole [DAPI]: blue, and merged signals) of the discs at 14-and 28-days post-injection in high-power fields (×400). Immunopositivity was calculated as a percentage relative to DAPI-positive cells. We designated the control without puncture (groups C and C ), and compared the four groups (C, P, A, G + AC) and (C , P , G , G + AC ). Data are expressed as the mean ± standard deviation (n = 6). One-way analysis of variance (ANOVA) with the Tukey-Kramer post-hoc test was used. Significant differences are set as * p < 0.050 and ** p < 0.010. (b) Effect of co-administration of GDF-6 and AC on aggrecan and type II collagen expression compared with GDF-6 alone in a rat tail puncture model. After establishment of the intervertebral disc (IVD) degeneration model, vehicle (PBS; 2 µL per disc, group P ), GDF-6 (20 µg in 2-µL PBS per disc, group G ), or GDF-6 (20 µg in 2-µL AC gel per disc, group G + AC ) was randomly injected. Extracellular matrix components were assessed using immunofluorescence (aggrecan: green, type II collagen: red, CD24: purple, 4 ,6-diamidino-2-phenylindole [DAPI]: blue, and merged signals) of the IVDs at 14 and 28 days after injection in high-power fields (×400). Immunopositivity was calculated as a percentage relative to DAPI-positive cells. We designated the control without puncture (groups C and C ), and compared the four groups (C, P, A, G + AC) and (C , P , G , G + AC ). Data are the mean ± standard deviation (n = 6). One-way ANOVA with the Tukey-Kramer post-hoc test was used. Significant differences are set as * p < 0.050 and ** p < 0.010.
Then, 28 days after puncture, the decrease in group G + AC compared with in groups P (aggrecan; p < 0.001, type II collagen; p = 0.001) and AC (aggrecan; p = 0.004, type II collagen; p = 0.003) (Figure 7a) and that in group G + AC compared with in groups P (aggrecan; p < 0.001, type II collagen; p < 0.001) and G (aggrecan; p = 0.001, type II collagen; p = 0.045) were significantly suppressed. The decrease in group G compared with in group P (aggrecan; p = 0.031, type II collagen; p = 0.029) was also attenuated (Figure 7b).

Effects of Co-Administration of GDF-6 and AC on p-p38 Expression in a Rat Tail Puncture Model
Immunopositivity was calculated as a percentage of DAPI-positive cells. In the control, an abundance of p-p38-positive cells was observed at both time points (C: 14 days, 15.5 ± 4.5%; 28 days, 15.8 ± 3.0%) (Figure 9a) and C : 14 days, 15.9 ± 3.0%; 28 days, 16.1 ± 4.5%) (Figure 9b). However, 14 days after puncture, there was a significant increase in the number of p-p38-positive cells in groups P and AC, although no significant difference was observed in group G + AC compared to group C (P: 28.0 ± 4.2%, AC: 26.1 ± 8.0%, G + AC: 23.9 ± 5.1%) (Figure 9a). Similarly, there was a predominant increase in the number of p-p38-positive cells in group P , although no significant increase was observed in groups G and G + AC compared to group C (P : 26.2 ± 6.3%, G : 23.1 ± 5.5%, G + AC : 22.1 ± 4.1%) (Figure 9b). Twenty-eight days after puncture, the increase in group G + AC compared to that in groups P (p < 0.001) and AC (p < 0.001) (Figure 9a) and that in groups P (p < 0.001) and G (p < 0.001) compared to that in group G + AC were significantly suppressed. Furthermore, the increase in group G was significantly suppressed compared to in group P (p = 0.001) (Figure 9b).  per disc, group P), AC (2-µL AC per disc, group AC), or GDF-6 (20 µg in 2-µL AC gel per disc, group G + AC) was randomly injected. Pro-inflammatory cytokines were assessed using immunofluorescence (p-p38: red, CD24: purple, 4 ,6-diamidino-2-phenylindole [DAPI]: blue, and merged signals) of the IVDs at 14 and 28 days after injection in high-power fields (×400). Immunopositivity was calculated as a percentage relative to DAPI-positive cells. We designated the control without puncture (groups C and C ), and compared the four groups (C, P, A, G + AC) and (C , P , G , G + AC ). Data are expressed as the mean ± standard deviation (n = 6). One-way analysis of variance (ANOVA) with the Tukey-Kramer post-hoc test was used. Significant differences are set as * p < 0.050 and ** p < 0.010. (b) Effect of co-administration of GDF-6 and AC on p-p38 expression compared with GDF-6 alone in a rat tail puncture model. After establishment of the IVD degeneration model, vehicle (PBS; 2 µL per disc, group P ), GDF-6 (20 µg in 2-µL PBS per disc, group G ), or GDF-6 (20 µg in 2-µL AC gel per disc, group G + AC ) was randomly injected. Pro-inflammatory cytokines were assessed using immunofluorescence p-p38: red, CD24: purple, DAPI: blue, and merged signals) of the IVDs at 14 and 28 days after injection in high-power fields (×400). Immunopositivity was calculated as a percentage relative to DAPI-positive cells. We designated the control without puncture (groups C and C ), and compared the four groups (C, P, A, G + AC) and (C , P , G , G + AC ). Data are the mean ± standard deviation (n = 6). One-way ANOVA with the Tukey-Kramer post-hoc test was used. Significant differences are set as * p < 0.050 and ** p < 0.010.

Discussion
The BMP family, including GDF-6, influences the IVD homeostasis between anabolic factors and pro-inflammatory cytokines. Disrupting this balance leads to the progression of IVD degeneration. In this study, we examined the effects of GDF-6 according to age and IVD degeneration on IVD NP cells and observed the decreased expression of GDF-6 in IVD with aging and severity of degeneration. Although no difference in the expression of BMPRII (GDF-6 receptor) has been reported in the IVD owing to the degeneration [22], our results exhibited decreased levels of BMPs and BMPRII. Okuda et al. reported that the metabolic/anabolic responsiveness of IVD cells to growth factors decreases with age, which is associated with decreased growth factor production and receptor synthesis in IVD cells [40]. Thus, we compared two groups of patients, those with Pfirrmann grades II or III (moderate degeneration group) and with Pfirrmann grade IV (severe degeneration group), and identified the different responses to GDF-6 in Pfirrmann grade-IV patients from Pfirrmann grade-II or -III patients. While there have been no reports using TASCL in IVD cell culture, we used TASCL for 3D culture in the current study [41]. The TASCL has a two-layered structure with a lattice-shaped device and microporous membrane that facilitates the production of many uniform cell clusters using simple cell suspensions. This was considered more advantageous than the 3D culturing system using alginate beads, which involves a complicated procedure to obtain uniform cell clusters.
Treatment of GDF-6 for moderately degenerated IVDs increased the expression of notochordal CD24 and anabolic factors, such as aggrecan and type II collagen, which is consistent with a prior study [42]. In addition, we previously reported the decreased levels of IL-6 and TNF-α by GDF-6 in rabbit IVD studies in vivo [21]. However, no prior studies have compared the effects of GDF-6 on IVD cells under IL-1β-induced pro-inflammatory conditions using a 3D in vitro culturing system. In the current study, we showed in realtime RT-PCR that GDF-6 reduced the levels of TNF-α and IL-6 in IVD NP cells under pro-inflammatory stimulation. To corroborate these results, we focused on the MAPK pathway enhanced by GDF-6 administration. Our results showed that GDF-6 inhibited the IL-1β stimulated p38 phosphorylation. Hence, although the underlying mechanism associated with the efficacy of GDF-6 under pro-inflammatory stimulation requires further investigation, GDF-6 may be a potent therapeutic target for the management of inflammation and pain during the early stages of IVD degeneration.
For the in vivo experiments, we selected the rat caudal annular puncture model, which is a simple model for creating degenerated IVDs. Moreover, we employed the 30 s fixation technique with a 360 • rotation by advancing a 20-G needle to 5-mm depth, as described previously [37]. The DHI after puncture with a 20-G needle was slightly lower than that reported with this method at two and four weeks [37]. However, a similar decrease was obtained with a less invasive puncture model using a 21-G needle with 180 • rotation and 5-s fixation [43]. Therefore, the puncture technique was appropriate and effective for the purpose of this study.
We hypothesized that locally administered drugs or growth factors via injection or surgery to IVDs, which can be nourished only by diffusion, would not be effective since they could easily flow out of the intradiscal space. The AC has been investigated as an injectable scaffold [26,44]; AC solution becomes liquid at nearly 4 • C and gel at 37 • C, which may be a significant advantage in keeping the drug or growth factor localized. Drug delivery systems (DDS) using collagen can overcome the shortcomings of water-soluble polymers in terms of instability. Collagen preparations have been reported to achieve the sustained release of various proteins and could enhance the effects of drugs locally [27]. Type II collagen-based AC gel has a similar composition to IVD matrix; therefore, we selected it as a solvent owing to its low immunogenicity and high safety [26]. In vitro studies on culturing chondrocytes in AC gel have resulted in greater matrix synthesis compared to monolayer cell culture [26]. Porous hydroxyapatite/AC composites for treating bone defects are also often used clinically. Additional matrix is synthesized more abundantly when chondrocytes are cultured with AC than when cultured in a monolayer [45]. In this study, we focused on whether this property of AC could enhance the effect of GDF-6 by acting as a scaffold for NP cells.
AC could be injected without clogging or coagulation using a microsyringe. GDF-6 was soluble in AC, and the concentration was equalized by repetitive pipetting. The drug dose was set to 20 µg/disc based on the body weight ratio of rabbits and rats.
In vivo experiments in sheep and rabbits have identified that GDF-6 not only affects ECM synthesis but also suppresses the levels of pro-inflammatory cytokines, and is effective for maintaining IVD structure and treating discogenic pain [21]. In addition, GDF6 can improve the structure of the IVD, inhibit the expression of inflammatory and pain-related factors, and improve pain behavior in rats [46]. However, there have been no reports on GDF-6 administration combined with AC as a DDS. In our study, GDF-6 with AC had a more positive prolonged effect on IVD disruption by downregulating the inflammatory response and upregulating ECM production than GDF-6 alone. Our findings suggest that GDF-6 administered with AC can provide an IVD-protective effect with a relatively low dose of GDF-6 [46]. The use of AC as a DDS prolonged the effect of GDF-6 (compared with PBS) and possibly reduced the mechanical stress between vertebrae.
Although the protective effects of GDF-6 on IVDs have been reported [20][21][22][23], there have been no reports evaluating the effects of GDF-6 under inflammatory conditions in vitro, simulating IVD degeneration in vivo, and actually examining these effects in an in vivo degeneration model. Thus, this study has a high novelty with consistent results in vitro and in vivo.
This study has several limitations. First, we did not determine why GDF-6 exerted its anti-inflammatory effect only under pro-inflammatory conditions. Second, this study adopted a single concentration/viscosity of AC with no other scaffold controls. We did not investigate how GDF-6 release was sustained. In the future, it will be necessary to clarify the optimal conditions under which GDF-6 exerts its effects and to adjust the optimal scaffold.

Conclusions
The co-administration of GDF-6 and AC could decelerate degenerative IVD disease, which could serve as a new treatment strategy for degenerative IVD disease.