Mechanosignaling activation of TGFβ maintains intervertebral disc homeostasis

Intervertebral disc (IVD) degeneration is the leading cause of disability with no disease-modifying treatment. IVD degeneration is associated with instable mechanical loading in the spine, but little is known about how mechanical stress regulates nucleus notochordal (NC) cells to maintain IVD homeostasis. Here we report that mechanical stress can result in excessive integrin αvβ6-mediated activation of transforming growth factor beta (TGFβ), decreased NC cell vacuoles, and increased matrix proteoglycan production, and results in degenerative disc disease (DDD). Knockout of TGFβ type II receptor (TβRII) or integrin αv in the NC cells inhibited functional activity of postnatal NC cells and also resulted in DDD under mechanical loading. Administration of RGD peptide, TGFβ, and αvβ6-neutralizing antibodies attenuated IVD degeneration. Thus, integrin-mediated activation of TGFβ plays a critical role in mechanical signaling transduction to regulate IVD cell function and homeostasis. Manipulation of this signaling pathway may be a potential therapeutic target to modify DDD.


INTRODUCTION
Degenerative disc disease (DDD) remains a common musculoskeletal disorder that brings an enormous socioeconomic burden. [1][2][3] Although numerous factors associated with DDD have been identified, the exact molecular pathogenesis of DDD has yet to be elucidated. The current treatments focus on symptomatic relief from pain through injections, physical therapy, and activity modification 4 or surgical intervention such as disc decompression, spinal fusion, and disc replacement. 3,5 However, none of these interventions halt the progression of degeneration nor restore the physiologic disc function.
Dysfunction of nucleus pulposus (NP) cells is the key in the onset of intervertebral disc (IVD) degeneration. 1,[6][7][8] It is known that NP cells are of notochord origin, [9][10][11] termed as notochordal (NC) cells at early age. NC cells are large with intracellular vacuoles making up at least 25% of the cell area. [7][8] The large vacuoles generate IVD space during spinal morphogenesis. 9,[12][13][14] During maturation and degeneration, the NC cells undergo morphologic and functional transition with the loss of their vacuoles. The resultant fibroblast-like cells have decreased the expression of extracellular matrix protein such as aggrecan, 15 which enables the NP to maintain height and turgor against compressive loads via its osmotic properties. [16][17] The mechanism driving NC cell transition is unclear, particularly how the mechanical load influences cell signaling.
Temporal-spatial activation of latent matrix transforming growth factor beta (TGFβ) has been shown to modulate chondrocyte anabolic activity in articular cartilage, maintain bone homeostasis during bone remodeling, and help with tissue repair. [18][19] The α v integrins in combination with -β 6 , -β 5 , and -β 8 have been shown to mediate the activation of TGFβ. [20][21][22][23][24] Integrins enable cells to transduce mechanical loads into biological signaling. As NP cells express α v and multiple β integrin subunits, integrin-mediated activation of TGFβ may play a critical role in IVDs. 25 In addition, active TGFβ is known to act upstream of connective tissue growth factor (CTGF/CCN2) and aggrecan, both of which are involved in DDD development. [26][27] Thus, we sought to understand the role of TGFβ in IVD homeostasis.
In this study, we systematically investigated the role of mechanical stress on the functional transition of NC cells and IVD homeostasis. Utilizing multiple rodent models, we found that mechanical stress resulted in integrin α v β 6 -mediated activation of TGFβ. Abnormal stress resulted in excessive TGFβ signaling and accelerated NC cells functional transition. Administration of RGD peptide and neutralizing antibodies against TGFβ and α v β 6 attenuated these changes. On the other hand, conditional knockout of TβRII or α v also impeded NC cells' transition and caused IVD degeneration by mechanical stimuli. Thus, precise integrin-induced activation of TGFβ is required to maintain IVD cell function and homeostasis.

Animal models
Lumbar spine instability mouse model: C57BL/6J (male, 8-week old) mice were purchased from Charles River, Wilmington, MA, USA. After anesthetized with ketamine and xylazine, they were operated by resection of the lumbar 3th-lumbar 5th (L 3 -L 5 ) spinous processes along with the supraspinous and interspinous ligaments to induce instability of lumbar spine. [28][29] Sham operations were carried out only by detachment of the posterior paravertebral muscles from the L 3 -L 5 vertebrae. The operated mice were intraperitoneally injected with either TβRI inhibitor (SB-505124, Sigma-Aldrich, St Louis, MO, USA) at a dose of 1 mg·kg − 1 (SB group) or the equivalent volume of vehicle (dimethyl sulfoxide; Veh group) once every 2 days. Mice (8-week old) were killed at 0, 1, 2, 4, and 8 weeks after the surgery (n = 6 per group). Caudal spine instability mouse model: The instability of caudal spine was induced by fully depth annular stab and NP removal of the caudal 7th-8th (C 7-8 ) IVD. 30 The adjacent C [8][9] IVDs were chosen for observation at 4 weeks post surgery (n = 6 per group). The treatments were the same as those described in the first model. Rat caudal IVD compression model: Twelve-week-old male Sprague-sDawley rats (Charles River) were attached at caudal vertebrae with a loading device. [31][32][33][34] In detail, two 0.7366 mm stainless steel Kirchner wires were inserted percutaneously into each of the C 8 and C 10 , and attached to 35 mm external diameter aluminum rings. Rings were precoated with a resin and connected longitudinally with four stainless steel threaded rods. Axial loads were applied by four calibrated springs (0.50 N·mm − 1 ) installed over each rod. In the vehicle (Veh) and 1D11 group, the axial stress loaded from the distal side produced a calculated compressive pressure of 1.3 MPa onto C 8-9 and C 9-10 IVDs through tightening calibrated springs. 31,[33][34] The C 8-9 of loaded IVDs was injected with an alginate bead (1 μL) containing 0.7 μg 1D11 (TGFβ-neutralizing antibody; R&D Systems, Minneapolis, MN, USA), 18 whereas the C 9-10 IVDs with that containing vehicle (n = 6) once during the surgery. In the sham group, the rats were also attached with the loading device at C 8 and C 10 , but no compressive pressure was exerted onto the IVDs. Both C 8-9 and C 9-10 were injected with vehicle-containing alginate bead. The rats were killed at 2 weeks after the surgery (n = 6 per group).
The genotype of the mice was determined by PCR analyses of genomic DNA isolated from mouse tails using the following primers: Noto-directed cre forward, 5′-ATACC GGCAGATCATGCAAGC-3′, and reverse, 5′-ATGCACATAT GCAACCCACA-3′. 10 The loxP TgfβRII allele was identified with the primers forward, 5′-TAAACAAGGTCCGGAGC CCA-3′, and reverse, 5′-ACTTCTGCAAGAGGTCCCCT-3′. 36 α v flox/flox mice mouse strain was obtained from the lab of Loading duration was 24 h (n = 6 per group). 37 We determined the sample size based on power analysis via website http://www.biomath.info (for two groups, 80% power, 5% significance, two-sided), or via G*Power (for more than two groups, 80% power, 5% significance). For animal model and TgfβRII knockout studies, each experiment was conducted twice: one pilot experiment in three samples and two independent experiments of six samples. For α v knockout mice, three independent experiments with six samples were conducted.

Micro Computed Tomography (μCT)
The lower thoracic and whole lumbar spine from mice were dissected, fixed in 10% buffered formalin for 48 h, transferred into phosphate-buffered saline, and then examined by high-resolution μCT (Skyscan1172). The ribs on the lower thoracic were included for the identification of L 4 -L 5 IVD localization. Images were reconstructed and analyzed using NRecon v1.6 and CTAn v1.9 (Skyscan company, San Jose, CA, USA), respectively. Threedimensional model visualization software, CTVol v2.0 (Skyscan company, San Jose, CA, USA), was used to analyze parameters of the L 4 -L 5 IVD with half height of L 4 and L 5 vertebrae. The scanner was set at a voltage of 49 kVp, a current of 200 μA, and a resolution of 6.8 μm per pixel to measure the IVD and endplate (EP). A resolution 16.8 μm of per pixel was set for the whole L 5 vertebral body measurement. Coronal images of the L 4 -L 5 IVD were used to perform three-dimensional histomorphometric analyses of IVD. IVD volume was defined by the region of interest to cover the whole invisible space between L 4 and L 5 vertebrae. Parameter: TV (total tissue volume) was used for three-dimensional structural analysis.
Quantitative histomorphometric analysis Quantitative histomorphometric analysis was conducted in a blinded manner with Image-Pro Plus Software version 6.0 (Media Cybernetics Inc, Rockville, Maryland, USA). EP and IVD scores were obtained as previously described. 1,38 The percentage of pSmad2/3-positive cells was obtained by counting the number of positive staining cells to the number of total cells in the NP region. The expression of integrin α V β 6 was calculated by the sum of integrated optical density in the NP region. The area of CCN2-positive staining was calculated in the whole L 4 -L 5 IVD in lumbar spine instability (LSI) mice (2-month old). In other experiments, the area of CCN2-positive staining was calculated only in the NP region. The percentage of Acan-positive staining was calculated by counting the positive staining area of region of interest that covers all cells in NP.
Quantitative reverse transcription-PCR Total RNA was extracted from NP tissue of IVD in ex vivo assay using TRIzol reagent (Sigma-Aldrich) according to the manufacturer's instruction. The yield and purity of RNA were estimated spectrophotometrically using theA260/ A280 ratio. Two micrograms of RNA was reversetranscribed into complementary DNA using the SuperScript first-strand synthesis system (Invitrogen). One microliter of Bone Research (2017) 17008 TGFβ in IVD homeostasis Q Bian et al complementary DNA was subjected to quantitative reverse transcription-PCR amplification using SYBR GREENPCR Master Mix (Promega, Madison, WI, USA) and sequence-specific primers for Acan: 5′-CAGATGGCACC CTCCGATAC-3′and 5′-GACACACCTCGGAAGCAGAA-3′. The value of gene expression was normalized relative to the mouse GAPDH: 5′-AATGTGTCCGTCGTGGATCTGA-3′ and 5′-AGTGTAGCCCAAGATGCCCTTC-3′. PCR reactions were performed in triplicates. The data were analyzed using the 2 − ΔΔCT method.

Western blot
Western blot analyses were conducted on the protein extraction from NP tissue in ex vivo assay. The protein extraction was centrifuged, and the concentration of supernatants was evaluated by DC protein assay (Bio-Rad Laboratories, Hercules, CA, USA), then the proteins were separated by SDS-polyacrylamide gel electrophoresis and blotted on a polyvinylidene fluoride membrane (Bio-Rad Laboratories). After incubation in specific antibodies, proteins were detected using an enhanced chemiluminescence kit (Amersham Biosciences, Pittsburgh, PA, USA). The target protein concentrations were examined by antibodies recognizing mouse pSmad2 (1:1 000, 3101, Cell Signaling Technology Inc., Danvers, MA, USA), Smad2 (1:1 000, 3103, Cell Signaling Technology Inc.), integrin α V (1:500, sc-6617-R, Santa Cruz), and GAPDH (1:1 000, 8884, Cell Signaling Technology Inc.).

Statistics
The data were expressed as mean ± s.d., and statistical significance was determined using a Student's t-test in time point or genetic mice comparison, or one-way analysis of variance followed by a post hoc Least-Significance-Difference (LSD) test (homogeneity of variance) or a Tukey's test (heterogeneity of variance) in treatment or ex vivo assay comparison. The level of significance was defined as Po0.05. All data analyses were performed using SPSS 15.0 analysis software (SPSS Inc, Chicago, Illinois, USA).

RESULTS
Activation of TGFβ associates with reduced NC cell vacuoles and increased extracellular proteoglycan in response to mechanical stress The IVD height increased in mice from birth through 1 month, and then remained stable before decreasing around 4 months. The NC cells' sphere-shaped vacuole observed at birth changed to a spindle-like shape by 4 months (Figure 1a and c). Extracellular aggrecan secretion was stimulated on day 3 after birth and accumulated around the NC cells to generate IVD extracellular space adjacent to endplates from day 7 (Figure 1b). To investigate that these changes were associated with mechanical stress, we employed a LSI mouse model by removing spinous processes and posterior supraspinous and interspinous ligaments (Figure 1d). [28][29] The effects of mechanical stress on NC cells were analyzed by immunostaining of the IVD sections collected from 2month-old LSI mice, in which the IVD space is peaked. The vacuole sizes of NC cells gradually decreased beginning at 2 weeks post surgery in LSI, whereas in the sham-operated mice NC size decreased at 4 weeks with Safranin O staining (Figure 1e). Significant IVD degeneration was observed in LSI mice relative to sham-operated mice, as shown by IVD score beginning as early as 1 week post surgery 1,38 (Figure 1g). Mechanical stress has been shown to activate latent TGFβ. [20][21][39][40] Immunostaining revealed that phosphorylated Smad2/3-positive cells (pSmad2/3 + ) in the NP were significantly increased 1, 2, and 4 weeks post surgery in the LSI mice relative to the sham controls ( Figure  1f and h). Further, the levels of CCN2, a TGFβ downstream factor that upregulates the synthesis of matrix proteins in IVDs, [26][27][41][42][43] were significantly increased 2 weeks post surgery in LSI mice and gradually decreased to sham control levels by 4 weeks, consistent with the increase of TGFβ activity (Figure 2a-c). Similarly, aggrecan expression, upregulated by CCN2, 26,44 was also increased in LSI mice (Figure 2d and e). Taken together, the data reveal that increase of TGFβ activity in response to mechanical stress stimulates secretion of extracellular proteins such as aggrecan. Simultaneously, a reduction in intracellular vacuoles was observed with a transition of IVD space maintained by intracellular vacuoles in early postnatal life versus extracellular matrix in adulthood or with increased mechanical stress.
Integrin α v β 6 induces TGFβ activation in response to mechanical stress to regulate NC cell function We then investigated the mechanism of mechanical stressinduced activation of latent TGFβ. The α v β integrins are one known mechanism that mediates cell-induced conformational change of TGFβ latent complex to release active TGFβ. [21][22][23][24][45][46] The α v integrin is the common α subunit for β integrins. Immunostaining of IVD sections revealed that the expression of one specific β integrin, α v β 6 , in the NP was significantly increased 2 and 4 weeks post surgery in LSI mice relative to sham control (Figure 3a and b). The pattern of elevation of α v β 6 expression was similar to the increase of pSmad2/3-positive cells (Figure 1f and h), whereas the expression of β 8 , α v β 5 , and α v β 3 did not correlate (Figure 3c-h).
To determine a causal relationship, active TGFβ, RGD peptide, neutralizing antibodies against TGFβ orα v β 6 , or Bone Research (2017) 17008 TGFβ in IVD homeostasis Q Bian et al vehicle were applied to an ex vivo IVD compression loading model. 37 Immunostaining demonstrated that RGD peptide, antibodies against TGFβ (1D11), or α v β 6 all inhibited stress-induced phosphorylation of Smad2/3 in NC cells (Figure 3i and k). The result was confirmed by western blot analysis (Figure 3l). Importantly, the morphology of NC cell vacuoles was altered in the active TGFβ and vehiclecompression treatment groups (Figure 3j, second and third    Bone Research (2017) 17008 TGFβ in IVD homeostasis Q Bian et al receptor (TβRI) inhibitor (1 mg·kg − 1 ) was intraperitoneally injected in the LSI mice post surgery. Degeneration of IVD was attenuated as assessed by IVD score (Figure 7a). The vacuolar morphology was preserved with the TβRI inhibitor treatment relative to shamoperated versus LSI-operated with vehicle treatment (Figure 7d). As expected, pSmad2/3 + cells were reduced by TβRI inhibitor (Figure 7b and e). Expression of aggrecan and CCN2 was maintained at similar levels of s ham-operated mice as opposed to the elevation of both seen in the LSI-operated vehicle-treated group (Figure 7c, f and g).
In parallel, we administrated the TGFβ-neutralizing antibody directly into the IVD in a rat caudal IVD compression model. [31][32][33][34] (Figure 8a) Neutralizing active TGFβ in the IVD prevented IVD degeneration as defined by preservation of the NC cells morphology under static compression relative to vehicle treatment (Figure 8c). The expression pSmad2/3 + in the NC cells was reduced with TGFβ antibody treatment compared to vehicle under static compression (Figure 8d and h). Similar results were obtained using a caudal spine instability mouse model by intraperitoneal injection of TβRI inhibitor 30,[47][48] (Figure 8b), in which IVD degeneration was effectively prevented (Figure 8e-g and 8i-k ).

DISCUSSION
The initiation of DDD by NP dysfunction is a hot topic in the spinal disease field. NP cells originate from the notochord. During development, NC cells regulate the spinal morphogenesis by control of intracellular vacuoles evenly distributed in the NP to generate intervertebral space. 6,10,[49][50] After birth, IVDs gradually change their content with a reduction of intracellular vacuoles of NC cells corresponding to increase extracellular proteoglycans. 15 Loss of vacuolar NC cells and their morphologic and functional transition are thought as early degenerative changes of IVD. 12 However, the mechanism of NC cells transition is not well understood. In this study, we revealed that integrin α v β 6 -induced activation of TGFβ controls the functional transition of NC cells in IVDs in multiple rodent models. Active TGFβ reduced intracellular vacuoles of NC cells, but increased extracellular proteoglycans, thereby shifting the burden of mechanical loads from the cells to the extracellular environment.
Mechanical loads are one of the most important etiologies of DDD. 50 In normal physiological situations, appropriate mechanical loading produced by body weight and muscle force is necessary to activate TGFβ in the NP to modulate anabolic activity of NC cells and maintain IVD homeostasis, similar to its role in homeostasis of the bone and cartilage. 40,[51][52][53] However, in pathological conditions, as we found in the present study, compression stresses activate excess TGFβ and result in accelerated functional transition of NC cells, leading to pathologic changes of IVD.
Central to this mechanosignal transduction is the activation of TGFβ by integrin α v β 6 . Several of integrin α and β chains have been found in IVD involving in cell-matrix interaction. 25,54 Integrin α 5 β 1 has recently been shown to be associated with mechanotransduction in IVD cells. 55 Mechanical stress has been shown to induce cell cytoskeleton changes, leading to integrin α v β 6 binding to the latent TGFβ complex for its conformational change to allow active TGFβ to bind to its receptor. 45 Similarly, we found it is integrin α v β 6 that correlates to active TGFβ signaling pathway and regulates NC cells function by mechanical stimili. TGFβ signaling activates transcription and secretion of CCN2, which in turn binds to CCN2 receptor to stimulate a cascade of changes in extracellular proteoglycan expression. 56 Aggrecan, the major proteoglycan in IVD, exerts osmotic pressure to resist compressive loads. 57 However, aggrecan also affects the osmotic pressure in the IVD. Besides shifting water from outside the IVD, our study suggests that aggrecan likely also results in a shifting of water from intracellular to extracelluar matrix and leads to NC cells morphologic transition by dehydration.
TGFβ is recognized as an anabolic factor in IVD and has been found to prevent DDD. 58 However, high levels of TGFβ1 have also been observed in the IVDs from DDD patients. [39][40][59][60][61][62][63] One group has previously shown that inhibition of TGFβ can prevent DDD, which was attributed to a change in pSmad1/5/8 to pSmad 2/3 downstream signaling. 64 The various effects of TGFβ are likely related to the concentration of TGFβ. Our results demonstrated that pathologic mechanical loading on the spine drove aberrant overactivation of TGFβ, resulting in DDD. Complete inhibition of TGFβ signaling in knockout of TβRII or α v mice also caused IVD degeneration by mechanical stimuli. Thus, TGFβ regulates IVD cell function and homeostasis, whereas either high or low concentrations of TGFβ led to DDD development.
DDD is a polygenic disease exacerbated by many different factors. 2,16,50,[64][65][66][67][68] In this study, we found that under mechanical stress, αvβ6 integrin is activated and can release TGFβ from its latent form. Activation of TGFβ signaling pathway in multiple mouse models increased the expression of CCN2, which then upregulated aggrecan. High levels of aggrecan likely increase the osmotic pressure of the extracellular environment to result in a shift in vacuole liquid volume, resulting in an apparent change in NC cell morphology, from vacuole-like to fibroblast-like cells (Figure 8l). Reduction of aberrant TGFβ overactivation in the IVDs through modulation of mechanical stress, or inhibition of either α v β 6 or TGFβ signaling, could have therapeutic potential for DDD.