The mechano‐response of murine annulus fibrosus cells to cyclic tensile strain is frequency dependent

Abstract The intervertebral disk (IVD) is a composite structure essential for spine stabilization, load bearing, and movement. Biomechanical factors are important contributors to the IVD microenvironment regulating joint homeostasis; however, the cell type‐specific effectors of mechanotransduction in the IVD are not fully understood. The current study aimed to determine the effects of cyclic tensile strain (CTS) on annulus fibrosus (AF) cells and identify mechano‐sensitive pathways. Using a cell‐type specific reporter mouse to differentiation NP and AF cells from the murine IVD, we characterized AF cells in dynamic culture exposed to CTS (6% strain) at specific frequencies (0.1 Hz, 1.0 Hz, or 2.0 Hz). We demonstrate that our culture model maintains the phenotype of primary AF cells and that the bioreactor system delivers uniform biaxial strain across the cell culture surface. We show that exposure of AF cells to CTS induces cytoskeleton reorganization resulting in stress fiber formation, with acute exposure to CTS at 2.0 Hz inducing a significant yet transient increase ERK1/2 pathway activation. Using SYBPR‐based qPCR to assess the expression of extracellular matrix (ECM) genes, ECM‐remodeling genes, candidate mechano‐sensitive genes, inflammatory cytokines and cell surface receptors, we demonstrated that exposure of AF cells to CTS at 0.1 Hz increased Acan, Prg4, Col1a1 and Mmp3 expression. AF cells exposed to CTS at 1.0 Hz showed a significant increase in the expression of Acan, Myc, and Tnfα. Exposure of AF cells to CTS at 2.0 Hz induced a significant increase in Acan, Prg4, Cox2, Myc, Fos, and Tnfα expression. Among the cell surface receptors assessed, AF cells exposed to CTS at 2.0 Hz showed a significant increase in Itgβ1, Itgα5, and Trpv4 expression. Our findings demonstrate that the response of AF cells to CTS is frequency dependent and suggest that mechanical loading may directly contribute to matrix remodeling and the onset of local tissue inflammation in the murine IVD.


| INTRODUCTION
According to the most recent Global Burden of Disease Study, back pain is the leading cause of years lived with disability worldwide. 1 Although the etiology of chronic back pain remains largely unknown, persistent back pain has been associated with magnetic resonance imaging (MRI) findings of lumbar intervertebral disk (IVD) degeneration in~30% of patients. 2 The pathophysiology of IVD degeneration involves progressive cell-mediated changes to the IVD microenvironment, including extracellular matrix (ECM) breakdown, altered matrix synthesis, and local tissue inflammation, ultimately resulting in structural and functional tissue failure. [3][4][5][6] Although the etiology of IVD degeneration is unclear, initiation and progression of the degenerative cascade involves multiple interdependent factors including altered mechanical loading, 7-10 reduced nutrient supply, [9][10][11][12] altered cellular composition, 3,13 and hereditary factors. 5,14,15 The IVD is a fibrocartilaginous connective tissue structure essential for spine load bearing and movement. Anatomically, IVDs consist of three distinct tissues: the central gelatinous nucleus pulposus (NP), the outer collagenous annulus fibrosus (AF) that circumferentially encapsulates the NP, and the cartilage endplates that anchor the disk to the adjacent vertebrae and allow for passive diffusion of nutrients to the IVD. 8,12,16,17 As such, IVDs are heterogeneous composite structures with each tissue having a unique structure and specific ECM composition that together form the complex microarchitecture of the IVD required for joint function. 17 Similar to other musculoskeletal tissues, mechanical loading poses an interesting dichotomy in IVD biology: while physiological levels of mechanical loading during moderate locomotive activities are essential for IVD health and tissue homeostasis, 10,18,19 mechanical stimuli due to high and low degrees of physical activities (overloading and immobilization) can contribute to tissue degeneration. [18][19][20] IVDs are subjected to various types of mechanical forces, including hydrostatic pressure, compressive, tensile and shear forces. [21][22][23][24] Within this dynamic microenvironment, NP cells are primarily exposed to compressive and hydrostatic loading, while AF cells are exposed to multidirectional deformation resulting in tensile strain. [21][22][23][24] Previous studies investigating the mechanical forces experienced by human IVDs reported that in performing daily activities, intradiscal pressures are predicted to range from 0.1 MPa (bedrest) to 2.5 MPa (heavy lifting). 25 Furthermore, studies characterizing the mechanical properties and strain profiling of human IVDs reported that the AF experiences 1% to 13% strain during daily activities, 26 with tissue strain reaching up to 26% during physiological compressive loading. 27 To better understand how the dynamic environment of the IVD regulates cell function, studies have examined the effects of cyclic tensile strain (CTS) on AF cells to model their exposure to tensile loading in vivo. In response to uniaxial CTS (2%, 1.0 Hz), human AF cells in 3D culture show increased aggrecan gene expression and decreased expression of matrix degrading enzymes. 28 Using the same model system, exposure of AF cells to 4% CTS (1.0 Hz) induced a modest increase in matrix gene expression, dependent on osmotic potentials. 29 In contrast, exposure of human AF cells to high levels of bi-axial strain (20% CTS, 0.001 Hz) induced catabolic responses, including upregulation of inflammatory cytokine production and cell death. 30 Studies using the Flexcell tension system reported that human AF cells exposed to 10% CTS at 1.0 Hz showed increased expression of matrix genes and decreased expression of matrix degrading enzymes, whereas 10% CTS frequency at 0.33 Hz shifted the mechano-response toward matrix catabolism. 31,32 Of note, application of either loading protocol to human AF cells from degenerate IVDs increased catabolic gene expression. 30 To date, reports on the mechano-sensory mechanism of AF cells have focused on the role of integrin-ECM interactions in CTS sensing 32,33 ; however, specific signaling pathways involved in mechanotransduction remain largely unexplored. Mitogen-activated protein kinases (MAPKs) are a family of highly conserved protein kinases that serves as intermediates in signal transduction pathways. 34 MAPKs are grouped into three subfamilies-extracellular signal-regulated kinases (ERKs), p38, and c-jun N-terminal or stress-activated kinases (JNK/ SAPK)-based on their sequence, sensitivity to activation and mechanisms of action. 35 Functionally, ERKs regulate cell survival and proliferation, 36 whereas p38 kinases regulate inflammatory responses, cell cycle control, and differentiation. 37 Interestingly, AF cells isolated from degenerative human IVDs show activation of all three MAPK subfamilies (ERK1/2, p38, JNK) upon mechanical stimulation. 38 The current study used a transgenic mouse model to genetically label and isolate AF cells from the murine IVD and employed an in vitro culture system to deliver acute bi-axial CTS. Bi-axial multidirectional load was chosen to model complex circumferential tensile load experienced by the AF cells in vivo when the tissue is subjected to multiaxial load constrained by the adjacent vertebral bodies (axial and radial load). 23,27 The study aimed to quantify the mechanoresponse of healthy murine AF cells to different loading conditions, focusing on cytoskeletal adaptation, changes in gene expression, and the identification of CTS-induced signaling pathways.

| Tissue isolation and culture of AF cells
Lumbar spines from 2-month-old mice were dissected, followed by microdissection of NP and AF tissues using a fluorescent stereo microscope (Leica M165 FC). Isolated tissues were immediately fixed for RNA extractions. For primary cell culture (overview in Figure 1A), intact IVDs were dissected from 2-month-old mice (cervical to caudal) and the AF tissues were microdissected (Leica M165 FC). Isolated AF F I G U R E 1 Experimental design and mechanical stimulation of murine annulus fibrosus cells. A, Schematic representation of the protocol used for the isolation and culture of primary murine annulus fibrosus cells. B, Images of the MechanoCulture B1 device inside the cell culture incubator. C, Schematic depicting the in vitro model used to deliver cyclic tensile strain. Cells were seeded on a deformable silicone membrane puncture mounted on a 24-pin mounting ring designed to generate radial motion from a linear motion input, resulting in the delivery of biaxial stretch. D, E, Schematic representations of the experimental procedures used to deliver 6% cyclic tensile strain at frequencies of 0.1, 1.0, or 2.0 Hz to investigate gene expression, C, and signal transduction, D, in mechanically stimulated annulus fibrosus cells tissues were transferred to a sterile 3 mm culture dish with 2 mL of type II collagenase (3 mg/mL; Worthington, NJ) in Dulbecco's modified Eagle's medium/Ham's F-12 medium (DMEM/F12) and incubated for 20 minutes at 37 C. AF tissues were then minced and further digested for 1 hour at 37 C. Digested tissues were triturated and filtered using a 70 μm cell strainer and cells were pelleted by centrifugation (1100 rpm for 5 minutes). Cells were plated at an initial density of 400 000 cells/cm 2 and cultured in DMEM/F12 supplemented with 10% fetal bovine serum (FBS) and 1% penicillin and streptomycin (Thermo Fisher Scientific, MA) at 37 C in a humidified atmosphere of 5% CO 2 . Media was changed every 2 days until cells reached 80% confluency. AF cells isolated from IVD tissues of two mice were pooled together and used for each experimental replicate.

| Mechanical stimulation
The MechanoCulture B1 (MCB1) device (CellScale Biomaterials Testing, Waterloo, Ontario, Canada) was used to deliver bi-axial multidirectional cyclic tensile strain to monolayer cultures ( Figure 1B,C). 41 The MCB1 device included an actuator chamber, housing a programmable circuit board and motor ( Figure 1B

| Motion tracking analysis
To quantify the magnitude and uniformity of the strain applied to the silicone membrane by the MCB1 device, motion tracking analysis was performed. Fine graphite powder was used to pattern the silicone membrane in order to quantify the motion of specific points during mechanical stretch. The patterned surface was imaged at 0.5 mm intervals using DMK 41AU02 Monochrome Camera (Imaging Source, Charlotte, NC) until the programmed displacement (3.5 mm) was reached. Using commercially available image tracking software (LabJoy, CellScale), four different regions of the membrane were tracked and the strains experienced in these regions were calculated based on pixel displacement.  imaged using the ChemiDOC XRS + System (Bio-Rad). Densitometric analysis was performed using ImageLab software (Bio-Rad). Levels of phospho-proteins were quantified and presented normalized to corresponding total protein levels.

| RNA extraction and gene expression analysis
Total RNA was extracted from IVD tissues, primary AF cells cultured on standard tissue culture plastic or silicone membrane (passage 1) or CTS-treated and unloaded AF cells 2, 6, 12, or 24 hours post-stimulation (n = 4-7) using Trizol reagent (Life Technologies) according to the manufacturer's protocol. RNA was quantified using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). Complementary DNA was synthesized from 150 ng of RNA using the Bio-Rad iScript cDNA synthesis kit. Gene expression was determined by SYBR-based real-time PCR using the Bio-Rad CFX384 thermocycler. PCR reactions were run in triplicate using 470 nM forward and reverse primers (primer sequences in Table 1) with 2x SsoFast EvaGreen Supermix (Bio-Rad). The PCR program consisted of the following: initial 2 minutes enzyme activation at 95 C, 10 seconds denaturation at 95 C, 30 seconds annealing/ elongation at 60 C, for total of 40 cycles. For cell phenotype characterization, transcript levels were determined relative to a sixpoint calibration standard curve made from pooled cDNA generated from wild type murine heart, brain, kidney, muscle, IVD, and mouse embryonic fibroblasts. For CTS experiments, gene expression values were calculated using ΔΔCt, normalized for input based on hypoxanthine quinine phosphoribosyl transferase (Hprt) expression and expressed relative to the time-matched unloaded controls within each trial.

T A B L E 1 Sequences of the primers used in the real-time PCR analysis
TTCGTAGGGATCGTTGGTCCT TACAGTGGGGCATCGTCCGT

| Statistical Analysis
All statistical analyses were performed with GraphPad Prism 6 Software (GraphPad Software, San Diego, CA). The analysis of data from stress fiber quantification and gene expression for cell phenotype characterization were analyzed using one-way analysis of variation To quantify the magnitude and uniformity of the strain applied to the silicone culture membrane, motion tracking analysis was performed. The device was programmed for a 3.5 mm linear displacement, a load translated into 24 points of radial motion by the device (Figure 1B, C).
Regions of interest (ROIs) on the silicone membrane were tracked and imaged every 0.5 mm during the displacement and the strain calculated based on pixel displacement (Figure 2A). This analysis demonstrated that the average strain experienced by the ROIs on the silicone membrane at full displacement (3.5 mm) was 6.075 ± 0.096%. The small variability in strain measured between ROIs suggests that strain across the membrane at full displacement is uniform ( Figure 2B).
Previous studies predicted that the AF experiences compressive tissue strains of 1-26%, radial tensile strain of 1-19%, and lamellar fiber strain of up to 13% when IVD is compressed with a load physiologically similar to that of walking. 26,27,42 Given such data, CTS of 6% was chosen as it falls within the physiological range of mechanical stimulation in vivo.

| Primary murine annulus fibrosus cells maintain an AF-like phenotype in culture
To allow cell type-specific isolation from the murine IVD, primary cells were isolated from the Noto cre ;ROSA mTmG/mTmG conditional reporter mice in which AF cells express red fluorescent protein (RFP), whereas notochord-derived NP cells express GFP ( Figure 3A). In all AF cell preparations maintained in monolayer culture, only RFP-expressing cells were observed, confirming the absence of notochord-derived NP cells ( Figure 3A).
To ensure that primary AF cells maintained their phenotype in our in vitro culture system, we quantified the expression of previously reported AF-and NP-associated markers [44][45][46][47][48][49][50] as well as fibroblast markers and compared expression levels to those of the intact AF and NP tissues. Primary AF cells showed robust expression of the AF-associated markers, type I collagen (Col1a1), growth differentiation factor (Gdf10), paired box 1 (Pax1) and cartilage intermediate layer protein (Cilp), compared to expression of these genes in the NP tissues ( Figure 3B). No significant differences were detected in the expression of AF markers in primary cells cultured on tissue culture plastic or silicone membranes compared to the intact AF tissues. In contrast, AF cells showed minimal or no detectable expression of the NP-associated markers, Cd24 and Brachyury, with no differences detected between AF tissue and primary cells cultured on either tissue culture plastic or silicone membranes ( Figure 3C). Moreover, we quantified the expression of the markers of the myofibroblast/activated fibroblast phenotype, alpha smooth muscle actin (Acta2), 51 and fibroblast activation protein (Fap). 52 Primary AF cells cultured on tissue culture plastic and silicone membranes showed significantly increased expression of Acta2 compared to intact AF and NP tissues; however, expression of Fap was not altered in cultured AF cells compared to intact AF tissues ( Figure 3D).

| Mechanical stimulation induces cytoskeletal rearrangement in annulus fibrosus cells
To confirm that AF cells sense and respond to the CTS delivered by the MCB1 device, cytoskeletal rearrangement was examined since cytoskeletal reorganization has been shown to play a pivotal role in AF mechanotransduction. 53 In AF cells under static culture (unloaded control), F-actin was localized near the cell periphery in a predominantly punctate distribution with weak fiber staining. This pattern of F-actin staining was similar following 30 minutes exposure of cells to CTS at 0.1 Hz ( Figure 4A solid arrow); however following exposure to CTS at 1.0 and 2.0 Hz, AF cells showed significantly increased stress fiber formation, which increased with higher frequency of loading ( Figure 4A hollow arrow). Upon quantification, 61% of cells and 74% of cells were positive for stress fibers in AF cells exposed to CTS at 1.0 Hz and 2.0 Hz protocol, respectively ( Figure 4B). The stress fibers formed in AF cells exposed to CTS at 1.0 and 2.0 Hz showed a random orientation. show RFP but not GFP expression, indicating the absence of NP cell contamination. B-D, Gene expression analysis of primary annulus fibrosus cells compared to intact IVD tissues using panels of AF-and NP-associated markers and fibroblast markers. The expression of selected genes was quantified using qRT-PCR in primary AF cells grown on standard tissue culture plastic or silicone membranes (n = 5) and compared to intact AF and NP tissues (n = 6 mice; 5-6 IVDs pooled per mouse). B, AF cells cultured on tissue culture plastic and silicone membranes maintained expression of the AF markers, Col1a1, Gdf10, Pax1, and Cilp at levels comparable to intact AF tissue, significantly higher compared to NP tissue. C, AF cells cultured on tissue culture plastic and silicone membranes exhibited significantly lower expression of the NP markers, Cd24 and Brachyury compared to NP tissues. D, AF cells cultured on tissue culture plastic and silicone membranes had significantly increased expression of Acta2 compared to both intact IVD tissues; however, Fap expression was not significantly altered in AF cell cultures compared to intact AF tissue. Grubb's outlier test was used to identify outliers. Data presented as mean ± SEM. Data analyzed using one-way analysis of variation (ANOVA) followed by Tukey's post hoc test. *P < .05; **P < .01, ***P < .001; ****P < .0001. Scale bars = tissue micrograph = 200 μm; cell culture = 100 μm was detected for p38 activation; however, changes in levels of phosphorylation did not reach significance ( Figure 5B).

F I G U R E 6
Real-time PCR analysis of gene expression in annulus fibrosus (AF) cells following acute exposure to CTS at 0.1 Hz. Primary AF cells were subjected to 6% CTS at 0.1 Hz for 30 minutes and RNA harvested following 2, 6, 12 or 24 hours to assess the expression of extracellular matrix genes, A, matrix remodeling genes, B, inflammatory cytokines, C, and candidate mechanosensitive genes, D. AF cells showed a significant increase in the expression of Col1A1 (2 hours postloading), Prg4 (2 hours, 24 hours postloading), Acan (2 hours, 24 hours postloading) and Mmp3 (2 hours postloading). Relative gene expression was calculated using ΔΔCt, normalized for input using the housekeeping gene Hprt and expressed relative to time-matched unloaded controls within each trial (control = 1; indicated as gray dotted lines). Data presented in mean ± SEM; n = 4 cell preparations. Data were analyzed using one-way analysis of variation (ANOVA) followed by either Dunnett's or Tukey's post hoc test. Grubb's outlier test used to identify outliers. *P < .05 vs unloaded control; Ф = P < .05 between fold changes at two time points  Figure 7D).

| Acute exposure of AF cells to CTS induces frequency-dependent changes in gene expression
Lastly, AF cells were exposed to acute (30 minutes) 6% CTS at 2.0 Hz showed a significant increase in the expression of the ECM genes Acan and Prg4 (6 hours and 2 hours postloading, respectively; fold change (Acan) = 2.4 ± 0.12, fold change (Prg4) = 3.2 ± 0.89) compared to unloaded controls ( Figure 8A). No significant differences were detected in the expression of Col10a1, Bgn, or Dcn ( Figure S1) or genes associated with matrix remodeling ( Figure 8B). Acute CTS at F I G U R E 9 Real-time PCR analysis of candidate cell surface receptor gene expression in annulus fibrosus (AF) cells following acute exposure to CTS at 2.0 Hz. Primary AF cells were subjected to 6% CTS at 2.0 Hz for 30 minutes and RNA harvested following 2, 6, 12 or 24 hours to assess the expression of integrin subunits, A, Toll-like receptors, B, and membrane ion channel receptors, C. AF cells showed a significant increase in the expression of Itgα5 (12 hours postloading), Itgβ1 (12 hours postloading), and Trpv4 (12 hours postloading). Relative gene expression was calculated using ΔΔCt, normalized for input using the housekeeping gene Hprt and expressed relative to timematched unloaded controls within each trial (control = 1; indicated as gray dotted lines). Data presented in mean ± SEM; n = 4 cell preparations.
Data were analyzed using one-way analysis of variation (ANOVA) followed by either Dunnett;s or Tukey's post hoc test. Grubb's outlier test used to identify outliers. *P < .05 vs unloaded control; **P < .01 vs unloaded control; Ф = P < .05 between fold changes at two time points  [66][67][68][69][70] Our validation demonstrated that the MCB1 device overcame these limitations by incorporating a design mechanism that transfers linear motion into 24 points of radial stretch, which resulted in delivery of uniform biaxial strain on the silicone membrane.
One of the earliest responses we detected in AF cells exposed to  57 Future studies using this model system will build on the characterization of the acute effects of CTS, and specifically assess the role of Rho GTPase and the changes in cell proliferation in AF cells following chronic exposure to CTS.
The pathophysiology of IVD degeneration is associated with alterations to the ECM due to imbalances between synthesis and degradation of ECM proteins. It has been hypothesized that ECM degradation fragments induce local inflammation driven by cells of the IVD. 79 However, the role of mechanical stimulation in either the initiation or propagation of local inflammation is largely unknown. Previous studies report that IVD cells exposed to high mechanical strain at low frequency (20% at 0.001 Hz) show increased gene expression of inflammatory receptors and cytokines. 30 In keeping with these findings, the present study showed that exposure of AF cells to CTS which showed no change in TLR expression in murine AF cells may be due to differences in loading parameters, such as strain percentage, frequency, and duration. Lastly, Trpv4 and P2rx7, candidate mechanosensitive channels studied in other musculoskeletal cell types [58][59][60] were upregulated in AF cells following exposure to CTS. Previous studies reported that the expression of both genes was regulated by cyclic compression in chondrocytes. 58,59 Functionally, both TRPV4 and P2X7 have been shown to elicit intracellular calcium transient upon mechanical compression. 58,59 In the context of the IVD, TRPV4 is activated by changes in osmolarity, 83 and a related purinoreceptor, P2X4, has been shown to mediate ATP-induced membrane potential response. 84 Taken together, the current study demonstrates that AF cells adapt to mechanical stimulation by regulating cell surface receptor gene expression, thereby potentially modulating the activation of intracellular signaling pathways.
A limitation of the current study is the examination of cellular responses to CTS in a cell culture system where cells were seeded onto silicone membranes; consequently, endogenous cell-matrix interactions present in the IVD microenvironment were not recapitulated.
In order to minimize these issues, AF cells were precultured on membranes in media supplemented with ascorbic acid to enable collagen production and secretion. To more fully explore the effect of ECM in mediating mechanical strain, future studies could incorporate specific ECM protein coatings in the MCB1 device. We also acknowledge that our protocols for primary murine cell isolation may have introduced heterogeneity in the AF cell population studied. To maximize the yield of our primary cell isolation from the murine IVDs, we pooled AF tissues from all anatomical regions (cervical through caudal) and included cells from both inner and outer AF. Subtle differences in the phenotype of AF cells in each of these regions, or in the mechanical loading environment from which they were isolated may impact their response to mechanical stimulation in vitro.

| CONCLUSIONS
Overall, our findings suggest that effects of CTS on healthy murine AF