Restoration of physiologic loading modulates engineered intervertebral disc structure and function in an in vivo model

Abstract Tissue‐engineered whole disc replacements are an emerging treatment strategy for advanced intervertebral disc degeneration. A challenge facing the translation of tissue‐engineered disc replacement to clinical use are the opposing needs of initial immobilization to advantage integration contrasted with physiologic loading and its anabolic effects. Here, we utilize our established rat tail model of tissue engineered disc replacement with external fixation to study the effects of remobilization at two time points postimplantation on engineered disc structure, composition, and function. Our results suggest that the restoration of mechanical loading following immobilization enhanced collagen and proteoglycan content within the nucleus pulposus and annulus fibrosus of the engineered discs, in addition to improving the integration of the endplate region of the construct with native bone. Despite these benefits, angulation of the vertebral bodies at the implanted level occurred following remobilization at both early and late time points, reducing tensile failure properties in the remobilized groups compared to the fixed group. These results demonstrate the necessity of restoring physiologic mechanical loading to engineered disc implants in vivo, and the need to transition toward their evaluation in larger animal models with more human‐like anatomy and motion compared to the rat tail.


| INTRODUCTION
Intervertebral disc degeneration manifests as a spectrum of compositional, structural, and functional changes to the disc and its subcomponents, the annulus fibrosus (AF), and nucleus pulposus (NP). 1 Disc degeneration is frequently associated with axial back pain and neurogenic extremity pain, thereby necessitating clinical intervention. 2 Severe degeneration is commonly treated with spinal fusion surgery, which removes the degenerative disc, decompresses the adjacent nerve roots, and immobilizes the pathologic motion segment. 3 Fusion does not restore native disc structure or function, and has been associated with degeneration of adjacent motion segments, potentially via alterations to whole spine kinematics. 4 Due to the limitations of currently available clinical treatments, new strategies that aim to functionally regenerate the disc have become a major area of research.
For end-stage degeneration, where all disc substructures have undergone degenerative changes, a biologic tissue-engineered disc replacement, which could integrate with the surrounding native tissues, restore healthy disc structure and function, and continually remodel during physiologic use, is an attractive treatment option. 5 Tissue-engineered constructs for whole disc replacement generally consist of an isotropic cell-seeded hydrogel for the NP region, surrounded by a cell-seeded biomaterial for the AF region. This AF region may further be refined to recapitulate aspects of the native AF anisotropy and organization, with recent formulations employing circumferentially oriented collagen and ordered angle-ply laminate structures. 5,6 While a variety of such whole, tissue engineered disc designs have been characterized in vitro, the in vivo evaluation of such constructs has been relatively limited to date. [7][8][9][10][11][12][13] An inherent challenge in the process of in vivo translation of these tissue engineered disc replacements is the balance between initially protecting and stabilizing the implant to allow for sufficient integration to retain the construct in the disc space, but eventually providing for the restoration of physiologic loading to maintain and promote disc homeostasis through mechanical loading. 14,15 Additionally, the degree to which the engineered implant matches disc functional properties at the time of implantation (vs matures in situ) will define the degree to which it can bear load from the outset. The first tissue-engineered discs evaluated in vivo were composed of an NP cell-seeded alginate hydrogel surrounded by an AF cell-contracted aligned collagen gel, and were implanted for up to 8 months in the rat caudal disc space without fixation. 7 While these constructs were retained and performed well in the rat caudal spine, when translated to a larger canine cervical disc replacement model, displacement of the implant occurred in nearly half the animals investigated. 16 It is not clear if this displacement was due to the larger loads placed on the implant, or whether it was due the greater degrees of freedom in motion about the cervical spine compared to the lumbar spine, or both.
Our group has developed a whole tissue engineered disc-like angle ply (DAPS) construct composed of a cell-seeded hyaluronic acid hydrogel for the NP region, surrounded by cell-seeded layers of electrospun poly (ε-caprolactone) (PCL) where the nanofibers within a layer are oriented at ±30 to the long axis of the construct to mimic the hierarchical structure of the native AF. 8 These tissue-engineered discs can be fabricated with NP and AF components only (DAPS), or can include acellular PCL foams as endplate analogs (eDAPS). 9,10 We previously demonstrated that for DAPS implanted in vivo in the rat caudal spine to remain in place, the motion segment had to be immobilized using an external fixation device to prevent extrusion. 17 However, long-term immobilization of the native intervertebral disc results in substantial alterations to disc height and mechanical properties indicative of degeneration, in addition to reduced disc cell anabolism and increased catabolism. [18][19][20][21] Long-term immobilization is therefore not only contrary to the goals of a tissue engineered disc replacement, but also would likely have a detrimental effect on engineered disc health.
Thus, the purpose of this study was to determine the effects of removal of external fixation (remobilization) on engineered disc composition, structure, and mechanical function, when the immobilization was removed at different times postimplantation. We hypothesized that after sufficient time of implantation to allow for boney fixation, restoration of physiologic mechanical loading would improve engineered disc composition and mechanical properties.

| Study design
The effects of remobilization (removal of external fixation) on eDAPS structure, composition, and function were assessed via the study design summarized in Figure 1. Thirty athymic rats underwent a surgical procedure to implant precultured eDAPS within the caudal spine and immobilize the implanted motion segment with an external fixator. Both early (5 weeks postimplantation) and late (10 weeks postimplantation) remobilization time points were investigated in this study. To do so, one subset of animals was designated for a 10-week study endpoint, where the external fixator was removed in 6 rats at 5 weeks postimplantation (10 W R), but left in place in the remaining 5 rats for the 10 week study duration (10 W F). A second subset of animals was designated for a 20-week study endpoint, where the external fixator was removed in 10 rats at 10 weeks postimplantation (20 W R), but left in place in 9 rats for the 20 week study duration (20 W F). At the study endpoints, animals were euthanized and vertebral body-eDAPS-vertebral body motion segments and adjacent native control motion segments were harvested for analysis. All motion segments in each experimental group underwent MRI T2 mapping, and were subsequently split for histological analysis (n = 2-3 per group), or mechanical testing followed by biochemical assays (n = 3-7 per group). A subset of the data generated for the 10 W F and 20 W F groups has been published previously in Gullbrand et al, and were utilized as controls for this study. 10 F I G U R E 1 Schematic illustrating the study design and time points for remobilization 2.2 | eDAPS fabrication and preculture Bovine NP and AF cells were isolated from caudal discs (~3 years old, 2 hours after sacrifice; JBS Souderton, Inc. Souderton, PA), as previously described. 22 Cells were expanded to passage two prior to use in Basal Media containing Dulbecco's Modified Eagle Medium (DMEM; Gibco, Invitrogen Life Sciences, Carlsbad, CA), 10% fetal bovine serum (Gibco), and 1% Antibiotic-Antimycotic (Gibco). eDAPS sized for the rat caudal spine (5 mm high, 4-5 mm diameter) were fabricated, cultured and seeded according to our previously established methods. 9,23 Briefly, the AF region of the eDAPS was fabricated by concentrically wrapping electrospun scaffolds of layered PCL and poly(ethylene oxide) (PEO) nanofibers to achieve a layered The cell-seeded AF and NP regions of the eDAPS were cultured separately for 2 weeks in chemically defined media, composed of high glucose DMEM supplemented with 1% PSF, 40 ng/mL dexamethasone (Sigma-Aldrich), 50 μg/mL ascorbate 2-phosphate (Sigma-Aldrich), 40 μg/mL L-proline (Sigma-Aldrich), 100 μg/mL sodium pyruvate (Corning Life Sciences), 0.1% insulin, transferrin, and selenious acid (ITS Premix Universal Culture Supplement; Corning), 1.25 mg/mL bovine serum albumin (Sigma-Aldrich), 5.35 μg/mL linoleic acid (Sigma-Aldrich), and 10 ng/mL TGF-β3 (R&D Systems). 23 The AF and NP regions were combined after 2 weeks, and two acellular PCL foam endplates were apposed to either side of the AF region using 31G needles. 25,27 Media were changed three time per week, and the needles were removed immediately prior to implantation.

| eDAPS implantation and remobilization
All in vivo studies were approved by the Corporal Michael J. Crescenz

Veterans Affairs Medical Center Institutional Animal Care and Use
Committee, and were performed according to the guidelines recommended by this committee. Following 5 weeks of preculture, eDAPS were implanted in the C8-C9 disc space of the athymic rat caudal spine (Foxn1 rnu retired breeders, Envigo), as previously described. 9,10 Under anesthesia and using standard aseptic techniques, two kirschner wires were passed through the C8 and C9 vertebral bodies, allowing for the placement of an external fixator designed to immobilize the implanted level. The native disc and portion of the vertebral endplates were removed using a high speed burr, and the eDAPS were press fit into the evacuated space. The incision was closed with suture, and the rats were returned to normal cage activity. As per Figure 1, at 5 or 10 weeks postimplantation, a subset of animals were anesthetized, and the external fixator and kirschner wires were removed (10 W R, 20 W R groups) to restore physiologic loading to the eDAPS implant. The external fixator was left in place in the remaining animals (10 W F, 20 W F groups). After recovery from anesthesia, the rats were returned to normal cage activity for the remainder of the study.

| Radiographic analysis and magnetic resonance imaging
eDAPS implanted motion segments from the 10 W R and 10 W F groups underwent μCT scans (Scanco μCT50, 10 μm resolution) to assess bone morphology and vertebral anatomic positioning. μCT scans were performed on isolated motion segments following euthanasia and immediate fixator removal in the 10 W F groups. The angle between the vertebral bodies adjacent to the eDAPS was measured from mid-coronal μCT slices, using ImageJ. In the 20 W R and 20 W F groups, serial radiographs of the same animal were taken using a fluoroscope over the duration of the study in order to capture dynamic changes in vertebral body angulation over time. Awake animals were briefly restrained for fluoroscopy, and the animals were positioned in the same orientation as for the eDAPS implantation surgery. Radiographs in the fixed group were obtained with the fixator in place.
Radiographs were obtained immediately postoperative, and at 10 weeks (immediately following fixator removal in the remobilized group), 15 weeks and 20 weeks. From these images, the angle between vertebral bodies was quantified in Image J.
After euthanasia, eDAPS implanted and native rat tail motion segments underwent MRI scanning at 4.7 T using a multi-echo-multi-spin sequence for T2 mapping (0.5 mm slice thickness, 117 μm in plane resolution, 16 echoes, TR/TE = 2000/11.13 ms), and a custom 17 mm diameter solenoid coil. 28 From the image series, average T2 maps for each experimental group were generated using a custom MATLAB code, as previously described, to quantify T2 relaxation times in the NP. 29 2.5 | Biomechanical testing and analysis eDAPS implanted and native motion segments were subjected to compressive mechanical testing, followed by tension to failure testing, as previously described. 10 The skin of the tail was removed and the vertebral bodies adjacent to the eDAPS were cleared of soft tissue.
Ink marks were placed on the vertebral bone immediately distal and proximal to the eDAPS to allow for optical displacement tracking during testing. Motion segments were potted in a low melting temperature indium casting alloy, and subjected to a testing protocol consisting of 20 cycles of compression from 0 to −3 N (0 to −0.25 MPa) at 0.05 Hz (Instron 5948) in a bath of phosphate-buffered saline at room temperature. The 20th cycle of compression was used to calculate toe and linear region modulus, and transition and maximum strain. After compression testing, a complete circumferential dissection of the muscle and tendon surrounding the eDAPS was performed, and tension to failure testing was performed at 0.025 mm/s. A bi-linear fit of the tension curves was used to quantify toe and linear tensile modulus and failure stress and strain. Mechanical properties in tension and compression were normalized to disc height and area obtained from MR images assuming circular geometry of the eDAPS.

| Evaluation of eDAPS structure and composition
Implanted eDAPS subjected to mechanical testing were removed from the motion segment and were manually dissected into NP, AF, and EP portions and individually digested overnight in proteinase K at 60 C. Collagen content of each region was quantified using the p-diaminobenzaldehyde/chloramine-T assay for ortho-hydroxyproline, and glycosaminoglycan (GAG) content was quantified using the dimethylmethylene blue (DMMB) dye binding assay. 30

| Statistical analyses
Statistical analyses were conducted in GraphPad Prism 6. Data are presented graphically as mean with SD. Data were first tested for normality using the Shapiro-Wilk test. If the sample size of the data set was insufficient to test for normality, data were assumed to be nonnormally distributed. Statistically significant differences between groups were assessed using a one-way ANOVA with Tukey's posthoc test or a Kruskal-Wallis test with Dunn's posthoc test for normally and non-normally distributed data, respectively. A two-way, repeated measures ANOVA with Sidak's multiple comparison test was utilized to determine statistically significant differences in vertebral angle from the serial radiographs. Statistical significance was defined as P < .05.

| RESULTS
In all experimental groups, eDAPS maintained near native NP T2 relaxation times for to 20 weeks in vivo, with no detectable difference in NP T2 occurring following remobilization at either time point. In the disc, T2 relaxation times are correlated with both water and proteoglycan content, 28 and in engineered discs, this is reduced with increasing deposition of matrix within the construct (Figure 2). Alcian blue and picrosirius red-stained histology ( Figure 3A)  This was primarily evidenced as an angulation of the tail away from the initial incision utilized to implant the eDAPS. Quantitatively, there was a 174% increase (P < .001) in the angle between the vertebral bodies at 10 weeks postimplantation in the remobilized group compared to the fixed group, as measured from microCT scans ( Figure 6A,   B). In an effort to track this angulation phenomena following fixator removal, serial radiographs were acquired in animals in the 20 week study, where the fixator was removed at 10 weeks ( Figure 6C). The angle between vertebral bodies in the remobilized group significantly increased over time (P < .05), while there was no detectable change in vertebral body angle in the fixed group from 10 to 20 weeks ( Figure 6D). There were no differences in vertebral body angle between fixed and remobilized groups at the 10-week time point, demonstrating that the change in vertebral body angle was not an acute phenomena due to fixator removal, but rather occurred gradually over time. At 20 weeks, the vertebral body angle was 98% higher (P < .001) in the remobilized group compared to the fixed group.
In order to assess the functional consequences of the structural and compositional alterations to the eDAPS in vivo following remobilization, we performed compressive and tensile biomechanical testing of the eDAPS implanted motion segments. There were no detectable differences in compressive toe and linear modulus, or maximum strain at 3 N compression between fixed and remobilized eDAPS, or compared to native rat tail motion segments at any time point ( Figure 7A-C,E). Transition strain ( Figure 7D) was significantly (P = .006) increased in the 10 W R group compared to native controls.
Tensile strength ( Figure 8A-E) trended to be lower in the remobilized group compared to the fixed group, particularly at 20 weeks, with respect to linear modulus (mean 80% reduction) and failure stress (mean 70% reduction). In all groups, failure of the eDAPS occurred at the interface between the PCL endplate and the engineered disc, rather than at the endplate/vertebral bone interface.

| DISCUSSION
Whole disc tissue engineering holds significant promise as an alternative to fusion surgery for patients with advanced disc degeneration and associated back pain. One of the challenges in the translation of tissue engineered discs is promoting the robust integration of a precultured, engineered construct with the native spinal tissues, primarily the adjacent vertebral bone. In order for this integration to occur, a period of immobilization of the motion segment is likely necessary to prevent displacement or excessive motion until such time that sufficient integration has occurred to support physiologic loading. Indeed, in animal models, immobilization has been necessary to retain tissue engineered discs within the intervertebral space. 16,17 However, longterm immobilization or unloading is known to cause degenerative changes to the native, healthy disc, 18,19 and would likely have a detrimental effect on the composition and function of an engineered disc as well. Here, we utilize our established model of total disc replacement in a rat tail model with external fixation to investigate the effects of restoring physiologic loading (remobilization) to the implant at both early (5 weeks postimplantation) and late (10 weeks postimplantation) time points.
Our results demonstrate that remobilization was beneficial for engineered disc matrix composition, leading to improved GAG and collagen II content in the NP, and higher collagen content in the AF compared to fixed groups. These results are consistent with the large body of work studying the effects of dynamic compression on the native intervertebral disc, where physiologic loading rates and magnitudes enhance anabolic remodeling in the disc, inclusive of increased matrix deposition. 14,15,20,32 Additionally, dynamic loading during in vitro culture increases the GAG and collagen content of tissue engineered discs, further highlighting the importance of anabolic loading. 33 Remobilization additionally enhanced the integration of the PCL endplate region of the eDAPS with the native vertebral body, as evidenced by increases in organized collagen content and mineralization in remobilized vs fixed groups. Mechanical loading can significantly increase bone volume formed in fracture healing models, and the timing of loading application of loading can be modulated to enhance bone vascularization as well. 34,35 This may have important implications for enhancing the remodeling of the PCL endplate region of the eDAPS, as the formation of a boney, vascularized interface will likely be critical for the long-term maintenance of tissue engineered disc homeostasis. Taken together, these results suggest that the disc cells initially seeded within the eDAPS, and infiltrating native cells at the endplate, are capable or responding to mechanical cues in vivo.
Despite the apparent positive effects of in vivo remobilization on eDAPS composition and integration, remobilization was detrimental to motion segment morphology and mechanics, as significant angulation of the vertebral bodies at the implanted level was observed following fixator removal. This angulation still occurred, but was mitigated by prolonging the duration of immobilization (5 vs 10 weeks, 174% vs 98% increase in vertebral body angle) prior to fixator removal. The altered motion segment morphology is likely due to the disruption of the dorsal tail tendons during the eDAPS implantation surgery, leading to asymmetric loading of the motion segment following remobilization. This asymmetric loading may be particularly impactful in this implantation site, as the lack of facet joints in the rat tail allow for multi-axial motion as the rat utilizes its tail for balance and proprioception. 36 The biomechanical function of the eDAPS implanted motion segment in compression was largely unaffected by this altered motion segment morphology, with the exception of an increase in transition strain in the early remobilization (10 W R) group.
Tensile mechanical properties, however, were impacted by remobilization, with lower tensile modulus and failure stress in the remobilized group compared to the fixed group at 20 weeks. Failure in all groups occurred at the interface between the PCL endplate and tissue engineered AF and NP components. Thus, although the integration of the PCL endplate with the vertebral body itself was enhanced with remobilization, the interface between the endplate and disc may have been weakened in the remobilized group as a consequence of the angulation of the vertebral bodies causing shearing at this interface.
Overall, the results from this study suggest that the restoration of mechanical loading to a tissue engineered disc implant in vivo will likely be essential to maintain construct homeostasis with long-term implantation. Indeed, remobilization is a requirement for functional restoration of the spine. However, it is clear from prior studies that some degree and duration of immobilization is necessary for sufficient integration to develop and to protect against displacement of the construct. The design and optimization of internal provisional fixation systems for engineered disc replacements may be necessary to facilitate the translation of this technology toward clinical use. For example, bioresorbable plates have already been investigated ex vivo to temporarily stabilize tissue engineered disc implants, 37 the stiffness of which could potentially be tuned to optimize bone formation within the endplate region. 38 Finally, the detrimental effects of remobilization observed in this study, namely the alterations to motion segment morphology, are likely exacerbated in the hypermobile rat tail, which lacks stabilizing posterior elements. Future work in large animal models that have more human-like morphology and motion, will be critical to fully understand the effects of remobilization on engineered disc structure and function.