A comprehensive tool box for large animal studies of intervertebral disc degeneration

Abstract Preclinical studies involving large animal models aim to recapitulate the clinical situation as much as possible and bridge the gap from benchtop to bedside. To date, studies investigating intervertebral disc (IVD) degeneration and regeneration in large animal models have utilized a wide spectrum of methodologies for outcome evaluation. This paper aims to consolidate available knowledge, expertise, and experience in large animal preclinical models of IVD degeneration to create a comprehensive tool box of anatomical and functional outcomes. Herein, we present a Large Animal IVD Scoring Algorithm based on three scales: macroscopic (gross morphology, imaging, and biomechanics), microscopic (histological, biochemical, and biomolecular analyses), and clinical (neurologic state, mobility, and pain). The proposed algorithm encompasses a stepwise evaluation on all three scales, including spinal pain assessment, and relevant structural and functional components of IVD health and disease. This comprehensive tool box was designed for four commonly used preclinical large animal models (dog, pig, goat, and sheep) in order to facilitate standardization and applicability. Furthermore, it is intended to facilitate comparison across studies while discerning relevant differences between species within the context of outcomes with the goal to enhance veterinary clinical relevance as well. Current major challenges in pre‐clinical large animal models for IVD regeneration are highlighted and insights into future directions that may improve the understanding of the underlying pathologies are discussed. As such, the IVD research community can deepen its exploration of the molecular, cellular, structural, and biomechanical changes that occur with IVD degeneration and regeneration, paving the path for clinically relevant therapeutic strategies.


| INTRODUCTION
Chronic low back pain has been reported as the leading cause of years lost to disability for the past three decades. 1,2 Intervertebral disc (IVD) degeneration is recognized in at least 40% of cases of symptomatic back pain. 3,4 Much effort has been aimed toward the development of more effective diagnostic, preventative, and therapeutic strategies for IVD degeneration and preclinical studies involving large animal models are still considered a critical translational tool. Advantages of large animal models include the use of whole-organ/whole-body systems, relative vertebra-disc geometry, 5 biomechanics, 6,7 similar pathology, and clinically relevant outcome measures (reviewed by Thorpe et al 8 ).
Most large animal models require an experimentally induced insult or injury to the IVD in order to create and study IVD degeneration, with each method for induction having specific advantages and disadvantages. [9][10][11][12] It is beyond the scope of this article to provide a detailed review of large animal IVD models, their strengths and weaknesses, and the extent to which these models reproduce all the characteristics of pathological human IVDs. Instead, the reader is referred to previous reviews on this topic. [9][10][11][12] Furthermore, there are speciesspecific biological considerations, including overall size, age of skeletal maturation, spinal anatomy, and cellular composition in the nucleus pulposus (NP) ( Table 1), and IVD biomechanics, and from a clinical perspective, differences in pain-related behaviors. Within the particular species employed, confounding factors including sex (hormones), genetics, and susceptibility to developing spontaneous IVD conditions all need to be considered. Preclinical findings from experimental animal models can be supplemented with clinical studies involving clientowned companion animals, for example, canine patients. Dogs develop spontaneous degenerative IVD diseases and manifest with well-recognized clinical presentations. The diagnostics and treatment are closely comparable to those for humans (reviewed by Lee et al 13 ).
Although these studies are emerging, as yet their implementation is There was a strong consensus that consolidating these methodologies and their specific application in large animal studies of IVD degeneration and regeneration would represent an invaluable resource for the spine research community. Therefore, by consolidating knowledge from literature as well as expertise, and experiences from the IVD research community, a basic set of tools was compiled allowing for complementary functional and anatomical evaluations in large animal models. Here, a workflow is presented based on general considerations pertaining to study design and tissue processing of spinal specimens, and considerations and suggestions for three scales of read outs: macroscopic (gross morphology, imaging, and biomechanics), microscopic (histological, biochemical, and biomolecular analyses), and clinical (neurologic state, mobility, and pain).

| General considerations for study designs
The ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines 14,15 were developed to increase reproducibility and scientific rigor by providing a checklist of information to include when reporting animal studies. Many publishers require manuscripts to closely follow the reporting guidelines and thereby support the 3Rs principles (Replacement, Reduction, and Refinement). To further reduce the number of animals, IVD degeneration can be induced at multiple levels employing nonadjacent levels to limit interference among the IVDs studied. In such a multilevel approach, within each animal/lumbar spine, the following control conditions may be, at a minimum, included for proper comparisons: a nonoperated intact (nondegenerated) IVD, an experimentally degenerated, but untreated IVD, and a sham operated IVD (eg, surgical exposure and vehicle only). The inclusion of these controls may facilitate assessment of the intrinsic repair capacity of the large animal model being employed.
The multilevel approach may complicate understanding the causative relationship between the clinical scale read out parameters (such as painful symptoms) and IVD degeneration or the intervention being studied. In order to remove a confounding factor of having more than one degenerative IVD causing pain and disability, these research questions may require models that involve inducing IVD degeneration at a single level per animal in contrast to the abovementioned multilevel approach.
Longitudinal imaging follows IVDs from their native state to induced degeneration and subsequent experimental therapeutic interventions, thereby allowing determination of changes from baseline measurements. Although the IVDs in each region (lumbar thoracic and cervical) are of relatively similar size, the lumbosacral IVD is larger and subjected to different biomechanical conditions compared to the other lumbar IVDs. 16 Thus, since it is a level where clinical pathologies are commonly found and treated, in both humans and canines, in the authors' opinion the lumbosacral IVD is preferably separated from the others in the experimental design of large animal studies and not pooled or randomized with other levels.

| General considerations for processing spinal specimens
The standard directional anatomical nomenclature for quadrupeds is shown in Figure 2A. By following the process described in Figure 2B, the macro-and microscopic parameters described here can be applied to each investigated IVD. As such, gross morphological grading can be combined with histopathological schemes that evaluate cellular and matrix microstructure and molecular compositional information. For all three scales, previously published, wellvalidated schemes to evaluate these parameters can be applied and possible drawbacks and opportunities are discussed in the future perspectives section.
Immediately after euthanasia, imaging (eg, plain radiography, computer tomography (CT) and magnetic resonance imaging (MRI)) can be performed in situ, prior to recovering the vertebral column from the animal and extracting the functional spinal units. A functional spinal unit (FSU, also referred to as spinal motion segment [SMS]) comprises one IVD between the adjacent vertebral bodies, intact interbody ligaments, and facet joints, but with paraspinal muscle tissue removed.
The pedicles and posterior column elements are typically removed to isolate the ventral column unit (VCU, equivalent to anterior column unit (ACU) in humans; Figure 2B). Isolated VCUs can be further divided transversely (axially) or sagittally for macroscopic evaluation and further processing. Transversely cutting the IVD will result in a T A B L E 1 Overview of experimental domesticated large animal models used in IVD studies Social hierarchy exists. Aggression and fighting within groups may result in welfare concerns. It is especially recommended intact boars be separated.
Young males are often castrated by 1 mo of age. Calm and docile with relative ease of handling Goats exhibit distinct behaviors from sheep and they tend to be much more active, curious, and orally investigative. a The age at skeletal maturity may range depending on the specific breed being utilized. 193,194 Furthermore, body weight and size of animal can widely vary depending on breed and breeding practices; this is particularly evident in sheep that are bread either for meat or wool (eg, sheep bred for wool are often <50 kg body weight).
b NCD, non-chondrodystrophic laboratory dogs (eg, Mongrel Hounds); CD, chondrodystrophic laboratory dogs (eg, Beagles). In laboratory animal sciences, the terms "Mongrel dogs" or "laboratory hounds" have become interchangeable in a sense that they refer to non-Beagle (CD) laboratory dogs. They represent the NCD dog population. These animals are purpose-bred in USDA Class A dealer facilities which certify clear pedigree and specific pathogen-free. The benefits of using these animals include avoidance of unclear pedigree and health status which can often be seen in livestock obtained for biomedical research purposes from a sale barn or open herd.
cross-section consisting of the AF and NP. This cut is acceptable for studies that focus on basic analyses of AF and NP; however, evaluation of the cartilaginous endplate (CEP) is not possible. Sagittal sectioning, which preserves the anatomic orientation and context between components of the FSU, is suggested for macroscopic and microscopical evaluation and grading of AF, NP, and CEP as shown in Figure 2B. The workflow proposed by the authors in Figure 2B and the Supporting Information (Appendix S1) allows for the complete evaluation of gross morphological analysis in combination with imaging, histology, biochemical, and biomechanical analysis (outlined in more detail in the Supplementary flowchart, Appendix S1). Transverse sectioning prior to fixation is not recommended as without constraint of the IVD by the vertebral bodies, the NP tissue may swell, distorting tissue architecture and allowing leaching out of extracellular matrix (ECM) components during subsequent tissue processing. 17  A sample of recent peer reviewed manuscripts employing four common large animal models (ie, canine, caprine, ovine, porcine; n = 10 per species) to study IVD degeneration or therapeutic strategies in the past two decades. The number of six main outcome measures concomitantly used (macroscopic, histologic, radiologic, biochemical, biomechanical, pain), A, and the detailed use of each type of outcome, B, were registered and demonstrated the majority of the studies employed 3-4 outcome measures; with limited concomitant biomechanical analysis and absent pain assessment. C, When IVD degeneration was induced ("Deg. Induction"; 92% of all studies), healthy and degenerate controls ("Deg. Control" may be either induced or naturally occurring disc degeneration) were regularly, but not consistently, reported. Spontaneous degeneration was only reported for canine species. Adherence to the ARRIVE guidelines was mentioned in one study. D, Available scoring systems in histological and radiological outcomes ("evaluation" refers to yes/no evaluation and "scoring" specifies within those studies whether or not a scoring scheme was employed), including quantitative MR imaging were seldom employed. The studies included are provided in the Supporting Information, Appendix S1 F I G U R E 2 Legend on next page. liquid nitrogen and stored at À80 C. 18 For biomechanical testing, the use of fresh specimens is preferred, but if this is not feasible, specimens can be properly stored in controlled conditions until testing as described in the respective section. The most commonly adopted gross morphological grading scheme for IVDs was published by Thompson et al. 19 This scheme is based on a refinement of the comprehensive descriptions of human IVD pathology by Vernon-Roberts, 20  Macroscopic evaluation schemes are also inherently subjective, with grades potentially subject to ambiguity and high interobserver variability even when performed by experts.

| Implementation
The parameters described herein for gross morphological grading of large animal IVDs are drawn in large part from the original study of Thompson et al 19 (to which readers are encouraged to refer).
IVDs should be bisected using a single, smooth cut using a straightedged trimming knife or similar implement, in order to minimize cutting artifact that could confound assessments. Following the scheme of Thompson et al, 19 IVDs are assigned a grade from 1 to 5, where 1 is healthy and 5 is most degenerate. Grading criteria for each IVD substructure (NP, AF, CEP, and vertebral bodies) are shown in Table 2 (example images in Figure 3). Both left and right hemi-IVDs should be graded and averaged for each of the assessors.
F I G U R E 2 A, Comparative Anatomy of biped and quadruped spines. B, Flowchart of possible port-mortem procedures for evaluation of all read out parameters from each functional spinal unit (FSU). A, Anatomic analogy between humans and quadrupeds. B, Once the spinal column is extracted, the spinal cord can be removed en bloc (a). Transverse processes can be removed by trimming the lateral aspects of the spine (b). To isolate FSUs, cuts are made transversely through the mid-vertebrae (c). Dorsal aspects can be removed by cutting sagittally through the spinal canal (d) resulting in individual ventral column units (VCUs) (e). The isolated VCU can then be transversely transected into two identical parts (e). Thereafter, both parts can be used to take digital photographs for macroscopic evaluation and subsequently, one part may be fixed for histopathology (f). From the second part, the nucleus pulposus (NP) and annulus fibrosus (AF) tissue can be isolated from the cartilaginous end plates (CEPs) and vertebra with a surgical blade for biochemical analysis (g-i). Note that prior to the described post-mortem procedures (advanced) imaging and non-destructive biomechanical analysis can be conducted as described in this manuscript

| Imaging
Imaging is frequently employed for the evaluation of IVD degenera-  19 where Grade 1 corresponds to healthy IVDs and Grade 5 the most severely degenerated. Noteworthy characteristics specific to each image series include: the translucent, notochordal NP region present in healthy dog and pig IVDs, compared to the more cartilaginous NP in healthy sheep, goat and human IVDs; and the presence of growth plates in sexual and skeletal maturity goat and sheep and in the immature pigs. Gross morphological grading includes assessments of the major substructures of the IVD (NP, AF, and CEPs) in addition to the adjacent vertebral body margins and as such, the gross changes may differ depending on the method of disc degeneration and the therapeutic approach tested. Examples in this figure are from induced disc degeneration models via chondroitinase ABC (goat) and partial nuclectomy (ovine) and from naturally occurring disc degeneration (canine). The worst grade of the different substructures is used to define the final score. Animal IVDs are oriented with the ventral side facing down. Human IVDs are oriented with the anterior side facing to the right and are from naturally occurring disc degeneration. n.a.: not available; Relative size differences need to be considered between the different species; grading is independent of the relative IVD size. Goat, dog, and pig images were kindly provided by Prof. Dr. Theo Smit, Dr. Niklas Bergknut, Prof. Dr. Hans-Joachim Wilke respectively. Images of human IVDs were adapted from Wilke et al 198 and Galbusera et al 199 centered in the radiograph as parallax errors increase at the borders of the image. Furthermore, it is important to maintain the vertebral column parallel and as close as possible to the x-ray cassette for a better evaluation of IVD spaces. General aspects to further consider are tight collimation to enhance detail. Next to the most common lateral view, ancillary views such as ventrodorsal, flexed, and extended views can be acquired when evaluating cervical IVDs. 27 Disc height index (DHI) quantifies changes in IVD height ( Figure 4) and is considered a more accurate method than absolute IVD height measurement as it corrects for positioning and animals of differing sizes. The DHI is calculated by taking the average of twodimensional measurements obtained from the dorsal, middle, and ventral portions of the IVD and dividing those by the average of the adjacent vertebral body heights. 28  can also be detected by MRI.
In 2001, a grading system using MRI was developed by Pfirrmann et al, 35 for the semiquantitative assessment of the human lumbar IVD degeneration condition. T2-weighted-MRI sequences without fat saturation were used for this purpose as the signal loss of the IVD on these sequences correlates with progressive degenerative changes.
The Pfirrmann scale ranges from 1 to 5 and considers: the structure and signal intensity of the NP, the height of the IVD, as well as the distinction of the NP and AF (Table 3, Figure 5). This grading system is still the most accepted and commonly used MRI grading system for the evaluation of IVD degeneration. 36,37 The Pfirrmann grading system was proposed using 1 T MRI images almost 20 years ago, and the quality of images of high and low magnetic fields in MRI have improved since then. As such, an updated MRI disc degeneration grading system may be considered. 38 For this purpose, the present manuscript provides representative images for the four commonly used large animal species in the field ( Figure 5).
Various quantitative MR sequences have been shown to reliably indicate biochemical changes in the IVD, including hydration status, proteoglycan content, and IVD degeneration as reported for goats, 39 sheep, 40 pigs, 41 and dogs. 42 Relying on signal intensity alone (eg, from single T2-weighted images) for quantitative assessments is problematic, as signal intensity will vary based on position in the magnet. In animal species in comparison to those of humans is shown in Figure 6.
The most common spine sample type for biomechanical evaluation is the FSU, as it is the basic functional repeating unit of the spine; however, two or more coupled FSUs are also sometimes evaluated. can be further processed for complementary readouts is ideal ( Figure 2; Supporting Information, Appendix S1).   (Table 6) Grade 1: The healthy IVD shows a homogenous structure with a hyperintense signal intensity and normal IVD height. Grade 2: The structure of the IVD is no longer homogeneous and the signal is still intense. Horizontal gray bands may be present in the IVD that is related to a beginning unclear distinction between NP and AF. Grade 3: Signal intensity is intermediate and the height of the IVD is slightly but visibly decreased with unclear distinction between NP and AF. Grade 4: Signal intensity is hypointense and there is no longer a distinction between NP and AF, the IVD height is moderately decreased. Grade 5: Inhomogeneous structure of the IVD with a hypointense signal intensity and a collapsed IVD space. Note that in naturally occurring IVD disease at these stages spondylosis occurs eventually, potentially fusing the segment with progression (eg, dog, Grade 5). NP: nucleus pulposus, AF: annulus fibrosus. Human and sheep MRIs were kindly provided by Frank Niemeyer 169 and Marion Fussilier, 178 respectively storing at 4 C and sealing to prevent dehydration. 52 The loading history of the sample influences the reproducibility of the biomechanical evaluations, for example, due to variations in hydration. [53][54][55] The IVD height varies during day and night (diurnal changes) due to water flux caused by the changing average load magnitude. This affects, among other parameters, the intradiscal pressure. 56 There may be water content variations from storage conditions, which are ideally balanced out prior to testing. This can be addressed by applying axial compression load ("pre-load") or displacement, or allow passive hydration without load with the goal to equilibrate to a physiologically realistic situation. 57 During testing, the specimens are frequently kept in a temperature and humidity-controlled test environment that best mimics the in vivo situation. Accordingly, temperature is kept constant at 37 C, or alternatively room temperature at a relative humidity of ca. 100% at all times. Humidity can be controlled, for example, by spraying 0.15 M PBS, in an environmental chamber, wrapping in soaked gauze or by creating a reservoir. 48,58 Of note, swelling will occur in a saline bath without axial loading. 59 At 37 C, catabolic enzymes are active, and degeneration is enhanced. Therefore, specimens cannot be tested for a longer duration than several hours without the use of protease inhibitors and antifungal agents. Biomechanical experiments are frequently carried out at room temperature, however, the viscoelastic properties of the IVD and adjacent structures are temperature dependent. 60,61 Specimens are frequently "preconditioned," usually for at least three cycles to achieve a consistent response in creep, 62 stress relaxation, and in load-displacement curves. 47 Furthermore, the strain rate, preload, and follower load can be optimized in preliminary studies, as they are, for example, dependent on animal, IVD size and level, age, biomechanical test, or degeneration state.

| Testing of FSU or VCU specimens
For multiple degrees-of-freedom (see Supporting Information, Appendix S1) FSU-testing and analysis recommendations, we refer to the review by Wilke et al. 47 In addition, individual VCU specimens may be tested in unconfined compression 63 to measure axial, timedependent characteristics, for example, creep-recovery. 49

| Testing the AF, NP, and CEP
The AF lamellae experience high circumferential and longitudinal tensile, shear, and radial compression forces. The orientation during testing is relevant as the tissue is anisotropic. 66 The AF can be tested in uni-or biaxial tension, in shear, or under unconfined or confined compression (ie, with more controlled boundary conditions). For each test, single or multiple lamellae may be used. 48,67,68 Due to a high fixed charge density, the NP can imbibe large amounts of water. If the glycosaminoglycans (GAGs) responsible for this swelling and the water content are altered (because GAGs leach out or water is imbibed in unloaded conditions 69 Table 4). For most applica- Abundantly present (>90% of total cells) within the entire NP 0 Present in moderate amounts (50%-90% of total cells) 1 Present in low amounts (<50% of total cells) 2

NP Matrix staining (AB-PSR)-Figures 7 and 10
Blue proteoglycan ECM stain dominates 0 Reduction in blue proteoglycan ECM staining (ie, fading) 1 Reduction in blue proteoglycan ECM staining with presence of collagen staining (up to 50% of area) 2 Loss of blue proteoglycan ECM staining and/or dominance of collagen staining (>50% of area) 3 AF morphology (AB-PSR)- Figure 13 Well-organized, well-defined, uniform collagen lamellae form concentric half-ring arcs throughout entire AF 0 Mild disorganization/delamination of collagen fiber lamellae with some disruption or loss of concentric layers (<25%) 1 Moderately disorganization/delamination of collagen fiber lamellae with progressive disruption or loss of concentric layer (25-75%) 2 Complete disorganization/delamination/collapse of AF; almost all concentric collagen lamellae are severely disrupted or lost (>75%) 3 AF Cellular (H&E) and ECM metaplasia c (AB-PSR)/distinction between AF and NP- Figure 14 and Figure 15 Clear distinction between AF and NP tissue with intense blue proteoglycan ECM staining in NP: spindle-shaped fibrocytic nuclei populate AF with no or rare individual metaplastic cells resembling NPCs in the AF 0 Distinction less clear: loss of annular-nuclear demarcation. Proliferation of cells resembling NPC's is focally restricted (ie, perilesional) or mild proliferation with >75% lacunae containing single fibrocytic nuclei (ie, rare to few small NPC-like clusters) 1 Distinction less clear: loss of annular-nuclear demarcation. Regionally extensive perilesional proliferation of cells resembling NPC's or moderate proliferation involving 25%-50% AF lacunae that contain more than one cell nucleus (ie, frequent NPC-like clusters).

2
Poorly or no discernable annular-nuclear demarcation. Regionally extensive perilesional proliferation of cells resembling NPC's or moderate proliferation involving >50% AF lacunae that contain more than one cell nucleus (ie, frequent NPC-like clusters).  Larger and more numerous clefts or formation of transverse clefts that bridge an adjacent AF region (ie, inner-to-mid, or outer-to-mid) 2 Abundantly present, large clefts; formation of transverse clefts that bridge more than one layer of the AF; or presence of cellular infiltration/neovascularization
Due NP tissue's tendency to swell during the rehydration steps, which can cause wrinkles and reduce its adherence to the slide resulting in either poor quality sections (eg, folds) or even loss of sections, adherence to the slide can be enhanced by drying the sections overnight at 37 C prior to further processing, or precoating the slides with gelatin. 90 The standard histological staining for the general assessment of tissue structure and cell morphology is hematoxylin/eosin (H&E) that stains cell nuclei blue/purple (hematoxylin) and the collagenous ECM shades of pink (eosin) (Figures 7 and 8). Areas with high proteoglycan content (eg, NP) stain blue-gray. 91   Larger vertebral exostoses that protrude cranio-caudally and impinge on outer layers of the AF 2 Abundant new bone formation with partial to complete bridging spondylosis 3 Notes: All categories apply to all species reviewed herein; however, modifications of the scheme or analysis protocols (eg, additional histochemical or immunohistochemical stains) may be warranted, depending on the induction method employed to create a lesion (eg, surgical AF disruption vs chemicalinduced nucleotomy) or the research question (eg, for processes concerning angiogenesis, nerve ingrowth and inflammation, repair strategy employed). The reader is referred to Figures 7-18 Figure S2). Grade 1 (mild disc degeneration) IVDs show loss of basophilic/turquoise staining and reduced definition between the NP and AF that corresponds to reduced NP glycosaminoglycan (GAG) content and chondroid metaplasia of the AF, respectively. Inner to mid lamellae of the AF in this region (arrows) contain fine clefts spanned by proteoglycan-rich fibrillated collagen corresponding to microtears (black frame, see Figure S2). CEPs retain their discrete, uniform contour but the trabecular bone of flanking BEPs has compacted (ie, endplate sclerosis). Grade 2 (moderate disc degeneration) changes show almost complete loss of NP basophilia (H&E) with poor definition of NP-AF interface. Inner to mid lamellae of the AF in this region (arrows) show larger, more extensive clefts in the AF (black frames, see Figure S2) and progressive compaction of flanking trabecular bone. In the sheep, although proteoglycan staining of NP persists, there is loss of the inner to mid AF layers with NP protrusion into this region (arrows) that coincides to narrowing of the disc space and regional endplate thickening. Grade 3 (severe disc degeneration) changes include complete loss of NP basophilia (H&E) and more severely depleted NP GAG staining (AB-PSR) with loss of NP architecture and poor discernment of NP-AF interface. There is a collapse of IVD space and clefts within the distorted, degenerate NP (black frame, see Figure S2) and extrusion of degenerate chondroid IVD material beyond the CEPs that extends to the flanking cartilage growth plates of the vertebral bodies (arrows). The chondroid material (white arrow) stains dark blue on AB/PSR (see Figure S2). loss of definition. Moderate shows progressive NP and AF matrix degeneration with the production of small nodular exostoses (ie, syndesmophytes) at the dorsal margins of the AF (arrows); the ventral aspect of the H&E panel contains section artifact (arrowheads) and cannot be evaluated. Severe shows collapse of the IVD with partial dorsal and ventral extrusion of degenerate NP and AF matrix (arrowheads) and ventral bridging exostoses (arrows) compatible with intervertebral ankylosis (eg, self-fusion). End-stage IVDs show more complete extrusion of degenerate IVD matrix dorsally (arrowheads) and ventrally (asterisks) with complete collapse of IVD space and foci of bone-to-bone contact; ventrally, a large exostosis (arrows) surrounds the extruded IVD material (arrows). Reprinted with permission from Spine 152 and further modified to serve the needs of demonstrating naturally occurring IVD degeneration changes. Note that these are representative images from a naturally occurring disc degeneration model and not from a large animal model where disc degeneration is induced either chemically or surgically. In the canine species the growth plates close and are as such absent in these sections indicating that they are from skeletally mature dogs uniform scheme include, in addition to inflammation, neovascularization, and nerve invasion/fiber characterization. These parameters will need to be evaluated on a case-by-case basis and will depend on the research question posed.
Immunostaining Immunohistochemistry (IHC) and immunofluorescence are applied to identify specific cellular or extracellular proteins in the tissue. Immunostaining is particularly challenging for large animal studies, as most commercially available antibodies are developed to react with antigens from human and rodent proteins. Cross-reactivity with the particular species, including appropriate positive and negative controls, needs to be confirmed to ensure specific immunostaining. Furthermore, decalcification and antigen retrieval present additional challenges for obtaining high-quality and consistent immunostaining.
Within this context, immunostaining procedures need to be optimized from the first step to achieve satisfactory and reproducible results.
The spine researcher undertaking immunostaining is referred to detailed information provided in the (Supporting Information, Appendix S1) and to Binch et al reporting on these protocols for human samples. 103

| Biochemical, gene, and protein analysis
For this analysis, the NP and AF tissue (from 1 /4 IVD; Figure 2B)  process. [107][108][109] In the case of limited starting material, unbiased preamplification of the target sequences may provide an alternative. 18,110 Although the authors anticipate that these techniques may also be suitable for other species, age groups and IVD levels than those employed in the reported work, the RNA isolation method would need to be validated to make final conclusions. To normalize target gene expressions, it is suggested to use more than one stably expressed reference gene. Gene expression profiling is a tool that is often used to address the findings from a mechanistic perspective and correct normalization of these profiles is essential. To date, while sets of stable reference genes have been studied for some human, dog and goat IVD tissues, 18 Figure 12. Of note, cell clusters are typically considered a hallmark of degeneration but have also been related to cellular proliferation. As such, within the context of therapeutic strategies studied, cell clustering may reflect an attempt for regeneration rather than a degenerative reactive response 180 generally accepted methods or objective criteria for assessing pain in large animal models employed for IVD regeneration. However, a detailed neurologic examination by a veterinary expert can help localize spinal pain. 1 Due to the highly specific nature of research investigations, the optimal pain scale for a given experiment should ideally be as objective, standardized, and repeatable as is possible based on species, methods, and resources as determined a priori using generally accepted or validated methods.  . AB/PSR stained low magnification bright-field photomicrographs of progressively severe AF tears in sagittally sectioned goat IVDs from the same model featured in Figure 7. Outer AF of the ventral aspect of the disc is oriented to the left and inner AF/NP to the right. Grade 0 AF fibrous lamellae are intact, uniformly aligned and stained; dark blue linear streaks (arrow) are staining artifacts. Grade 1 AF tears form irregular clefts within the inner annulus in areas of mild matrix degeneration (arrows). Grade 2 AF tears show clefts extending from the outer NP through inner (black arrow) to mid (white arrow) annulus. Grade 3 AF tears comprise large irregularly divergent clefts (asterisk) within a large region of degenerate matrix that extends from inner to outer lamellae. One cleft extends along the interface with the outer endplate (EP) and is partially filled with proteoglycan-rich fibrillated matrix (white arrow). High (10Â) magnification inset of H&E stained AF from the framed area show remnants of AF lamellae (arrowheads) separated by a fibrous stroma containing many vascular profiles (arrows); scale bar = 100 μm  Table 6. Researchers are encouraged to discuss with veterinary specialists in identifying the most appropriate clinical entity to be employed in their translational endeavor once the intended approach has been proven to be safe and have sufficient efficacy.
In addition to the pain assessment methods, in clinical patients, gait/posture analysis can be performed to monitor changes due to IVD perturbations. Pain associated with cervical and thoracolumbar IVD disease (spontaneous or induced) can be assessed via neurologic grading of IVD disease. Table 7

| Perspective-future outlook
The current manuscript presents a comprehensive tool box for conducting in vivo large animal studies of IVD treatment with the focus on methodology and complementary outcome parameters (key points summarized in Table 8). Moving forward, key areas for refinement and expansion with specific focus on the comprehensive tool box are discussed below to facilitate comprehensive analysis and a better understanding of the underlying pathology and model-specific differences.
The most significant species-specific confounders affecting outcomes and their utilization include sex and breed-dependent genetic background. IVD (patho)physiology is potentially influenced by sex hormone differences and as well as related to the species-dependent reproduction cycle, particularly considering the well described effects of sex on bone (patho)physiology within the field of osteoporosis. 134 As such, where possible involvement of both sexes in studies will allow discernment of this effect. In small animal models, strains of different genetic backgrounds are beginning to be employed for a better understanding of IVD aging and pathology 135,136 and genetic background has been shown to play a role in rodent models 137 ; however, these remain relatively unexplored in large animal models. There is the well described role of the FGFR4 retrogene in the canine species, 138 and livestock animal models typically involve genetic polymorphisms or mutations affecting skeletal developments. The IVD may potentially show similar genetic traits, but only their effects on bone phenotype have been described so far. An example is the quantitative trait locus mapping to the growth differentiation factor-8 (GDF8 or myostatin) gene reported in Texel sheep 139,140 and in crossbreeds like the Swifters, with GDF8 shown to inhibit chondrogenesis by suppressing SOX9 and collagen type II expression. 141 Within this context, there are reports of spontaneous NP 18,142-146 and AF 147,148 repair in large animal models. While those may partly relate to IVD size and as such to nutritional status of the IVD, 149 to date there is limited understanding of the differences in intrinsic repair capacity F I G U R E 1 6 Sheep-cartilage end-plate (CEP) morphology (AB-PSR). AB-PSR stained brightfield photomicrographs of progressive CEP disruption in the goat IVD. Grade 0 shows intact CEP of uniform contour and thickness. Grade 1 shows regional thinning or the CEP (arrow). Grade 2 shows multifocal disruption of the CEP <10% of the total area (arrows) with limited extrusion of IVD matrix into the bony endplate. Grade 3 shows a focal small disruption (arrowhead) adjacent to a large regional disruption of the CEP involving >30% of total area with extrusion of IVD matrix deep within the bony endplate (arrows). Note the Grade 3 image was obtained at half magnification of Grades 0-2 in order to demonstrate the extent of the endplate disruption across large animal species and how these translate into changes in the outcomes of the three scales, that is, macroscopic, microscopic, and clinical, displayed in this manuscript.
With the exception of dogs, IVD geometries have been reported for goats, pigs and sheep [5][6][7]150 and summarized by Fusellier et al. 151 Furthermore, varying measurement methodologies (eg, application of axial load) makes comparison between the available studies difficult.
Although basic geometries have been documented, anatomical details are still insufficient. For example, CEP-thickness for dog, 152 pig, 153 and human 154 were reported, but not for goat and sheep, and these may be important when selecting models for investigation of IVD nutrient transport. Furthermore, potential damage to the CEP (eg, due to freezing or enzymes 155,156 ) and its role in degeneration needs to be further investigated. For future studies, one might also consider freezing specimens comparable to cryopreservation for total IVD transplantation. 51 Macroscopic grading based on gross morphology and MRI imag-   Thoracolumbar-Acute herniation-T11-L3 Acute pain that may be accompanied by motor deficits in the hindquarters depending on the level of spinal cord trauma. Damage to the upper motor neuron causes the urinary bladder to be full and tense which leads to difficulty with expression of the urinary bladder eventually leading to urinary incontinence.
Cauda Equina Syndrome-L6-L7-S1, chronic protrusion 197 Chronic pain with acute episodes, progressive hind limb weakness and muscle wasting. May lead to progressive sensory and motor deficits in the hind limbs. Nerve root signature (pain apparent on palpation or traction of the limb). Damage to the lower motor neurons will influence bladder function that will cause urinary bladder to be flaccid, eventually leading to urinary incontinence and fecal incontinence.
Notes: Dog breeds can be categorized into chondrodystrophic (CD) or non-chondrodystrophic (NCD) breeds. CD breeds have distinctive conformational differences that are characterized by short limbs and long torso length compared to limb strengths. The clinical signs relate to the fact that the diseased IVD is affecting the peripheral nervous system (PNS) or central nervous system (CNS). Localization is done based on thorough clinical and neurological examination complemented by imaging modalities.
T A B L E 8 Highlights of the comprehensive tool box for large animal studies in IVD degeneration and regeneration

Macroscopic scale
• Gross morphological grading is relatively low cost, easy to implement, and provides overall information on IVD macrostructure.
• Imaging via radiography is a rapid and inexpensive modality • Radiography can be complemented by MRI to obtain the Disc Height Index and detailed information on degenerative changes in vivo.
• T2-weighed images are species-dependent as notochordal cell-rich IVDs have a higher signal intensity.
• Non-destructive biomechanical tests allow functional evaluation of the IVD.

Microscopic scale
• A new four-point histological grading scheme with consensus for all large animal IVD models has been compiled, based on validated schemes for individual species. • On mid-sagittal sections all relevant IVD components: the nucleus pulposus, the annulus fibrosus, and the cartilage endplates can be evaluated • H&E for cell morphology and extracellular matrix strains enable microscopic IVD grading.

Clinical scale
• Following the ARRIVE guidelines and consultancy with veterinary and animal welfare experts helps determine • pain-related outcome measures, analgesia rescue protocols, and humane endpoints.
• Client-owned dog patients represent different clinical entities that could contribute to translational research and lead to the development of therapeutic strategies. • Post-operative evaluations for neurologic state, pain, and other clinical parameters may be species-specific. • It is beneficial to have a standard assessment scheme that will decrease inter-evaluator variability.
F I G U R E 1 9 Bone modeling at external annulus fibrosus (AF)-bone interface. Low magnification bright-field photomicrographs of AB-PSR stained hemisections of intervertebral IVDs from goat (induced IVD degeneration) and dog specimens (naturally occurring IVD degeneration). Grade 0 shows a smooth bony contour (white arrows) between the AF-bone interface without peripheral nodular exostoses. Grade 1 shows a focal nodular exostosis (black arrows) at the periphery of the AFbone interface. Grade 2 shows regionally extensive nodular exostosis that protrudes from the bony endplate into the AF (white arrows). Grade 3 shows a large nodular exostosis surrounding the extruded IVD. Similar histopathological changes have also been extensively described in the annular lesion ovine model of IVD degeneration, 204 including bony remodeling of the annular rim 205 and are also depicted in Figure 7, right panel differences in clinical presentation and lesion progression between CD and NCD dogs. Further exploration of the correlation between ECM composition, IVD biomechanics, and lesion morphology between these two spontaneous dog models could help elucidate the biomechanical effects of IVD lesions and ECM composition in humans.
Changes at the macroscopic level have been reported to be more prevalent in patients suffering from back pain 189 but they are also present in asymptomatic patients. 190 As such, in the authors opinion, study designs addressing only the macroscopic and microscopic parameters of IVD degeneration/regeneration do not necessarily determine the clinical relevance of the therapeutic strategy being studied. Herein, behavioral and/or pain outcome measures are reliable indicators of how effective candidate therapeutic interventions may be. There are better tools becoming available with the advancement of technology such as motion analysis and telemetry, which will permit access to highly detailed and quantifiable data without interfering with the wellbeing of the laboratory animals. Furthermore, collaboration with veterinary experts to establish clinical parameters relevant to the spinal pathology being studied, for example, acute thoracolumbar IVD herniation in the CD dog or progressive lumbosacral IVD protrusion in the large breed working dog, that are relevant to the individual studies is valuable.
In conclusion, this comprehensive tool box for large animal IVD models comprised of macroscopic, microscopic and clinical outcome parameters can assist in generating a comprehensive data set to address the full spectrum of changes associated with IVD degeneration and regeneration in studies with large animal models. Within this context, there are opportunities for further refinement and expansion, including the implementation of the emerging innovative techniques we have touched on throughout this article. By employing such an algorithm, metadata analysis will be less complicated, even across species, and provide insight into model-dependent differences and how they could translate to understanding human pathology, improving translatability and clinical relevance which will ultimately serve patients suffering from pain and disability due to IVD degeneration.

ACKNOWLEDGMENTS
Research and salary support for NNL is provided through NIH T32 grant