The human IVD is comprised of three main components: the inner nucleus pulposus (NP), the surrounding annulus fibrosus (AF), and the superior and inferior cartilage endplates (CEP) (21). The healthy NP is highly gelatinous and comprised of proteoglycans, namely aggrecan [i.e., glycosaminoglycan (GAG)], and collagen II which provides load distribution and compressive force absorption (22, 23). In contrast, the surrounding AF region is comprised of an aligned matrix of collagen I fibers and functions to contain the NP and anchor the disc to adjacent vertebrae to resist torsional and tensional loads (24, 25). While under characterized, the CEP plays a critical role in the diffusion of nutrients to the largely avascular IVD from the vertebral bodies as well as the removal of metabolic waste (26). However, with advancing degeneration, the IVD demonstrates decreases in cellularity and proteoglycan synthesis along with increased catabolism, inflammation, immune cell infiltration, changes in IVD structure, altered mechanics, and neurovascular invasion, all of which are critical factors to assess in IVD degeneration models (27–30).

In comparison to the human IVD, mice, despite their small stature, have many commonalities with humans as summarized in Figure 1. On the cellular level, humans and mice IVDs both exhibit decreased cellularity with increased apoptosis, senescence, and immune infiltration during aging or induced IVD degeneration (13, 31). At a molecular level, nutrient deficiency, and reduced extracellular matrix (ECM) production have been found in mouse models of IVD degeneration and degenerated human IVDs. Immune infiltration has been identified in painful degenerate human and mouse IVDs as evidenced by the presence and recruitment of mast cells and macrophages (32–37). Structurally, human and mouse IVDs are avascular and aneural in their healthy state with similar components (NP, AF, CEP) and ECM tissue structure. Interestingly, rodent lumbar IVDs are most geometrically analogous to humans compared to other animals based on percent deviation of normalized disc height, anterior-posterior width, and NP area (38). In IVD degeneration, changes at the structure/function level in both humans and mice include decreased disc height/volume, neurovascular invasion, limited nutrition, reduced hydration, AF disorganization, and a more fibrous NP (38–42). In terms of pain and pain-like behaviors, humans and mice can experience impaired gait, decreased activity time, and reduced range of motion due to IVD degeneration along with mechanical and thermal hypersensitivity and changes in neuronal plasticity (43–47).

In contrast, drawbacks of using mice as a translational model of LBP are mainly due to their small size, low forces and reduced diffusion distances across their IVD (0.35 ± 0.09 mm²), quadrupedal nature with altered mechanical forces, higher cellularity, and retention of notochordal cells throughout adulthood (13, 38). Ideally, research involving therapeutic studies should include quantification of pain, but the measurement of pain differs from human patients to mice as animals cannot communicate pain levels and are not affected by other risk factors of LBP such as lifestyle and career occupation (48). Instead, behavioral assessments for “pain-like” behaviors are used and inferred. Nonetheless, mouse models allow for pre-clinical validation of disease pathogenesis and evaluation of potential regenerative therapies and therefore remain critical in vivo research models. The existing mouse models can be categorized into aging and genetic, mechanically induced and puncture models.

Age-related changes representative of degeneration are not as severe in mice, as they retain their notochordal cell population throughout adulthood, compared to animals without notochordal cell retention such as primates and chondrodystrophic dogs (49). While less severe, mice do develop age related changes at later stages of life. Specifically, wild-type C57BL/6J mice at 2 years old (old age) demonstrate features of IVD degeneration such as reduced disc height, bulging, and neurovascular invasion in both caudal and lumbar IVDs along with upregulation of inflammatory cytokines tumor necrosis factor-alpha (TNFα) and interleukin-1 beta (IL-1β) compared to younger mice (50). Alvarez et al. also showed degeneration at 2 years in wild-type mice associated with downregulated Forkhead box O (FOXO) expression (51). Taken together, aging in wild-type mice may mimic IVD degeneration that occurs with age in humans and has high clinical relevance. However, laboratory mice typically have a lifespan of 2 years, potentially limiting the utility of using aged mice for studying regenerative therapies as they may not survive the length of treatment.

As such, genetically modified mice have been used in place of wild-type mice for “accelerated aging” or for studying specific mechanisms involved in the pathogenesis of IVD degeneration. The bile duct ligation (BDL) strain of mice, a result of an autosomal mutation in the ky gene, develop kyphoscoliosis with structural changes in the vertebrae and cervical IVDs coupled with degeneration and herniation (52). Growth differentiation factor 5 (GDF5) deficient mice also demonstrate signs of IVD degeneration with changes in collagen and proteoglycan content (53). In addition, secreted protein acidic and rich in cysteine-null (SPARC-null) transgenic mice demonstrate age-related changes in IVD composition and structure such as decreased disc height index (DHI) and GAG with increasing severity. SPARC-null mice also exhibit similar human pain-like behaviors such as decreased tolerance to axial stretch, thermal sensitivity, and impaired locomotion (54). More mouse models that stimulate an “age-related” degenerative phenotype include Excision Repair 1 (ERCC1)-mutant mice that lack important DNA repair mechanisms, suggesting aberrant DNA repair may contribute to IVD degeneration (55). Most recently, SM/J mice have been shown to exhibit spontaneous IVD degeneration due to poor cartilage healing (56). As degradation of the ECM plays a key role in the pathogenesis of IVD degeneration, many ECM genetic knockout models have also been assessed (i.e., collagen II, collagen IX, aggrecan, and biglycan) and as expected demonstrate significant ECM loss/disorganization with impaired disc and vertebrae development (57–62). A review by Jin et al. highlights additional genetic knockouts with degenerative IVD phenotypes (12). While specific mutations have been identified in some populations and global deletions remain rare, genetic models demonstrate accelerated disease/aging which is of clinical relevance and value for animal studies given how long degeneration can take to progress in humans (63–65).

Age-related models are advantageous as they share many similarities with disease pathogenesis that occurs in degenerate human IVDs. However, it is important to note that IVD degeneration does not happen solely due to age, but results from multiple factors such as genetics, environment, and lifestyle; furthermore, IVD degeneration can also occur in the younger population (63, 65, 66). Limitations of using aged mice as a model for regenerative therapies include the long-term care of aging mice and the short lifespan of the mice. Genetic mouse models are excellent tools allowing the investigation of specific genes/proteins/pathways contributing to IVD degeneration. Limitations of these models may include the multifactorial disease process that occurs in humans which may not be gene-specific, however, these models still allow for screening of novel regenerative therapies for IVD degeneration.

Mechanical loading is an important regulator of IVD development and homeostasis and altered biomechanics plays a significant role in IVD degeneration in humans as evidenced by the increased risk of IVD degeneration in manual laborers and changes due to physical exercise (9). As such, a wide array of mechanically induced models have been developed that alter or modify the mechanical environment experienced by mice IVDs to induce degeneration. Instability models such as surgical resection of posterior elements (i.e., facet joints and spinous processes), tail bending, static/dynamic compression, and axial loading models have demonstrated that altering the mechanical loading environment applied to the IVD can lead to progressive disc degeneration characterized by cell death, decreased disc height, decreased collagen II and aggrecan expression, and disorganization of AF structure (67–74). Other models which alter spinal loading via induced bipedal behaviors exhibit accelerated degenerative characteristics due to abnormal mechanical stress compared to quadrupedal mice (75, 76). A recent bipedal mice model was established by placing mice in water to promote standing as, mice are aqua-phobic, and have demonstrated decreased DHI with increasing degeneration grade (76). Reduced spinal loading due to microgravity, investigated through space flight, has also been shown to alter the viscoelastic behaviors of caudal mouse IVDs upon return to earth (77). A model of whole-body vibration has also induced structural changes in mice IVD indicated by structural NP and AF deficits (78). Collectively, these models are capable of inducing degeneration by manipulating the external loads applied to the IVD and these models demonstrate high clinical relevance given the significant changes and enhanced stresses and loads experienced by humans with degeneration and aging.

Another common method to induce IVD degeneration is via needle puncture, which induces a direct injury to the IVD and is thought to depressurize the NP, thereby changing the normal internal mechanisms of load support (79, 80). Needle puncture models have shown similar hallmarks of IVD degeneration as in humans including decreased DHI, reduced disc hydration and cellularity, matrix disorganization, and reduced torsional and compressive stiffness (79, 81–86). The severity of degeneration can be influenced by multiple factors such as needle gauge, surgical procedure, depth of puncture, and region of injury (i.e., caudal vs. lumbar) (40). There has been considerable work in multiple animal models to determine how the size of the puncture relative to IVD height influences degeneration (79, 87–90). Across models, there appears to be a threshold in which a smaller injury ratio (i.e., needle diameter to disc height ratio) induces less severe degeneration and above which severe degeneration is induced, however, the exact threshold varies based on the species and region of the puncture (87, 90). In mice, an injury ratio of 90% (26G needle) significantly altered IVD compressive and torsional mechanics while a 65% injury ratio (29G needle) had no significant differences (79). It is important to note that puncture models are useful due to the acute or rapid degeneration of the punctured discs. However, the puncture itself is generally not intentionally induced in humans in the present day.

Mechanical and puncture induced models are advantageous as they do not require significantly aged mice or genetic modification to induce accelerated degeneration. However, it is worth noting that the severity of degeneration relies heavily on the degree of mechanical loading or injury. To date, puncture models have primarily been induced in caudal level IVDs due to surgical accessibility. However, lumbar level puncture models may be more representative of LBP and give the capability to assess pain-like behaviors. In addition, injury induced degeneration often results in post-injury inflammatory processes which may be mechanistically different from spontaneous IVD degeneration and more representative of acute trauma (91). Thus, the type and degree of degeneration that the therapeutic strategy aims to target are critical for determining the relevant animal model utilized.

Other models exist such as dietary-induced diabetic or Tobacco smoke inhalation models which demonstrate increased disc degeneration with increases in cell senescence and reduced proteoglycan synthesis compared to control mice not exposed to a high-fat diet or tobacco products (92, 93). Multiple studies have also shown that diabetes and obesity are significant risk factors for IVD degeneration which may serve as models if studying diabetic/obesity-related IVD degeneration (94–98).

To focus the scope of this comparative review on translational mice models of LBP for studying regenerative therapies, articles were collected via PubMed and Scopus with multiple combinations of the search terms: “Intervertebral Disc,” “Low Back Pain,” “Mice,” “Mouse,” “Model,” “Discogenic back pain” and “degeneration.” Articles included in the review met the following criteria: characterization of a mouse model, direct/indirect induction of IVD degeneration (lumbar or caudal), and potential for use as an IVD degeneration model for regenerative medicine (i.e., no genetic deficits that greatly alter other tissues/organs). Exclusion criteria included: studies with a focus on specific pathologies/mechanisms or spinal diseases beyond IVD degeneration (i.e., Scoliosis, kyphosis, arthritis) or LBP unrelated to IVD degeneration (i.e. direct neuronal trauma without IVD degeneration). Respective articles assessing Molecular and Cellular changes (Table 1), Tissue & Structure/Function (Table 2), and Pain-like behavior (Table 3) were compared for their induction methods (aging, injury type, degeneration level (i.e., caudal vs lumbar), length of study, model demographics (age, sex, strain), along with study results. The following sections will review different assessments to identify key parameters for translation to the human condition.

Most mouse models that included molecular and cellular assessments described in this section have based their assessment of disc degeneration on changes in cellularity and ECM markers, with few papers examining changes in phenotypic and inflammatory markers (Table 1). This may be due to the limited size of mice IVDs and availability of tissue for multiple assessments as previously mentioned. Multiple strains of mice have been used including wild-type, CD-1, SPARC-null, SM/J, and Swiss Webster mice (55, 56, 69, 73, 76, 78, 84, 86, 99); interestingly only one study compared both male and female mice in this section (56). Parameters have been evaluated at different time points ranging from pre-op (if mechanically/puncture induced), 0 days (immediately after puncture) and up to 2 years. These cell/molecular level comparisons can be categorized into cell morphology & cellularity, cell-specific Extracellular Matrix Genes, and phenotypic & inflammatory markers. Despite immune infiltration as a contributor to LBP, only expression of inflammatory cytokines was assessed in the following studies.

Molecular and cellular changes that characterize IVD degeneration in humans include decreased cellularity (with increased apoptosis and senescence), immune infiltration, loss of proteoglycan/GAG & collagen 2, and increased fibrosis/collagen 1 in the NP (109). Many of the mouse models of IVD degeneration reviewed here demonstrated similar molecular and cellular changes as the human. However, one major difference was inconsistent changes in collagen 1 across different studies. In humans, large vacuolated notochordal cells predominate before 10 years of age and transition into chondrocyte-like cells during adolescence. In degeneration, the NP decreases in cellularity with more degenerate cells (110). This characteristic change in cell phenotype and number was observed in puncture models along with the SM/J model, providing clinical relevance to the human condition on the cellular level for using these models to study IVD degeneration and regenerative therapies. Healthy NP phenotypic markers identified in the human IVD include GLUT-1, Aggrecan/collagen 2 ratios>20, sonic hedgehog, Brachyury, KRT18/19, CA12, and CD24 (111). Markers such as GLUT-1, aggrecan, KRT-19 were also decreased in mice with IVD degeneration. Unlike the NP, the AF and CEP are under characterized and some studies have suggested COL1A1, Elastin (ELN), Mohawk (MKX), and Scleraxis (SCX) as potential markers of the healthy AF (112–115). In addition, there is an increase in catabolic and pro-inflammatory markers (i.e., TNFα, IL-1β, IL-6, MMPs, ADAMTs) in degenerate human discs compared to healthy IVDs (116, 117). Increased expression of TNFα, MMP13, and IL-1β in mice IVDs was consistent with the human condition while many mouse models found no differences ADAMTs genes expression. Importantly, infiltration of immune cells, notably macrophages and mast cells, have been found in degenerate human IVDs and mouse IVDs post-injury, warranting the need to assess immune cell infiltration in mice models to account for immune-related systemic effects (32–34, 36, 37).

Given the molecular & cellular parameters highlighted here, gene expression seemed to vary between studies and was often inconsistent when comparing relevant human markers while histological and IHC assessments for protein level assessments were more consistent when compared with the human condition. Thus, in terms of molecular markers, TNFα and IL-1β are critical inflammatory markers while MMP13 may be a good catabolic marker for assessment of degeneration in mice IVDs as they are most relevant to the human condition. However, due to size limitations, the temporal expression on the gene level, and difficulty isolating high-quality RNA from single-level IVDs, transcriptome level changes may be more challenging to corroborate between mouse models. Histological stains such as H&E, Safranin O, are good for the assessment of cellularity while TUNEL assay can give information related to apoptosis consistent with human IVD degeneration.

In mouse models of IVD degeneration (Table 2), most studies assessed changes in disc structure & degeneration grade via histology, and DHI via radiographs or microCT. Interestingly, despite being critical parameters of disc structure/function, few studies assessed mechanics or disc hydration. Multiple genetic strains of mice were used, including the wild-type, CD-1, SPARC-null, SM/J, ERCC1-XPF, and Swiss Webster mice with only 2 studies comparing both male and female mice. Parameters were assessed from pre-op to 2 years of age and included imaging parameters (X-ray, microCT, MRI) for DHI and hydration, histology for disc structure and grade, neurovascular invasion, and mechanics.

On the structural level, human IVD degeneration is characterized by dehydration of the NP due to proteoglycan loss, AF disorganization resulting in a loss of AF-NP boundaries, decreased DHI, and sclerosis of the CEP. In the case of herniated IVDs, the IVD contains AF fissures leading to NP protrusion (109, 117, 132, 133). These changes in the NP and AF were all observed in the mice models discussed here however degenerative changes in the CEP region were mostly unquantified (Table 2). Additionally, this loss of IVD architecture in degeneration creates a permissive environment for vasculature and nerve ingrowth into the usually avascular and aneural healthy IVD which may be a source of pain in DBP (134–136). This correlates with the increased nerve fiber ingrowth in degenerative mice IVD as seen in the assessed studies. In the clinic, structural changes in the IVDs of living patients are often detected via MRI and CT or x-ray which correlates with the use of MRI and microCT in mouse models to detect structural changes in the IVD, spinal cord, and adjacent tissues. Furthermore, human IVDs can also be assessed macroscopically ex vivo using the Thompson grading scheme (137). However, this grading scheme is limited clinically as it is typically performed on cadaveric tissue. Instead, the Pfirrmann grading scale can be used to grade human IVDs imaged via T2-weighted MRI assessing DHI, hydration, and NP/AF compartmentalization (125). The present mouse studies primarily utilized histological assessments which may be attributed to the accessibility of histological techniques compared to imaging modalities such as MRI. It is also important to note that degenerate IVDs imaged as black discs on MRI do not always correlate with LBP; patients can have several degenerate IVDs and experience little or no pain (95, 138). This highlights a need to also quantify measures of pain as changes on the structural level alone cannot always be used as a proxy for diagnosis or treatment of LBP.

These structural changes that develop during degeneration induce corresponding changes in IVD mechanical behaviors & properties. Mechanical properties of tissues can vary substantially based on the size and dimensions of the tested material; similar to how a thick piece of rope is stiffer than a thin piece of rope of the same material. This concept also applies to the IVD as the physical dimensions and associated axial and torsional properties vary between species due to IVD size (38, 139, 140). To account for these geometric differences the properties can be normalized to the geometry of the specimen, similar to material properties in material science, and thereby allow comparison of properties independent of geometric biases. For example, if the mechanical behaviors of IVDs of various species and anatomical locations (i.e., lumbar and caudal) are normalized to their respective geometries, the differences between species and region is reduced and in some cases are insignificant which allows direct comparison between mice and human IVDs (140, 141). For example, torsional stiffness (K) and torque range (TR) are often used to characterize the rotational behaviors of IVDs. Comparing the raw measurements there is a difference of ~4 orders of magnitude between human (K_Human = 3.18 ± 0.89 Nm/degree) and mouse (K_Mouse = 1.1^*10⁻⁴ ± 1.83^*10⁻⁵ Nm/degree) IVDs. However, after normalization to the disc height and polar moment of inertia (i.e., the shape's resistance to torsional deformation) there are no significant differences between the two species (139). Studies have further compared the normalized axial and viscoelastic creep properties across species and demonstrated that normalized material properties are largely conserved across species (139, 140, 142). This suggests that when comparing normalized mechanical parameters, mouse models are reasonable models of the human condition. In addition to comparing normalized mechanical parameters extracted directly from mechanical testing, other studies have compared how analytically derived parameters determined from fitting mechanical data with theoretical models under similar loading conditions compare between regions and with degeneration. For example, the response of human and mouse IVDs to creep loading has been fitted with a fluid-transport model, and similar decreases in the parameter corresponding to the strain dependence of swelling pressure decreases in both mouse models and degenerated human IVDs (77, 99, 143).

For the structural and functional parameters assessed, histological staining was consistently used across most mouse models with limited differences between models and correlated well with changes observed in human IVD degeneration. Thus, histological assessment of reduced proteoglycan in the NP, reduced NP/AF demarcation, and disorganization of the AF matrix are all critical factors to assess when confirming degeneration in mouse models and the regenerative potential of therapies. DMMB assessments are consistent between all studies not only within these models but across the IVD field. Collectively, the combination of these observations in addition to standardized histopathological scoring systems will improve the comparability across studies and their translation to the human condition (129, 130). T2-weighted MRI to assess Pfirrmann grade, IVD hydration, and microCT for DHI is also strongly recommended as these assessments are used clinically in human patients and parameters are consistent across mouse models. Notability, assessment of mechanical stiffness is important given the consistent increase in compressive stiffness observed in diseased mice and human IVDs, yet the expertise and access to equipment to do these tests may warrant collaborations across disciplines.

Pain can be categorized broadly into neuropathic or nociceptive pain and LBP likely has neuropathic and nociceptive contributors. Neuropathic pain results from direct injury to neuronal tissue. In the case of LBP, the pain can arise from IVD herniation which causes pressure on the nerve root and innervating DRGs, resulting in inflammation, radiculopathy, and damage to the nervous system directly. Nociceptive pain results from injury to non-neuronal tissues such as from the muscles surrounding the IVD or the IVD joint itself, where neoinnervation can occur and peripheral nociceptive neurons are stimulated/excited and transmit signals to the central nervous system (146, 147). Although not widely understood, an emerging third type of pain known as “nociplastic pain” may also play a role in LBP with a distinct mechanism unlike neuropathic or nociceptive pain (148). LBP can also be classified by the duration of pain (acute or chronic). Acute LBP occurs due to tissue trauma with patient recovery within a month, is typically self-limiting, and becomes subacute in the 1–3 month range (149). Chronic LBP lasts more than 12 weeks and patients who present with acute pain may develop chronic LBP over time (150, 151). Acute pain is often protective and alerts individuals to potentially damaging environmental stimuli, thereby aiding in the healing process. Meanwhile, chronic pain serves no protective role and can be debilitating (152). LBP can also be classified into spontaneous or movement evoked discomfort with localization to the lower back and spine, or, radiating pain which also affects the legs due to injury or inflammation of the nerve root (153, 154).

Anatomically, the IVD is innervated bilaterally by neurons with cell bodies residing in the DRGs. DRGs are heterogeneous (including proprioceptors, nociceptors, Schwann cells, fibroblasts and satellite glial cells) and are responsible for transmitting signals from the periphery to the central nervous system via projections into the dorsal horn (155, 156). Functional changes and sensitization of the sensory neurons due to IVD degeneration can lead to neuropathic or nociceptive pain via multiple mechanisms (147). During IVD degeneration, nerve endings expand from the outer AF to the inner AF and NP regions of the disc due to decreases in chondroitin-sulfated proteoglycans and elevated levels of neurotrophic factors [i.e., neurotrophins such as NGF and brain-derived neurotrophic factor (BDNF)] and this can lead to nerve sensitization (29, 136, 147, 157–160). In addition, the degenerate IVD is largely innervated by small nociceptors that express voltage-gated sodium channels (VGSCs) or transient receptor potential cation channel subfamily V member 1 (TRPV1) that regulate neuronal activity (161). Persistent inflammation in the IVD can sensitize these neurons and induce changes, causing altered action potential duration, hyper-excitability, lowered thresholds to stimuli, and enhanced pain as observed in rodent models (50, 161–163). The activation of Protease-activated receptor 2 (PAR2) on DRG sensory neurons can regulate acute and chronic pain by activating the extracellular signal-regulated protein kinase (ERK 1/2) signaling pathway (162, 164, 165). However, the exact mechanisms driving LBP remain unclear with a limited number of mice models investigating changes at the DRG level.

In vitro pain assessments in this section are related to neuronal function/activity of innervating DRGs and spinal cord taken from mice models ex vivo while neurovascular invasion (nerve ingrowth and neo-angiogenesis) into IVD tissue were addressed in the previous section. Ex vivo characterization of isolated neurons has been performed to determine the role of DRGs in discogenic neuropathic/nociceptive pain. Electrophysiology and IHC have provided insight into the function and expression patterns of the ion channels/receptors and have highlighted their role in the pain signaling pathway in rat models but few ion/channels have been assessed in mice (161, 166). In addition, calcium imaging is a standard method to quantify changes in ion channel activity in the neurons (167). At the gene level, transcriptome analysis of the DRG neurons has been used to identify key genes associated with pain perception (168). In addition, IHC on the DRGs from a lumbar IVD puncture model demonstrated increased key pain markers, CGRP, Tropomyosin receptor kinase A (TrKA), and NGF from 2 to 12 weeks post-op and this is further supported by increased gene expression of CGRP, substance P, TrKA, BDNF, TRPV1, Neuropeptide Y (NPY), VGSCs Nav1.7/1.8 (84). In SPARC-null mice, CGRP reactivity was also elevated in addition to NPY in DRG neurons (103). IHC on the spinal cord showed increased expression of astrocyte marker, Glial fibrillary acidic protein (GFAP), and also microglia expression at 2 weeks in the puncture model, potentially due to inflammation from injury, while SPARC-null mice found increasing numbers of astrocyte and microglial cells with age (84, 169). Thus, the DRGs are key determinants in the induction and maintenance of both neuropathic and nociceptive pain, making their assessment in LBP models critical.

Pain-like behavioral assessments can be classified into evoked or spontaneous pain as illustrated in Figure 2. Spontaneous pain occurs in the absence of specific stimuli and is more indicative of clinical chronic pain conditions (170). Examples of spontaneous pain in mice may include audible vocalization, avoidance, and self-mutilation (171). Meanwhile, in evoked behavioral assessments, mice are presented with a stimulus representing sensory modalities that allow for the measurement of pain thresholds (172). Signs of evoked pain may include motor reflexes such as limb withdrawal from the stimulus, reduced locomotion, or agitation. Evoked pain can further be categorized into hyperalgesia and allodynia. According to the International Association for the Study of Pain (IASP), allodynia is “pain due to a stimulus that does not normally provoke pain” while hyperalgesia is “increased pain from a stimulus that normally provokes pain” (173). The following sections describe the different behavioral assessments used in rodent models of LBP and their associations with human pain.

Unlike ex vivo assessments of DRGs which have been widely utilized, behavioral assessments of pain in mice models of DBP are less well characterized. Only 35% of the papers included in this review assess pain as a major outcome and a majority of these use the SPARC-null mouse (54, 103, 104, 189). Thus, published models measuring pain-like behaviors are limited to the ventral lumbar puncture model, SPARC-null, and aged wild-type mice as shown in Table 3. Male or female mice were used ranging from 6 weeks to 2 years old with one study comparing both sexes (50) in terms of behavioral assessments. Results for spontaneous pain & locomotor capacity, mechanical & thermal sensitivities, and axial/lateral pain are described below.

A major difference between pain assessments in humans and animal is the inability of animals to communicate their pain. Humans can verbally communicate pain while for animals we rely on changes in pain-like behaviors as a proxy for pain. Unlike mouse models, humans have environmental influencers to pain and pain is also subjective as pain tolerances are different across patients. Physicians have clinical patient history and they provide in-depth physical examinations with pain scales (Numeric Pain Rating Scale (NPRS), Pain disability index (PDI), Visual Analog scale (VAS), and Oswestry index). Humans also possess psychosocial aspects which can affect pain while mice may not. However, some studies suggest mice have the potential for high-level emotional pain (190). As reviewed by Mogil et al., evoked assessments often fail in their translation to the clinic and thus spontaneous pain may be more clinically relevant to the human condition (191). It is evident from closer observation of mouse behavioral assessments, especially in evoked assessments, that the variability is heightened, and multiple factors can attribute to differences in pain-like behavior such as testing environment, the force of pull such as in grip tests, experimenter, and animal stress. Such factors need to be considered when assessing pain-like behaviors.

Neuropeptides such as CGRP have been identified in small DRG neurons involved in pain perception (192), and NPY is also upregulated in nerve injury/inflammation and present in the lumbar IVDs in humans and mice (193, 194). In addition, sodium ion channels are phosphorylated post-injury which allows for increased nociceptive signaling due to greater current density (195, 196). In direct molecular comparison of mice and human DRGs, TRK receptors present in small nociceptive neurons and TRPV1 were similar between humans and mice, while human neurons have a larger average size (~1.5–3 times larger than the mouse) (197). The mice models above demonstrated increased expression of neuropeptides along with increased sodium channels as assessed by IHC and RT-qPCR. Interestingly, receptor tyrosine kinase (RET), a pivotal protein in neuronal development, and Nav1.8/1.9 are significantly more abundant in TRKA+ cells of human DRGS compared to mice (197). Localization of CGRP and TRPV1 differs between humans and rodents (198) as does Nav1.8 activity (198, 199). This suggests that, while changes in specific markers may translate to human pain, their localization with in neuronal populations between species may be distinct.

Synonymous with open field and rotarod assessments, human patients with radiculopathy have slower gait speed, shorter travel distances, and greater standing time similar to aged old mice (44–47, 200). Patients with LBP have increased movement evoked fatigue, decreased physical activity, as well as reduced flexibility (201). In humans, mechanical hyperalgesia may also be measured similar to algometers in mice with the application of force with gradual intensity to the patient's lower back (43). Interestingly, mechanical allodynia is more relevant in patients with radicular pain due to nerve compression or inflammation which may explain the lack of differences when conducting von-Frey on mice models with IVD injury or genetic disposition that lack direct neuronal trauma (202). In terms of thermal hyperalgesia, humans with LBP may experience coldness, radiating pain, and cold allodynia in one or both legs which correlates with findings in mice using the cold plate and acetone tests along with the mouse models of IVD degeneration in this review (24, 203). Grip force and tail suspension tests showed axial discomfort which is also relevant in humans and sensitivity to stretching in mice models may be predictive of lumbar stiffness as perceived in humans (204). Additionally, mice models of IVD degeneration under tail suspension tend to have decreased time immobile with more time spent mitigating pain which is similar to the human condition where humans are more likely to modify movements to avoid pain than patients without pain (200, 205, 206). It is noteworthy to mention that “pain” is complex and multifaceted and cannot be directly measured in animal models. However, surrogate indirect measures such as observations of pain-like behaviors can be used to provide insight into DBP and the potential translation of therapies for LBP. Overall, the direct translation of pain in mice to human pain requires further interrogation, and advances such as multi-institutional collaborations involving multiple disciplines are critical before we can fully interpret the translation between models (207).

Based on the review of pain-like behaviors described above, we have identified key ex vivo and in vivo pain-like behaviors that can be utilized in mouse models. Given their similarity to humans and consistency across models, ex vivo assessment of CGRP, TRKA, TRPV1, and sodium channels levels are recommended. For in vivo assessments, the recommendations are derived from the small number of studies that assessed pain-like behaviors due to their consistent results and an increase in the number of studies may further elucidate more critical behavioral parameters. In vivo assessments could include open field for spontaneous pain-like behaviors, grip force & tail suspension for axial discomfort, and cold plate/acetone due to their similarity to human pain outcomes and consistency across models. Thermal hypersensitivity measures are less common in humans and have demonstrated inconsistencies across studies. While mechanical sensitivity is relevant in humans, the lack of significant difference in von-Frey in mice limits its clinical potential.

In this review, we have highlighted several mouse models of IVD degeneration with the potential for understanding disease mechanisms and for screening regenerative therapies for LBP in vivo. Methods of assessment on the molecular, cellular, tissue, structure/function, and pain level were discussed and highlighted in light of their clinical translation to LBP in humans. Several prominent induction methods and downstream parameters stood out between the assessed mouse models. Mechanically induced models such as those which use compression, instability, whole-body vibrations, and bipedalism along with needle puncture models demonstrated IVD degeneration at early time points compared to naturally aging mice and may be more clinically relevant as an acute model of IVD degeneration following herniation/trauma with controlled level-specific effects. These models also indicate that the mode of puncture is critical with changes observed with different sized needle gauges which alters the degree of degeneration (79). Genetic/aging models such as ERCC1, SM/J, and SPARC-null mice exhibited hallmarks of IVD degeneration at early ages compared to aged wild-type mice without forced trauma on the IVD and may be good models for spontaneously occurring chronic LBP. Clinical limitations to note include the fact that cause of accelerated aging in these mice may be due to developmental/global defects. IVD degeneration was present across all lumbar IVDs in genetic models (varying in degree of degeneration) such as the global SPARC-null mice while in humans, IVD degeneration related to LBP is more prevalent in lower lumbar levels (L4/L5, L5/S1) (54).

In consensus, model selection criteria are largely dependent on the research question and regenerative therapy assessed. For example, critical differences between mice and human IVDs are their small size and reduced diffusion of nutrients across the CEP. Therefore, they may not be optimal for studying cell-based therapies as the cell viability in a mouse model will likely differ in larger human IVDs. Rather, mice may serve as a good tool for studying small molecule therapies, drug evaluation, and non-viral gene delivery (83, 101). For example, the SPARC-null mice has been used for studying potential therapies targeting the IL-8 pathway and toll-like receptor 4 inhibition (189, 208). Another important limiting factor is the presence of notochordal cells in mouse models, however, accelerated aging models do demonstrate a shift of notochordal to NP chondrocyte-like cells such as in SM/J mice (56). Another research question relates to the stage/severity of degeneration the therapy is targeting. For example, mice models of IVD degeneration caused by mechanical manipulation or direct puncture typically develop degenerative changes relatively quickly (which can vary depending on the degree of mechanical forces/puncture, needle size or injection of pro-inflammatory cytokines) while genetic models may be more representative of slower developing IVD degeneration as a result of aging. Thus, the target population should be considered as in humans, elderly patients present with lumbar spinal stenosis due to a decrease in NP hydration and narrowed IVD height while younger patients are more prone to AF rupture and NP herniation (30, 209). Despite some limitations, these mice models are advantageous in their wide array of molecular assessments and genetic phenotypes and may serve as an excellent intermediate model between in vitro and more clinically relevant but logistically challenging large models such as the chondrodystrophic dog (49). They provide the ability to assess IVD structure/function and pain parameters in living animals whilst providing a more efficient way to screen regenerative therapies without using large animals, further contributing to the 3R principles (Reduce, Replacement, Refinement).

This review highlights critical cellular/molecular, tissue structure/functional, and pain assessments for determining the validity and efficacy of regenerative therapies in the mouse model as illustrated in Figure 3. Of significance, sex differences in pain perception and IVD degeneration have been found in humans as well as animal models inclusive of articles within this review, which warrants the need to include both male and females studies to accurately represent the clinical patient population (19, 210–212). It is also imperative as sex bias in animal testing has resulted in clinical limitations and only 10% of the studies presented here assessed sex differences as shown in Table 4, which highlights the percentage of models in this review assessing each parameter. Evidently, the parameters that were included most assessed IVD morphology (85%) and DHI & hydration (66%) which are critical parameters used to determine IVD structure and function. Spontaneous pain, neuronal function, mechanics, and locomotor capacity were parameters that were least assessed (<25%). While these measures are important, there under use in mouse models of LBP may be a result of a lack of investigator access to the proper tools/expertise. A solution maybe to enhance collaborations amongst research groups with differing areas of expertise so that these outcome measures can be included in more mouse models. In terms of timeline, IVD models saw effects as early as 1-week post-induction/injury or as early as 8 weeks in accelerated aging mice. However, results at earlier time points in mechanically induced or puncture models may be more related to acute inflammation and thus may not accurately represent IVD degeneration in the human. Longer time points are recommended with incremental time points if logistically possible (i.e., 4, 8, 12, weeks vs. end time point) to study the progression of the disease model over time and therapeutic outcomes both short and long term.

The review of specific outcome measures provides a comprehensive overview of IVD degeneration in the mouse model as well as the potential to screen of regenerative therapies. As illustrated in Figure 3, these include (1) Histology and IHC to determine changes in disc structure and ECM, including nerve and immune cell infiltration, (2) Molecular and biochemical changes in ECM (proteoglycan and collagen) via DMMB or Hydroxyproline/Sircol assays, (3) Imaging IVD joints using microCT and MRI to assess changes in DHI, disc hydration, and grade, (4) Mechanical assessments for compressive stiffness, and (5) Pain-like behaviors such as open field, grip force, tail suspension, and cold plate/acetone tests. Histological assessments were used in most mice models to determine histopathological degeneration scores, structural integrity, disc height, and cellularity/cell morphology. IHC allows for the spatial assessment of specific proteins of interest and can be used to quantify changes in ECM matrix proteins, catabolic proteins, and even immune infiltration and neurovascular invasion into the IVD joint or neuronal structures such as innervating DRGs and spinal cord ex vivo as proxies of pain. Assays such as DMMB allow for quantitative assessments of proteoglycans in mice IVDs which is directly related to IVD hydration while TUNEL assays give information on changes in cellularity and cell death. Imaging methods such as microCT and MRI can be performed in vivo or ex vivo to quantify changes in IVD joint structure including DHI, DWI, hydration by disc intensity, as well as assessment of bone health adjacent to the IVDs. Lastly, as a major goal of regenerative therapies for IVD degeneration is to treat clinically relevant LBP, pain parameters in vivo are extremely critical in addition to ex vivo assessments.

In conclusion, this review has highlighted existing models of IVD degeneration for translational research and the treatment of LBP (acute vs. chronic), compared/contrasted induction methods (Mechanical, injury, genetic, age), and has outlined critical methods/parameters for both characterizing disease and downstream assessment of regenerative therapies in the mouse model. We hope this review will assist with model selection and critical parameters for assessments within the LBP research community to further push the translatability of clinically relevant therapies.