Therapeutic potential of growth differentiation factors in the treatment of degenerative disc diseases

Intervertebral disc (IVD) degeneration is a major contributing factor to chronic low back pain and disability, leading to imbalance between anabolic and catabolic processes, altered extracellular matrix composition, loss of tissue hydration, inflammation, and impaired mechanical functionality. Current treatments aim to manage symptoms rather than treat underlying pathology. Therefore, IVD degeneration is a target for regenerative medicine strategies. Research has focused on understanding the molecular process of degeneration and the identification of various factors that may have the ability to halt and even reverse the degenerative process. One such family of growth factors, the growth differentiation factor (GDF) family, have shown particular promise for disc regeneration in in vitro and in vivo models of IVD degeneration. This review outlines our current understanding of IVD degeneration, and in this context, aims to discuss recent advancements in the use of GDF family members as anabolic factors for disc regeneration. An increasing body of evidence indicates that GDF family members are central to IVD homeostatic processes and are able to upregulate healthy nucleus pulposus cell marker genes in degenerative cells, induce mesenchymal stem cells to differentiate into nucleus pulposus cells and even act as chemotactic signals mobilizing resident cell populations during disc injury repair. The understanding of GDF signaling and its interplay with inflammatory and catabolic processes may be critical for the future development of effective IVD regeneration therapies.

Intervertebral disc (IVD) degeneration is a major contributing factor to chronic low back pain and disability, leading to imbalance between anabolic and catabolic processes, altered extracellular matrix composition, loss of tissue hydration, inflammation, and impaired mechanical functionality. Current treatments aim to manage symptoms rather than treat underlying pathology. Therefore, IVD degeneration is a target for regenerative medicine strategies. Research has focused on understanding the molecular process of degeneration and the identification of various factors that may have the ability to halt and even reverse the degenerative process. One such family of growth factors, the growth differentiation factor (GDF) family, have shown particular promise for disc regeneration in in vitro and in vivo models of IVD degeneration. This review outlines our current understanding of IVD degeneration, and in this context, aims to discuss recent advancements in the use of GDF family members as anabolic factors for disc regeneration. An increasing body of evidence indicates that GDF family members are central to IVD homeostatic processes and are able to upregulate healthy nucleus pulposus cell marker genes in degenerative cells, induce mesenchymal stem cells to differentiate into nucleus pulposus cells and even act as chemotactic signals mobilizing resident cell populations during disc injury repair.
The understanding of GDF signaling and its interplay with inflammatory and catabolic processes may be critical for the future development of effective IVD regeneration therapies.

K E Y W O R D S
annulus fibrosus, bone morphogenetic protein, cartilage derived morphogenetic protein (CDMP), growth differentiation factor (GDF), intervertebral disc degeneration, nucleus pulposus, mesenchymal stem cell

| INTRODUCTION
Low back pain places a significant socioeconomic burden on society, with~632 million people affected globally. 1 Approximately, 84% of people will experience low back pain during their lifetime, leading to associated annual expenditure of £12 billion in the United Kingdom, with similar costs reported in other developed countries (eg, $85.9 billion in the United States and €16.5-50 billion in Germany). 2,3 This cost arises from direct medical expenses, work absences and wage compensation 1,4,5 and surpasses that of many other causes of disability, including arthritis. 6,7 The incidence of low back pain and associated cost are rising dramatically as the current global demographic shifts toward an increasingly aged population. 8 Although low back pain is multifactorial and complex in etiology, intervertebral disc (IVD) degeneration has long been identified as a major underlying cause. [9][10][11] The IVDs are fibrocartilaginous tissues positioned between the vertebrae, contributing to about one-third of total spinal length. 12 Functionally IVDs are crucial structural components responsible for conferring mechanical strength and flexibility to the vertebral column. 13,14 IVD degeneration is thought to arise from cell driven changes to the extracellular matrix (ECM) of the central portion of the disc, the nucleus pulposus (NP), which results in mechanical failure of the NP and annulus fibrosus (AF; a collagenous tissue circumferentially enclosing the NP), progressive AF fissure formation and eventual NP herniation. 15 This process is concurrent with an in-growth of blood vessels and nociceptive nerve fibers into the inflamed disc, facilitating immune cell infiltration and increasing associated pain. 16,17 The progressive obstruction of the IVDs ability to absorb and disperse spinal loads through the motion segment (the structural unit comprising the IVD, facet joints and adjacent vertebral bodies) in degeneration is secondarily linked with facet joint arthritis, spur/osteophyte formation, and vertebral body deformation. These have been associated with degenerative spinal conditions such as spinal cord stenosis, spondylolysthesis, degenerative scoliosis, and other painful pathologies resulting from nerve compression, such as sciatica. 9,18 IVD degeneration can be exacerbated by excessive manual labour, underlying genetic factors, and the aging process. 6 As a natural phenomenon of aging, some aspects of IVD degeneration may be difficult to prevent. 10,19 Indeed, the majority of adults over 30 years show some form of structural IVD degeneration without any accompanying symptoms or pain. 6 This makes diagnosis and effective early intervention in cases of emerging pathogenic degeneration a priority. Current treatment options are limited and provide predominately symptomatic relief without addressing the underlying pathology. These can be broadly grouped into, first, conservative treatments, ranging from painkillers and anti-inflammatory medication to physiotherapy, and second, surgical interventional. Surgery is utilized as a last resort, with procedures such as discectomy and spinal fusion costly to perform and resulting frequently in suboptimal healing outcomes and recurrence. Therefore, there is great demand for a biological treatment aimed at restoring IVD homeostasis and regenerating damaged tissue. Of importance to such strategies is the restoration of both structure and function of the NP and AF tissues. To this end, biological therapies have shown promise in preclinical studies. These could include cellular and acellular therapies delivered with and without instructive biomaterials and in conjunction with bioactive molecules or growth factors (see 20 for recent in-depth review). One such family of factors, growth differentiation factors (GDFs), appear to be an exciting prospect due to their crucial role in chondrogenesis (including differentiation to NP cells, namely, discogenesis) and cartilaginous tissue homeostasis. [21][22][23][24] As such, the focus of this review is directed on the continuing development of regenerative strategies for IVD repair employing GDF family members and the potential therapeutic role of GDF6.

| IVD STRUCTURE, FUNCTION AND DEGENERATION
The IVD can be described as three distinct regions-the NP, AF, and cartilaginous endplates (CEPs)-making up the largest avascular structure in the human body. The properties and functionality of the IVD are dependent on the specific microstructures of its component tissue regions, which in turn are produced by distinct cell populations.
The NP is characterized by an ECM that is rich in anionic proteoglycans (PGs), predominately aggrecan. These are arranged within an irregular type II collagen lattice and are present at a ratio of 27:1 (aggrecan: collagen). 18,25 Additional matrix components consist of small amounts of other collagen types (I, VI, IX, and XI) along with other hydrophilic aggregating PGs, such as versican, and nonaggregating small leucine-rich PGs such as biglycan, decorin, fibromodulin, keratocan, and lumican. [26][27][28][29][30] The high density of negatively charged PG molecules draws in and retains water allowing the NP to resist compressive loads.
Circumferentially, the NP is enclosed by the AF, a ligamentous structure composed of highly organized collagen fibers arranged in concentric lamellae, with superior and inferior fiber ends rigidly anchored in the CEPs. 31 The AF is composed of more than 2 /3 collagen and unlike the NP has a small PG component. 32,33 More than 95% of the collagen in the outer AF is type I, decreasing in an almost linear fashion to less than 5% with proximity to the NP. An opposing gradient exists for type II collagen going from the center of the NP to the outer AF. 34 In successive lamellae, type I collagen fibers are obliquely oriented at angles of approximately 62 to 47 to the spinal axis, preventing IVD deformation under load. 35 The CEPs physically confine the NP and AF to their anatomical boundaries and act as semipermeable barriers, supporting nutrient and fluid exchange. The anchorage of AF fibers to the CEPs superiorly and inferiorly is also critical to the integrity of the motion segment and is strengthened by fiber bundle splitting to increase force distribution. [36][37][38][39] During IVD degeneration, type II collagen synthesis by the NP cells is gradually replaced by type I collagen, while PG synthesis is decreased resulting in the boundaries between the NP and AF becoming less distinct. The high PG content is central to healthy NP function, and the ratio of PG-to-collagen is one of the defining features sets NP cells apart from chondrocytes. This is important to note for regeneration of the NP, where implantation of chondrocytes rather than NP cells may result in a cartilage-like matrix with insufficient PG concentrations. 40 Deficient water-retention by the NP results in a decrease in disc height, which in turn leads to a loss of ability to uniformly distribute compressive forces to the AF, creating areas of high pressure.
The cells of the NP are highly specialized, now recognized through detailed transcription profiling and murine cell tracing studies to be developmentally and phenotypically distinct from chondrocytes. 41  As all surgical procedures are invasive and irreversible, they are utilized in less than 2% of symptomatic patients 59 and are associated with long-term issues. Where herniation of the IVD has occurred, microdiscectomy remains the gold standard surgical treatment. 73 Spinal fusion has been extensively used and shown variable success rates of 32% to 98%. 74 Despite good short-term outcomes, fusion may result in an accelerated degeneration of the IVDs adjacent to the fusion site due to adjusted load bearing in the spine. [75][76][77] Alternatively, to take load off the AF a nucleoplasty can be performed, allowing the IVD to return to a normal size and decreasing pressure on symptomatic nerve endings by NP tissue removal. However, this technique is frequently linked to subsequent IVD instability. 78 In short, the current available surgical treatment options lack the ability to interrupt and correct the degenerative cascade and inflam- Despite its role in osteogenesis, BMP2 has been shown to increase ECM production in rat, bovine, and human IVD cells in vitro without increasing osteogenic marker expression. [99][100][101][102] Similarly, BMP7 was shown to enhance PG synthesis and cell proliferation in human NP and AF cells, 103 with others reporting similar results in rat and rabbit IVD cells. 104,105 However, in human cells, BMP7 was found to be less effective at increasing ECM secretion at similar doses than in rabbit and bovine NP and AF cells. 104,106 In a recent in vitro and organ culture model, BMP2/7 was also shown to increase glycosaminoglycan (GAG) synthesis in bovine NP cells without increasing osteogenic markers. 107 However, the majority of these studies failed to investigate NP-specific or AF-specific marker genes, often used mixed IVD cell populations and frequently employed only single-gene analysis to determine osteogenic response. Thus, while they demonstrate the potential of BMP-based approaches the findings suggest that responses may be cell type specific and be influenced by species, model system, or even local microenvironment, which is further supported by contradictory evidence from in vivo studies. In a rabbit model of degeneration, both BMP2 (adenoviral vector injection) and BMP7 (100-μg protein injection)-based therapies demonstrated positive regenerative effects. [108][109][110] However, a large animal study in goats comparing the efficacy of BMP2, BMP7, and BMP2/7 (1-5 μg) conjugated to a fibrin/hyaluronic acid carrier showed no evidence of IVD regeneration. 111 Similarly, in a canine study of spontaneous IVD degeneration, up to 250-μg BMP7 was injected per IVD but no regeneration was observed. 112 Such data further suggest that species, cell type, and microenvironment may be important factors in growth factor-based, particular BMP-based, regenerative approaches. Concerns also exist with the use of extreme supraphysiological doses of BMPs, for example, up to 12 mg BMP2, which has been linked to adverse effects including heterotopic ossification and increased risk of malignancy 113,114 meaning a more sophisticated solution combining  knockouts. 23 The incomplete nature of these carpal and tarsal fusions in GDF6 knockout mice and in some familial cases of KFS in humans indicates a degree of redundancy in GDF signaling. While variability in spinal fusions between regions of the spine, individuals and species may indicate that GDF6 requirements for complete vertebral segmentation may vary, particularly between different regions of the developing spine. 23 A recent related report provides evidence that GDF6 is an antiaging factor secreted by young bone marrow-derived mesenchymal stem cells (MSCs) and its expression can be regulated by miR-17, a microRNA known to decline with age. 130 This could have important implications for the progression of IVD degeneration, which is closely associated with the aging process.
During development, cellular secretion of GDFs forms directional morphogenetic gradients. [143][144][145][146][147] Recently, the expression and localization of GDF6 in developing spinal column of human fetus have been examined. 24 The strong expression of GDF6 is especially shown throughout cartilaginous region of vertebrae at early developing period (8-13 weeks of gestation) and in the developing IVDs (between 8 and 19 weeks of gestation) and is diminished from ossification areas. GDF6 expression is localized to the NP and inner AF, but could not be detected in the outer AF by immunohistochemical staining. 24 In adults, GDF5 and GDF6 are detectable in the inner and outer AF 129 but are most strongly expressed in the NP. 148 The precise role of GDF family members in adult IVD homeostasis is unresolved, though there is strong evidence of their importance for anabolic gene expression.

| THE EFFECT OF SUPPLEMENTATION WITH GDFS ON DISC CELLS AND CHONDROCYTES IN VITRO
Aside from driving cell differentiation, GDFs have been shown to have an anabolic effect on IVD cells (

| THE EFFECT OF GDFS ON THE DIFFERENTIATION OF MSCS TOWARD AN NP-LIKE PHENOTYPE
Numerous growth factors have been investigated as inducers of NPlike differentiation in MSCs including TGFβ, IGF-1, fibroblast growth factor 2 (FGF-2), and platelet derived growth factor (PDGF). 156,157 However, these studies focus on general chondrogenic markers rather than those markers now associated with specific NP differentiation.
Both GDF5 and GDF6 have also been shown to induce the expression   166 This lack of hypertrophy and progression toward endochondral ossification when using GDF6, coupled with the enhanced expression of NP markers, PG production and the higher aggrecan to type II collagen ratio observed in comparison to GDF5 strongly suggests that GDF6 is the most promising candidate to produce implantable NP cell phenotypes from MSCs or particularly ASCs.

| IN VIVO EFFECTS OF GDFS ON IVD REGENERATION AND DELIVERY METHODS
In preclinical animal models of IVD degeneration treatment with GDF5 and GDF6 have shown promising results. Initial studies investigated delivery of GDF5 and GDF6 through intradiscal injection into models of IVD degeneration. For example, in murine models, IVD recovery after application of static compression was improved with a single GDF5 injection. 167 After 4 weeks, a significant increase in disc height and cell number was observed in the NP and inner AF, with cells expressing both aggrecan and type II collagen. Similarly, in stab models of degeneration in mouse 168 and rabbit 150 delivery of GDF5 improved disc height and histological appearance. Interestingly, GDF5 was found to colocalize with proliferating cells adjacent to the epiphyseal plate in a rabbit model. 169 Similarly, in an ovine model of annular injury, GDF6 was found to improve defect healing in the AF and improve the hydration and cellularity of the NP. 81 Recently, Miyazaki and coworkers reported the development and use of a novel rat xenograft radiculopathy model, where rabbit NP tissue was transplanted   (Figure 1).
In a proof of principle mouse annulus needle puncture model of IVD degeneration, Liang et al injected an adenoviral vector carrying the GDF5 gene into lumbar discs. 168 The study demonstrated that the GDF5 gene was successfully expressed and active GDF5 produced, which leads to significant restoration of disc height, histology, and improved disc hydration as assessed through magnetic resonance imaging. 168

| CONCLUSION AND FUTURE DIRECTION
Small molecule therapy, especially in combination with cell implantation, holds promise for the treatment of degenerative disc diseases.
The link between ECM composition and tissue functionality in the NP means that it is critical to select the correct bioactive molecule, or combination of molecules. GDF family members have shown great promise both as direct anabolic factors when delivered to NP cells and as NP-specific differentiation-stimuli when delivered to MSCs or ASCs. In either case, GDF6 has been shown to increase GAG production relative to type II collagen and upregulate healthy NP-specific marker gene expression, and critically, to do so to a greater extent than other previously defined chondrogenic factors. Future work will focus on the delivery and controlled release of GDF family members to the disc. The combination of GDF with cell therapy should also be a focus going forward, with work underway to determine optimal cell populations for therapy. In vivo data hinting at a role for GDF6 in modulation of degeneration-associated pain and the mobilization and localisation of tissue resident stem cells during healing of artificially induced degeneration is exciting. The precise nature of these interactions should be explored in more detail but perhaps they offer a potential avenue for effective cell-free delivery of GDF family members. Given the proinflammatory environment of the degenerative IVD and the convergence of noncanonical GDF and cytokine signaling on kinase cascades, further research is needed to determine the nature of this interaction, which will likely lead to the identification of novel therapeutic targets that may optimize GDF therapy. Stephen M. Richardson https://orcid.org/0000-0002-7637-4135