Stem Cell Approaches to Intervertebral Disc Regeneration: Obstacles from the Disc Microenvironment

Age-related change of IVD

Studies in embryogenesis have revealed that the NP originates from the axial notochord, while the AF and CEP are formed from the paraxial mesoderm [28 –30]. During growth and maturation, IVD undergoes significant histocytological changes that include transition of the notochord cells to the chondrocyte-like NP cells, thinning of the cartilaginous layers in the end plate, and fibrocartilaginous metaplasia of the organized AF (Fig. 1). Accompanying this, the gel-like NP reduces the content of aggrecan and collagen type II in the hydrophilic extracellular matrix (ECM) [31], thus weakening its capacity to withstand axial compressive loadings. Disorganization of AF often starts at the transition area between NP and inner AF, where the cartilaginous matrix accumulates and compromises tensile strength of the fibrotic AF (Fig. 1). Since IVD is avascular and its nutrient supply is limited by the inefficient permeation of the end plate [17,32], the thinning and calcification of CEP is supposed to cause a further nutritional deficiency within aged NP.

Although a widely accepted definition of IVD degeneration is not available as yet, this process is currently known to mimic many of those age-related changes and is substantially accelerated and/or exacerbated by multiple degeneration-inducing stressors [6,8,16]. To date, promoters of IVD degeneration have extended from aging to excessive mechanical loading [33,34], disc injury [35,36], oxidative stress [37,38], diabetes [39,40], chronic tobacco smoking [41,42], obesity [43,44], and genetic inheritance [7]. These etiological factors are overlapping and may work in a coordinated manner to disturb disc homeostasis.

Molecular basis of IVD degeneration

In contrast to other musculoskeletal tissues, adult IVD contains a small cell population that occupies <1% of the total disc volume [45]. These cells are sparsely located in the ECM and function to maintain a balance between anabolism and catabolism. During aging and degeneration, excessive apoptosis [46] and cellular senescence [8,16] contribute primarily to a reduction in viable disc cells. A deficiency in anabolic factors, such as transforming growth factor-β (TGF-β) [47] and insulin-like growth factor-1 (IGF-1) [48], may cause further reduction in cellular viability and ECM synthesis. Meanwhile, increased expression of proinflammatory cytokines [1,49], aggrecanases [50], and matrix metalloproteinases [51,52] promotes substantial matrix degradation, thereby enhancing the imbalance between anabolism and catabolism. By generating reactive oxygen species (ROS) and catabolic cytokines [16,37,38], the accumulated senescent disc cells may further deteriorate the microenvironment within aged and degenerated IVD. Based on these observations, an ideal biological regeneration should not only provide disc with sufficient functional cells but also be dedicated to improving the survival conditions in a degenerating IVD. Additionally, the safety concern, user-friendliness, and durability of therapeutic effects are of equal importance when translating stem cell-based disc regenerations from basic research into clinical practice.

Regenerative effects of stem cells

In the disc treated with stem cell supplement, the content of aggrecan and type II collagen was reserved in most cases [12,35,57,61,71 –73]. Accompanying this was the preserved disc height and the magnetic resonance imaging (MRI) T2 signal, which in our observation correlated well with improved histopathological evaluations [35]. The results of one study suggested that the restored ECM content of NP also contributes to maintain perfusion and permeability of the vertebral end plate [58].

In addition to matrix restoration, significant pain relief has been reported in clinical observations [13,74 –76]. However, increased disc height was seldom evidenced in those pain-relieved human discs, suggesting the nerve root decompression might not originate directly from the restored matrix content. The longest observation of stem cell transplantation was 6 months in animal discs [73]; however, the pain relief may last longer than 1 year under clinical trial conditions [13,74]; thus, additional case–control studies are required to validate active ingredients and therapeutic significance of stem cell therapy.

Regarding the issue of safety, complications of acute discitis and epidural abscess were reported in a 64-year-old patient who received intradisc injection of bone marrow and adipose tissue [77]. In a 41-year-old man with cervical herniated IVD, a pulmonary embolism occurred after intravenous autologous adipose stem cell therapy [78]. Similar concerns may arise from the back flow of stem cells into epidural space, which compromises disc regeneration and increases the risk of ectopic ossification [79]. More importantly, despite the improved radiological and histochemical markers, no study has reported a disc with stem cell injection that could be equally normal to or more rejuvenated than a normal control disc.

Regenerative mechanisms of stem cells

After transplanted into IVD, stem cells have been found to migrate, relocate, and even home the inner AF [80 –83]. In the NP tissue, the stem cells have been shown to demonstrate chondrocyte-like round morphology, whereas in the inner AF, they were much elongated and fibroblastic, suggesting differentiation into the disc-like cells [80]. In support of this finding, several chondrocytic markers, such as aggrecan, type II collagen, and Sox-9, were induced in the MSC that were transplanted into NP [35,57,72], cocultured with NP cells [64,68,84,85], or maintained in the NP-like hypoxic and mechanical conditions [86 –88]. When cocultured with AF cells [89] or seeded onto structure mimicking AF lamellae [90,91], the MSC tended to produce fibrotic ECM that was rich in type I collagen. Within a degenerating disc, the stem cell-derived disc-like cells may compensate for the cellular loss caused by apoptosis and senescence, enhance anabolism, and therefore contribute to restore ECM content and disc homeostasis.

In addition to differentiation, the stem cell itself is a reservoir of various growth factors and cytokines. Paracrine secretion of these bioactive factors exerts trophic effects onto the neighboring cells [92,93]. A large number of coculture studies have shown that disc cells increase proliferation and activate matrix synthesis after being cocultured with MSC [94,95]. The enhanced disc cell viability might originate from the elevated growth factors in the coculture system, which includes but is not limited to TGF-β, IGF-1, endothelial growth factor, basic fibroblast growth factor (bFGF), and various bone morphogenetic proteins (BMPs) [94,95]. Most of these growth factors have been proven to be not only as anabolic to disc cells [10] but also as antisenescence (TGF-β, IGF-1) [96,97], antiapoptosis (IGF-1, BMP-7) [98,99], anticatabolic (TGF-β, BMP-7) [100,101], and anti-inflammatory (TGF-β, BMP-7) [100 –102]. Given that discogenic pain is closely related to proinflammatory cytokines, such as tumor necrosis factor-alpha (TNF-α), interleukin (IL)-1β, and IL-6 [1,49], the pain relief effects of stem cell therapy might also result from these trophic molecules [60].

Stem cells can affect disc cells through paracrine secretion [95], direct cellular contact [94,103 –106], and even cell-to-cell fusion [107]. Direct contact with disc cells has been known to enhance the biological functions of stem cells [105,106], probably by amplifying the secretion of trophic factors. Although the regenerative process involves both differentiation into disc-like cells and trophic cocktails, it remains elusive whether or not the two mechanisms occur simultaneously and contribute equally to disc regeneration. Our previous study showed that the expression of TGF-β1 and IGF-1 in the GFP-labeled MSC peaked at 2-week postintradisc transplantation, whereas the type II collagen and hypoxia inducible factor-1α (HIF-1α) showed a steady increase for at least 8 weeks [108]. Collectively, it is probable that stem cell-based disc regeneration is a dynamic process that initiates an early trophic effect and then a gradual differentiation into new disc cells.

Identifying IVD stem cells

Emerging evidence has shown that cells isolated from NP [14,15,109 –111], AF [15,109,112,113], and chondrocytic end plate [114,115] also express most of the phenotype markers that define mesenchymal stromal cells [27], thus supporting the existence of resident stem cells within IVD. To date, endogenous stem cells have been detected in human [15,109,110,116], rhesus macaque [117], porcine [118], murine [14], rat [111], and rabbit IVD [113,119]. These cells maintain the potential of multilineage differentiation, but are diversified in phenotype signatures and biological capacities (Table 1). Compared to bone marrow MSC, the MSC derived from degenerated NP demonstrated much lower adipogenic differentiation ability [110], whereas the end plate MSC exhibited an enhanced potential of osteogenesis and chondrogenesis [114]. When maintained in the disc mimicking microenvironment, the NP MSC exceeded the adipose-derived MSC in tolerating hypoxic [120], hypertonic [121], and acidic culture conditions [111].

Since stem cell markers change with disc aging and degeneration [118,119], the altered components of stem cell pools might account for the varied biological behaviors in those MSC isolated from degenerated IVD. This notion is, in part, supported by the discovery of tyrosine kinase receptor (Tie2) and disialoganglioside 2 (GD2), double-positive NP progenitors in the human and mice disc [14]. In contrast to the Tie2-positive but GD2-negative ancestors, the double-positive cells preserve multipotency and are superior to form chondrocytic clones with increased accumulation of aggrecan and type II collagen. Moreover, the survival and functionality of the NP progenitor cells have been revealed to be maintained by interacting with angiopoietin-1, which is the Tie2 ligand and coexpressed in the Tie2-positive cells [14]. Although the identification of disc stem cells has provided new approaches to understand and regenerate IVD degeneration, many questions remain unanswered regarding the origin and spatiotemporal regulation of these multipotent cell resources.

Origin of IVD stem cells

Given that IVD itself is avascular and originates from the axial (notochord) and paraxial mesoderm (somite) [28 –30], it is possible that the endogenous stem cells are remnants of the multipotent mesendoderm cells during embryogenesis. The stem markers are ubiquitously detected across species both in the young and the aged IVD [15,111,113,117,118]. Furthermore, cells isolated from the IVD of fetuses have been shown to maintain multipotency and inherit enhanced chondrogenic differentiation capacity [122,123]. Based on this perspective, the stem cell pools, especially those chondrocytic progenitors, may be stored in the thick cartilaginous layers of the young end plate (Fig. 1). The large and vacuolated notochord cells in the gelatinous NP, which yield offspring of smaller chondrocyte-like cells and giant cells [124], are probably precursors of the NP progenitor cells [14,125,126]. This inference is in accord with the positive staining of notochord makers both in the smaller bovine NP cells [127] and human degenerated NP tissues [128]. Since the end plate chondrocytes can be attracted and can migrate into NP tissues [129 –131], the NP stem cell pools might also contain a proportion of those end plate stem or progenitor cells. A recent study lent strength to this notion by showing that the migration of end plate-derived stem cells is regulated by the macrophage migration inhibitory factor produced by NP cells [132]. Consistent with this, stem cells isolated from the CEP and the NP demonstrated similar mesenchymal phenotypes [110,114].

In addition to the remnants of embryonic multipotent cells, there is increasing evidence supporting the infiltration or migration of exogenous stem cells into the IVD [133 –135]. To date, specific stem cell niches have been identified in the outer AF borders that connect the ligament or perichondrium of adjacent vertebrae [133]. Cells expressing progenitor markers (Notch1, Delta4, Jagged1, C-kit, Ki-67, Stro-1) are detected there and can migrate along certain routes [positive for growth differentiation factor-5 (GDF-5), Sox-9, SNAIL homolog 1 and 2, β1-integrin] toward the inner parts of IVD [134]. Given that similar patterns of cellular migration exist in the rabbit knee joint, recruiting cells from these stem cell niches seems to be a highly conserved self-repair mechanism for cartilage tissues [135].

Despite the avascular nature, MSC from the neighboring vertebra bone marrow might be another resource of disc stem cells. In a new degeneration model induced by imbalanced loading, the intravenous delivered bone marrow-derived cells can be detected in the end plate capillary, inner AF, and NP of the degenerated tail disc [136]. However, the limited number of infiltrated cells suggests that the self-pair of IVD is probably a chronic and inefficient process [136,137]. Although elevated chemokines such as Chemokine (C-C motif) ligand 5 (CCL5) and stromal cell derived factor-1 (SDF-1) contribute to attract stem cells and facilitate migration [138,139], it remains to be determined whether a degenerating disc actually benefits from these multipotent cell resources. Several differentiation studies provide evidence that disc-derived stem cells can also be induced to express neural or endothelial markers [112,140]. By causing revascularization and/or reinnervation [1,45], the multipotent disc cells may contribute as well to disturb IVD homeostasis and to irritate discogenic pains.

Altered cellular and matrix content

In coculture systems, the stem cells contacting directly with disc cells have demonstrated most efficient trophic effects and chondrogenic differentiations [94,105,106]. There are studies suggesting the NP-like differentiation of human MSC can also be enhanced by the conditioned media from notochord cells [143,144]. Unfortunately, these ideal ex vivo conditions cannot be translated directly into the aged or degenerated disc, where few notochord cells are available [125 –127], and the resident disc cells are sparsely located within thick ECM [45].

After implantation into IVD, stem cells are supposed to be interacting primarily with various ECM components. A large number of scaffold studies have revealed that the proliferation and differentiation of stem cells are maintained by macromolecules, such as type II collagen [145 –147], hyaluronan (HA) [147,148], leucine-rich proteoglycans [117], elastin [145], and laminin [69]. In human umbilical cord mesenchymal stromal cells, the laminin-receptor integrin α3 and β4 subunits have been detected [69]. Similarly, the HA receptor CD44 is widely expressed in the disc-derived stem cells [14,113,114,117,140,149]. Although the biological significance of their interactions remains to be determined, breakdown of these matrix contents may contribute to disturb cellular adhesion, thus threatening the survival of both endogenous and transplanted stem cells.

Nutrient deficiency

The IVD is avascular, and the end plate capillaries are responsible for the nutrient supply to the disc cells [17]. As is evidenced by the MRI tracing of intravenous-injected contrast medium, solute diffusion through the cartilaginous layer is much less efficient than the adjacent vascularized tissues [32]. Furthermore, the nutrient transport can be further slowed down by disc aging and excessive mechanical loads [150,151], probably by promoting ossification of CEP. To maintain viability under nutritional deficiency, disc cells may adaptively reduce the demand for nutrient supplies. This is supported by the findings that NP cells enhance proliferation in low-glucose cultures, while AF cells accelerate senescence with glucose abundance [36,152]. The restricted glucose consumption of resident disc cells is in line with their anaerobic glycolysis under hypoxia, which generates energy at a lower cost of oxygen and ROS production [18,19,38].

Interestingly, in the IVD-like low-glucose conditions, MSC derived from bone marrow and adipose can also maintain viability and increase matrix synthesis [141,142]. The metabolic adaptation to glucose deficiency may facilitate the temporary survival of stem cells into their new disc niches. However, for longer maintenance of stem cell viability and functionality, a stable supply of growth factors is also required. It has been revealed that the NP-like differentiation of MSC can be enhanced by supplementation with TGF-β1 [86,153], TGF-β3 [154], GDF-5 [155], GDF-6 [156], BMP-2 [153,154], or cocktails of IGF-1, FGF-2, and platelet-derived growth factor [157]. Although disc cells express most of these factors [158,159], their paracrine actions are likely to be crippled by the low cellularity, thick ECM, as well as the age-related decrease of several anabolic signals [45,47,48]. In light of end plate calcification, diffusion of circulating anabolic factors may be also attenuated in aged or degenerated discs [150,151]. Endeavors are in progress to provide stem cells with scaffolds capable of growth factor release [160]. However, to date, no injectable scaffold has met the requirements for both cellular retention and nourishment support.

Hypoxia

One overriding aspect of disc cell biology is that they are removed from the blood supply and resided under conditions of hypoxia [17 –19]. An increasing number of studies have demonstrated that NP cells enhance proliferation in low oxygen tension and stabilize the expression of HIF-1α even under normoxia [161 –165], indicating a metabolic adaptation to the local hypoxic niches.

With regard to MSC, hypoxia is not only favorable for maintaining viability but also a potent inducer for the NP-like differentiation and matrix synthesis [65,67,86,155,166,167]. Moreover, hypoxia can also antagonize the inhibitory effects of IL-1β on the chondrogenesis of bone marrow-derived MSC [168]. For stem cell-based IVD regeneration, it is suggested that the entire ex vivo process of MSC expansion and modulation be performed under the disc-like hypoxic conditions [166]. However, given that disc injection breaches tissue integrity and promotes revascularization [35], the transplanted stem cells are possibly experiencing fluctuated oxemic tensions within a degenerated IVD. This may initiate oxidative damages since a higher level of mitochondrial-derived superoxide anions is detected in the disc cells under normoxia [38].

By stabilizing HIF-1α independent of oxygen tension [161 –164], the resident NP cells might have metabolically tolerated the oxemic shift during disc aging and degeneration. Although stem cells can be induced to express HIF-1α within disc niches [86,108], it is unknown whether these new NP-like cells also demonstrate an adaptive stabilization of HIF-1α and survive the disturbed oxemic homeostasis. To avoid the revascularization and oxemic fluctuation that complicate annulus puncture injuries, transpedicular cell delivery has been proved technically feasible in ovine and human lumber IVD [169]. Since this approach damages the end plate, more studies are needed to validate its safety and user-friendliness. In the alternative, the side-firing holmium laser in transforaminal endoscopic discectomy may provide quick concretion of annulus lesions [2]. Nevertheless, it remains to be determined whether a disc with a cracked AF and herniated NP is indicated for stem cell therapies [79,170].

Mechanical loadings

Healthy IVDs, especially those at the cervical and the lumber segments, function to absorb and conduct varied mechanical forces, including compression, shear, torsion, and flexion [25,26]. Depending on the loading pattern and intensity, the disc-specific mechanical features have been known to exert a wide range of impacts on both disc cell biology and ECM turnover [25,26]. In scoliotic disc, degeneration is potentially accelerated by the unbalanced mechanical loadings that promote calcification of CEP and deteriorate nutritional deficiency [151]. Transition of the notochord cells to the smaller chondrocyte-like NP cells, which signifies an early onset of disc degeneration, can be significantly enhanced by static compressive loadings or dynamic hydrostatic pressures [34,171].

Taking into account that the overwhelming majority of cells within adult NP, including the NP progenitors, are derived from the embryonic notochord resources [124 –126,172], the accumulated mechanical stress might contribute to exhaust, at least, the NP stem cells during disc aging and degeneration [14,118,119]. Studies in the foreleg-amputated rats suggest that premature senescence can be induced by mechanical injuries [33], thereby potentially accelerating the loss of both disc cells and their precursors. Additionally, the enhanced differentiation driven by mechanical stimuli might also account for the exhaustion of multipotent disc cells. For example, radial compressive loadings promote the AF-like differentiation in bone marrow MSC [173], whereas the NP-like differentiation of adipose-derived MSC can be significantly activated by dynamic compressions [88].

Given that the energy metabolism in resident disc cells responds differently to the static and dynamic compressive stimulations [174], the AF-like or NP-like differentiation of MSC may suggest a mechanical adaptation into the local disc microenvironment. Although this potentially strengthens the differentiation role of stem cell therapy [175], the trophic effects are unlikely to benefit from these compressive loadings, which increase intradisc pressures and promote backflow of stem cells into epidural spaces [79]. To tackle this problem, improved disc regeneration was achieved by combined application of stem cell implantation and spine distraction [176]. In addition, an injectable annulus plug was also designed to block the injection portal that alleviated cell leakage during intradisc cell delivery [177]. Since most symptomatic IVD has lost disc height and yielded segmental instability [1], future studies are needed to provide injectable and expendable scaffolds capable of immediate mechanical supports.

Acidity

Owing to the anaerobic glycolysis, lactic acid is generated and accumulated within the hypoxic disc niches [18,19,163]. It is reported that the pH levels average 7.0–7.2 in normal IVD, but can be as low as to 6.5 in the severely degenerative IVD [178]. The extracellular acidity has been known to be sensed by acid-sensing ion channels (ASICs) that include six subunits: ASIC1a, ASIC1b, ASIC2a, ASIC2b, ASIC3, and ASIC4 [179], and most of which are expressed in human IVD [20,21,180]. Of these, ASIC1a mediates intracellular Ca²⁺ influx and is proapoptotic to end plate chondrocytes [181]; activated ASIC3 is contributing to the survival of both AF and NP cells under acidic stresses [21].

In contrast to disc cells, the viability and matrix synthesis of MSC are significantly inhibited when cultured in low pH conditions [111,141,142,178]. Since normal disc cells express mainly ASIC2 and ASIC3 [180], while bone marrow MSC are positive for ASIC1a and ASIC3 [182], the varied ASIC signatures might account for their distinct adaptations into the acidic disc niches. A recent comparative study with normal and degenerative human discs shows that the expression of ASIC1, ASIC2, and ASIC3 is increased in the degenerated NP, whereas the ASIC1 and ASIC4 are elevated in the AF tissues [180]. Given that the basal ASIC3 activity of disc cells is maintained by the neural growth factor (NGF)/p75/ERK signals [21], the elevated NGF expression may contribute to tolerate the increased acidity within degenerative IVD [183,184], possibly by enhancing the activity of ASIC3.

However, being a major neurotrophic factor, the activated NGF is generally associated with inflammation [185,186], innervation [187,188], as well as pain induction [1,184]. Similarly, under the neural differentiation conditions, the NP-derived stem cells also upregulate ASIC3 and gain adaptation to the acidic disc niches [140]. These observations indicate that tolerating disc acidity is probably a painful process that promotes neurogenetic differentiation and facilitates nerve ingrowth. This might weaken the pain relief effects of stem cell therapies [13], although the cellular survival within a degenerating disc could be enhanced by upregulating ASIC3 and/or downregulating ASIC1a [21,140,189]. There are studies suggesting that the ASIC3 overexpression in dorsal root ganglions contributes to evoke radicular pains [190,191]. In addition to matrix preservation and trophic cocktails, it remains to be determined whether the stem cell-based pain relief also involves the regulation of ASIC components in those disc-related nerve fibers.

Hypertonicity

Within the avascular IVD, the gelatinous NP maintains hypertonic osmotic pressures to facilitate nutrient diffusions and resist compressive loadings [22 –24]. It has been known that the hypertonicity originates mainly from the glycosoaminoglycan side chains of aggrecan molecules and can be sensed by NP cells to activate the tonicity enhancer-binding protein (TonEBP) [23]. Once activated, the transcription factor TonEBP binds to the TonE motif of tonicity-sensitive genes, transactivating the expression of heat shock protein 70 [22], β1, 3-glucuronosyltransferase-I [192], aquaporin2 [193], as well as the aggrecans [23]. Although most of these TonEBP targets function to synthesize and preserve the water-bounding matrix [194], the hypertonic stimuli have been revealed to irritate DNA damage response and arrest cell proliferation in the bovine NP cells [195]. For the MSC, the cellular viability and matrix synthesis are also reduced by disc-like high osmolarity [121,141,142].

Conversely, by restoring the osmotic balance across the cell membrane, the notochord cells have demonstrated tolerance to hypotonic stresses [196]. This is attributed to the large cytoplasmic vacuoles that release their low-osmolar content to dilute osmotic pressures [196]. However, to date, it is unknown whether these osmoregulatory vacuoles contribute as well to tolerate the hypertonicity. There is a study showing that stem cell transplantation can restore the notochord content in a degenerative IVD [197]. Future experiments are needed to understand the osmotic regulation of TonEBP and its targets in both the notochord cells and the transplanted stem cells.